Treatment of Multidrug-Resistant Leukemia Cells by Novel Artemisinin-, Egonol-, and Thymoquinone-Derived Hybrid Compounds

Two major obstacles for successful cancer treatment are the toxicity of cytostatics and the development of drug resistance in cancer cells during chemotherapy. Acquired or intrinsic drug resistance is responsible for almost 90% of treatment failure. For this reason, there is an urgent need for new anticancer drugs with improved efficacy against cancer cells, and with less toxicity on normal cells. There are impressive examples demonstrating the success of natural plant compounds to fight cancer, such as Vinca alkaloids, taxanes, and anthracyclines. Artesunic acid (ARTA), a drug for malaria treatment, also exerts cytotoxic activity towards cancer cells. Multidrug resistance often results from drug efflux pumps (ABC-transporters) that reduce intracellular drug levels. Hence, it would be interesting to know, whether ARTA could overcome drug resistance of tumor cells, and in what way ABC-transporters are involved. Different derivatives showing improved features concerning cytotoxicity and pharmacokinetic behavior have been developed. Considering both drug sensitivity and resistance, we chose a sensitive and a doxorubicin-resistant leukemia cell line and determined the killing effect of ARTA on these cells. Molecular docking and doxorubicin efflux assays were performed to investigate the interaction of the derivatives with P-glycoprotein. Using single-cell gel electrophoresis (alkaline comet assay), we showed that the derivatives of ARTA induce DNA breakage and accordingly programmed cell death, which represents a promising strategy in cancer treatment. ARTA activated apoptosis in cancer cells by the iron-mediated generation of reactive oxygen species (ROS). In conclusion, ARTA derivatives may bear the potential to be further developed as anticancer drugs.


Introduction
The active compound artemisinin is derived from Artemisia annua L. (qinhao, sweet wormwood), a medical plant used in traditional Chinese medicine, and its semisynthetic derivatives artesunate and dihydroartemisinin exert not only antimalarial activity, even to otherwise drug-resistant Plasmodia [1], but also display inhibitory activity towards other diseases, including cancer in vitro [2][3][4][5][6][7][8] and in vivo [9,10]. Artemisinin and its derivatives also revealed anticancer activity in clinical pilot trials with human and veterinarian cancer patients [11][12][13][14][15][16][17][18]. More recently, it turned out that the bioactivity spectrum is much broader, and that artemisinin and its derivatives may also be valuable to treat other Scheme 1. Chemical structures of artemisinin-, egonol-, and thymoquinone-based hybrids, dimers, and trimers studied in this work.

Cell Culture
Human leukemic CCRF-CEM and P-glycoprotein-expressing CEM/ADR5000 cells were obtained from the University of Jena (Department for Pediatrics, University of Jena, Jena, Germany). Cells were cultured in RPMI 1640 medium (Life Technologies, Schwerte, Germany), supplemented with 10% FBS (Life Technologies) and 1% penicillin (1000 U/mL)/streptomycin (100 µg/mL) (Life Technologies). They were maintained in a humidified, 5% CO 2 supplied atmosphere in an incubator at 37 • C. CEM/ADR5000 cells were treated with 5000 ng/mL doxorubicin (Medical Center, Johannes Gutenberg University, Mainz, Germany) once per week to retain their resistance phenotype. The MDR profile has been previously reported [63][64][65]. Cells were passaged twice a week and then used for experiments in the logarithmic phase.

Resazurin Reduction Assay
The resazurin reduction assay was used to detect the cytotoxicity of ARTA derivatives towards sensitive and multidrug-resistant cells. Viable cells are able to reduce resazurin to the highly fluorescent resorufin, and this fluorescence can be measured with the Tecan Reader Infinite m200 Pro (Crailsheim, Germany). The method was previously described [66,67]. Briefly, CCRF-CEM and CEM/ADR5000 cells were seeded in the appropriate density (10,000 cells/well) in 96-well cell plates, with a total volume of 200 µL per well. Compounds were added in varying concentrations (0.001, 0.003, 0.01, 0.03, 0.1, 0.3, 1, 3, 10, and 100 µM). Each concentration was tested six times per experiment, and each experiment was repeated three times. Additionally, we tested the CEM/ADR5000 cells with doxorubicin alone and in combination with three different ARTA derivatives (10 µM) or verapamil (0.1, 0.3, 1, 3, 10, and 100 µM). After 72 h at 37 • C and 5% CO 2 , 20 µL resazurin (Sigma-Aldrich, Taufkirchen, Germany) 0.01% w/v in ddH 2 O was added to each well and further incubated for 4 h. The plates were measured using an excitation wavelength of 544 nm and an emission wavelength of 590 nm. The test compound concentrations required to inhibit 50% of cell proliferation were represented by IC 50 values calculated from dose-response curves.

