A total of 130 patients who were diagnosed with serous ovarian cancer at the Department of Gynecology and Obstetrics in Tianjin Medical University General Hospital (Tianjin, China) according to a pathological report between January 2009 and December 2016 were selected for the current study. The mean age of the patients was 57 years (range, 35-76 years). This study was approved by the Ethics Committee of Tianjin Medical University General Hospital. All participants signed consent forms prior to the surgical procedure. Pathological specimens were collected during the primary surgery. The ovaries of patients were resected during the surgery, and the adjacent non-tumor tissues were collected 1.0 cm away from the tumor. The specimens were fixed in formalin in room temperature for 24 h and embedded in paraffin. Then, 5-μm continuous sections were prepared for HE staining, in situ hybridization of miR-508-3p and immunohistochemical staining of CCNA2, MMP7 and proliferation marker protein Ki67. Each slide was re-evaluated by an expert pathologist without knowledge of patient clinical data before the experiments were performed. Only specimens containing >70% of tumor tissue were used for the study. The cases were classified according to the FIGO stage system (2013) (14). The clinicopathological data were collected, including age, FIGO stage and grade (Table SI). In this cohort, the median follow-up duration was 36.8 months (range, 1.2-94 months). A total of 78% of patients were alive at the last follow-up. Overall survival (OS) times were calculated from the initial cytoreductive surgery and considered censored for patients who were alive at or died after the last follow-up.

To further evaluate the results of the survival data from Tianjin Medical University General Hospital, the ovarian carcinoma cohort was obtained from The Cancer Genome Atlas (TCGA; https://portal.gdc.cancer.gov/). The association between miR-508-3p expression and survival of 452 patients with ovarian cancer was analyzed using TCGA data.

The ovarian cancer cell lines SKOV3, HeyA8 and A2780, and the cervical cancer cell line HeLa were obtained from the American Type Culture Collection and maintained in RPMI-1640 medium (Gibco; Thermo Fisher Scientific, Inc.) supplemented with 10% FBS (Gibco; Thermo Fisher Scientific, Inc.) without any antibiotics. The cells were incubated at 37°C in an atmosphere containing 5% CO₂. Human miR-508-3p mimic (hsa-miR-508-3p) and scrambled negative control (miR-ctrl) were obtained from GE Healthcare Dharmacon, Inc. Small interfering RNA (siRNA) targeting CCNA2 or MMP7 and scrambled negative siRNA control were purchased from Sigma-Aldrich; Merck KGaA. The pcDNA3.1 plasmids (Promega Corporation) expressing CCNA2 or MMP7 and empty vectors (EV) were constructed by Shanghai GenePharma Co., Ltd.

The ovarian cancer cell lines SKOV3, HeyA8 and A2780 (2×10⁵ per well) were seeded in 6-well plates and allowed to attach for at least 16 h. miR-508-3p mimic (miR-508-3p), miR-ctrl or siRNA were transfected into cells using Lipofectamine^® RNAiMAX (Invitrogen; Thermo Fisher Scientific, Inc.) at a final concentration of 50 nM. Plasmids (500 ng) were transfected into cells using Lipofectamine^® 2000 (Invitrogen; Thermo Fisher Scientific, Inc.). The transfection reagents were used according to the manufacturer's instructions, and the transfections were performed at 37°C for 48 h. Total RNA and protein were collected 48 h post-transfection. Cell supernatants were collected at 24 h for ELISA.

Total RNA were isolated from the ovarian cancer cell lines SKOV3, HeyA8 and A2780 using the mirVana™ miRNA Isolation kit (Ambion; Thermo Fisher Scientific, Inc.) according to the manufacturer's instructions. A total of 50 ng RNA was reverse transcribed to cDNA using a Taqman MicroRNA Reverse Transcription kit (cat. no. 4366596; Applied Biosystems; Thermo Fisher Scientific, Inc.) for the reverse transcription of miRNA and U6 (16°C for 30 min, 42°C for 30 min, 85°C for 5 min and store at 4°C) or Taqman Reverse Transcription reagents (cat. no. 8080234; Applied Biosystems; Thermo Fisher Scientific, Inc.) for the reverse transcription of CCNA2 and MMP7 mRNA (25°C for 10 min, 42°C for 50 min, 70°C for 15 min and store at 4°C). TaqMan RT-PCR assay probes for CCNA2 (cat. nos. A15629 and A15630) and MMP7 (cat. nos. A15629 and A15630) were purchased from Thermo Fisher Scientific, Inc. TaqMan miRNA assay kits were purchased from Applied Biosystems; Thermo Fisher Scientific, Inc (cat. no. 4427975).

