Nongenotoxic ABCB1 activator tetraphenylphosphonium can contribute to doxorubicin resistance in MX-1 breast cancer cell line

Hyperactivation of ABC transporter ABCB1 and induction of epithelial–mesenchymal transition (EMT) are the most common mechanism of acquired cancer chemoresistance. This study describes possible mechanisms, that might contribute to upregulation of ABCB1 and synergistically boost the acquisition of doxorubicin (DOX) resistance in breast cancer MX-1 cell line. DOX resistance in MX-1 cell line was induced by a stepwise increase of drug concentration or by pretreatment of cells with an ABCB1 transporter activator tetraphenylphosphonium (TPP+) followed by DOX exposure. Transcriptome analysis of derived cells was performed by human gene expression microarrays and by quantitative PCR. Genetic and epigenetic mechanisms of ABCB1 regulation were evaluated by pyrosequencing and gene copy number variation analysis. Gradual activation of canonical EMT transcription factors with later activation of ABCB1 at the transcript level was observed in DOX-only treated cells, while TPP+ exposure induced considerable activation of ABCB1 at both, mRNA and protein level. The changes in ABCB1 mRNA and protein level were related to the promoter DNA hypomethylation and the increase in gene copy number. ABCB1-active cells were highly resistant to DOX and showed morphological and molecular features of EMT. The study suggests that nongenotoxic ABCB1 inducer can possibly accelerate development of DOX resistance.


Results
Generation of DOX-resistant MX-1 cell lines. A series of DOX-resistant MX-1 cell sublines were established during the selection of parental cells for survival in the media with increasing concentrations (10 to 80 nM) of DOX. Cytotoxicity measurement with the MTT assay revealed a 38-fold increased resistance to DOX in MX-1/D80 cells as compared to the parental cell line. The IC50s in parental MX-1 cells and MX-1/D80 were 0.5 µM and 19 µM respectively (Fig. 1B). Development of DOX-resistance was accompanied by the evident morphology changes from cobblestone-like with relatively strong cell-cell adhesion to a spindle-like appearance (Fig. 1A).
Followed by repeated treatments of parental cells with increasing concentrations (up to 128 µM) of TPP + , cell subline MX-1/T was obtained. In the cell survival assay, IC50 of DOX in MX-1/T cells was 38 µM (Fig. 1B) and the resistance index has reached 76. Moreover, MX-1/T cells acquired resistance to other chemotherapeutics, particularly paclitaxel and cisplatin, as well (Additional file 1, Fig. 1). Similarly, to MX-1/D, MX-1/T cells became elongated in shape and were dissociated from surrounding cells (Fig. 1A). These TPP + pretreated cells was exposed to 1280 nM of DOX for a week, and MX-1/TD cell subline was generated which was able to maintain the similar resistance index (40) as in the case of MX-1/D80 cells.

Different mRNA expression pattern in chemoresistant MX-1 cells.
For the characterization of the mechanisms of chemoresistance, the microarray-based genome-wide gene expression analysis of MX-1/D, MX-1/T and MX-1/TD, as well as parental MX-1 cells, was performed. Comparison of gene expression between the parental cells and their chemoresistant derivatives revealed numerous changes in gene expression. Cluster analysis showed that TPP + -treated cells (with or without DOX exposure), tended to group separately from the parental cell line or DOX-only treated cells (Fig. 1C).  www.nature.com/scientificreports/ A total of 605, 662 and 1516 genes were significantly deregulated (FC ≥ 2, P < 0.05) in MX-1/D, MX-1/T, and MX-1/TD cells, respectively, as compared to parental MX-1 cells (Fig. 1D). All chemoresistant cell sublines shared 37 up-and 11 downregulated genes (Fig. 1D). MX-1/T and MX-1/TD had the largest number of overlapping genes with deregulated expression (251 up-and 145 downregulated genes). The most abundant differences in gene expression (457 up-and 508 downregulated) was identified between MX-1/D and MX-1/TD cells, and the number of overlapping deregulated genes was small (Fig. 1D).

