The Molecular Mechanisms of Oleanane Aldehyde-β-enone Cytotoxicity against Doxorubicin-Resistant Cancer Cells

Simple Summary Currently, the main reason for the ineffectiveness of systemic chemotherapy is the formation of cancer cells with the phenotype of multiple drug resistance (MDR), which are able to survive at high doses of chemotherapy drugs. One of the MDR mechanisms involves the overproduction of proteins of the ABC transporter family, including P-glycoprotein, which are responsible for the transport of a wide range of drugs from the cell. This paper focuses on a new derivative of the natural triterpenoid betulin, oleanane aldehyde-β-enone (OA), having low toxicity toward normal cells and cytotoxicity toward cancer cells and their doxorubicin-resistant variants. The calculations and experimental methods revealed that this compound is not released from cells like most known drugs. The analysis of molecular cellular targets showed that the mechanism of OA toxic action is associated with the activation of the external and/or internal pathways of the apoptotic death of cancer cells and their drug-resistant variants. Thus, OA, as a highly cytotoxic compound toward doxorubicin-resistant cancer cells, can be a promising candidate for drug development. Abstract Oleanane aldehyde-β-enone (OA), being the semi-synthetic derivative of the triterpenoid betulin, effectively inhibits the proliferation of HBL-100 and K562 cancer cells (IC50 0.47–0.53 µM), as well as the proliferation of their resistant subclones with high P-gp expression HBL-100/Dox, K562/i-S9 and K562/i-S9_Dox (IC50 0.45−1.24 µM). A molecular docking study, rhodamine efflux test, synergistic test with Dox, and ABC transporter gene expression were used to investigate the ability of OA to act as a P-gp substrate or inhibitor against Dox-resistant cells. We noted a trend toward a decrease in ABCB1, ABCC1 and ABCG2 expression in HBL-100 cells treated with OA. The in silico and in vitro methods suggested that OA is neither a direct inhibitor nor a competitive substrate of P-gp in overexpressing P-gp cancer cells. Thus, OA is able to overcome cellular resistance and can accumulate in Dox-resistant cells to realize toxic effects. The set of experiments suggested that OA toxic action can be attributed to activating intrinsic/extrinsic or only intrinsic apoptosis pathways in Dox-sensitive and Dox-resistant cancer cells, respectively. The cytotoxicity of OA in resistant cells is likely mediated by a mitochondrial cell death pathway, as demonstrated by positive staining with Annexin V–FITC, an increasing number of cells in the subG0/G1 phase, reactive oxygen species generation, mitochondrial dysfunction, cytochrome c migration and caspases-9,-6 activation.


Introduction
Despite significant progress in cancer therapy, including traditional methods of surgery and radiation therapy as well as modern immunotherapy and gene therapy, chemotherapy is still the most commonly used therapy in cancer treatment. There are three major morphological types of cancer cell death, including apoptosis, autophagy and necrosis [1].

Drugs and Reagents
Oleanane aldehyde-β-enone (OA) used in the present investigation was synthesized from the available triterpenoid betulin, as previously described [21]. The NMR spectral data of OA coincided with the description.

Cell Line and Culture Conditions
The following cell lines were used in this study: HBL-100 (human mammary epithelial cells immortalized by SV-40 virus, which, due to long-term cultivation, became tumorigenic) [22]

P-Glycoprotein Overexpressing Cell Cultures and Culture Conditions
Several cell cultures with a multidrug-resistance phenotype developed, owing to the overexpression of P-glycoprotein. The K562/i-S9 subclone with the functionally active P-gp was obtained as the result of stable ABCB1 (MDR1) gene transfection [23]; the HBL-100/Dox cell line and K562/iS9_Dox subclone were derived from HBL-100 and K562/iS9 cells, respectively, after continuing cell selection in the presence of a rising concentration of Doxorubicin (Dox; TEVA, Tel-Aviv, Israel). P-gp expression in the obtained sublines was previously assessed; in HBL-100 and K562 cells, P-gp is not expressed; in the K562/i-S9 subline, 75% of cells have a high P-gp expression; and in HBL-100/Dox and K562/iS9_Dox, about 95% of cells have highly expressed P-gp [24]. The HBL-100/Dox, K562/i-S9 and K562/iS9_Dox cells were cultivated in RPMI1640 medium (PanEko, Moscow, Russia) with the addition of 10% fetal bovine serum (Biosera, Nuaillé, France) and gentamycin (40 µg/mL).

MTT Cell Viability Assay
Cell viability was evaluated via an MTT cell viability assay [25]. MTT (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide) was used as a colorimetric substrate for measuring cell viability. Non-viable cells, with altered cellular redox activity, were unable to reduce the MTT dye. The cells (5000 cells/well in 100 µL of medium for adherent cell lines and 25,000 cells/well in 150 µL of medium for suspension cell lines) were incubated in 96-well plates in a 5% CO 2 incubator at 37 • C. Compound OA and Dox were dissolved in DMSO (AppliChem, Council Bluffs, IA, USA) to a concentration 10.0 mM. After cultivating the cells in a pure medium for 24 h, Dox and OA were added to the cells at ten different concentrations ranging from 0.2 to 100.0 µM. The cells were cultivated with the compounds for 72 h; then, MTT (5 mg/mL) was added (20 µL/well). After 2 h of incubation, the medium with the compounds was removed, and the resulting formazan precipitate was dissolved in 100 µL of DMSO. The absorbance in the relevant wells was measured at 540 nm using a microplate photometer Multiskan FC (Thermo Scientific, Waltham, MA, USA). The Biology 2023, 12, 415 4 of 24 IC 50 value (the concentration of a substance causing 50% of cell death) was determined graphically based on the results of the MTT reaction.

P-Gp Substrate and P-Gp Inhibitor Prediction
We used four online services to predict whether triterpenoid OA is a P-gp substrate or an inhibitor. First, the new version of admetSAR (version 2.0), mainly focusing on the in silico prediction of chemical ADMET properties, is available at https://admetmesh.scbdd. com/ (accessed on 1 February 2023) [26]. ADMETlab 2.0 is an extended version of the widely used ADMETlab for systematically evaluating ADMET properties, as well as some physicochemical properties and medicinal chemistry friendliness [27]. Predictions of a substance as being a P-gp substrate or an inhibitor can be freely accessed in the Absorption section at http://lmmd.ecust.edu.cn/ admetsar2 (accessed on 1 February 2023).
A novel approach of the Bio21 Molecular Science and Biotechnology Institute for predicting pharmacokinetic properties, called pkCSM, which relies on graph-based signatures, is provided online at http://biosig.unimelb.edu.au/pkcsm/ (accessed on 1 February 2023). These encoded distance patterns between atoms are used to represent small molecules and to train predictive models [28].
The PgpRules prediction models, based on the classification and regression of a tree algorithm, were provided by Wang et al. [29] as a web application, which can be freely accessed at https://pgprules.cmdm.tw/ (accessed on 1 February 2023).