The probes of PCR for miR-508-3p (cat. no. AP01571) and U6 (cat. no. AP01501) were purchased from HaiGene Co., Ltd. The qPCR instrument used was Applied Biosystems Real-Time PCR System (BioRad Laboratories, Inc.) with the following thermocycling conditions: 95°C for 10 min, followed by 50 cycles of 95°C for 15 sec and 60°C for 1 min. U6 and GAPDH were used as normalization controls. Data were analyzed by the 2^−ΔΔCq method (15).

The 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphen-yltetrazolium bromide (MTT) substrate (Sigma-Aldrich; Merck KGaA) was used to assay cell proliferation according to the manufacturer's instructions. Briefly, SKOV3, HeyA8 and A2780 cells transfected with miR-508-3p, miR-ctrl, siRNA-ctrl, siRNA-CCNA2 or siRNA-MMP7 were seeded into 96-well plates at a density of 1×10³ cells/well with 200 μl RPMI-1640 medium supplemented with 10% FBS. Then, 5 mg/ml MTT reagent was added to the medium (without FBS) at a ratio of 1:9 and was used to replace the medium in each well (100 μl/well) at different time points (1, 2, 3 and 4 days), followed by incubation at 37°C for 4 h. After removal of the medium, 150 μl dimethylsulfoxide was added to each well, and the cells were gently agitated for 5 min at 37°C. Absorbance values were determined using a microplate reader at 590 nm. Each experiment contained three replicates.

The ovarian cancer cell lines SKOV3, HeyA8 and A2780 were harvested 24 h after transfection with miR-508-3p or miR-ctrl, seeded in a fresh 6-well plate (1,000 cells/well) and incubated for 14 days. Colony formation was analyzed by staining cells with 0.1% crystal violet at room temperature for 30 min. The colonies were observed under an inverted phase contrast microscope (Olympus corporation), and a cluster containing >50 cells was considered a colony. The efficiency of colony formation was calculated as follows: Colony formation rate = (No. of colonies / No. of seeded cells) x 100%.

SKOV3, HeyA8 and A2780 cells were transfected with miR-508-3p, miR-ctrl, siRNA-ctrl or siRNA-CCNA2. At 48 h post-transfection, the cells were harvested by trypsinization, washed with PBS and fixed in ice-cold 70% ethanol at 4°C overnight. The cells were then washed in PBS and incubated with 50 μl 100 g/ml ribo-nuclease (Thermo Fisher Scientific, Inc.) for 20 min at room temperature to ensure that only the DNA was stained. Then, the cells were incubated in 200 μl 50 g/ml propidium iodide for 30 min at 37°C in the dark. Cell cycle distribution was assayed using a FACSCalibur™ Flow Cytometer (BD Biosciences). The data were analyzed with the ModFit LT v3.3 software (BD Biosciences).

For Transwell experiments, 24 h post-transfection, SKOV3 and HeyA8 cells (5×10⁴ cells/well) were seeded into the upper wells of the Transwell chamber in RPMI-1640 medium without serum. The Transwell chamber was pre-coated with Matrigel (BD Biosciences) for 30 min at 37°C for the invasion assay or left uncoated for the migration assay. The cells were allowed to migrate or invade toward RPMI-1640 medium containing 10% FBS for 24 h at 37°C. Then, the cells on the lower surface were fixed with 4% paraformaldehyde at room temperature for 30 min and stained with 0.1% crystal violet at room temperature for 15 min. The cells on the upper surface were removed using a cotton swab. The migrated or invaded cells were counted in six randomly selected fields (×400 magnification) under a positive phase contrast microscope (Olympus Corporation).