Cellular responses deregulated in chemoresistant cell sublines.
In order to understand the most enriched pathways and to identify deregulated gene signatures in chemoresistant cells compared to their parental counterparts, GSEA and IPA software were used. Analysis of deregulated genes revealed that despite the differences in gene expression patterns, almost identical response pathways were involved in the chemoresistance development of all cell sublines. Seven major pathways were deregulated during chemoresistance development: epithelial-mesenchymal transition, cancer stem cells, cell adhesion and motility, immune response, chemoresistance-related genes, various channels and transporters, as well as epigenetic regulators ( Fig. 2A-H; Additional file 1, Table S1).
Among chemoresistance-related genes, the top upregulated genes were aldo-keto reductase AKR1B10, aldehyde dehydrogenase ALDH1A3, and xanthine dehydrogenase XDH (Additional file 1, Table S1). Marked upregulation of various ion channels and transporters was determined in all chemoresistant cells, predominantly including calcium and potassium channels as well as SLC family transporters, while intense (FC > 6 × 10 3 ) upregulation of ABCB1 was observed in MX-1/T and MX-1/TD, but not in MX-1/D sublines.
The epithelial-mesenchymal transition pathway, as well as the cell adhesion and motility CSC pathway, were enriched in all chemoresistant sublines, but involved a wide spectrum of different genes ( Fig. 2 and Additional file 1, Table S1). In contrast, chemokine encoded by CXCL8 (FC ≥ 39), which is also considered as a CSC-mark, was upregulated in all chemoresistant sublines.
Some pathways were selectively activated as a response to DOX or TPP + . Immune response-related genes were predominantly upregulated in TPP + -exposed cell lines, with a > 200-fold increase in expression levels of interleukins IL6 and IL1B in MX-1/T subline (Additional file 1, Table S1). In contrast, marked changes in expression and the spectrum of histones and epigenetic regulators encoding genes prevailed in all DOX treated cells (MX-1/D and MX-1/TD).

Expression of EMT and cell stemness genes.
In an attempt to validate the involvement of the EMT process and to determine the sequence of events, the expression levels of key EMT-related (CDH1, CDH2, SNAI1 and ZEB1) and cell stemness genes (POU5F1, SOX2 and NANOG) were measured in MX-1/D sublines resistant to increasing concentrations of DOX, and compared to MX-1/T and untreated MX-1 cells (Fig. 3).

Genetic and epigenetic regulation mechanisms of ABCB1.
To clarify the mechanism of this atypical ABCB1 production in MX-1/T and MX-1/TD cells, changes in the promoter DNA methylation status and gene copy number were analyzed. Pyrosequencing of 15 CpGs in the downstream promoter of the gene revealed a marked reduction of DNA methylation levels in MX-1/T and MX-1/TD sublines (from ≥ 83% down to 6%) as compared to MX-1/D80 or untreated cells (all P < 0.001; Fig. 4C). Furthermore, > 40 copies of ABCB1 gene were detected in both MX-1/T and MX-1/TD sublines, while only two copies were present in untreated cells (Fig. 4D). In comparison, neither gene amplification nor promoter demethylation occurred in MX-1/D80 cells. www.nature.com/scientificreports/  Table S1) and quantitative PCR results. The different coloring and directions of the arrows reflect the possibly different sequence of the activation of particular response pathway. In DOX-only exposed cells (red arrows) the deregulated expression of canonical EMT related genes and CSC markers was observed, what possibly caused cell migration and chemoresistance with an upregulated expression of drug metabolism-related enzymes (e.g. DOX metabolismrelated AKR1B10) and later activation of ABCB1 mRNA expression. In TPP + -exposed cells (green arrows) marked activation of ABCB1 and immune response related genes were observed what supposedly caused cell migration, chemoresistance and possibly later activation of non-canonical EMT and CSC-related genes.

