Investigation of the Role of CpG Methylation in Epithelial Mesenchymal Transition Master Genes in a Chemoresistant Ovarian Cancer Cell line


 Ovarian cancer is one of the commonly diagnosed cancers among women. Chemoresistant is an essential reason for treatment failure and high mortality. Emerging evidence connects epithelial mesenchymal transition (EMT) like changes and acquisition of chemoresistance in cancer. DNA methylation influences cellular processes including EMT. Here, we investigate EMT like changes in cisplatin-resistant A2780 ovarian cancer cells (A2780cis), and we study the DNA methylation role in EMT master genes regulation. Cell viability assay was carried to test the sensitivity of A2780, and A2780cis human cancer cell lines to cisplatin compared to other cancer cell lines. Differential mRNA expression of EMT markers using qPCR was conducted to investigate EMT like changes. The role of CpGs methylation in gene expression regulation was investigated by 5-azacytidine (5-aza) treatment. DNA methylation changes in EMT genes were identified using Methylscreen assay between A2780 and A2780cis cells. A2780cis maintains its cisplatin tolerance ability and exhibits phenotypic changes congruent with EMT. Methylscreen assay and qPCR study revealed DNA hypermethylation in promoters of epithelial adhesion molecules CDH1 and EPCAM in A2780cis compared to the cisplatin-sensitive parental cells, these changes were concomitant with gene expression down-regulation. DNA hypomethylation associated with transcription up-regulation of the mesenchymal marker TWIST2 was observed in the resistant cells. Azacytidine treatment confirmed the DNA methylation role in the regulation of gene expression of CDH1, EPCAM and TWIST2 genes. A2780cis cell line undergoes EMT like changes, and EMT master genes are regulated by DNA methylation. A better perception of the molecular alterations which correlate with chemoresistance may lead to therapeutic benefits such as chemosensitivity restoration.


Introduction
Ovarian cancer is one of the deadliest gynecologic cancer in females 1 . Patients with ovarian cancer have a poor prognosis, and a high mortality rate 2 . The low survival rate of ovarian cancer is due to metastasis and acquiring of chemotherapies resistance 3 . Cisplatin (cis-diamminedichloroplatinum [II]) is a platinum compound used to treat many solid tumors including ovarian cancer 4,5 . Cisplatin induces DNA lesions and activates several signaling pathways such as DNA repair mechanisms 6 . When DNA damage fails to be repaired, cells undergo apoptosis and die 7 . Drug resistance is the main obstacle that limits treatment effectiveness 8 . The mechanisms that underlie platinum drug resistance are multifactorial. Several cellular processes have been identi ed as responsible for the resistant phenotype, including alterations in drug in ux or e ux, DNA repair, cell cycle and apoptosis 9,10 . Recent evidence suggests that epithelialmesenchymal transition (EMT) processes may have a role in the development of chemoresistance. EMT is a cellular procedure where epithelial cells miss their cell-cell adhesion, cell polarity and gain metastasis capability 11 . The molecular signatures of EMT are loss of epithelial cell markers CDH1 and mesenchymal markers up-regulation including EMT transcription factors SNAIL and TWIST 12, 13 . EMT like changes has been reported in various chemoresistant cancer such as gastric cancer 14 non-small cell lung cancer 15,16 , nasopharyngeal carcinoma cells 17 , breast cancer cells 18, 19 , and ovarian cancer cells 20, 21 .
Multiple epigenetic events, such as DNA methylation, histone modi cations, and non-coding RNA, have been described to contribute to the acquisition of chemoresistance [22][23][24] . DNA methylation is the most well-known epigenetic mechanism that occurs at CpGs islands that in uence cellular processes by regulating gene expression 25 . Commonly, gene promoter hypermethylation is associated with reduced expression, while hypomethylation increases gene expression 26,27 . DNA methylation changes at CpG islands associated with transcriptional silencing have been stated in cisplatin-resistant cancer cell lines [28][29][30] . For example; DNA methylation of several genes including (ARMCX2, COL1A1, MDK, MEST, MLH1, KLF4, ST3GAL5, SYNE1, CXCL8, HERC5, FOSL1, and ARRDC4) was linked to ovarian cancer initiation and chemotherapy resistance 31,32 . A study in ovarian and breast cancer cell lines with doxorubicin tolerance described methylation changes in genes that contributed to chemoresistance and identi ed hyper-methylation of CDH1, BRCA1, SULF2 and DNAJC15 besides hypo-methylation of APC, ABCB1 and HIC1 genes 23 . Based on these considerations, the need to study the linkage between EMT and DNA methylation regulation has become pivotal to understand the chemoresistant phenotype.
The present study aimed to assess the morphological and transcriptional changes in cisplatin resistant A2780 ovarian cancer cell line in relation to EMT, and comparing these changes with different types of cisplatin-resistant cell lines. Our study focused on CpG methylation changes that regulate EMT master genes in A2780 chemoresistant cells.