SKOV3 or HeyA8 cells were solubilized in RIPA Lysis and Extraction Buffer (Thermo Fisher Scientific, Inc.) supplemented with Halt Protease Inhibitor Cocktail (Thermo Fisher Scientific, Inc.). Protein quantification was performed using BCA Protein Assay kit (Abcam). Equal amounts (40 μg) of protein from each sample was subjected to SDS-PAGE on a 10% SDS-acrylamide gel and loaded onto a PVDF membrane. The membrane was blocked by 5% skimmed milk TBS + Tween-20 (1,000:1) buffer at room temperature for 1 h and probed with the following primary antibodies: CCNA2 (1:200; cat. no. sc-596; Santa Cruz Biotechnology, Inc.), MMP7 (1:200; cat. no. sc-80205; Santa Cruz Biotechnology, Inc.), β-actin (1:1,000; cat. no. sc-8432; Santa Cruz Biotechnology, Inc.), MMP2 (1:1,000; cat. no. ab37150; Abcam), pRb (1:1,000; cat. no. 9307; Cell signaling Technology, Inc.) CDK4 (1:1,000; cat. no. sc-53636; Santa Cruz Biotechnology, Inc.), CDK6 (1:1,000; cat. no. sc-53638; Santa Cruz Biotechnology, Inc.) and Ki67 (1:1,000; cat. no. AF0198; Cell Signaling Technology, Inc.) overnight at 4°C. The membrane was washed three times by TBS + Tween-20 (1000:1) buffer at room temperature for 10 min. The secondary antibodies included horseradish peroxidase (HRP)-conjugated goat anti-rabbit IgG (1:4,000; cat. no. ab205718; Abcam) and HRP-conjugated horse anti-mouse IgG (1:6,000; cat. no. 7076; Cell Signaling Technologies, Inc.). The secondary antibody incubation was performed at room temperature for 1 h. The proteins were visualized using the SuperSignal West Pico Chemiluminescent Substrate (Pierce; Thermo Fisher Scientific, Inc.).

The concentration of MMP7 in the conditioned media from cultured cells was determined by a MMP7 ELISA kit (cat. no. PDMP700; R&D Systems, Inc.) according to the manufacturer's instructions. Briefly, the collected conditioned media from SKOV3 or HeyA8 cells were added to a well coated with an MMP7 polyclonal antibody and immunosorbed by a biotinylated monoclonal anti-human MMP7 antibody at 37°C for 1 h. The color development catalyzed by horseradish peroxidase was terminated with 2.5 mol/l sulfuric acid. The protein concentration was determined by a microplate reader at 450 nm.

The 3′-untranslated regions (UTRs) of the CCNA2 and MMP7 genes, each of which contained one putative miR-508-3p binding site identified by TargetScan (http://www.targetscan.org/vert_71/) and miRBase (http://www.mirbase.org/) databases, were amplified by PCR from cDNA derived from HeyA8 cells and inserted into the multiple cloning sites of the pmirGLO vector (Promega Corporation). The primers used were as follows: CCNA2 forward, 5′-AAGTATATGGTGTACAGTTTTTAACTTAGGTTTTAATTTTTTCTGAATACAGAAGTTGTG-3′ and reverse, 5′-CACAACTTCTGTATTCAGAAAAAATTAAAACCTAAGTTAAAAACTGTACACCATATACTT-3′; MMP7 forward, 5′-CGGCTAGCTCAGGCAGAACATCCATT CATTC-3′ and reverse, 5′-GCTCTAGACATTTATTGACATCTACCCACTGCA-3′. Two mutant (mt) CCNA2 and MMP7 3′-UTR reporter vectors that lacked the binding sites for miR-508-3p were created through site-directed mutagenesis using a QuikChange kit (Stratagene; Agilent Technologies, Inc.). The primers used for site-directed mutagenesis were as follows: CCNA2 3′-UTR-mt forward, 5′-TACTTGTCAATATTTAAACATGGTTTGCTGAAAATGGTATTTTCCCCC-3′ and reverse, 5′-GGGGAAAATACCATTTTCAGCAAACCATGTTTAAATATTGACAAGTA-3′; MMP7 3′-UTR-mt forward, 5′-GGATTGTATATCATTGTTGCGAATTGATAAGCACTGTTCC-3′ and reverse, 5′-GGAACAGTGCTTAT CAATTCGCAACAATGATATACAATCC-3′. All constructed clones were verified by DNA sequencing performed by Beijing Genomics Institute (Shenzhen, China).

