Changes in DNA methylation are associated with the development of drug resistance in cervical cancer cells

Human mesenchymal stem cells (MSCs) were isolated and cultured as described by Lee et al. [41], and cell expansion was as described by Hsiao et al. [17, 41]. MDA-MB-231, SiHa, and S3 cells were cultured with L-15, Minimum Essential Medium (MEM; Invitrogen), and MEM with 2 μg/ml oxaliplatin, respectively. For all cells, the medium was supplemented with 10 % fetal bovine serum (Invitrogen), 100 mg/ml penicillin/streptomycin (Invitrogen), and 2 mM l-glutamine (Invitrogen).

Cells were treated with 5 μM 5-aza or an equal volume of DMSO as a control for 5 consecutive days.

Primers for the human Casp8AP2 promoter are listed in Additional file 1: Table S1. Human MSC genomic DNA was used as a polymerase chain reaction (PCR) template. Purified PCR products were ligated into the pyT&A cloning vector (Yeastern Biotech) according to the manufacturer’s protocol. Inserts were confirmed by restrictions and sequencing.

Methylated DNA showing resistance to methylation-sensitive restriction enzymes (HpaII) was considered to indicate completed conversion (Additional file 1: Figure S1).

PCR products (0.4 μg/well, unmethylated as a control) were denatured at 95 °C and then transfected into 5 × 10⁵ cells/well in a 6-well plate using DMRIE-C (Invitrogen) according to the manufacturer’s instructions. Cells were transfected three times, on day 1, 3, and 5 [17, 42]. The transfection efficiency and the localization of the transfected DNA were tracked as in Additional file 1: Figure S2.

Genomic DNA (0.5 μg) was bisulfite-converted and purified as described by Yan et al. [43].

The qMSP was performed as described by Yan et al. [43]. Bisulfite-converted genomic DNA was subject to real-time PCR with methylation-specific primers (Additional file 1: Table S1). A SYBR Green I PCR Kit (Toyobo) was used to conduct qMSP in an iQ5 PCR instrument (Bio-Rad). After reactions, analysis of melting temperature was performed to ensure that a specific amplicon was generated. Col2A1 (NM_033150) was used for standard curve construction and as a loading control. Methylation percentage was calculated as: (mean of target gene)/(mean of Col2A1). Fold change was calculated as: (targeted DNA methylation percentage)/(mock methylation percentage).

The DMH procedure was performed as described by Leu et al. [44] using a human CpG microarray (Agilent). Treated and control genomic DNA (2 μg) was restricted into small fragments by MseI and ligated with designated primers. Methylation-sensitive restriction enzymes (BstUI and HpaII) were used to discriminate between methylated and unmethylated DNAs, and DNAs was then amplified by PCR using adaptors as primers. PCR-amplified DNA from mock-treated S3 cells was labeled with Cy5 and from SiHa or 5-aza-treated S3 cells was labeled with Cy3 and then co-hybridized onto slides. After scanning, the ratio between Cy5 and Cy3 dyes was normalized by locally weighted scatterplot smoothing. Significant methylation differences were identified.

RNA isolation, first-strand cDNA synthesis, and detection of transcripts were carried out as previously described [44]. Total RNA (2 μg) was reverse transcribed using SuperScript II reverse transcriptase (Invitrogen). qRT-PCR was performed using a SYBR Green I PCR Kit (Toyobo) in an iQ5 Real-Time instrument (Bio-Rad). A serial dilution of GADPH-amplified (NM_002046) cDNA was used as a control to generate a standard curve, and GAPDH from each sample was used as a loading control. The primers used are listed in Additional file 1: Table S1.

Cells (5 × 10⁴) were plated into each well of a 96-well assay plate and allowed to attach. Cells were then treated with different concentrations of drugs and incubated at 37 °C overnight. 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) solution (20 μl, 5 mg/ml; Sigma) was added to each well and incubated at 37 °C for 5 h. The reaction was terminated by adding 100 μl DMSO, and absorbance was measured at 595 nm.

Cells were harvested in RIPA buffer, and proteins were separated in 10 % polyacrylamide gel and trans-blotted onto a membrane. After blocking with skim milk, the membrane was hybridized with designated antibodies. After washing, secondary antibody conjugated with horseradish peroxidase was used to detect hybridization. Results were visualized by chemiluminescence. The film was then scanned and analyzed.

Treated or control cells (5 × 10⁴) were plated into 4-well chamber slides and allowed to attach. After washing and fixing in 2 % formaldehyde, cells were again washed and permeabilized by 0.5 % NP40 in phosphate-buffered saline (PBS). After another wash, horse serum in PBS (1:100) was used to block, and the slides were washed. Specific antibodies were used to stain the cells, and fluorescein-conjugated secondary antibodies were used to detect the staining. The slides were mounted, and cells were visualized using a fluorescent microscope.

