DNA Methylation of miR-7 is a Mechanism Involved in Platinum Response through MAFG Overexpression in Cancer Cells

Abstract

One of the major limitations associated with platinum use is the resistance that almost invariably develops in different tumor types. In the current study, we sought to identify epigenetically regulated microRNAs as novel biomarkers of platinum resistance in lung and ovarian cancers, the ones with highest ratios of associated chemo-resistance. Methods: We combined transcriptomic data from microRNA and mRNA under the influence of an epigenetic reactivation treatment in a panel of four paired cisplatin -sensitive and -resistant cell lines, followed by real-time expression and epigenetic validations for accurate candidate selection in 19 human cancer cell lines. To identify specific candidate genes under miRNA regulation, we assembled "in silico" miRNAs and mRNAs sequences by using ten different algorithms followed by qRT-PCR validation. Functional assays of site-directed mutagenesis and luciferase activity, miRNAs precursor overexpression, silencing by antago-miR and cell viability were performed to confirm their specificity in gene regulation. Results were further explored in 187 primary samples obtained from ovarian tumors and controls. Results: We identified 4 candidates, miR-7, miR-132, miR-335 and miR-148a, which deregulation seems to be a common event in the development of resistance to cisplatin in both tumor types. miR-7 presented specific methylation in resistant cell lines, and was associated with poorer prognosis in ovarian cancer patients. Our experimental results strongly support the direct regulation of MAFG through miR-7 and their involvement in the development of CDDP resistance in human tumor cells. Conclusion: The basal methylation status of miR-7 before treatment may be a potential clinical epigenetic biomarker, predictor of the chemotherapy outcome to CDDP in ovarian cancer patients. To the best of our knowledge, this is the first report linking the regulation of MAFG by miRNA-7 and its role in chemotherapy response to CDDP. Furthermore, this data highlights the possible role of MAFG as a novel therapeutic target for platinum resistant tumors.

Keywords: Cisplatin-resistance; DNA methylation; MAFG; cancer.; miR-7.

Figure 1

Selection of candidate miRNAs under epigenetic regulation and candidate target genes. The flowchart indicates the steps and criteria used for the selection of the final 7 candidate miRNAs under epigenetic regulation and the final 4 candidate genes under possible regulation of miR-7.

Figure 2

miRNA relative expression on CDDP-sensitive and -resistant ovarian cancer cell lines. (A) Viability curves showing the acquired resistance of A2780 and OVCAR3 cell lines; Cells were exposed for 72 h to each drug concentration. Data were normalized to the untreated control, which was set at 100% and represent the mean + SD of at least 3 independent experiments performed in quadruplicate at each drug concentration tested for every one cell analyzed. IC50, is the inhibitory concentration that kills 50% of the cell population. Resistant index (RI) calculated as IC50 resistant / IC50 sensitive cell line. p<0.001 was considered as a significant change in drug sensitivity (Student's t-test). (B-C) Relative expression levels of the selected miRNAs measured by qRT-PCR. Data are represented in log10 scale and are expressed using the corresponding sensitive (S) line as a calibrator. Each miRNA level was normalized to RNU48 as an endogenous control. Assays were made in the ovarian cancer cell lines A2780 (B) and OVCAR3 (C) in all experimental conditions: S, R and RT. S: sensitive; R: resistant; RT: resistant treated with epigenetic reactivation drugs (5-Aza and TSA). The expression number assays for each miRNA are indicated in Supplementary Table 2.

Figure 3

Bisulfite sequencing of miRNA-7 regulatory CGIs. Chromosomal location of miR-7 and their nearby CGIs, as well as representative images of corresponding bisulfite sequences (BS). CGIs are represented in red boxes; each CpG position is characterized by vertical black lines inside the boxes. The first nucleotide of each miRNA is indicated by +1. Facing arrows mark the primer positions used for BS. It is shown the methylation analysis of the two CGIs closely related to the encoded miR-7 region. For the first CGI, the 3 different fragments (left half of the Figure) corresponding to the most frequently methylated positions are shown. A representation of 5 of the 11 additional tumor cell lines interrogated, BT474, SKOV3, LoVo, IMIMPC2 and SW780, is also shown. All CpG positions interrogated at the second CGI were fully methylated in all the samples analyzed (right half of the Figure), as indicated by the presence of C preceding a G in the sites indicated by the asterisks. OC: ovary control; LC: lung control. Asterisks indicate methylated positions.