For the luciferase reporter assay, 0.5 μg pmirGLO, pmirGLO-3′-UTR-wild-type or pmirGLO-3′-UTR-M was transfected into HeLa cells that were cultured in 24-well plates together with 50 nM miR-508-3p or miR-ctrl using Lipofectamine^® 2000 (Invitrogen; Thermo Fisher Scientific, Inc.). At 48 h post-transfection, the cells were subjected to lysis, and firefly and Renilla luciferase activities were determined using a Dual-Luciferase Reporter Assay system (Promega Corporation) as previously described (16). Firefly luciferase activity was normalized to Renilla luciferase activity for each construct and compared with that of the miR-ctrl group.

miRNA ISH was performed as previously described (15). Briefly, the patient tissue sections were deparaffinized and incubated for 30 min at 37°C with 20 μg/ml proteinase K. Then, the slides were dehydrated, followed by hybridization with 100 nM DIG-labeled miRCURY LNA™ detection probe hsa-miR-508-3p (cat. no. 611776-350; Qiagen GmbH) for 2 h at 55°C. Next, anti-DIG reagent (sheep anti-DIG-AP at 1:500 in antibody diluent; cat. no. 11093274910; Roche Applied Science) was added, and the sections were incubated for 60 min at room temperature. Then, freshly prepared alkaline phosphatase (AP) substrate [nitro blue tetrazolium chloride (NBT)/5-bromo-4-chloro-3-indolyl-phosphate, toluidine salt (BCIP); cat. no. 11697471001; Roche Applied Science) was applied to the sections, and the sections were incubated for 2 h at 30°C away from light. The U6 probe was used as a positive control for each section. The stained sections were reviewed by two independent pathologists under an inverted phase contrast microscope (Olympus Corporation) at ×200 magnification. Signals in tumor cells were visually quantified using a scoring system between 0 and 9; the scores were assigned based on the intensity of the signal and the percentage of positive cells as follows: i) Signal: 0, no signal; 1, weak signal; 2, intermediate signal; and 3, strong signal; and ii) percentage: 0, 0%; 1, <25%; 2, 25-50%; and 3, >50%. The scores were multiplied to achieve a final score. The low and high miR-508-3p expression groups were based on the mean of miR-508-3p ISH score and, the low and high CCNA2 or MMP7 expression groups were based on the mean of the IHC scores.

IHC staining was performed on the tumor tissues from 130 patients with serous ovarian cancer. The sections were immersed in eBioscience™ IHC Antigen Retrieval solution (cat. no. 00-4956-58; Thermo Fisher Scientific, Inc.) and heated in a microwave for 15 min. Subsequently, the sections were blocked with goat serum (cat. no. 16210064; Thermo Fisher Scientific, Inc.) at room temperature for 1 h and incubated with rabbit anti-human Ki-67 (1:200; cat. no. AP101M; EMD Millipore), mouse anti-human CCNA2 (1:200; cat. no. sc-596; Santa Cruz Biotechnology, Inc.) and mouse anti-human MMP7 (1:100, cat. no. sc-80205; Santa Cruz Biotechnology, Inc.) primary antibodies overnight at 4°C. The secondary antibody (anti-rabbit IgG; cat. no. SPN-9001; OriGene Technologies, Inc.) was then added at room temperature for 1 h. Finally, the sections were incubated with diaminobenzidine (30 mg dissolved in 100 ml Tris buffer) for 5 min, rinsed with running water and counterstained with hematoxylin for 1 min at room temperature.

Ki-67-positive cells were defined as those with brown staining in the nucleus, whereas MMP7-positive cells were defined as those with immunoreactivity in the cytoplasm. CCNA2-positive cells were considered as those with immunoreactivity in the nucleus and cytoplasm. The expression of these proteins was evaluated using the percentage of positive tumor cells in 1,000 tumor cells. The results were analyzed by two independent pathologists, and scores were assigned based on the intensity and the percentage of the stained tissues described above.

