Gene profiling data of ovarian normal surface epithelia and ovarian cancer epithelial samples were downloaded from the GEO dataset (http://www.ncbi.nlm.nih.gov/geo). GSE14407 dataset was selected for bioinformatic analysis (18). The differential analysis was performed using R package ‘limma’ (19). Differential genes obtained from GSE14407 were visualized using the R package ‘pheatmap’ (https://CRAN.R-project.org/package=pheatmap) and ‘ggplot2’ (20). RNA sequencing data of KDM2A was achieved from the TCGA data portal (https://cancergenome.nih.gov/), containing 374 ovarian cancer samples. Corresponding clinical data were also downloaded and filtered out for useful information. Kaplan-Meier survival curves were conducted to assess the prognostic value of KDM2A using the R package ‘survival’ (https://CRAN.R-project.org/package=survival).

Human specimens were obtained between 01 January 2005 and 31 December 2011 from 27 patients, with a mean age of 46 years (range, 18–73 years), who underwent primary tumor resection at Renmin Hospital of Wuhan University (Wuhan, China). Among the 27 cases, the specimen groups consisted of EOC (n=9), borderline ovarian tumors (n=9) and normal ovary tissues (n=9). All specimens were confirmed by at least 2 pathologists. In the present study, the patients accepted no chemotherapy or radiotherapy before surgery. Before conducting our scientific investigation, consent was obtained from all the patients and the study was approved by the Ethics Committee of Wuhan University (Wuhan, China).

The immunohistochemical analysis of KDM2A expression in human EOC was performed as previously described (21). The rate of KDM2A-positive cells was scored semi-quantitatively in each section.

The human OC cell lines A2780 and SKOV3 were obtained from the State Key Laboratory of Molecular Biology, Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences (Shanghai, China). The cells were respectively cultured in MEM/F12 and RPMI-1640 medium supplemented with 10% fetal bovine serum (FBS) (both from Gibco; Thermo Fisher Scientific, Inc., Waltham, MA, USA), 1% penicillin and 1% streptomycin (Sigma-Aldrich; Merck KGaA, Darmstadt, Germany) in a CO₂ incubator under standardized conditions. Antibodies KDM2A (cat. no. ab191387), GAPDH (cat. no. ab181602), and β-actin (cat. no. ab8227) were obtained from Abcam plc. (dilution 1:500; Cambridge, UK). Antibodies phospho-PI3K (cat. no. sc4257), PI3K (cat. no. sc4292), phospho-Akt (cat. no. sc4060), Akt (cat. no. sc4691), phospho-mTOR (cat. no. sc5536), mTOR (cat. no. sc2983), MMP2 (cat. no. sc10736), MMP9 (cat. no. sc10737), Bcl-2 (cat. no. sc2872), Bax (cat. no. sc14796), E-cadherin (cat. no. sc14472), N-cadherin (cat. no. sc13116) and vimentin (cat. no. sc5741) were purchased from Cell Signaling Technology (dilution 1:500; Danvers, MA, USA). LY294002 was reported as the inhibitor of the PI3K, which was purchased from Nanjing KeyGen Biotech Co., Ltd. (Nanjing, China) (22).

In order to knockdown the expression of endogenous KDM2A, a lentivirus containing an shRNA sequence targeting KDM2A was designed and synthesized by Shanghai GenePharma Co., Ltd. (Shanghai, China). The shRNA sequence was as follow: GGTGGGCAGTAGGAATCAA. The cells were seeded at ~1.0×10⁵ cells/well into 6-well plates and cultured at 37°C overnight under standard conditions. After 50% confluence was reached, the number of cells in a well was counted using a hemocytometer. KDM2A shRNA was transfected into the cells in Opti-MEM (Invitrogen; Thermo Fisher Scientific, Inc.) at a multiplicity of infection (MOI) of 10 [MOI = transducing units per cell (TU) number/cells], according to the manufacturer's instructions. The culture medium was replaced after a 24-h incubation. A total of 48 h after transfection, the cells were observed and photographed under a fluorescence microscope. After successful transfection, the shRNA sequence was stably expressed. Untransfected cells were used as a blank control, while cells transfected with scrambled shRNA were considered as the negative control (shControl).

Cell Counting Kit-8 (CCK-8; purchased from Dojindo Laboratories, Kumamoto, Japan) was used to determine the inhibitory effect of the silencing of KDM2A on the proliferation of A2780 and SKOV3 cell lines according to the manufacturer's instructions. Cells were seeded at ~10.0×10³ cells/well into 96-well plates with 10% FBS and cultured for 24, 48 and 72 h. Then, 10 µl of CCK-8 was added and subsequently incubated for 1 h at 37°C. The absorbance value (OD) at 450 nm of every well was measured with a spectrophotometric plate reader. Assays were performed in triplicate on 3 independent experiments.