Molecular Docking
Two-dimensional structures of ARTA and its derivatives were drawn and converted to 3D structures using the Corina Online Demo, and were saved in PDB format. Using the X-ray crystallography-based structure of a mouse P-glycoprotein as a template (PDB code: 5KOY), the homology structure of human P-glycoprotein was modeled as described [68]. The PDB file was converted to the PDBQT format using AutodockTools-1.5.6rc3. A grid box (coordinates of three dimensions: [grid center]: X: 21.092, Y: 92.594 and Z: 24.0; number of grid points in the three dimensions [npts]: X: 120, Y: 98 and Z: 100; spacing: 0.375) was constructed to define the transmembrane docking spaces of every type of atom in the ligand energies, which are used to predict the binding energies of the ligand; transmembrane docking spaces were calculated with the Autogrid 4.2 (The Scripps research Institute, Molecular Graphics Laboratory, La Jolla, CA, USA). Docking parameters were set to 250 runs and a 2,500,000 energy evaluation was set for each cycle. Using the Autodock 4.2 (Molecular Graphics Laboratory), we docked every ligand via the Lamarckian algorithm. The binding energies and interacting amino acids were received from DLG files, and the images were obtained using Visual Molecular Dynamics VMD (University of Illinois at Urbana Champaign, Champaign, IL, USA).

Flow Cytometry
CCRF-CEM and CEM/ADR5000 cells were exposed to doxorubicin (10 µM) (in the presence and absence of verapamil) and ARTA and its derivatives (10 µM). After incubation for 24 h, cells were harvested by centrifugation at 1500× g for 5 min. The supernatant was removed and the cells were suspended in a RPMI colorless medium. The fluorescence intensity of the intracellular doxorubicin was determined using a flow cytometer FACScalibur (Becton-Dickinson, Heidelberg, Germany), equipped with an ultraviolet argon laser (excitation at 488 nm, emission at 530/30 and 570/30 nm band-pass filters). The experiment was repeated thrice. Viable cells were gated, and we obtained the log fluorescence of single cells in forward and side light-scatter, based on the acquisition of data from 20,000 cells.

Single Cell Gel Electrophoresis (Alkaline Comet Assay)
DNA single-strand breaks were determined and calculated by single-cell gel electrophoresis. We used the OxiSelect™ Comet Assay Kit (Cell Biolabs-BIOCAT, Heidelberg, Germany). The alkaline comet assay detects both single and double DNA strand breaks. Radical molecules formed by ARTA derivatives generate DNA lesions and strand breaks. The DNA fragments migrate in the electrophoretic gel and form comets. With increasing DNA damage, the tail intensity increases rather than its length; tail length is determined primarily by the length of the loops. Here, we used a recently described protocol [69]. Exponentially growing cells were exposed to ARTA derivatives (10 µM) for 24 h, and then processed according to the manufacturer's instructions. We varied both lysis times and electrophoresis times by 30 min. Fifteen cells per sample were analyzed with casplab 1.1.3b.2 software (http://casplab.com), (version 1.1.3 beta 2, CASPLAB, Wroclaw, Poland).

Results
To analyze the cytotoxic effects of ARTA derivatives, CCRF-CEM and P-glycoprotein-overexpressing CEM/ADR5000 cells were treated with the compounds for 72 h, and then the cell viability was evaluated using the resazurin reduction assay. Since drug resistance is a major problem in cancer chemotherapy, we assessed whether or not the derivatives bypassed MDR. The results are shown in Table 1. Most of the derivatives inhibited proliferation by 50% at concentrations below 10 µM. With the exception of TF26 and TF29, CEM/ADR5000 cells showed no or negligible cross-resistance towards the derivatives. Except for TF26 and TF29, the degree of resistance of the CEM/ADR5000 cells was much lower when compared to doxorubicin. To measure the P-glycoprotein inhibitory activity of the ARTA derivatives, doxorubicin efflux assays were performed; relative intracellular doxorubicin concentrations were measured by flow cytometry. For comparison, the doxorubicin efflux assays were executed in the presence of a known P-glycoprotein modulator, verapamil. Figure 1 shows the results of the flow cytometric measurements. We measured low fluorescence intensities (MFIs). The MFIs of doxorubicin-treated cells were the lowest because of the efflux transport by P-glycoprotein. Concurrent treatment with verapamil inhibited P-glycoprotein, hence, these MFI values were the highest. Treatment with doxorubicin and the ARTA derivatives was found to be between both standards. Some of the derivatives did not interact with P-glycoprotein, and some inhibited the ABC-transporter (e.g., REI235, REI259, and TF19).
To confirm the flow cytometry results, we carried out resazurin reduction assays by combining doxorubicin with the P-glycoprotein-modulating ARTA derivatives. We tested doxorubicin alone, in combination with verapamil (a known modulator of P-glycoprotein), and each of the three ARTA derivatives. Table 2 summarizes the IC 50 values and the degrees of resistance reversal, which were calculated by dividing the IC 50 values of doxorubicin alone by the IC 50 values of doxorubicin in combination with the modulators.  To consider the mode of binding of the ARTA derivatives, we performed molecular docking on human P-glycoprotein at the transmembrane domain (TDM). Table 3 summarizes the results for each compound, providing the lowest binding energies. The total number of interacting amino acids and the amino acids involved in hydrogen bonds are also displayed in Table 3.  Figure 2 demonstrates the binding modes of dihydroartemisinin, verapamil, and doxorubicin with the respective interacting amino acids. Furthermore, we examined the induction of DNA damage triggered by the artemisinin-, egonol-, and thymoquinone-based hybrids. For this reason, we performed the alkaline comet assay. CCRF-CEM cells were incubated with the derivatives for 24 h. As shown in Figure 3a, the tail lengths varied between 102.91 (± 0.53) and 65.15 (± 0.20) µm. Representative images of CCRF-CEM cells subjected to the alkaline comet assay upon treatment with ARTA derivatives are shown in Figure 3b.