Cancer cells display different morphological phenotypes
We observed the morphological characteristics of the different cancer cell lines sensitive or resistant to chemotherapy during exponential growth, i.e. shape, size and growth pattern of the cells during culture. A2780 cells formed clusters with small round-shaped cells consistent with an epithelial phenotype. In contrast, A2780cis cells were larger, more elongated and were frankly polygonal (Fig. 1A).
These observations were compared with the morphological of MDA-MB-231 and T98G cell lines as they were known to be resistant to chemotherapy [33][34][35] . Figure 1B shows that MCF7 cells have small squamous epithelial appearance, whereas MDA-MB-231 cells have spindle-like morphology. U-87 MG and T98G cells exhibit spindle like morphology, T98G cells showed a larger and more elongated cell shape compared to U-87 MG (Fig. 1C). These observations suggest that this dissimilar cellular shape is possibly associated with alteration in the EMT process.
Dissimilar shaped cancer cell lines have different cisplatin tolerance capacity MTT assay was performed to assess the effect of cisplatin on different human cancer cell lines. Cells were treated with increased concentrations of cisplatin for 24 hrs. Concentration-dependent effect of cisplatin on different cells was observed. IC 50 value in A2780cis cells was 4 times higher than the parental cell line's IC 50 , which suggested that the A2780cis cells were more resistant to cisplatin-induced cytotoxicity compared with A2780 cells and had still maintained cisplatin resistance ( Fig. 2A). Results in Fig. 2B show that the IC 50 value for MDA-MB-231 cells is six times greater than IC 50 for MCF7 cells. Figure 2C shows the differences in cisplatin sensitivity between U-87 MG and T98G cell lines, the difference in IC 50 values was the least among other cell line models used in this study. IC 50 in T98G cells was 1.3 times higher than U-87 MG's IC 50 . The IC 50 results were summarized in (Fig. 2D).

MDR1 Expression increases in the resistant variant A2780cis cell line
To determine whether cisplatin resistance in the A2780cis cells is associated with increased expression of ABC transporters genes, mRNA expression level of the following genes was assessed: MDR1(multidrug resistance protein 1 gene), MRP1 (MDR-related protein 2) and MRP2 (MDR-related protein 2) using qRT-PCR. Figure 3 showed that there was a statistically signi cant increase of MDR1 transcript level (5.8 fold) in A2780cis cell line compared to the parental cell line. On another hand, MRP1 and MRP2 expression did not signi cantly change in A2780 and A2780cis cell lines.

