Background

Epithelial ovarian cancer (EOC) is a highly lethal and heterogeneous disease characterized by a distinctive propensity for peritoneal spread, whereas metastasis to distant organs only tends to occur in the late stages [1]. The standard EOC treatment includes cytoreductive surgery followed by combination chemotherapy with carboplatin and paclitaxel; however, some patients relapse due to chemoresistance, and can even die from the disease [2]. Therefore, it is important to identify markers that predict the patient outcome, which may enable the development of prognostic or therapeutic biomarkers.

MicroRNAs (miRNAs), a class of small noncoding RNAs (approximately 22 nt), have a post-transcriptional regulation function of binding to the 3′-untranslated region (3′-UTR) of target mRNAs, resulting in mRNA degradation or translation inhibition [3,4,5,6]. An altered miRNA expression has been reported in almost all human cancer types. miRNAs can function as oncogenes or tumor suppressor genes, being involved in multiple pathways and cell functions related to cancer development and progression, such as proliferation, apoptosis, invasion, and resistance to therapy [3,4,5,6,7,DNA isolation and bisulfite sequencing analysis

Genomic DNA was isolated using a QIAamp DNA Mini Kit (Qiagen). Sodium bisulfite modification of the DNA was performed using an EZ DNA Methylation-Gold Kit (Zymo) according to the protocol of the manufacturer. The two CpG sites of the miR-509-3p promoter region were amplified via PCR using the bisulfite-modified DNA template. The methylated allele (M1) was amplified using the primers miR-509-3p-F (5′-GGTATAGAATATTTAGTATGTGG-3′) and miR-509-3p-R (5′-TTTCTATTTTATTTCTCTTTT-3′) and the methylated allele (M2) was amplified using the primers miR-509-3p-F (5′-AGGAAGAAAGAATAAGTTATTTA-3′) and miR-509-3p-R (5′-AAAACAATTA TTTCTTATATT-3′). The PCR product was analyzed using a BigDye Terminator cycle sequencing kit (Applied Biosystems, Foster City, CA) and an ABI 3730 automated capillary sequencer.

Plasmid constructs and transfection

Small interfering RNAs (siRNAs) directed against human COL11A1 (sc-72956-SH), and a non-targeting negative control target (sc-108060) were purchased from Santa Cruz Biotechnology (Dallas, TX, USA). COL11A1 cDNA plasmid (BC117697, GE Healthcare) was cloned into a pCMV6-AC-GFP vector (PS100010, OriGene), followed by verification via sequencing. SUMO-3 (HG12782-UT) cDNA plasmid was purchased from Sino Biological (Bei**g, China). miR-509-3p mimics (MC12984), mimics negative control (4464058), miR-509-3p inhibitor (MH12984), and inhibitor negative control (4,464,076) were purchased from Ambion (Foster City, CA, USA). A2780CP70 or OVCAR-8 cells were transfected with miR-509-3p mimics and SUMO-3 in combination using Lipofectamine 3000 (Thermo Fisher Scientific).

Western blot analysis, antibodies, and reagents

Following protein extraction, equal amounts of protein were separated using 8–15% sodium dodecyl sulphate–polyacrylamide gel electrophoresis [12]. Antibodies against COL11A1 (GTX55142), DNMT1 (GTX116011), DNMT3A (GTX129125), and DNMT3B (GTX129127) were obtained from GeneTex (Irvine, CA, USA). An anti-β-actin antibody (sc-47778) was purchased from Santa Cruz Biotechnology (Dallas, TX, USA), whereas antibodies against Akt (9272), phospho-Akt (Ser473, 9271), ubiquitin (58,395), mouse IgG (7076), and rabbit IgG (7074) were obtained from Cell Signaling Technology (Danvers, MA, USA). An antibody against phospho-DNMT1 (Ser84) was purchased from Affinity Biosciences (Melbourne, Australia). An antibody against phospho-DNMT1 (Ser154) was purchased from Bioss Antibodies (Woburn, MASS, USA). An antibody against SUMO-3 was purchased from Abcam (Cambridge, UK). An antibody against p16 (AF5484) was purchased from Affinity Biosciences (Bath, UK). 5-aza-2′-deoxycytidine (5-aza) and MG132 were obtained from Sigma-Aldrich. Cisplatin (Fresenius Kabi Oncology, Ltd.) was provided by the Cancer Center of National Cheng Kung University Hospital.