Molecular Docking of P-Gp
The molecular docking studies were carried out on a laptop PC with an installed Intel ® Core i3-6100 QM CPU with 3.70 GHz and RAM 8 GB operating under the Windows 8 Professional OS. Briefly, the 3D structures of human P-gp with the bound substrate (PDB ID: 6QXE) and inhibitor (PDB ID: 6QEE) were loaded into GOLD, version 5.0.1 (CCDC Software, London, UK) with the addition of all the hydrogen atoms. All the types of atoms, charges and bond hybridization were carefully checked. The structure of OA was optimized using the MM2 force field integrated in the ChemBio3D Ultra 14.0 software (Perkin Elmer, Waltham, MA, USA). The docking was performed using the GOLD wizard (400 docking poses), and the Chemscore values were determined. The conformation of the ligand with the highest Chemscore value was selected as the best one and was then compared with the conformation of conventional and bound P-gp substrates (verapamil and paclitaxel) and inhibitors (tariquidar and zosuquidar) with Maestro 11.6 software (Schrödinger LLC, New York, NY, USA).

Evaluation of Cell Sensitization to Doxorubicin
The combined effect of OA and Dox was assessed using the MTT test. The cells were seeded into a 96-well plate (5000 cells/well in 100 µL of medium for adherent cell lines, and 25,000 cells/well in 150 µL of medium for suspended cell lines). After 24 h of incubation in a 5% CO 2 incubator at 37 • C, OA was added to the wells at a concentration of 0.2 µM (for HBL-100, K562 and K562/i-S9) and 0.6 µM (for HBL-100/Dox). This concentration was non-toxic to all the cells, and cell survival exceeded 90%. Concurrently with OA, Dox at different concentrations was added to the wells. The cells were incubated for another 72 h. After that, the MTT reagent was added to the wells, and 2 h later, all manipulations were performed as described above. The IC 50 values were compared for Dox alone and when combined with OA.

Evaluation of P-Gp Functional Activity
The functional activity of P-gp was evaluated using a previously described method [30]. The K562/i-S9 cells were incubated for 20 min in a culture medium containing 5.0 µg/mL Rhodamine 123 (Rh123; Sigma-Aldrich, St. Louis, MO, USA). After incubation, the cells were washed twice and divided into several fractions (5 × 10 5 cells per point). One fraction was incubated in a pure medium, and the others were incubated with the addition of 1.0 µM and 20.0 µM of the investigated substance. The well-known P-gp inhibitor verapamil (TEVA, Tel-Aviv, Israel) served as a reference compound. Incubation was carried out in culture medium RPMI1640 without FBS at 37 • C for 30 min. Cell fluorescence was evaluated via a FACScan flow cytometer (Becton Dickinson, Franklin Lakes, NJ, USA). The results were analyzed using FlowJo software (ver. X 10.0.7r2, FlowJo Software, San Jose, CA, USA) and were represented with the geometric mean fluorescence intensity (gMFI) and MAF coefficient [31], which was calculated using the following formula: MAF = (Mean (inh) − Mean (free))/Mean (inh), where MAF is the MDR Activity Functional, Mean (free) is the mean value of cell fluorescence without the inhibitor, and Mean (inh) is the mean value for cell fluorescence with the inhibitor. The effect of OA on the expression of ABC-transporters (ABCB1, ABCC1 and ABCG2) was studied in pairs of cell lines K562 and K562/i-S9, as well as HBL-100 and HBL-100/Dox. The cells were seeded 2 × 10 6 /dish, and OA and Dox were added immediately to suspension cultures and on the next day to adherent cultures. Concentrations of substances were used corresponding to their IC 50 values, the cells were incubated for 48 h, and then RNA was isolated.

RNA Isolation, Reverse Transcription and Real-Time PCR
PureZOL RNA isolation reagent (Bio-Rad Laboratories, Hercules, CA, USA) was used for RNA isolation, according to the manufacturer's protocol. RNA concentration was measured using a NanoVue Plus spectrophotometer (GE Healthcare, Chicago, IL, USA) and was calculated on the basis of its optical density at 260 nm. The integrity of RNA was checked with 1% agarose gel electrophoresis containing 0.01% ethidium bromide. RNA reverse transcription was carried out with random hexanucleotide primers and M-MLV reverse transcriptase (Thermo Fisher Scientific, Waltham, MA, USA). Then, real-time PCR was performed on a CFX Connect Real-Time PCR Detection System (Bio-Rad Laboratories, Hercules, CA, USA). The amplification steps were as follows: 95 • C-3:00 min, (95 • C-0:10 min, 60 • C-00:10 min, and 72 • C-00:30 min) with 39 cycles and a melt curve of 65-95 • C. The housekeeping gene RPL0 was used to normalize gene expression data. The primers for real-time PCR are given in Table S1.
The relative expression level for each gene was calculated as [∆C t = C t (investigated gene) − C t (RLP0), with C t being the minimal baseline cycle in the exponential phase of the amplification curve]. The relative difference in gene expression levels was determined as 2 −∆∆Ct , where ∆∆C t = ∆C t (a treated sample) − ∆C t (an untreated control).

Annexin V-FITC/PI Assay
This assay was conducted according to the manufacturer's protocol (Annexin V-FITC/PI Apoptosis Detection Kit 1; Elabscience, Houston, TX, USA). Briefly, the cells were seeded into a 6-well plate and incubated for 24 h at 37 • C in a humidified atmosphere containing 5% CO 2 . The medium in each well was subsequently replaced with a fresh medium containing IC 50

Cell Cycle Analysis
The HBL-100 and HBL-100/Dox cells were seeded into a 6-well plate and incubated for 24 h at 37 • C in a humidified atmosphere containing 5% CO 2 . The medium in each well was replaced with a fresh medium containing the IC 50 values of the test compounds. Dox was used as a positive control of cell cycle alteration. After a 16 h incubation period, all the detached/dead and viable cells were collected, washed with cold PBS and re-suspended in 300 µL of cold PBS before being treated with 4.5 mL of cold 70% ethanol. The cells were then incubated for 2 h at 4 • C. At the end of the incubation period, the cells were centrifuged (Model 5417 R, Eppendorf, Hamburg, Germany) at 1500× g for 5 min, and the resulting pellet was washed with cold PBS and re-suspended in 300 µL of PBS.

Confocal Fluorescent Microscopy
The HBL-100 and HBL-100/Dox cells were seeded on confocal dishes with glass bottoms and incubated for 24 h at 37 • C in a humidified atmosphere containing 5% CO 2 . The max cell density was 1 × 10 5 cells/cm 2 . The medium was replaced with fresh medium containing 0.5 or 1.