Cisplatin resistant cancer cells have molecular changes consistent with EMT
To determine whether the gaining of cisplatin resistance promotes speci c molecular alterations corresponding with EMT in ovarian cancer, qRT-PCR was performed to investigate EMT-related biomarkers expression. Results showed that the expression of epithelial markers, CDH1 and EPCAM, was signi cantly reduced by 0.02 and 0.013 fold, respectively in the A2780cis cells compared with A2780 cells. The expression of mesenchymal markers, SNAIL, and TWIST2, were higher by 4.9 and 20.3 fold respectively in the A2780cis cells compared with the parental cell line (Fig. 4A). Based on these observations, A2780cis cells may be considered to have a mesenchymal like phenotype.
We also compared these changes in transcriptional levels in MDA-MB-231 and T98G cells, to gure if the changes observed in A2780cis are common in other cancer cells resistant to therapy. The same molecular changes of EMT markers observed in A2780cis cells were detected in MDA-MB-231 in comparison with MCF7, CDH1 and EPCAM genes were decreased 0.5 and 0.006 fold respectively. These decreases were associated with up-regulation of EMT genes SNAIL and TWIST2 (2.6 and 40732 fold) (Fig. 4B). The same molecular changes were observed in EMT markers; CDH1 and EPCAM were down-regulated (0.046 and 0.2 fold, respectively) in T98G the more resistant cell line, whereas SNAIL and TWIST2 (88.5, 76 fold respectively) were up-regulated (Fig. 4C). We notice a signi cant overlap in expression pro les of EMT genes between resistant cancer cell lines, proposing a shared mechanism associated with resistance to therapy.

5-azacytidne treatment up-regulates expression of EMT-related genes in A2780 cell line
To identify the epigenetically regulated genes from the gene set that expressed differentially between A2780 cell line and the resistant variant, we analyzed gene expression by qPCR after azacytidine treatment. As shown in Fig. 5 azacytidine treatment signi cantly increased the expression of CDH1, EPCAM, SNAIL, and TWIST2 genes, by 7.8, 9.26, 5.24, and 4 fold respectively. This suggests that DNA methylation may be essential in the regulation of the expression of these genes in this cell line.