Cell proliferation and 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) cytotoxicity assay

Cells (104/well) were seeded on 96-well flat-bottomed microtiter plates, and then transfected with miR-509-3p mimics or miR-509-3p inhibitor and cultured for 24, 48, 72, and 96 h. For co-transfection, before miR-509-3p mimic treatment, the cells in the exponential growth phase were pretreated with SUMO-3 overexpression plasmids for 24 h and cultured for 24, 48, 72, and 96 h. For cisplatin treatment, after transfection with miR-509-3p or miR-509-3p inhibitor for 24 h, cells were treated with different cisplatin doses. After 48 h of incubation, the in vitro cytotoxic effects of these treatments were determined using an MTT assay (at 570 nm) and the cell viability was expressed as the percentage of control (untreated) cells (% of control). The MTT analysis was conducted as previously reported [12].

Transwell invasion assay

The Transwell cell invasion assay was performed using polycarbonate membranes with 8 μm pores (Costar, Cambridge, MA, USA). Cells (5 × 104) were seeded on the membrane of the upper chamber of the Transwell pre-coated with rat collagen I (60 µg/Transwell). Fibronectin in medium (0.6 mL) was added to the lower chamber as the chemoattractant in a 24 h assay at 37 °C under 5% CO2. The remaining cells in the upper chamber that did not migrate were removed using a cotton swab. The filters were fixed in 95% ethanol and stained with 0.005% crystal violet for 1 h. Migrated cells were counted using a phase-contrast microscope (Olympus, Lake Success, NY, USA). The mean of 10 contiguous fields represented the cell number. Each experimental condition was assayed in triplicate. used. The invasive capacity of cells was normalized to that of each corresponding control. One-sample unpaired Student's t-test was conducted to analyze the differences between the normalized invasive capacities obtained from the three independent experiments and the hypothetical value (which was set to 1).

Luciferase reporter analysis

The SUMO-3 3′-UTR fragments with wild-type miR-509-3p binding sites (Wt) or mutated binding sites (Mut) were inserted into a pGL4 vector (Promega). The SUMO-3 3′-UTR PCR product was cloned into the SacI/EcoRV site of the pGL4 vector. The following primers were used to target the SUMO-3 3′-UTR: forward, 5′-TTCACCACGATGATTTTCCT-3′ and reverse, 5′-GCACACAAAAGTACCCACAATATC-3′. The resultant construct was confirmed using DNA sequencing. Site-directed mutagenesis was performed to generate SUMO-3 3′-UTR constructs containing miR-509-3p mutant-binding sites by using the following complementary oligonucleotides: forward, 5′-CTGTAACTTAAATTGGGTTAATCAG-3′ and reverse, 5′-CTGATTAACCCAAT TTAAGTTACAG-3′. A2780CP70 or OVCAR-8 cells were transfected with the vector and the miR-509-3p mimics in combination. We performed luciferase assays 48 h post-transfection using a dual-luciferase reporter assay system (Promega). The normalized luciferase activity was reported as the ratio of luciferase activity to β-galactosidase activity. The activities of firefly luciferase and Renilla luciferase were measured as described previously [12].

Chromatin immunoprecipitation (ChIP) assays

Native protein–DNA complexes were cross-linked via treatment with 1% formaldehyde for 15 min, and ChIP assays were performed as previously reported [13]. Briefly, equal amounts of isolated chromatin were subjected to immunoprecipitation using anti-DNMT1, anti-DNMT3A, anti-DNMT3B, and IgG monoclonal antibodies. Primers with the following sequences were used for the ChIP assays: miR-509-3p forward, 5′-GGTACAGAACATTCAGCATGTGG-3′ and reverse, 5′-AGAAAACTAGAAAAC TGTACAAA-3′.

Statistical analysis

Data were analyzed using SPSS statistical software (version 21.0, IBM Corp., Armonk, NY, USA). Categorical variables are presented as frequencies and percentages and were analyzed using Chi-square test or Fisher’s exact test. Continuous variables are expressed as the mean ± standard deviation or as the median ± interquartile range. Interval variables were analyzed using Student’s t-test or Mann–Whitney U test. The cut-off values obtained based on the receiver operating characteristic curve for miR-509-3p, COL11A1 and miR-335 were optimized for their diagnostic sensitivity and specificity in predicting cancer progression or death. Survival was estimated using the Kaplan–Meier method and was compared using the log-rank test. Two-sided P-values < 0.05 were considered statistically significant. Cox proportional hazards models were implemented to estimate the hazard ratios (HRs) and 95% confidence intervals (CIs).