Western Blotting
The HBL-100 and HBL-100/Dox cells were cultured in a 6-well plate at a density of 2 × 10 6 cells/well. After treatment with IC 50 and 2 × IC 50 concentrations of OA for 16 h, all the detached/dead and viable cells were harvested and lysed in the Mitochondria Isolation Kit for Cultured Cells (Thermo Fisher Scientific, Waltham, MA, USA). The cytosol and mitochondrial fractions were separated via centrifugation at 12,000× g for 15 min at 4 • C. The supernatant (cytosol fraction) and mitochondrial pellets were then collected, and the protein concentration was determined via UV fluorescence at 260 nm (UV-2600i spectrophotometer; Shimadzu, Tokyo, Japan).
To assess the caspase activity in dynamics, cell cultivation was carried out as described above, and the caspases were analyzed after 20, 24, 30, 36 and 48 h of incubation of the cells with OA at IC 50 . All the detached/dead and viable cells were harvested, and full cell lysates were prepared with RIPA buffer (Cell Signaling Technology, Danvers, MA, USA) with 1.0 mM PMSF as a protease inhibitor (Cell Signaling Technology, Danvers, MA, USA). The cell debris was separated via centrifugation at 14,000× g for 10 min at 4 • C. The supernatants were then collected, and the protein concentration was determined using the Lowry method [32].
Then, the same protein amounts (20.0 µg in each line) were loaded and separated with 4-20% Mini-PROTEAN TGX Stain-Free Protein Gels (Bio-Rad Laboratories, Hercules, CA, USA), followed by transferring onto polyvinylidene difluoride (PVDF) membranes. The membranes were blocked with 1% (w/v) casein in Tris-buffered saline containing 0.1% Tween-20 (TBS-T) and were then incubated with primary antibodies (1:1000) of Cytochrome c and β-actin (Cell Signaling Technology, Danvers, MA, USA) for cytosol, and mitochondrial fractions, cleavage caspase-3,-6,-7,-8,-9, PARP and β-actin (Cell Signaling Technology, Danvers, MA, USA) for full cell lysates at 4 • C overnight. On the next day, the PVDF membranes were washed thrice in TBS-T and incubated with HRP-conjugated secondary antibodies (1:1000) (Cell Signaling Technology, Danvers, MA, USA) for 1 h at RT. Immunoreactive proteins were detected with Clarity Western ECL Substrate (Bio-Rad Laboratories, Hercules, CA, USA) and were then analyzed on the ChemiDoc MP Imaging system (Bio-Rad Laboratories, Hercules, CA, USA). The proteins were quantified via densitometry using Image Lab software (version 6.0; Bio-Rad Laboratories, Hercules, CA, USA). To normalize protein expression data, a blot with β-actin was used.

Statistical Analysis
All the experiments were performed in triplicate. The results are presented as Mean ± SD values. The differences were considered statistically significant at p < 0.05. For statistical analysis and graph design, GraphPad Prism 5.0 was used. The statistical significance was evaluated with an unpaired t-test and the Mann-Whitney test.
Calculations of IC 50 were performed in the "Program for the (half-minimal inhibitory concentration of a substance) chemotherapeutic substances" written in the Laboratory of tumor cell genetics of the N.N. Blokhin National Medical Research Center of oncology of the Health Ministry of Russia (Moscow, Russia).

The Cytotoxic Effect of OA against Human Cancer and Non-Cancerous Cell Lines
We previously described the synthesis of a triterpenic derivative OA with high cytotoxicity (IC 50 0.66-2.35 µM) against seven cancer cell lines (Hep-2, HCT 116, MS, RD TE32, A549, MCF-7 and PC-3) [21]. The structure of OA is presented in Figure 1.
inhibitory concentration of a substance) chemotherapeutic substan Laboratory of tumor cell genetics of the N.N. Blokhin National Medic of oncology of the Health Ministry of Russia (Moscow, Russia).

The Cytotoxic Effect of OA against Human Cancer and Non-Cancerous
We previously described the synthesis of a triterpenic deriva cytotoxicity (IC50 0.66-2.35 µM) against seven cancer cell lines (Hep-2 TE32, A549, MCF-7 and PC-3) [21]. The structure of OA is presented i The new compound with potential anticancer activity should b cancer cells, with minimal damage being caused to normal human evaluate the selectivity of OA against cancer cells, the cytotoxic HBL-100 and K562 cell lines and their Dox-resistant subclones was action against the non-cancerous human breast epithelial cell line M are shown in Table 1.  The new compound with potential anticancer activity should be toxic against the cancer cells, with minimal damage being caused to normal human cells. In order to evaluate the selectivity of OA against cancer cells, the cytotoxicity of OA against HBL-100 and K562 cell lines and their Dox-resistant subclones was compared with its action against the non-cancerous human breast epithelial cell line MCF10A. The results are shown in Table 1. As is apparent from Table 1, the tested cancer cells were susceptible to the toxic action of OA. In contrast to Dox, OA was 6.5-17.8 times more toxic to human cancer cells than it was to non-cancerous MCF10A cells. Thus, OA showed certain selectivity against the cancer cells.
The drug resistance of cancer cells is often attributed to the activity of P-gp, the protein participating in the efflux of drugs from the cells. High P-gp expression was demonstrated in resistant subclones of HBL-100/Dox, K562/i-S9 and K562/i-S9_Dox [19,23]. Briefly, P-gp expression in parental HBL-100 and K562 cell lines was negligible, and P-gp-positive cells were practically absent in parental cell lines. P-gp-positive cells accounted for 100%, 80.0 ± 8.2% and 94.7 ± 2.3% of HBL-100/Dox, K562/i-S9 and K562/iS9_Dox-resistant subclones, respectively. Thus, it can be suggested that the resistance of HBL-100/Dox, K562/i-S9 and K562/iS9_Dox cells toward Dox is a result of P-gp activation.