Discussion
Ovarian cancer ranks as one of the most common causes of cancer deaths among females 1 . Patients suffering from ovarian cancer have a poor prognosis, with a low survival rate 3 . High mortality of ovarian cancer is mainly due to metastasis and the evolution of resistance to chemotherapies 2 . Cisplatin is a cornerstone of the treatment regime for many solid tumors including ovarian cancer 4,5 . However, its clinical effectiveness is in uenced by acquiring tumor cells chemoresistant 8 . Many studies concentrated on the molecular mechanisms mediating the development of cisplatin resistance and lots have been characterized including decreased cellular uptake of the drug, increased drug e ux, enhanced DNA damage repair capacity, and anti-apoptotic signaling 9,10 .
In this study, we demonstrated that the A2780cis cell line still has the tolerance capacity to cisplatin and its resistance accompanied by an increase in the expression of MDR1 (Fig. 3). This marker seems to be a universal cellular response marker to chemotherapy in various cancer. A high expression level of MDR1 was detected in the cisplatin, paclitaxel and doxorubicin resistant variant of A2780 cells and different resistant cell lines derived from ovarian cancer [36][37][38][39] . Several studies have shown con icting results about the differential expression of MRP1 and MRP2 genes between sensitive and resistant ovarian cancer cell lines, and the ability of these changes to represent the chemoresistance 38,40 . Recently, epithelialmesenchymal transition (EMT) was implicated as a core mechanism mediating drug resistance 41 . EMT is de ned as a biological mechanism characterized by loss of cell adhesion, as well as loss of cell polarity and gain motility 11 . EMT results in changes in cell morphology associated with alterations in epithelial and mesenchymal markers expression 42 19 . In this present study, we demonstrate that the resistant variant A2780cis cell line underwent EMT. This was con rmed by noticing a morphological change from small round shaped to elongated and polygonal shaped cells and changes in molecular markers of EMT; signi cant reduction in CDH1 and EPCAM and upregulation of the transcription factors, SNAIL and TWIST2 ( Fig. 1 and Fig. 4).
Multiple studies have shown that naïve cancer cell lines and clinical tumor samples can be divided according to their mesenchymal/epithelial phenotype, this sorting could determine the sensitivity to chemotherapy in various cancers including ovarian, breast, and lung cancers [47][48][49][50][51][52] . From this observation, we compared the EMT like changes detected in A2780cis with changes in breast and glioma cell line models known as resistant cancer cells [33][34][35] . We used MDA-MB-231 and T98G cancer cells as other models of cell lines to gure if the acquired changes observed in A2780cis are common in other cancer cells, resistant to therapy. We detect the same molecular alteration consistent with morphological in cisplatin resistant MDA-MB-231 cells compared with the sensitive cells MCF7. The same transcripts changes in CDH1, EPCAM, SNAIL and TWIST2 genes also detected in chemoresistant T98G cell line in comparison with the less cisplatin tolerance cells U-87-MG (Fig. 2). This indicates the importance of CDH1, EPCAM, SNAIL and TWIST2 genes in EMT mechanism associated with cisplatin resistance. CDH1 is Ca 2+ dependent adhesion molecule that binds by its extracellular domain to CDH1 on the adjacent cell creating a bridge between the cell's cytoskeletons 53 . Many studies revealed that CDH1 downregulation may be associated with cancer cells resistant to chemotherapy that can be attributed to the EMT mechanism activation 23,54 . Acquisition of paclitaxel chemoresistant induces EMT phenotypic changes and CDH1 in down-regulation in NOS-PR and A2780/PTX ovarian cancer cell line 20,21 . EPCAM is an epithelial cell surface transmembrane glycoprotein that mediates homophilic cell-cell adhesion without Ca 2+ dependent 55,56 . In ovarian cancer cells, EPCAM upregulation is connected to a more favorable prognosis and more effective platinum-based therapy 57  DNA methylation is one of the best described mechanisms of epigenetic that regulate gene expression.
Aberrant DNA methylation is observed in cancers in CpG dinucleotides clustered around the TSS of genes, called CpG islands, leading to gene expression dysregulation 65 . Upon initiation of EMT, DNA methylation of the genome selectively undergoes CpG site methylation changes, which regulate transcription of EMTrelated genes 23,30,66,67 . In our study, we investigate the role of methylated CpG islands in the modulation of gene expression of EMT regulated genes in A2780 cancer cell lines. DNA methyltransferase inhibitor 5aza induces gene expression of CDH1, EPCAM, SNAIL, and TWIST2 which indicate that these genes may be regulated by DNA methylation. Chang et al, used gene expression pro ling after cancer cells treatment with 5-azadeoxycytidine, they identi ed genes that were dysregulated in cisplatin resistant cancer cells and reactivated by the DNA methyltransferase inhibitor 68 . Here, we found that epithelial gene promoters CDH1 and EPCAM became signi cantly more methylated in A2780cis compared to the parental cell line. These promoter methylation changes correlate with signi cant gene expression down-regulation.
Boettcher et al, pro led DNA methylation of 800 selected CpG islands and identi ed hypermethylation in CDH1 CpG islands in breast and ovarian doxorubicin resistance cancer cells 23 . EPCAM overexpression has been linked to promoter hypomethylation EPCAM-negative cells treated with a DNA methyltransferase inhibitor prompted EPCAM expression in various cancer types including ovarian cancer 69-71 . A recent study reported consistent methylation changes across multiple cancer cell lines differed in chemoresistant. Speci cally, hypermethylation of epithelial marker genes such as CDH1 and EPCAM promoters and hypomethylation of mesenchymal marker genes such as SNAIL in resistant versus parental cell lines 30 . Analysis of SNAIL promoter region predicted a CpG island surrounding the TSS, we examined the DNA methylation status in the genomic region (-688 to -165bp), we cannot observe any methylation in CpGs located in this region of SNAIL promoter in A2780 and its resistant variant cells, although there were changes in SNIAL expression between the two cell lines and after azacytidine treatment. Literature has described changes in the histone modi cations regulating SNAIL gene expression 72 . Single study described changes in the methylation of CpG island located in the rst intron after 1000 pb from TSS in EMT models of cancer cells 73 . Maybe the differential methylation in SNAIL gene could be identi ed in the intron region in A2780 cell line. CpGs island of mesenchymal transcription factor TWIST2 promoter showed DNA hypomethylation in A2780cis compared to the parental cells, this hypomethylation coincides with gene expression down regulation due to EMT activation. TWIST2 methylation changes were observed in various cancer such as colorectal cancers, prostate cancer, and chronic lymphocytic leukemia, this epigenetic event might be the underlying mechanism for TWIST2 transcriptional regulating 74,75 .
In conclusion, we have shown in this study that, the gain of cisplatin resistance in cancer cells is  Table 1.