The results are presented as the mean ± standard deviation. The cell culture experiments were performed at least three times and in triplicate. Wilcoxon test was performed using GraphPad Prism version 6 software (GraphPad Software, Inc.). Pearson's χ² test, Kruskal-Wallis test with Dunn's post hoc test, Kaplan-Meier analysis, log-rank test and Cox's proportional hazards regression model were performed using SPSS 21.0 software (IBM Corp.). P<0.05 was considered to indicate a statistically significant difference.

miR-508-3p mimics were transfected into HeyA8, SKOV3 and A2780 cells. The results of the RT-qPCR analysis demonstrated increased expression of miR-508-3 compared with the miR-ctrl group in all the three cell lines (P<0.01; Fig. S1), and the MTT assay revealed that miR-508-3p overexpression resulted in the suppression of cell proliferation at all time points compared with the miR-ctrl group (Fig. 1A). At 96 h post-transfection, miR-508-3p inhibited the viability of HeyA8 cells by 35.3% (P<0.05), SKOV3 cells by 34.8% (P<0.05) and A2780 cells by 33.4% (P<0.05) compared with miR-ctrl (Fig. 1A). In addition, miR-508-3p impaired the colony formation efficiency of ovarian cancer cells; cells transfected with miR-508-3p displayed lower colony formation capacity compared with those transfected with miR-ctrl (P<0.05), further confirming the long-term anti-proliferative effect of miR-508-3p (Fig. 1B).

When miR-508-3p was transfected into HeyA8 and SKOV3 cells, the proportion of cells in the S phase decreased compared with the corresponding miR-ctrl-transfected cells (P<0.05), and concomitantly, the proportion of cells in G1 phase significantly increased (P<0.05; Fig. 1C). To examine whether miR-508-3p was involved in apoptosis, Annexin V apoptosis assay was performed after transfection of miR-508-3p or miR-ctrl. SKOV3 and HeyA8 cells exhibited a mild non-significant increase in apoptotic rates at 48 h post-transfection with miR-508-3p compared with those in the miR-ctrl groups (data not shown). These results suggested that the suppressive effect of miR-508-3p on ovarian cancer cell proliferation was mediated by cell cycle arrest.

To examine the suppressive effects of miR-508-p on ovarian cancer cell invasion and migration, transient mimic transfections were performed, and the results demonstrated that miR-508-3p suppressed the migration and invasion of SKOV3 and HeyA8 cells (P<0.05; Fig. 1D). These results provided evidence that miR-508-3p may be an important inhibitor of the viability, migration and invasion of ovarian cancer cells.

To identify genes potentially regulated by miR-508-3p, the TargetScan and miRBase databases were searched, and CCNA2 and MMP7 were identified as potential target genes of miR-508-3p. TargetScan analysis predicted one miR-508-3p binding site in the 3′-UTR of the CCNA2 gene and one in the 3′-UTR of the MMP7 gene (Fig. 2E), suggesting that miR-508-3p may directly target CCNA2 and MMP7. To confirm this, SKOV3 and HeyA8 cells were transfected with miR-508-3p or miR-ctrl followed by measurement of CCNA2 and MMP7 mRNA and protein levels 48 h post-transfection. The results demonstrated that CCNA2 and MMP7 mRNA (Fig. 2A and C) and protein (Fig. 2B and D) levels were significantly downregulated following miR-508-3p transfection compared with those in the miR-ctrl groups in SKOV3 and HeyA8 cells.

Luciferase reporter assays were performed to examine whether miR-508-3p directly targeted CCNA2 and MMP7. The 3′-UTRs of CCNA2 or MMP7 were cloned into the pmirGLO vector to generate pmirGLO-CCNA2 and pmirGLO-MMP7 constructs. Co-transfection of pmirGLO-CCNA2 with miR-508-3p into HeLa cells resulted in 65.7% lower level of luciferase activity compared with that following co-transfection with miR-ctrl, suggesting that miR-508-3p directly targeted CCNA2 (Fig. 2F). Similarly, co-transfection of pmirGLO-MMP7 with miR-508-3p resulted in 37.3% less luciferase activity compared with co-transfection with miR-ctrl, suggesting that miR-508-3p directly targeted MMP7 (Fig. 2F).