Molecular Docking with P-Gp
The substances could overcome P-gp-dependent drug resistance due to not being P-gp substrates, and thus, P-gp could not transport them from the cells. On the other hand, the interaction of a substance with P-gp could disturb the P-gp-structure, followed by impairing the transport capability of this protein. Therefore, the FDA recommends that new cytotoxic compounds be tested as P-gp substrates and inhibitors [33]. Many computer models based on the structural similarity of ligands have been developed to predict whether the developed anticancer agents are either P-gp substrates or inhibitors [34]. The web-based analysis confirmed (Table S2) that compound OA and well-known P-gp inhibitor verapamil can be classified as P-gp inhibitors. Moreover, OA, in accordance with the protocols, was more often predicted as not being a P-gp substrate, in contrast to verapamil. A detailed analysis of individual molecular properties that identified hydrophobicity (MlogP) as a key descriptor for identifying P-gp inhibitors [35] suggested that the hydrophobic triterpenoid OA (MlogP 6.79) most likely acts as an inhibitor of the P-gp transporter.
Keeping in mind that the successful interaction of a compound with P-gp can be complicated by space limitations, we performed the docking of compound OA into the binding cavity of P-gp as a substrate or as an inhibitor. Because the binding pocket of P-gp is large enough to have multi-binding sites and multi-drug specificities, we performed statistical analysis with 400 docking poses with GOLD 5.0 (CCDC Software), as well as with the best docking pose to find the residues localized in binding pocket.
First, molecular docking with reconstituted human P-gp (PDB ID: 6QEX) was used to calculate whether the compound can act as a P-gp substrate. We found that OA can occupy the binding cavity of P-gp similarly to the known substrates (verapamil and co-crystallized taxol (chemical name: paclitaxel)) ( Figure 2). However, the highest values of the Chemscore parameter for OA did not exceed those for verapamil and taxol (38.16, 40.97 and 43.93, respectively). The analysis of intermolecular interactions of the studied compounds with amino acid residues within 5Å from the binding site showed that OA, unlike the known substrates, did not form hydrogen bonds within the binding cavity (Figure 2a), whereas verapamil and taxol formed H-bonds with Tyr307 and Gln990 residues (Figure 2b,c). In addition, taxol formed a Pi-Pi bond with Phe336 and Phe994 residues (Figure 2c). 40.97 and 43.93, respectively). The analysis of intermolecular interactions of the studied compounds with amino acid residues within 5Å from the binding site showed that OA, unlike the known substrates, did not form hydrogen bonds within the binding cavity (Figure 2a), whereas verapamil and taxol formed H-bonds with Tyr307 and Gln990 residues (Figure 2b,c). In addition, taxol formed a Pi-Pi bond with Phe336 and Phe994 residues (Figure 2c).  Figure 3 illustrates the P-gp structure used in the docking studies with the bound inhibitor zosuquidar and OA. For comparison, tariquidar, as a well-known P-gp inhibitor [36], was also docked. As seen in Figure 3, OA and tariquidar could occupy the binding cavity of P-gp, simultaneously overlaying the structure of zosuquidar. However, the highest Chemscore values for OA did not exceed those for tariquidar and zosuquidar (46.71, 58.49 and 55.00, respectively). OA formed two H-bonds with Tyr302 and Tyr306. Strong Pi-pi interactions were also observed for tariquidar and zosuquidar with Phe769 and Phe302 residues, respectively. Thus, the docking suggested that OA is unlikely to be a substrate or inhibitor of P-gp.  Figure 3 illustrates the P-gp structure used in the docking studies with the bound inhibitor zosuquidar and OA. For comparison, tariquidar, as a well-known P-gp inhibitor [36], was also docked. As seen in Figure 3, OA and tariquidar could occupy the binding cavity of P-gp, simultaneously overlaying the structure of zosuquidar. However, the highest Chemscore values for OA did not exceed those for tariquidar and zosuquidar (46.71, 58.49 and 55.00, respectively). OA formed two H-bonds with Tyr302 and Tyr306. Strong Pi-pi interactions were also observed for tariquidar and zosuquidar with Phe769 and Phe302 residues, respectively. Thus, the docking suggested that OA is unlikely to be a substrate or inhibitor of P-gp.  To determine how OA overcomes drug resistance induced by P-gp overexpression in cells, we experimentally investigated the influence of this compound on P-gp efflux activity. Rhodamine 123 (Rh123) is a well-known substrate of P-gp, as the transport of Rh123 from cells measured with its fluorescence can indicate the level of P-gp activity. Rh123 efflux from cells could be decreased by another P-gp substrate competing with Rh123 for the binding site of this protein or as a result of P-gp interactions with the compound inhibitor that impairs its structure.
We studied the potential effect of OA on Rh123 transport by P-gp to the outside of

The Effect of OA on Rhodamine 123 Efflux from K562/iS9_Dox Cells
To determine how OA overcomes drug resistance induced by P-gp overexpression in cells, we experimentally investigated the influence of this compound on P-gp efflux activity. Rhodamine 123 (Rh123) is a well-known substrate of P-gp, as the transport of Rh123 from cells measured with its fluorescence can indicate the level of P-gp activity. Rh123 efflux from cells could be decreased by another P-gp substrate competing with Rh123 for the binding site of this protein or as a result of P-gp interactions with the compound inhibitor that impairs its structure.
We studied the potential effect of OA on Rh123 transport by P-gp to the outside of K562/iS9_Dox cells, with the subclone of K562 cells expressing the largest amount of P-gp. Verapamil, a well-known competitive transport inhibitor of P-gp, was used as a positive control. Concentrations of 1.0 µM (a concentration equal to 2 × IC 50 of OA for K562/iS9_Dox cells) and 20.0 µM (a standard concentration for verapamil in such experiments) were used in our experiments. The results are presented in Figure 4.

The Effect of OA on Rhodamine 123 Efflux from K562/iS9_Dox Cells
To determine how OA overcomes drug resistance induced by P-gp overexpression in cells, we experimentally investigated the influence of this compound on P-gp efflux activity. Rhodamine 123 (Rh123) is a well-known substrate of P-gp, as the transport of Rh123 from cells measured with its fluorescence can indicate the level of P-gp activity. Rh123 efflux from cells could be decreased by another P-gp substrate competing with Rh123 for the binding site of this protein or as a result of P-gp interactions with the compound inhibitor that impairs its structure.
We studied the potential effect of OA on Rh123 transport by P-gp to the outside of K562/iS9_Dox cells, with the subclone of K562 cells expressing the largest amount of P-gp. Verapamil, a well-known competitive transport inhibitor of P-gp, was used as a positive control. Concentrations of 1.0 µM (a concentration equal to 2 × IC50 of OA for K562/iS9_Dox cells) and 20.0 µM (a standard concentration for verapamil in such experiments) were used in our experiments. The results are presented in Figure 4.  After 30 min of incubating the cells stained with Rh123 in a fresh medium (control), the majority of Rh123 was released from the cells, and the gMIF value was only 155 ± 23. Verapamil competed with Rh123 for P-gp binding, diminishing the Rh123 efflux from the cells both at 1.0 and 20.0 µM concentrations (MAF was 0.69 and 0.78, respectively) ( Figure 4a). OA substantially decreased Rh123 efflux from the cells when used at a highly toxic concentration of 20.0 µM (gMIF value of 454 ± 52 vs. 155 ± 23 for control), but the effect of the 1.0 µM concentration of this substance, which was also rather toxic for these cells, was smaller (gMIF value of 308 ± 28 vs. 155 ± 23 for control) (Figure 4b). Therefore, it seems unlikely that OA strongly interacts with P-gp. The concentrations of OA used in these experiments-exceeding the IC 50 value for resistant K562/iS9_Dox cells-indicated that OA overcomes cell resistance, not owing to interactions with P-gp. The fact that the addition of OA failed to lead to Rh-123 accumulation inside the cells suggests that OA is neither a direct inhibitor nor a competitive substrate of this efflux pump.

The Combined Treatment of Cancer Cells with Dox and OA
In order to determine whether OA can modify the toxic action of Dox against resistant cells, a series of MTT assays was carried out using combinations of subtoxic concentrations of OA with different concentrations of Dox. By analyzing the obtained data (Table 2), it could be concluded that adding OA at subtoxic concentrations to Dox did not change the toxic effect of Dox against HBL-100, HBL-100/Dox or K562 cells, which confirms the almost unchanged IC 50 values for Dox. The unchanged IC 50 values for Dox in resistant subclones confirm that OA is not a P-gp inhibitor.