5-Azacytidine treatment
In order to select the candidate genes for the methylation study, A2780 cells were cultured and treated with 0.1 mM 5-Azacytidine (Sigma-Aldrich, USA). Culture medium was removed every 24 hrs and replaced by a fresh medium containing 0.1 mM 5-aza. Treated and mock treated cells were collected after 7 treatment days and total RNAs were extracted as described above.

DNA extraction
Genomic DNAs from A2780 and A2780cis cells were extracted using the QIAamp DNA Mini kit (Qiagen, Germany) according to the manufacturer's instructions. Isolated DNAs were quanti ed using NanoVue Plus (GE Healthcare Life Sciences, Germany).

Methylscreen assay
Quantitative PCR-based methylation analysis (Methylscreen assay) was performed to analyze DNA methylation of genes that have differential expression between A2780 and the resistant variant cells and that expression increased after 5-aza treatment. Methylscreen assay is based on combined restriction digestion of DNA with methylation sensitive and methylation dependent restriction enzymes, MSRE and MDRE respectively 76 . Genomic DNA of A2780 and A2780cis cells were divided into four parts and treated with different digestions: (1) Rs: two methylation-sensitive enzymes MSRE (HhaI + AciI) or (HpaII + AciI) depending on the frequency of their restriction sites within the studied fragments, which are cutting only unmethylated DNA, (2) Rd: one methylation-dependent restriction enzyme McrBC (MDRE), which is cutting only methylated DNA or (3) Rsd: both MSRE and MDRE enzymes (double digest, DD), and (4) R0: neither MSRE nor MDRE (mock control). Each 50 ml reaction contained 1 mg of gDNA, 1x CutSmart Buffer, 100 μg/mL BSA, 1 mM guanosine-5'-triphosphate, 3 % glycerol and 10 U of each enzyme used in restriction reaction, 50 % glycerol was used instead of enzymes in mock reaction in order to keep restriction digest cocktail homogeneity. Digestions were incubated at 37 °C for 6 hrs followed by inactivation of the enzymes at 65 °C for 20 min. The enzymes, CutSmart Buffer, BSA, and guanosine-5'triphosphate were purchased from New England Biolabs, USA. Restricted samples were analyzed by qPCR with locus-speci c PCR primers and SYBR Green dye. An in-silico analysis was performed using EMBOSS Cpgplot sequence analysis tool (https://www.ebi.ac.uk/Tools/seqstats/emboss_cpgplot/) from European Bioinformatics Institute (EMBL-EBI) to identify the CpG sites associated with the proximal promoter and transcription start site (TSS) for four genes. Sets of locus-speci c PCR primers were designed to amplify gDNA at proximal CpG located within 1000 bp (±) of the transcription start site for each gene. Primers sequences, genomic loci, numbers of CpGs nucleotides and number of restriction sites contained in ampli ed amplicons are listed in Table 2. The PCR ampli cation was performed in 20 ml volume with 10 ml Maxima™ SYBR™ Green/ROX 2x qPCR Master Mix (Thermo Scienti c, USA), 300 nM of each primer and 2 ml (40 ng) of digested template DNA using the qPCR System. The PCR conditions were as follows: 95 °C for 10 min, and 45 cycles of 95 °C for 1 min and temperature for optimized annealing for 1 min.
Ampli cation for each sample was performed in triplicate in a 48-well plate. All primer pairs were tested to identify the annealing temperature for optimal e ciency and melting curve analysis was conducted after the reaction to verify the ampli cation of the desired products.

Calculations of DNA Methylation Occupancy
The Ct values from R0, Rs, Rd and Rsd, reactions were used to calculate the initial amount of DNA in each digest before PCR as following:  Tables   Table 1: Primer sequences used for quantitative real-time PCR in this study. Table 2: Primer sequences used for methylation study using methylscreen method.        (C) Charts display the result of TWIST2 assay as a percentage of each portion of DNA.