Figure 4

miRNA-7 methylation analysis in primary tumors and survival analysis. (A) Representative MSPs of miR-7 nearby CGI in DNA obtained from ovarian tumor tissues, normal ovarian tissues, non-tumor cell line and PBMCs from healthy donors. For each sample, the PCR product in the M lane was considered as the presence of methylated DNA, whereas the amplification product in the U lane was considered as the presence of unmethylated DNA. In vitro methylated DNA was used as a positive control (+). Uncropped gels of Figure 4a are included in Supplementary Figures (Supplementary Figure 4). (B) and (C) Kaplan-Meier comparison between cisplatin treatment and miR-7 proximal island methylation in ovarian cancer patients treated with platinum in terms of progression free survival (B) and overall survival in months (C). LogRank, Breslow and Tarone-Ware tests were used for comparisons and p<0.05 was considered as a significant change in OS or PFS. p values in (B) represent the significant difference between sensitive-unmethylated and sensitive-methylated patients

Figure 5

mRNAarrays data validation and effect of miRNA-7 over-expression on candidate target genes in the ovarian cancer cell line A2780. (A) Relative expression levels of the selected genes measured by qRT-PCR. Assays were made in all experimental conditions: S, R and RT. S: sensitive; R: resistant; RT: resistant treated with epigenetic reactivation drugs (5-Aza and TSA). Sensitive cells were used as calibrator. (B) Relative expression levels of MAFG, ELK-1 and miR-7 measured by quantitative RT-PCR after miR-7 overexpression. The sensitive cell line transfected with the mimic negative control was used as a calibrator (S miR-NC, white). A2780R cells were transfected with same negative control (R miR-NC, stripped) or with miR-7 precursor (R miR-7, grey). For both (A) and (B), data are represented in Log10 scale obtained from the combined relative expression of 2 independent experiments measured in triplicate. Each gene expression level was normalized to GAPDH or B-actin as an endogenous control.

Figure 6

Site-directed mutagenesis for luciferase activity assay and effect of miR-7 silencing over MAFG expression. (A) Chromosomal localization of miR-7 predicted binding sites at 3'UTR of MAFG. Regions 2 and 8 were identified by six or more bioinformatical algorithms. Sanger sequencing showed that the seed sequence of miR-7 was fully mutated at regions 2 and 8 of the 3' UTR of MAFG. (B) Co-transfection of mimic miR-7 (miR-7) or mimic control (miR-NC) with the 3' UTR of MAFG WT, mutated on region 2 (MAFG 2*) and mutated on region 8 (MAFG 8*). Experiments were performed at 15nM and data was analyzed after 24h of co-transfection. (Upper panel) Relative luciferase activity. The figures represent the mean ± SD of at least 3 independent experiments after data normalization with Renilla and the data from the negative control 3'-UTR; p<0.01 was considered as significant change in Luciferase activity (Student's t-test). (Lower panel) Relative miR-7 expression levels measured by qRT-PCR after co-transfection, as an internal control for the mimic transfection. Each bar represents the combined relative expression of 2 independent experiments measured in triplicate. The miR-NC co-transfected with the 3'-UTR of each tested group was used as calibrator. (C) Relative miR-7 and MAFG expression levels measured by qRT-PCR after silencing of miR-7 with antago-miR in A2780S cells. Two different concentrations of Anti-miR-7 were tested, 20nM and 40 nM. Data was analyzed at 48h after transfection. Each bar represents the combined relative expression of 2 independent experiments measured in triplicate. A2780S cells transfected with the Anti-miR Negative Control was used as calibrator.

Figure 7

Effect of overexpression of MAFG and ELK-1 on cell sensitivity to CDDP in A2780 cell lines. (A-B) Viability curves of A2780 cell lines transfected with pCMV6 (S-Ø and R-Ø) and with the overexpression vectors (S-MAFG and S-ELK-1). Each experimental group was exposed for 48 h to 6 different CDDP concentrations, and data were normalized to each untreated control, set to 100%. The data represent the mean ± SD of at least 3 independent experiments performed in quadruplicate at each drug concentration for each cell line analyzed. The CDDP-RI (Resistant Index to CDDP) was calculated as “IC50 from the R-Ø / IC50 from the S-Ø" and “IC50 from the S-transfected with the gene / IC50 from the S-Ø” ± SD. p<0.01 was considered as significant change in drug sensitivity (Student's t-test). (C) Validation of the transfection efficacy at mRNA and protein levels. Top, Relative expression levels of MAFG and ELK-1 measured by quantitative RT-PCR, in the cell line A2780, represented in Log10 scale; In each experimental group, the sensitive cell line transfected with pCMV6 plasmid was used as a calibrator. Each bar represents the combined relative expression of 2 independent experiments measured in triplicate. Bottom, total cell protein (20µg) at 24 and 72 hours was subjected to WB, membranes were hybridized with antibodies against c-Myc and β-tubulin as loading control. S: Sensitive; S-G: Sensitive transfected with the gene; R: Resistant; β-tub: β-tubulin. (D) Stable overexpression of ELK-1 in A2780S cell line. Top, Viability assay after the nonsilencing (NS) plasmid transductions (S-NS white; R-NS stripped) and overexpressing ELK-1 plasmid (S-ELK-1 grey). Each experimental group was exposed for 72 h to 6 different CDDP concentrations, and data were normalized to each sensitive subtype. Data represents the mean ± SD of at least 3 independent experiments performed with 4 wells at each drug concentration for each cell line analyzed. Bottom, relative ELK-1 expression levels measured by qRT-PCR and represented in Log10 scale. The sensitive cell line with nonsilencing vector was used as a calibrator. Each bar represents the combined relative expression of 2 independent experiments measured in triplicate.