The Effect of OA on ABC Transporter Gene Expression
The possible effect of OA on ABC transporter gene expression (ABCB1, ABCC1 and ABCG2) in HBL-100, K562 and the resistant subclones of these cells was studied and compared with that of Dox. Concentrations of OA and Dox approximately corresponding to their IC 50 values were used. OA had no statistically significant effect on ABC transporter gene expression in K562 or K562/iS-9 cells ( Figure 5). In HBL-100 cells, incubation with OA resulted in a 2-3-fold decrease in the expression of all ABC transporter genes (p < 0.05), but in HBL-100/Dox cells, only a tendency toward decreases in ABCB1 expression was noted (p = 0.1).
significance of gene expression values as compared to fresh medium (p < 0.05).
After 30 min of incubating the cells stained with Rh123 in a fresh medium (control), the majority of Rh123 was released from the cells, and the gMIF value was only 155 ± 23. Verapamil competed with Rh123 for P-gp binding, diminishing the Rh123 efflux from the cells both at 1.0 and 20.0 µM concentrations (MAF was 0.69 and 0.78, respectively) ( Figure 4a). OA substantially decreased Rh123 efflux from the cells when used at a highly toxic concentration of 20.0 µM (gMIF value of 454 ± 52 vs. 155 ± 23 for control), but the effect of the 1.0 µM concentration of this substance, which was also rather toxic for these cells, was smaller (gMIF value of 308 ± 28 vs. 155 ± 23 for control) (Figure 4b). Therefore, it seems unlikely that OA strongly interacts with P-gp. The concentrations of OA used in these experiments-exceeding the IC50 value for resistant K562/iS9_Dox cells-indicated that OA overcomes cell resistance, not owing to interactions with P-gp. The fact that the addition of OA failed to lead to Rh-123 accumulation inside the cells suggests that OA is neither a direct inhibitor nor a competitive substrate of this efflux pump.

The Combined Treatment of Cancer Cells with Dox and OA
In order to determine whether OA can modify the toxic action of Dox against resistant cells, a series of MTT assays was carried out using combinations of subtoxic concentrations of OA with different concentrations of Dox. By analyzing the obtained data (Table 2), it could be concluded that adding OA at subtoxic concentrations to Dox did not change the toxic effect of Dox against HBL-100, HBL-100/Dox or K562 cells, which confirms the almost unchanged IC50 values for Dox. The unchanged IC50 values for Dox in resistant subclones confirm that OA is not a P-gp inhibitor.

The Effect of OA on ABC Transporter Gene Expression
The possible effect of OA on ABC transporter gene expression (ABCB1, ABCC1 and ABCG2) in HBL-100, K562 and the resistant subclones of these cells was studied and compared with that of Dox. Concentrations of OA and Dox approximately corresponding to their IC50 values were used. OA had no statistically significant effect on ABC transporter gene expression in K562 or K562/iS-9 cells ( Figure 5). In HBL-100 cells, incubation with OA resulted in a 2-3-fold decrease in the expression of all ABC transporter genes (p < 0.05), but in HBL-100/Dox cells, only a tendency toward decreases in ABCB1 expression was noted (p = 0.1).   Considering that the main mechanism of the anticancer action of triterpenoids is most often associated with the activation of apoptosis in cancer cells [12,14], the possibility of OA inducing apoptotic cell death in adherent HBL-100 and suspension K562 cells and their resistant subclones HBL-100/Dox and K562/iS9 was further analyzed via flow cytometry using Annexin V-FITC/PI double staining (Figures 6 and S1).
The HBL-100 and HBL-100/Dox cells were treated with OA at concentrations close to the IC 50 and 2 × IC 50 values (0.5/1.0 and 1.0/2.0 µM, respectively) for 16 h. The results showed OA to induce the deaths of both parental and resistant cells through apoptosis ( Figure 6). The processing of the Annexin V-FITC/PI staining findings demonstrated an increase in the percentage of early apoptotic (EA) cells from 2.15% to 22.6 and 20.10% for parental cells, and a dose-dependent increase from 1.76% to 5.06 and 8.55% for resistant cells as compared with the control. The content of the late apoptotic (LA) cells was also dosedependent and tended toward increasing, as compared with the untreated cell culture (from 0.36% to 5.03 and 9.91% for parental cells and from 0.8% to 2.98 and 11.90% for resistant cells). The above-mentioned trend was observed after the action of OA on K562 and K562/iS9 suspension cells ( Figure S1). Moreover, after incubation with Dox, a characteristic difference was observed for Dox-sensitive and Dox-resistant cancer cells, when the content of early and late apoptotic populations of sensitive cells equaled 25.00 and 9.45%, whereas for resistant cells, these values were significantly lower-4.88 and 1.57%, respectively. Thus, OA induced the apoptotic death of both adherent and suspension human cancer cells, as well as their resistant variants, and all types of cells were characterized by significant positive Annexin V staining. The following experiments were carried out for a more detailed assessment of the mechanism of apoptotic cell death upon OA treatment using adherent HBL-100 and HBL-100/Dox cells with a normal karyotype. Considering that the main mechanism of the anticancer action of triterpenoids is most often associated with the activation of apoptosis in cancer cells [12,14], the possibility of OA inducing apoptotic cell death in adherent HBL-100 and suspension K562 cells and their resistant subclones HBL-100/Dox and K562/iS9 was further analyzed via flow cytometry using Annexin V-FITC/PI double staining (Figures 6 and S1).  analysis after 16 h of incubation of HBL-100 and HBL-100/Dox cells with OA or Dox using flow cytometry and staining with DAPI (Figure 7). In HBL-100 cells incubated in a fresh medium, the following signal distribution was observed: G0/G1 phase-51.76%; S phase-30.12%; and G2/M-18.12%. Incubation of these cells with Dox at a concentration of 0.2 µM resulted in cell cycle arrest in the S phase-88.03% of the HBL-100 cells. No appreciable changes were observed in the cell cycle distribution after the 16 h treatment of the cells with OA at 0.5 µM, as follows: G0/G1 phase-45.80%; S phase-32.80%; and G2/M-21.31%. The percentage of the cells in the sub-G0 phase increased from 1.85% in the untreated cells to 5.27 and 5.55% in the cells incubated with 0.5 and 1.0 µ M OA, respectively. The cell population in the subG0/G1 phase for Dox-treated cells also increased and equaled 4.16%.
In the case of HBL-100/Dox cells, we observed a similar situation for Dox at a concentration of 25.0 µM, which led to cell cycle arrest in the S-phase (73.96%). The OA treatment of HBL-100/Dox cells at 2.0 µ M led to slight changes in the cell cycle, causing an increase in the proportion of the cells in the S-phase of the cell cycle (45.15% vs. 28.73% for the control). In addition, we observed an appreciable increase in the number of cells in the sub-G0/G1 phase for 1.0 and 2.0 µ M OA (6.93 and 8.59% vs. 1.54% for the control). In the case of HBL-100/Dox cells, we observed a similar situation for Dox at a concentration of 25.0 µM, which led to cell cycle arrest in the S-phase (73.96%). The OA treatment of HBL-100/Dox cells at 2.0 µM led to slight changes in the cell cycle, causing an increase in the proportion of the cells in the S-phase of the cell cycle (45.15% vs. 28.73% for the control). In addition, we observed an appreciable increase in the number of cells in the sub-G0/G1 phase for 1.0 and 2.0 µM OA (6.93 and 8.59% vs. 1.54% for the control).