An appropriate number of cells were plated on a 6-well plate (500 cells/2 ml/well) and cultured in 2 ml medium with 10% FBS. Then, cells were changed every 3 days with 10% mycoplasma-free FBS for 2 weeks until the cells in the plates had formed colonies that were of substantially right size (50 cells/colony or greater). The cells were fixed with 2 ml 75% ethanol at room temperature for 15 min, and then stained with 0.5% crystal violet. The cloning efficiency was calculated by dividing the number of wells containing proliferating cells with the total number of cell-plated wells.

For apoptosis analysis, cells were treated for 48 h. The cells were suspended with 100 µl of 1X binding buffer (0.1 mM HEPES/NaOH, 1.4 M NaCl, 25 mM CaCl₂ and pH 7.4) and stained with 5 µl of PE Annexin V and 5 µl of 7-amino-actinomycin (7-ADD) for 15 min at room temperature and then 400 µl 1X binding buffer was added to each tube (Annexin V-PE Apoptosis Detection kit; BD Pharmingen; BD Biosciences, San Diego, CA, USA). Analysis of the results was carried out by BD FACSAria (BD Biosciences, Franklin Lakes, NJ, USA). Data were quantified using FlowJo Software (FlowJo, LLC, Ashland, OR, USA).

The 3 groups of A2780 and SKOV3 cells were fixed using methanol acetic acid for 15 min and then stained with Hoechst 33342 for 5 min. Being washed twice with PBS, the cells were immediately photographed under an Olympus BX51 inverted microscope at ×400 magnification (Olympus Corp., Tokyo, Japan).

Untransfected and transfected cells were seeded at 5.0×10⁵ cells/well in 6-well plates and cultured routinely. After reaching 90% confluence, the cell monolayer was scratched with a sterile pipette tip. After washing 3 times with PBS for 5 min each to clear the floating cells, 1.5 ml MEM/F12 and RPMI-1640 medium supplemented with 1% FBS were added into each well. Photographs were taken by an Olympus BX51 inverted microscope at ×100 magnification (Olympus Corp.) at 0, 24 and 48 h after scratching. Results were indicated as the relative width of scratch-the distance migrated relative to the original scratched distance. The experiment was conducted 3 times.

The invasive ability of cells was measured using the Corning^® Matrigel^® Basement Membrane Matrix (cat. no. 356234; Corning Inc., Corning, NY, USA) and a 24-well Transwell chamber (Corning Inc.) according to the manufacturer's protocol. The number of cells that passed through an 8-µm polycarbonate membrane was calculated. The polycarbonate surface of each chamber was covered with 50 µl Matrigel (1:8 dilution) to create an artificial basement membrane. Cells were cultured at 37°C in FBS-free MEM/F12 and RPMI-1640 medium for 24 h. After serum starvation, the cells were seeded at 1×10⁵ cells/well in the upper Transwell chamber, which contained ~200 µl serum-free medium. The lower chamber was filled with 500 µl of the medium supplemented with 10% FBS. After an incubation of 48 h at 37°C, the chambers were fixed at room temperature with paraformaldehyde for 30 min. Cells that had attached to the upper surface of the chambers were removed with a sterile cotton swab, and cells that adhered to the lower surface were stained with 0.1% crystal violet (Guangfu Institute of Superfine Chemical Industry, Tianjin, China) for 20 min at room temperature. The numbers of stained cells were counted using an inverted microscope (Olympus IX 70–142; Olympus Corp.) in 8 random fields. The experiment was repeated 3 times.

Cell lysates containing equal amounts of protein was subjected to 12% SDS-PAGE and then transferred to 0.45 µm Immobilon-P transfer membranes (EMD Millipore, Billerica, MA, USA) by electroblotting. Membranes were blocked at room temperature for 1 h in TBST (50 mmol/l Tris-HCL, pH 7.6, 150 mmol/l NaCl and 0.1% Tween-20) containing 5% non-fat dry milk and then incubated with primary antibody at 4°C overnight. The membranes were then incubated with appropriate anti-rabbit secondary antibody (dilution 1:10,000; Sigma-Aldrich; Merck KGaA) for 1 h at room temperature and visualized with a chemiluminescence substrate kit (Pierce^™ ECL Western Blotting Substrate; Thermo Scientific Fisher, Inc.). The intensity of the target protein bands was normalized to the loading control, GAPDH/β-actin.

Statistical analyses were performed using SPSS v. 17.0 software (SPSS, Inc., Chicago, IL, USA). Quantitative data are expressed as the mean ± standard deviation (SD). The log-rank test, one-way ANOVA and and the Tukey's test were used. The statistically significant difference was indicated in the figures and legends as follows: *P<0.05, **P<0.01, ***P<0.001.