Discussion
Drug resistance is the major obstacle to the successful treatment of cancer. Prolonged exposure to chemotherapeutic drugs causes cellular changes leading to the acquired ability of cancer cells to survive drug effect; however, in some instances, the resistance develops quite rapidly. Recent studies 19 suggest that inherent features of cancer cells can accelerate chemoresistance development and these features exist even in chemotherapy-naive cancer cells. The inherent tissue-specific capacity to extrude xenobiotic compounds, including chemotherapeutic drugs, is markedly enhanced in cancer cells through various pathways that include ABC transporter hyperactivation. In our study, at least two possible scenarios of chemotherapeutic drug DOX-resistance were described in unique cell sublines with chemoresistance developed under pressure of DOX and the nongenotoxic activator of ABCB1 transporter-TPP + . The gradual activation of the canonical EMT pathway was shown during the development of DOX-induced resistance without significant changes in P-glycoprotein expression. Meanwhile, in the case of ABCB1-hyperexpression, observed in TPP + -exposed cells, EMT pathway was likely dispensable. Besides, the mechanism of this nongenotoxic xenobiotic pressure-induced overproduction of ABCB1, was quite unique and combined the two (epi)genetic phenomena, i.e. the increase in gene copy number and DNA hypomethylation in the promoter region. DOX is an anthraquinone-type genotoxic chemotherapeutic drug commonly used for the treatment of a wide variety of cancers, including breast cancer 20,21 . The mechanism of DOX action consists of DNA intercalation www.nature.com/scientificreports/ followed by topoisomerase 2 inhibition and generation of free radicals 22,23 . Under the pressure of increasing concentration of the drug, the DOX-resistant MX-1 cell line was derived, showing a 38-fold higher resistance than the parental cell line. The process was accompanied by the evident morphology and gene expression changes specific to EMT. During the EMT, polarized epithelial cells of carcinoma undergo major phenotypic changes by acquiring mesenchymal morphology and CSC features 24,25 . Recent studies 26,27 revealed that the chemotherapyinduced EMT is mainly responsible for the chemoresistance development, not only by the cancer dissemination as suggested previously 24,28 , but increasing expression of ABC transporters has been related to this process as well, owing to the activity of the EMT-inducing transcription factors (TFs) on the promoters of ABC transporters 2,29,30 . In our study, the development of DOX-resistance in the MX-1 cell line, nicely conformed to this scenario. First, the key EMT TFs SNAI1 and ZEB1 together with stem cell markers SOX2 and NANOG were activated, and then this was followed by increased ABCB1 expression, however, without significant changes at the protein level. Moreover, this process involved multiple changes in the EMT, cell stemness, and chemoresistance pathways as observed in genome-wide gene expression analysis. However, the process of chemoresistance development can be led not only by the EMT. www.nature.com/scientificreports/ There are reports showing the ability of ABCB1 to influence the EMT, and inhibition or decreased expression of ABCB1 can lead to its suppression 31,32 . In our study, the treatment of MX-1 cells with a typical substrate for ABC transporters-TPP + -resulted not only in significant upregulation of the ABCB1 gene, but also more than a 70-fold increase in resistance to DOX-and induced EMT-specific morphology changes. TPP + is a non-genotoxic compound that tends to accumulate in mitochondria 33 . Attempts are made to use this lipophilic cation as a tracker moiety to deliver antioxidants to mitochondria for the prevention of senescence and age-related diseases, and also for targeted delivery of drugs in cancer chemotherapy [33][34][35] . In our study, DOX-resistance activated by this xenobiotic was accompanied by the most abundant gene expression changes and dysregulation of various cellular pathways, but the strongest effect of this compound was observed on ABCB1 expression. ABCB1 is considered as the key efflux pump, extruding various drugs and xenobiotics from the cells 36 and the contribution of this protein to chemoresistance was discovered years ago 37 . Our study suggests that treatment of MX-1 cells with increasing concentrations of DOX can eventually activate expression of ABCB1, which is in agreement with the previous study 29 . However, the cell pretreatment with nongenotoxic ABC transporter substrate TPP + can considerably enhance this process. Moreover, TPP + -induced overexpression of ABCB1 is driven by both genetic and epigenetic phenomena. Previous studies have shown the positive correlation between ABCB1 promoter hypomethylation 13 or the gene amplification 14 and its expression level. In our study, ABCB1 hyperexpressing cells were characterized by both gene amplification and DNA demethylation in the downstream promoter; the latter mechanism potentially is decisive, because tremendous > 7 × 10 5 -fold expressional upregulation of ABCB1, observed in this study, cannot be explained only by 21-fold increased gene amplification. To the contrary, no such genetic or epigenetic alterations were found in DOX-only treated cells, where ABCB1 expression was possibly enhanced by another possible mechanism involving the EMT-specific TFs.