The Effect of OA on DNA Migration and Topoisomerase I Activity
DNA damage is one of the known cellular signals that induce apoptosis. To study the possible ability to interact with DNA, OA was incubated for 30 min with the supercoiled plasmide DNA (DNApHOT1, TopoGen) (scDNA) at concentrations from 10.0 µM to 200.0 µM. The anticancer drug carboplatin was used as a positive control. In our study, in contrast to carboplatin, which changed scDNA migration, incubation with OA, at concen-trations significantly exceeding IC 50 , did not influence scDNA migration, thus indicating that the interaction of OA with DNA is unlikely ( Figure S2).
Topoisomerases (Topo) modify and regulate the topological state of DNA. They participate in processes such as DNA replication, transcription, recombination and repair [37]. Topo I acts by making a transient single-stranded nick and remaining covalently attached to the 3 -end of the transient break, forming a Topo I cleavage complex. This reaction is reversible, and under normal conditions, only a small fraction of the cleaved DNA is detectable. Topo I inhibitors stabilize cleavage complexes by preventing the DNA regulation step, thus producing irreversible DNA lesions. Topo I unwinds the scDNA to form topoisomers. Therefore, when Topo I is inhibited, a certain part of scDNA remains intact, depending on the degree of inhibition. In our experiments, only the highly cell-toxic concentrations of >20.0 µM OA clearly inhibited the activity of Topo I ( Figure S3), whereas OA concentrations close to IC 50 only weakly inhibited Topo I. Taking into account that the IC 50 of OA for the human cancer cells tested is less than 2.0 µM, we suggest that OA has only a negligible effect on Topo I activity at non-toxic concentrations.

The Effect of OA on p53 Gene
The functionally active nuclear transcription factor p53 responds to upstream signals by activating the transcription of genes that are important for cell cycle arrest, DNA repair and apoptosis [38]. In addition to its transcriptional activity, p53 also induces apoptosis via non-transcriptional mechanisms, including the inhibition of Bcl-2 and Bcl-XL at the mitochondrial membrane [39]. Mutations in the p53 tumor-suppressor gene are found in over 50% of human tumors, which sometimes display a chemo-resistant phenotype [36]. In this study, the effect of OA was investigated by comparing its cytotoxicity against five human cancer cell lines (HCT116, HCT116 p53-/-, MCF-7, MDA-MB-231 and MDA-MB-453) with different p53 statuses (Table 3). As is apparent from Table 3, the IC 50 ratio (IC 50 against HCT116 p53 −/− to IC 50 against HCT 116 wt) clearly shows that p53 knockout HCT116 cells were similarly sensitive to OA as p53 wild-type HCT 116 cells (0.72 vs. 0.76 µM), suggesting that the cytotoxic effect of OA (in contrast to Dox) is independent of the p53 status of the cell lines. Moreover, OA was equally highly cytotoxic toward three breast cancer cell lines (1.4-2.2 µM), and there was no clear association observed with the p53 statuses of the cell lines.

Confocal Fluorescence Microscopy
Cell apoptosis is tightly associated with chromatin condensation, the loss of mitochondrial transmembrane potential and ROS formation. To clarify the mechanism of apoptosis in HBL-100 and HBL-100/Dox cells after 16 h of OA treatment, cellular morphology, mitochondrial permeability and ROS formation were analyzed via confocal fluorescence microscopy using Hoechst 33342, TMRE and H 2 DCFDA staining, respectively.
As shown in Figure 8, the untreated HBL-100 and HBL-100/Dox cells had typical oval nuclei with non-condensed chromatin visualized with the DNA-specific dye Hoechst 33342. Despite the fact that all potentiometric dyes are good substrates for MDR carriers, which remove the dye from cells and lead to a decrease in its accumulation [40], we observed a high intensity of red fluorescence in the untreated HBL-100 and HBL-100/Dox cells stained with TMRE. Moreover, we failed to detect any green fluorescence of the native cells after staining with H 2 DCFDA, which measured hydroxyl, peroxyl and other ROS activity within the cell.
contained condensed or fragmented nuclei as indicators of apoptosis. Furthermore, the incubation of HBL-100 and HBL-100/Dox cells with OA (0.5 and 1.0 µ M, respectively) was also characterized by a significant decrease in mitochondrial fluorescence (ΔΨm loss), indicating their inability to load the positively charged mitochondrial TMRE indicator. Moreover, OA caused ROS formation in HBL-100 and HBL-100/Dox cells, as indicated bythe high intensity of green fluorescence as compared with the control after staining with H2DCFDA, whereas the Dox treatment of HBL-100 and HBL-100/Dox cells caused only the wrinkling of nuclei and the light depolarization of the mitochondrial membrane and ROS activity. The staining pattern of the Hoechst 33342 dye revealed that OA-treated cells contained condensed or fragmented nuclei as indicators of apoptosis. Furthermore, the incubation of HBL-100 and HBL-100/Dox cells with OA (0.5 and 1.0 µM, respectively) was also characterized by a significant decrease in mitochondrial fluorescence (∆Ψ m loss), indicating their inability to load the positively charged mitochondrial TMRE indicator. Moreover, OA caused ROS formation in HBL-100 and HBL-100/Dox cells, as indicated bythe high intensity of green fluorescence as compared with the control after staining with H 2 DCFDA, whereas the Dox treatment of HBL-100 and HBL-100/Dox cells caused only the wrinkling of nuclei and the light depolarization of the mitochondrial membrane and ROS activity.

Proteins Responsible for the Induction of Apoptosis
Based on the results that OA led to mitochondrial outer membrane permeabilization (∆ ΨM ), we assessed cytochrome c migration following OA treatment. Figure 9 evinces the release of cytochrome c from mitochondria to cytosol as being more of note in HBL-100/Dox cells.
Based on the results that OA led to mitochondrial outer m (ΔΨM), we assessed cytochrome c migration following OA tre release of cytochrome c from mitochondria to cytosol a HBL-100/Dox cells.  The activation of caspase-6,-8 and PARP was recorded in release of cytochrome c from mitochondria to cytosol as being more of n HBL-100/Dox cells. Next, we identified whether OA-induced apoptosis was associated wit activation of the caspase cascade, and the results showed that OA inc caspases-3,-6,-7,-8 and -9 activities and cleaved-poly-ADP ribose polymerase (P ( Figure 10). The activation of caspase-6,-8 and PARP was recorded in HBL-100 cells after 3 of OA treatment, whereas caspase-7 activity was observed only after 48 h of incu with OA. We also observed time-dependent activation of caspases-3,-9 starting wit of incubation with OA.
Western blot analysis of HBL-100/Dox cell lysates revealed the expression of c caspase-9 to be time-dependently stimulated by OA, starting with 30 h of incub Cleaved-PARP and caspase-6 appeared in HBL-100/Dox cells after 48 h of OA trea whereas caspase-3,-7,-8 activity did not differ from that in the control. The activation of caspase-6,-8 and PARP was recorded in HBL-100 cells after 30-36 h of OA treatment, whereas caspase-7 activity was observed only after 48 h of incubation with OA. We also observed time-dependent activation of caspases-3,-9 starting with 24 h of incubation with OA.
Western blot analysis of HBL-100/Dox cell lysates revealed the expression of cleaved caspase-9 to be time-dependently stimulated by OA, starting with 30 h of incubation. Cleaved-PARP and caspase-6 appeared in HBL-100/Dox cells after 48 h of OA treatment, whereas caspase-3,-7,-8 activity did not differ from that in the control.