Differential analysis was performed for the GSE14407 dataset. The results showed that expression of KDM2A was significantly upregulated in 12 serous ovarian cancer epithelial samples (SOCE) compared to that noted in the 12 healthy ovarian surface epithelial samples (HOSE) (Fig. 1A and B). Differences in the immunoreactivity of KDM2A staining were observed in EOCs, borderline epithelial ovarian tumors, and normal epithelial ovary tissues, predominantly expressed in the nucleus and light cytoplasmic staining. EOCs showed strong positive staining, while borderline epithelial ovarian tumors and normal ovary tissues showed low-level expression (Fig. 1C). Kaplan-Meier survival analysis suggested that patients with high KDM2A expression had a poor prognosis in overall survival (Fig. 1D). In addition, the expression level of KDM2A was higher in 7 fresh EOC tissues compared with that in 3 fresh normal ovary tissues (Fig. 1E).

To test the function of KDM2A in EOCs, shRNA transfection was used for KDM2A knockdown. The knockdown efficiency was assessed by western blot analysis. The results showed that the expression of KDM2A was significantly decreased after shRNA-KDM2A transfection in human ovarian cancer cell lines SKOV3 and A2780 (Fig. 2A and B).

We found that KDM2A inhibition in A2780 and SKOV3 cells prevented proliferation. We tested our hypothesis to confirm whether cell viability and the colony-forming capabilities were determined by cell proliferation. As shown in Fig. 2C, CCK-8 was used to measure the viability of EOC cells transfected with sh-KDM2A. The viability of the sh-KDM2A cells was significantly lower than that for cells transfected with the sh-Control and the control group. The colony formation assay indicated that the colony-forming capabilities of the SKOV3 and A2780 cells were significantly impaired due to KDM2A depletion (Fig. 2D). To explore whether the effects of KDM2A depletion on cell behaviors were related to cell cycle distribution, we performed flow cytometric analysis. The evidence showed that the cell cycle was related to the proliferative capabilities of cells; therefore, the effect of KDM2A in EOC was further investigated. The percentage of G2/M phase cells was increased after silencing of KDM2A (Fig. 2E and F). These results demonstrated cell cycle arrest in the G2/M transition due to KDM2A knockdown.

Flow cytometric analysis with PE/7-ADD staining showed that the early apoptotic cell population in the sh-KDM2A group was higher than that in the other 2 groups in the 2 cell lines (Fig. 3A and B). Moreover, western blot analysis showed that the expression level of Bcl-2 was downregulated, while the expression level of Bax was upregulated in the same conditions (Fig. 3C). Hoechst 33342 staining showed that the sh-KDM2A group had a greater number of apoptotic cells compared to the number of apoptotic cells in the 2 control groups (Fig. 3D). Additionally, cells of the sh-KDM2A group showed marked morphological changes in the nuclear size than cells of the control groups.

The serum-stimulated Matrigel invasion assay demonstrated that the percentages of human ovarian cancer cells that migrated through the polycarbonate membrane in the sh-KDM2A transfection group were significantly lower than those in the control groups (Fig. 4A and B). Therefore, downregulation of KDM2A decreased cell invasion of the human ovarian cancer cell lines after transfection.

The monolayer wound healing assay was used to assess cell migration. The transfected cells showed a significant deceleration in closure of the wound scratch width after 24 and 48 h compared with the sh-Control and control groups (Fig. 4C and D).

To study the effect of KDM2A depletion on cellular response, the expression of MMP2 and MMP9 were measured in SKOV3 and A2780 cell lines. As shown in Fig. 4E, the 2 transfected proteins, showed decreased expression in the sh-KDM2A group compared to the negative control and the control group.

Since EMT has been increasingly recognized as a crucial event in cancer metastasis, we measured the expression of EMT markers and EMT transcription factors before/after KDM2A knockdown by western blotting. The expression level of E-cadherin was markedly upregulated, while those of N-cadherin and vimentin were significantly decreased in A2780 sh-KDM2A and SKOV3 sh-KDM2A cells compared with the sh-Control and the control cells (Fig. 5A).

To elucidate the molecular mechanism involved in KDM2A-mediated EMT in EOC cells, we analyzed the key components of the PI3K/AKT/mTOR signaling pathway. Western blot analysis showed that KDM2A knockdown decreased the expression of phospho (p)-PI3K, p-AKT and p-mTOR compared to the control groups (Fig. 5B). Using the PI3K signaling pathway inhibitor, LY294002 (0, 5, 10 and 20 µmol/l), we treated the A2780 and SKOV3 cells for 48 h. The results showed that inactivation of the PI3K signaling pathway led to suppression of EMT in a dose-dependent manner, while the expression level of KDM2A was not significantly changed (Fig. 5C). The migration ability of SKOV3 and A2780 cells gradually decreased with 10 µmol/l LY294002 (Fig. 5D). In addition, LY294002 inhibited the colony formation of SKOV3 and A2780 cells. Fewer colonies were formed in the treated group compared with the control group (Fig. 5E).