Discussion
In the present investigation, we demonstrated the cytotoxic activity of 12 novel ARTA-based derivatives on sensitive CCRF-CEM and multidrug-resistant CEM/ADR5000 cells. The IC 50 values ranged from 0.0018 to 43.685 µM, and for most of the compounds we achieved low resistance indices. It is known for ARTA and its derivatives that their cytotoxicity is attributed to the cleavage of the endoperoxide bridge and subsequent ROS generation. One of the compounds, REI220+26, displayed higher cytotoxicity in CEM/ADR5000 than in CCRF-CEM cells. This ability of compounds to kill multidrug-resistant cells with greater efficacy than their parenteral drug-sensitive cells is termed collateral sensitivity [70,71]. The mechanism is still not completely understood, but it has been proposed that substances discharged by the ABC transporter deplete ATP [72], subsequent replenishment of ATP generates ROS by oxidative phosphorylation [73], and many collateral sensitive substances that are lipophilic re-enter the cell. Hence, a futile cycling [74] may develop, leading to ATP depletion and preferential killing of multidrug-resistant cells. Previously, it has been shown that twice the amount of ATP is consumed in multidrug-resistant cell lines in comparison to sensitive cells [74]. CEM/ADR5000 cells overexpress P-glycoprotein [63,65,75] and exhibit MDR with cross-resistance between anthracyclines, Vinca alkaloids, taxanes, and other drugs [64]. However, treatment of the CEM/ADR5000 cells with ARTA derivatives resulted in considerably enhanced intracellular retention of doxorubicin through inhibition of P-glycoprotein activity. The results obtained by flow cytometry were confirmed by resazurin assays that assessed the effect of the three most potent compounds REI235, REI259, and TF19 in combination with doxorubicin.
Furthermore, we used in silico molecular docking to investigate the possible interactions of these compounds with P-glycoprotein. With the exception of TF19, all derivatives revealed a similar binding site at the transmembrane region of P-glycoprotein. This binding site was close to the one of verapamil, and similar degrees of resistance reversal were observed. For this reason, we conclude that both the ARTA derivatives and verapamil may share a similar mechanism of inhibiting P-glycoprotein. Verapamil inhibits P-glycoprotein-mediated drug efflux in a competitive manner.
Most anti-cancer drugs kill tumor cells by the induction of programmed cell death, and it has been first shown by Efferth et al. that ARTA induces apoptosis in leukemia cells [4]. This result was subsequently corroborated by several other groups (for review see [6]). ARTA induces apoptosis in a different manner than doxorubicin. This may be the reason why ARTA induced apoptosis in doxorubicin-resistant cells [76]. In this study, we demonstrated by comet assays that the ARTA derivatives caused DNA damage. Recently, it was shown that ARTA induced DNA damage, presumably by ROS [59]. Furthermore, protein alkylation is a reason for ARTA-induced cytotoxicity [77,78]. The cleavage of the endoperoxide moiety of ARTA in the presence of ferrous iron by a Fenton-type reaction leads to the formation of ROS, as well as carbon-centered radical molecules. In conclusion, we described a panel of novel ARTA derivatives, most of which were not cross-resistant or collateral sensitive in multidrug-resistant cells. Some of the derivatives inhibited doxorubicin efflux by P-glycoprotein, leading to the reversal of doxorubicin resistance. Our data might inspire further investigations to clarify the exact signaling pathways of ARTA in cancer cells. ARTA derivatives may be useful drugs to fight MDR in cancer cells and improve the treatment of refractory tumors in clinics.

Supplementary Materials:
The following are available online. Physical-chemical characterization of selected compounds and representative flow cytometric histograms of doxorubicin uptake assay in CEM/ADR5000 cells.