To confirm that miR-508-3p specifically regulated CCNA2 and MMP7 through the predicted binding sites, the constructs pmirGLO-CCNA2-mt and pmirGLO-MMP7-mt, in which the miR-508-3p binding site sequences on the 3′-UTRs of the two genes were deleted, were generated and co-transfected with miR-508-3p or miR-ctrl into HeLa cells. Deletion of the miR-508-3p binding sites from the 3′-UTR of CCNA2 or MMP7 abolished the effects of miR-508-3p on the luciferase activity (Fig. 2F). These results indicated that the CCNA2 and MMP7 genes were direct targets of miR-508-3p.

The possible mechanism of the suppressive effects of miR-508-3p on cancer cell proliferation may be induced by CCNA2/pRb/CDK4/6 (17); thus, the expression levels of pRb, CDK4 and CDK6 were determined in the present study. At 48 h post-transfection of miR-508-3p into SKOV3 and HeyA8 cells, the protein levels of pRb and CDK6 were decreased compared with those in cells transfected with miR-ctrl, but CDK4 levels did not change (Fig. 3A). These results indicated that miR-508-3p may suppress the proliferation of ovarian cancer cells by decreasing CCNA2, pRb and CDK6 levels.

Cell mobility may be induced by MMPs, which are known as direct regulators of tumor metastasis (18). In addition to MMP7, miR-508-3p also downregulated MMP2 protein levels (Fig. 3A). MMP7 and MMP2 levels in the culture supernatant were also examined, and the results demonstrated that the level of MMP7 was decreased after transfection of miR-508-3p compared with that in cells transfected with miR-ctrl (Fig. 3B), indicating that miR-508-3p suppressed ovarian cancer cell migration and invasion by decreasing MMP7 levels.

To investigate whether the proliferation suppressive effect of miR-508-3p on ovarian cancer cells was mediated by repression of CCNA2, SKOV3 cells were transfected with siRNA-ctrl or siRNA-CCNA2. At 48 h post-transfection, knockdown of CCNA2 led to decreased expression of pRb, mimicking the suppressive effect of miR-508-3p (Fig. 3C). At 4 days post-transfection, knockdown of CCNA2 led to a significant suppression of cell proliferation (P<0.05; Fig. 3D). Cell proliferation was also suppressed at 2 and 3 days post-transfection, but no significant differences were observed between siRNA-CCNA2 and siRNA-ctrl when analyzed by the Kruskal-Wallis test (P>0.05; Fig. 3D). Knockdown of CCNA2 also led to cell cycle arrest and increased G1 phase distribution, which was similar to the results of miR-508-3p overexpression; however, no statistical differences of G1 phase distribution were observed between siRNA-CCNA2 and siRNA-ctrl when analyzed by the Kruskal-Wallis test (P>0.05; Fig. 3E). These findings indicated that suppression of CCNA2 may be required for proliferation inhibition and cell cycle arrest.

Based on the above results, SKOV3 cells were co-transfected with miR-508-3p mimics and a plasmid overexpressing CCNA2 to further study whether ectopic expression of CCNA2 may rescue the inhibitory effects of miR-508-3p. As presented in Fig. 3F, overexpression of CCNA2 partially rescued the decreased expression of pRb. At 48 h post-transfection, co-transfection of miR-508-3p and an empty vector led to significant suppression of cell proliferation compared with that of miR-ctrl and empty vector, which was attenuated by CCNA2 plasmid (P<0.05; Fig. 3G). Thus, ectopic expression of CCNA2 partially rescued cell proliferation suppression mediated by miR-508-3p. In addition, co-transfection of miR-508-3p and an empty vector induced G1 phase arrest compared with that of miR-Ctrl and empty vector, but the difference was not statistically significant (P=0.05; Fig. 3H); this was also attenuated by CCNA2 plasmid. Taken together, these findings indicated that miR-508-3p may induce the proliferation suppression and cell cycle arrest of ovarian cancer cells by targeting CCNA2.