Conclusions
The results of our study suggest two possible pathways towards chemoresistance: through activation of the key EMT TFs without the need of ABCB1 activation and through ABCB1 hyperactivation by both, genetic and epigenetic mechanisms. According to our observation, the combination of demethylation of the ABCB1 downstream promoter and gene amplification plays the crucial role for acquisition of stronger chemoresistance. This study showed the possibility, that controlling at least one of these mechanisms may lead to a way to overcome the chemoresistance. Besides, the data of our study warn about the possibility of accelerated development of chemoresistance by exposure to nongenotoxic ABCB1 inducers. Further, in vivo studies could assist to elucidate the exact response pathways activated by such ABCB1 inducers.

Materials and methods
Cell cultivation and establishment of resistant sublines. The MX-1 cell line was maintained as an attached monolayer culture in RPMI 1640 medium, supplemented with 10% (v/v) heat-inactivated fetal bovine serum (FBS), 100 U/mL and 100 μg/mL penicillin-streptomycin (all from Biochrom AG, Germany). The cells were grown in cell culture bottles (growth surface 75 cm 2 ) with filter screw caps (Carl Roth, Germany) at 37 °C in an atmosphere with 90% humidity and 5% CO 2. For the investigation of the mechanisms of chemoresistance, cell models-DOX-exposed and TPP + -exposed MX-1 cells-were established and named MX-1/D and MX-1/T, respectively. Drug resistance was derived by a stepwise selection method as described previously 38 . Briefly, initially DOX-sensitive MX-1 cells were incubated with 10 nM DOX (Teva, UK), and doubling of the drug concentration was performed until the final concentration of 80 nM DOX was applied. Ultimately, four MX-1 cell sublines resistant to increasing concentrations of DOX were established and named MX-1/D10, MX-1/D20, MX-1/D40, and MX-1/D80, accordingly. These MX-1/D cell sublines were cultured in the indicated concentrations to maintain the resistance. For generation of ABCB1-active MX-1 cells, increasing concentrations (from 2 to 128 μM) of TPP + (chloride salt, Fluka, St. Gallen, Switzerland), as a model substrate of ABC transporters 18 , were used and cells resistant up to 128 μM of TPP + were generated (MX-1/T). Finally, ABCB1-active cells were exposed to high concentrations (up to 1280 nM) of DOX for a week in order to generate MX-1/TD cell sublines. The morphological changes in all derived chemoresistant MX-1 cell sublines were observed by an inverted fluorescence microscope Motic AE31 (Xiamen, China). The images were captured at a 200-fold magnification.
Cell viability measurements. Cell chemoresistance was determined as cell survival in the exposure to various DOX concentrations by standard colorimetric MTT assay as previously described 39 . Cells were seeded into 96-well plates with each well receiving a volume of 200 μL at a density of 10 4 cells/well. Then DOX at a final concentration of 0.05-51.2 µM, paclitaxel (Pharmachemie B.V., Netherlands) and cisplatin (Accord healthcare, Estonia) at final concentrations of 0.065-64.0 µM was added to the test cells and no drugs was added to the control. Cells were cultured in plates at 37 °C under a humidified atmosphere with 5% CO 2 for 72 h. MTT was dissolved in PBS at 5 mg/mL and then medium with DOX was exchanged by 20 μL of this solution. The cells were incubated at 37 °C for another 1 h, after which cells were washed 3 times with PBS, and 50 μL of 2-propanol was added to each well. The absorbance at 585 nm was measured using a Tecan GeniosPro plate reader with 612 nm as a reference wavelength. The absorbance of the background was determined using 2-propanol but no cells. The results are presented as the percentage viability of cells exposed to DOX relative to non-exposed cells. www.nature.com/scientificreports/ (RIN) was ≥ 9.4, as evaluated with 2100 Bioanalyzer system using RNA 6000 Nano Kit (Agilent Technologies, Santa Clara, CA, USA).
Statistical analysis. Statistical analysis was performed using STATISTICA v8.0 (StatSoft, Tulsa, OK, USA) and GraphPad Prism software v0.0 (GraphPad Software, La Jolla, CA, USA). Data are statistically presented as the mean (± Standard Error of the Mean (SEM)) for at least two separate experiments. Comparisons between groups were analyzed via t-test 29 . Probability values of P < 0.050 were considered statistically significant.