Discussion
The toxicity of the triterpenic derivative OA with pronounced cytotoxic activity against cancer cells of various origins [21] was additionally studied against breast cancer cells HBL-100 and leukemia cells K562, as well as against their Dox-resistant subclones HBL-100/Dox, K562/i-S9 and K562/iS9_Dox. As is apparent from Table 2, OA confirmed its anticancer potential against the cancer cell lines. Concurrently, OA showed equal toxicity toward parental HBL-100 and K562 cells and their Dox-resistant subclones. As compared with the non-cancerous human breast epithelial cell line MCF10A, the compound OA was 6.5-17.8 times more toxic toward the tested cancer cells. Therefore, we must conclude that OA is relatively selective against cancer cells.
We previously showed that, in the resistant HBL-100/Dox, K562/i-S9 and K562/iS9_Dox cells, the expression of P-gp, the transmembrane protein participating in the efflux of drugs from the cells, was elevated [20,24]. Several possibilities could exist to overcome P-gp-dependent cell resistance under OA action. The compound could interact with P-gp in such a way that the native structure of this protein is disturbed, and its transporting function is damaged [7,41]. On the other hand, if a compound is not a P-gp substrate, then it is not able to be transported by a protein out of the cells, which enables this compound to exert its toxic effect independently of cellular P-gp activity.
Numerous P-gp classification structure-activity relationship models based on structurally similar ligands have been developed to predict whether the developed anticancer agents are P-gp substrates or inhibitors [34,42]. In silico approaches are usually swift, inexpensive, less labor-intensive and less time-consuming for drug discovery/prediction and ADME/Tox profiling comparing to in vitro and in vivo assays [43]. The performed analysis based on four available web services [26][27][28][29] enabled us to infer that OA unlikely belongs to P-gp substrates (Table S1).
Because P-gp can undergo substantial conformational changes upon binding with various ligands and has multiple substrate binding sites [44], it is highly challenging to accurately model P-gp-substrate interactions [45][46][47]. The recently available cryo-electron microscopy structure of the human-mouse chimeric P-gp in a complex with substrates and inhibitors [48] avoids the use of homologous modeling that we previously used [20]. Molecular docking with the human-mouse chimeric P-gp revealed that OA may occupy a part of the transmembrane-binding cavity of P-gp but not as successfully as known substrates verapamil or PTX do, without hydrogen bond formation within the binding cavity ( Figure 3). Interestingly, another betulin derivative, betulinic acid (BA), was also determined as not being a substrate of any of the three ABC transporters (MDR1/ABCB1, ABCG2 and ABCB5), but the cancer cells overexpressing these transporters were efficiently responded to by BA [18].
Docking studies have also shown the OA binding site to overlap with the binding sites of the known P-gp inhibitors tariquidar and zosuquidar in the transmembrane domain of the P-gp molecule [36]. However, it is unlikely that OA acts as an inhibitor due to its inability to form strong bonds within the P-gp binding pocket, similar to tariquidar and zosuquidar ( Figure 3). Most likely, a strong bond with P-gp and, as a result, high P-gp inhibitory activity are provided by the presence of aromatic rings in the tariquidar and zosuquidar structures, which are capable of forming Pi-pi staking with Phe302 and Phe769 residues of the protein.
It was advisable to check whether OA can interact with P-gp directly. For this purpose, we evaluated its capability to inhibit the efflux of the well-known P-gp substrate Rh123 from K562/iS9_Dox, the cells with a high level of P-gp expression. Incubation of the cells with Rh123, together with a typical P-gp substrate, verapamil, resulted in competitive inhibition and a dose-dependent decrease in the release of Rh123, as assessed by its fluorescence. As can be seen from Figure 4, only the highly toxic concentration of OA caused the minor inhibition of Rh123 efflux from K562/iS9_Dox cells. In turn, the OA concentrations that were close to the IC 50 had a negligible effect on Rh123 efflux from cells, thus indicating that OA is most likely not a substrate of P-gp at the IC 50 . According to Kannan et al. [49], the overall effect of inhibitors on the function of transporters, like the action of enzymes and receptors, depends on the relative concentrations of the inhibitor and its target protein; for example, at low concentrations, tariquidar acted as a P-gp inhibitor as well as a substrate of BCRP, whereas at much higher concentrations (≥100 nM), it acted as an inhibitor of both P-gp and BCRP.
The ability of pentacyclic triterpenoids to sensitize the susceptibility of human cancer cells to chemical drugs was recently shown [50]. In our previous study, the addition of semisynthetic triterpenoids at subtoxic concentrations significantly increased the sensitivity of Dox-resistant HBL-100/Dox and K562/iS9 cells to Dox action, indicating a synergistic effect of triterpenoids with Dox against MDR cells [20]. A similar effect of an OA structural analog was previously found with the semi-synthetic triterpenoid CDDO-Me, which passed phase I/II clinical trials for cancer treatment [51] and which induced an antiproliferative effect in Pgp1, over-expressing KHOS R2 and U-2OS TR cells synergistically with Dox [52]. Therefore, we also investigated the combined action of Dox and OA on cell viability in HBL-100, HBL-100/Dox, K562 and K562/iS9_Dox cells, but no effect of the subtoxic concentrations of OA on Dox toxicity was found. Our findings confirm the suggestion that OA cannot successfully interact directly with P-gp. The assumption that OA is not a substrate of P-gp leads to an inference that OA is not released from HBL-100/Dox cells and is able to accumulate in the intracellular space.
In contrast to CDDO-Me, which had no genotoxic effect on Pgp1 expression [52], the triterpenic derivatives previously synthesized by us suppressed not only the transport function of the P-gp protein but also the expression of ABC transporter genes in MDR cells [19]. Therefore, it was logical to study the OA effect on ABC transporter gene expression in HBL-100, HBL-100/Dox, K562 and K562/iS9_Dox cells. P-gp is encoded by the ABCB1 (MDR1) gene, and the ABCC1 (MRP1) and ABCG2 (BCRP) genes encode two other members of ABC transporters. OA had no statistically significant effect on ABC transporter gene expression in K562, K562/iS9_Dox and HBL-100/Dox cells ( Figure 5). In the Dox-resistant HBL-100/Dox culture, there was only a trend toward a decrease in the expression of ABCB1. However, in HBL-100 cells, OA statistically significantly inhibited the expression of all ABC transporter genes that were studied. It should be noted that incubation with OA did not lead to an increase in the expression of ABC transporter genes in any of the cell lines studied, which may indicate that OA does not activate drug resistance via the mechanism associated with ABC transporters.
We then showed that the decrease in viability of both the original K562 and HBL-100 cells and their resistant variants after OA treatment was caused by apoptosis with the characteristic externalization of phosphatidylserine, leading to significant positive Annexin V staining ( Figure 6 and Figure S1). There was a dose-dependent increase at the levels of total (early and late) apoptotic cells in the OA-treated HBL-100 and HBL-100/Dox cells. Thus, OA treatment for 16 h caused an increase in the subG0/G1 phase cell cycle in HBL-100 and HBL-100/Dox cells (up to 9%), and the number of HBL-100 cells in the G0/G1, S or G2/M phases did not change as compared with the number of untreated cells. Moreover, OA as well as Dox caused an increase in the number of the cell population in the S phase cell cycle for the HBL-100/Dox cell line ( Figure 6).