Green fluorescent protein (GFP) ELISA was performed using an ELISA Kit (Cell Biolabs) according to the manufacturer’s instructions. Starting from 2 × 10⁴ cells per assay, the cells were harvested and lysed, and the collected proteins were quantified. Proteins (0.1 μg/ml per assay) were compared with the provided standard after binding to GFP antibody, secondary antibody, and substrate solution with vigorous washes between steps. After stopping the reaction, absorbance was measured at 595 nm.

A more than 40-fold decline in sensitivity was found for the oxaliplatin-resistant cervical cancer cell S3 compared with its parental SiHa cervical cancer cell line (Fig. 1a). The expression of several death-related genes decreased during the development of oxaliplatin resistance (Fig. 1b). This decreased expression of death-related genes, such as Casp8AP2, the detoxification gene GSTp1, and the repair gene MLH1, was associated with increased methylation within their promoter regions (Fig. 1c). These findings suggest that changes in epigenetic and DNA expression states are involved in the development of drug resistance.

Genome-wide changes in methylation during the development of drug resistance in S3 cells were detected by DMH. Changes in methylation were compared between S3 and SiHa cells (Fig. 2a, S1 and S2) as well as before and after 5-aza treatment in S3 cells (Fig. 2a, S3 and S4). After hierarchical clustering analysis of repeated DMH results, we found both increases (Fig. 2a, block I) and decreases (Fig. 2a, block III) in methylation within the S3 genome. Some of the increases in methylation were inhibited by 5-aza treatment (block I), suggesting that these methylation changes are associated with the development of drug resistance. On the other hand, some increases in DNA methylation were observed after 5-aza treatment (block III), suggesting that these methylation changes are not involved in the development of drug resistance. Three primary target loci (NEUROG2, PVT1, and DLX2) identified from DMH analysis were validated together with the three known targets (Casp8AP2, GSTP1, and MLH1) as controls (Fig. 2b). S3 showed increased methylation within the promoter region and lower expression of these genes compared with SiHa. This increased methylation (Fig. 2c, upper panel) and lower gene expression (Fig. 2c, lower) in S3 cells was reversed by 5-aza treatment. Treatment with 5-aza also increased Casp8AP2 expression in S3 cells as detected by RT-PCR (Fig. 2c) and immunostaining (Fig. 2d, also detected by Western blot, Additional file 1: Figure S3).

If increased DNA methylation in drug-resistant cancer cells is necessary for the maintenance of drug resistance, then the reversal of DNA methylation in S3 should restore their sensitivity to drug treatment. After 5-aza treatment, S3 lost their resistance to cisplatin and taxol (Fig. 3a, upper and lower panels, respectively). Also, after 5-aza treatment, the sensitivity of S3 cells to oxaliplatin was restored to the same level as that in untreated SiHa (Fig. 3b). Together, these findings indicate that the maintenance of methylation may be critical for the maintenance of drug resistance in cancer cells.

To confirm that methylation of specific genes is sufficient to increase cellular drug resistance, we developed a two-component system to monitor methylated Casp8AP2. To visualize the targeted methylation of Casp8AP2, the Casp8AP2 promoter was cloned in front of the Tet repressor gene, and, on another vecor, Tet repressor binding sites TetO2 (Tet operator) were cloned in front of the enhanced GFP (EGFP) reporter (Fig. 4a). If Casp8AP2 is not methylated, the Tet repressor is expressed, binds to TetO2 and suppresses EGFP expression. By contrast, the targeted methylation of Casp8AP2 suppresses Tet repressor gene expression and releases the suppression of EGFP expression. Using this system, Casp8AP2 methylation is reflected by increased levels of EGFP.

We chose both an upstream and downstream regions to targeted methylate Casp8AP2 (Fig. 4b, upper panel). We observed increases in Casp8AP2 methylation (Fig. 4b, center panel) and EGFP expression (Fig. 4b, lower panels) regardless of whether the far or near ends were targeted. The targeted CAsp8AP2 methylation was also evidenced by bisulfite sequencing (Additional file 1: Figure S4) Successful targeted DNA methylation increased the methylation (Fig. 4c, upper panels) and reduced gene expression (Fig. 4c, lower panels) of Casp8AP2 in both MSCs and the breast cancer cell line MDA-MB-231. Both cell lines became less sensitive to cisplatin and taxol after Casp8AP2 methylation (MDA-MB-231 cells: Fig. 4d; MSCs: Additional file 1: Figure S5).