To elucidate the possible mechanism by which miR-508-3p suppressed ovarian cancer cell migration and invasion, SKOV3 cells were transfected with siRNA-ctrl or siRNA-MMP7. At 48 h post-transfection with siRNA-MMP7, the MMP7 levels in the whole cell lysate and cultured supernatant were decreased compared with those in the siRNA-ctrl transfected cells (P<0.05; Fig. 4A and B). In addition, SKOV3 and HeyA8 cells transfected with siRNA-MMP7_1 or siRNA-MMP7_2 exhibited significantly decreased migratory capacity compared with those transfected with siRNA-ctrl, which was in accordance with the effect of miR-508-3p (P<0.05; Fig. 4C and F).

To study whether ectopic expression of MMP7 may rescue the inhibition of migration and invasion, cells were co-transfected SKOV3 with miR-508-3p mimics and the plasmid expressing MMP7. As demonstrated in Fig. 4D and E, the suppressive effects of miR-508-3p were partially reversed by ectopic expression of MMP7. At 48 h post-transfection, miR-508-3p and empty vector led to a significant suppression of migration (P<0.05; Fig. 4E and G) and invasion (P<0.05; Fig. 4E and G) compared with that in the miR-Ctrl and empty vector group, which was attenuated by the MMP7 plasmid. These results demonstrated that the effect of the miR-508-3p mimics was consistent with that of siRNA-MMP7 on SKOV3 cells, suggesting that miR-508-3p may inhibit the migration and invasion of ovarian cancer cells by directly targeting MMP7.

Considering that miR-508-3p expression impaired tumor aggressiveness in the cellular models, its expression was evaluated in a cohort of 130 clinically annotated serous tumor samples. Expression of miR-508-3p and the control U6 small RNA was evaluated by ISH, whereas CCNA2 and MMP7 were evaluated by IHC in the same tissue samples.

To compare the levels of miR-508-3p, as well as CCNA2 and MMP7 in ovarian cancer with normal ovarian tissue, their expression was also evaluated in adjacent non-tumor tissue. The results demonstrated that miR-508-3p expression in ovarian cancer was significantly lower compared with that in adjacent non-tumor tissue (P<0.001; Fig. 5A). By contrast, CCNA2 expression in cancer tissue was significantly higher compared with that in adjacent non-tumor tissue (P<0.001; Fig. 5B). The results of MMP7 analysis were similar to those of CCNA2 (P<0.05; Fig. 5C).

A significant decrease in miR-508-3p expression was observed with disease progression from stages I-II to III-IV (P=0.028; Table SI). No significant differences in clinicopathologic characteristics were observed between the CCNA2 low and high groups (Table SII) or between the MMP7 low and high groups (Table SIII). In addition, both CCNA2 and MMP7 IHC scores in the tissues from the miR-508-3p low group were significantly higher compared with those from the miR-508-3p high group (P<0.001; Fig. 5D and 5E). The Ki67 expression score in the miR-508-3p low group was also significantly higher compared with that in the miR-508-3p high group (P<0.001). Univariate log-rank analysis demonstrated that older age, advanced FIGO stage, residual disease >1 cm, high CCNA2 and MMP7 and low miR-508-3p expression were significantly associated with a shorter OS time (Table I). A multivariate model revealed that age, residual disease size, miR-508-3p and CCNA2 were independent predictors for OS in ovarian cancer (Table II). Kaplan-Meier analysis demonstrated that patients with high expression levels of miR-508-3p exhibited longer OS times compared with those with low levels of miR-508-3p (median OS 43.91 vs. 27.90 months, respectively; log-rank P=0.006; Fig. 6A). Accordingly, high levels of CCNA2 or MMP7 predicted a shorter OS time compared with low expression levels in a univariate analysis of patients with ovarian cancer (median OS 45.36 vs. 22.61 months, respectively; log-rank P=0.0158 for CCNA2; median OS 45.15 vs. 28.16 months, respectively; log-rank P=0.0332 for MMP7; Fig. 6B and C).

To evaluate these results, ovarian carcinoma data from TCGA were analyzed. Among 452 patients with ovarian cancer, Kaplan-Meier analysis demonstrated that compared with patients with low miR-508-3p expression, those with high expression levels of miR-508-3p did not exhibit a longer OS time (log-rank P=0.07; Fig. 6D). The same conclusion was reached for the 423 patients with stage III-IV ovarian cancer (log-rank P=0.46; Fig. 6E).