Apoptosis is associated with cellular DNA damage, which can be both a consequence of apoptosis and its cause [53]. DNA-destructive factors include compounds that directly interact with DNA or those that impair the function of proteins participating in DNA reparation, or the relaxation of supercoiled DNA required during its replication and transcription. DNA damage induces cell cycle arrest through the activation of checkpoints allowing the time for DNA reparation; otherwise, the cell undergoes apoptosis [51]. Therefore, to clarify the mechanism of OA toxicity, we investigated the interaction of this compound with DNA and its possible influence on Topo I activity. In our study, no interaction of OA with plasmid supercoiled DNA was found ( Figure S2). Interestingly, a number of triterpenoids, including the aforementioned BA, are known as potent inhibitors of eukaryotic Topo I and Topo II [54]. Along with that, OA clearly inhibited Topo I activity only at highly toxic concentrations exceeding 20.0 µM (Figure S3), and the IC 50 for the tested cancer cell cultures ranged from 0.45 to 2.20 µM (Tables 1 and 3). Therefore, the IC 50 of OA is likely to produce a minor effect on Topo I. Therefore, it can be inferred that apoptosis induced by OA in parental and resistant cancer cells is not mediated by DNA damage. In turn, the p53 tumor suppressor continues to be distinguished as the most frequently mutated gene in human cancer and plays a major role in the response of cancer cells to DNA damage anticancer substances [36]. However, we did not observe a clear association with p53 expression (p53-dependent apoptosis) and the cytotoxicity of OA by using a panel of five cancer cell lines with different p53 statuses ( Table 3).
The relationship between increased ROS production and the inhibition of P-gp function and the expression of ABC transporter genes in MDR cells has been previously confirmed [55,56]. Moreover, Dox, a potent exogenous ROS generator [57] and ROS hyper generator after Dox treatment with cells or under hypoxic conditions, can lead to the activation of HIF-1α and can cause an increase in the activity and expression of P-gp, promoting Dox resistance [58]. Therefore, we tested the ability of OA to induce ROS production in parental and resistant cells. As can be seen from Figure 8, ROS formation after OA treatment was more pronounced in HBL-100 and HBL-100/Dox cells than that after Dox-treatment. Kim et al. showed [59] that the ability of CDDO-imidazolide to induce DNA damage in tumor cells is associated with ROS generation. Native betulin and betulonic acid have also been found to directly affect mitochondria and their membranes, inhibiting the activity of the complexes of the respiratory chain of organelles, thereby initiating ROS overproduction and mitochondrial dysfunction [60]. In this study, OA disrupted the oxidative balance of cells, but it was difficult to establish whether an increase in cellular ROS was an initiating event, or, for example, if it might be a consequence of mitochondrial perturbation. We showed OA to effectively depolarize mitochondrial membranes (Figure 8), the increased permeability of which can lead to the leakage of the number of pro-apoptotic factors, including cytochrome c, thus triggering apoptosis [61,62]. The biological effects described for pentacyclic triterpenoids are partly explained by their structural similarity with sterols, their ability to interact with artificial cholesterol-enriched membrane domains and, therefore, their ability to influence the heterogeneity of the lateral domains of cell membranes [63]. Therefore, BA exhibited a strong ability to interact with and damage synthetic membranes, disturb mitochondrial membranes and destabilize lysosomes [64].
Apoptosis, one of the better-known forms of regulated cell death, is activated when cell surface death receptors, such as Fas, are engaged by their ligands (the extrinsic pathway) or when BCL-2-family pro-apoptotic proteins cause the permeabilization of the mitochondrial outer membrane (the intrinsic pathway) [65]. Both pathways lead to the activation of caspases, a family of cysteine proteases responsible for cell death; they primarily activate caspase-8 (the extrinsic pathway) or caspase-9 (the intrinsic pathway) to further induce caspase-3 activation and subsequently lead to the cleavage of various substrates, such as PARP [66]. As can be seen from Figures 8-10, OA induced ROS generation, the depletion of MMP (∆Ψ m ) and cytochrome c translocation, and it promoted the activities of caspases in parental and resistant cancer cells. It is of interest that OA treatment led to early apoptotic changes in HBL-100 cells, and that these changes were simultaneously associated with extrinsic and intrinsic pathways mediated by the activation of caspases-8,-9,-3,-6,-7. On the basis of the band intensity ( Figure 10), it is likely that both pathways, the death receptor and the mitochondrial pathways, synergize and amplify each other's signals in parental HBL-100 cells. OA-induced apoptosis in HBL-100/Dox cells was found to increase only the expression of the caspase-9 protein associated with the intrinsic pathway of cell death (Figure 10). After OA treatment, the activation of caspase-6 and the emergence of cleaved-PARP in HBL-100/Dox cells, without caspase-3 activation, were also recorded. The functions of other executioner caspases, such as caspases-6,-7, have been proposed to partially replace caspase-3 activity [67], and, although PARP is primarily a substrate for caspases-3,-7, it is also recognized as a minor substrate for caspase-6 [68]. Similarly, several pentacyclic triterpenoids act as inducers of intrinsic and extrinsic apoptotic signaling pathways. For example, CDDO-Me was shown to activate both extrinsic and intrinsic apoptosis pathways in PC-3 cells [69]. Keto-beta-boswellic acid and acetyl-keto-beta-boswellic acid also induce apoptosis in HT-29 cells through caspase-8, -9 and -3 activation [70]. Although caspase-8 was shown to activate caspase-9 indirectly in type II cells via cleavage of the Bid protein, a member of the bcl-2 family, and the subsequent release of mitochondrial cytochrome c [71], there has been no evidence to-date of the direct processing of caspase-9 in vivo by this initiator caspase [72]. In this study, cross communication between the two pathways was unlikely, because caspase-9 was activated after 24 h, and caspase-8 was activated after 36 h of OA treatment.
Thus, parental HBL-100 cells were characterized by OA-induced apoptosis with the release of mitochondrial cytochrome c into cytosol and by the subsequent activation of the signaling proteins of the extrinsic and intrinsic pathways, accompanied by PARP cleavage, whereas only the mitochondrial-associated intrinsic apoptosis pathway drove cellular apoptosis in resistant HBL-100/Dox cells.

Conclusions
Overall, this study demonstrates that OA is neither an inhibitor nor a substrate of P-gp, and it can overcome P-gp-dependent cell drug resistance and induce apoptosis by provoking ROS generation and mitochondria membrane depolarization upregulating caspase activation. To better understand the multiple mechanisms of OA action, further detailed studies of the relationship between ROS generation and apoptosis induction in MDR cancer cells are required.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/biology12030415/s1: Table S1: Primers for real-time PCR; Table S2: Prediction of OA and verapamil as P-pg substrate or P-gp inhibitor via web services; Figure S1: Annexin V-FITC/PI double staining for detection of apoptosis in (a) K562 and (b) K562/iS9 cells after treatment with OA for 24 h; Figure S2: Interaction of OA with supercoiled DNA; Figure S3: Effect of OA on the activity of Topoisomerase I; Figure S4: Expression of cytochrome c in HBL-100 and HBL-100/Dox cells after OA treatment for 16 h, Figure S5