The differential genes expressed in ovarian cancer and metastatic ovarian cancer were downloaded from GEO database (https://www.ncbi.nlm.nih.gov/geo/) under the accession number GSE38734. Differential expression genes of GSE38734 were analyzed by R software using limma package. We downloaded the raw data of TCGA database from UCSC Xena (https://xena.ucsc.edu/). LinkedOmics is a publicly available platform for analyzing multiomics data from TCGA database (http://www.linkedomics.org/login.php). We searched for interacted proteins of EPYC on website of HitPredict (http://www.hitpredict.org/). Genes that correlated with EPYC in OCs were obtained from this website, then the corresponding genes were subjected to Gene Ontology (GO) functional analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway by R software using clusterProfiler package, as well as the GSEA analysis. Genes correlated with EPYC was analyzed by R software using Pearson correlation analysis.

All ovarian cancer cell lines and normal human ovary cells were purchased from Fuheng Biology and cultured in a humidified atmosphere of 5% CO2 at 37°C. Cell lines were cultured in RPMI 1640 medium. A total of 10% fetal bovine serum (FBS) was supplemented in the culture medium. EPYC-siRNAs, as well as the corresponding negative control, were synthesized from RibBio (Guangzhou, China). siRNA transfection were performed using the matching siRNA transfection kit from RibBio (Guangzhou, China).

TRIzol (Invitrogen, Carlsbad, CA, USA) was used to extract the total RNA from cells. According to the protocol, total RNA was reversely transcribed into cDNA using the two-step reverse transcription reagents (Promega, UK). SYBR Green Master (ROX) (Bimake.com) was used for qRT-PCR according to the protocol. The relative gene expression of EPYC, SNAI2, CDH1 was calculated by the method of 2−ΔΔCt with ACTB gene expression as a control. The primer sequences of EPYC, SNAI2, CDH1 and ACTB were shown below ( Table 1 ).

RIPA lysis buffer (beyotime, catlog# P0013B) was used to extract total protein of cells. BCA assay kit (Thermo Fisher Scientific, catlog#23225) was used for protein quantification. Then, equal amount of proteins were loaded for SDS-PAGE electrophoresis. After transferring to PVDF membranes, the membranes were blocked with 5% skim milk for 1.5 hours. After milk blocking, the membrans were incubated with EPYC antibody or β-actin antibody at 4°C overnight (EPYC antibody: neuromics, catlog# MO15127, β-actin antibody: proteintech, catlog# 60008-1-Ig). After incubating with horseradish peroxidase (HRP)-conjugated secondary antibody (CWBIO, catlog# CW0102) for 2 hours, protein bands were visualized using the enhanced chemiluminescence detection kit (Thermo Fisher Scientific, catlog# 34095) under the Chemiluminescence system.

1*10^5 cells/well were seeded into 24-well plates and incubated for 12 hours to make cells adhere to the wall. Then EPYC-siRNAs and negative control were added into culture medium according to the transfection protocol. After interfering for 24 hours, we used a 200 ul pipette tip to make a “cross-shaped” wounds. After gently washing 3 times with PBS, cells were cultured in RIPM 1640 medium containing 2% FBS for another 24 hours and images were captured under a phase-control microscope.

SKOV3 cells were seeded and transfected as described above. 24 hours later, the transfected cells were resuspended in serum-free RIPM1640 medium and added to the upper chambers of the transwell chamber (Corning, NY, USA). Add RIPM1640 containing 10% FBS to the lower chamber to induce cell invasion. After culturing for 24 hours, the bottom of upper chambers were fixed by 4% paraformaldehyde for 30 minutes and then stained with 0.5% crystal violet for 30 minutes, and the cells on the upper surface of the membrane were removed by a cotton swab. After washing and the chamber is air-dried, the membranes of the bottom of upper chamber were captured under a microscope for analysis.

Cell counting kit-8 (CCK8) (Beyotime, Shanghai, China) was applied in detecting proliferation of SKOV3 cells after transfecting EPYC-siRNAs and negative control for 0, 24, 48, 72 hours. We operated according to the instructions.

Statistical analysis was performed using R software. 2-3 experiments were repeated in our study. Data was presented as mean ± standard deviation (SD), and two-tailed t-test was utilized to compare the difference between two independent samples. The log-rank test was used for Kaplan–Meier survival analysis. P < 0.05 was considered statistically significant.

To identify the differential expressed genes in metastatic OCs, we analyzed the expression profiles of GSE38734. The datasets contained the mRNA sequencing of 4 primary OCs and their metastatic OC tissues. Differentially expressed genes (DEGs) were analyzed by using R software, and found out that EPYC was the most elevated gene in metastatic OC ( Figures 1A–C ).

To explore the role of EPYC on progression of OCs, we compared the expression of EPYC in OCs and normal ovary. Totally, 7 studies comparing the expression of EPYC in OCs and normal ovary tissues were identified in an online website ONCOMINE. EPYC expression was significantly up-regulated in ovarian cancers (p=0.02) ( Figure 2A ). Furthermore, we compared the gene and protein expression of EPYC in normal ovarian cells and several ovarian cancer cells. We found that the endogenous EPYC expression was higher in OC cells than normal ovary cells, IOSE80. The expression of EPYC mRNA was significantly higher in A2780 and SKOV3 than the other OC cells ( Figure 2B ). By using western blot, we detected the protein expression level of EPYC in normal ovarian cell and OC cells. The result showed that EPYC protein level was higher in OC cells than in normal ovarian cells ( Figures 2C, D ).

The human protein atlas (HPA) is a Swedish-based program initiated in 2003 for mapping all the human proteins in cells, tissues and organs (https://www.proteinatlas.org/). From the HPA website, we learned that EPYC protein was barely expressed in normal ovarian tissue (0/3), but could be moderately expressed in a third of OC tissues (4/12, 2 serous cystadenocarcinomas and 2 mucinous cystadenocarcinomas). All three endometroid OCs were negative ( Figure 2E ).

To explore the function of EPYC in ovarian cancer, we interfered the expression of EPYC in SKOV3 cells by EPYC-siRNAs. We verified the interference efficiency by qRT-PCR experiment and western blot ( Figures 3A–C ). Subsequently, we explored the effect of EPYC on the motility of OC cells. In the scratch test, we found out that after EPYC interfering, the migration ability of SKOV3 was obviously impaired ( Figures 3D, E ). In the transwell-matrigel assays, EPYC-siRNA significantly impaired the invasion ability of SKOV3 cells ( Figures 3F, G ).

We also explored the function of EPYC on cell growth of OC cells by CCK-8 assays. EPYC-siRNA suppressed the growth of SKOV3 cells after interfering for 24 hours, but then the effect diminished ( Figure 3H ).

On the website LinkedOmics (24), we analyzed the most related genes of EPYC using 581 ovarian cancer patients data from TCGA database. As shown in the association curve ( Figure 4A ), 4,443 genes (red dots) showed significantly positive correlation with EPYC, whereas 5,121 genes (green dots) showed significantly negative correlation with EPYC(p-value<0.05). The top 50 genes of both were shown in the heatmap ( Figures 4B, C ). Significant GO term analysis using ClusterProfiler package (25) by R software showed that top 200 genes significantly positively correlated with EPYC primarily participated in extracellular matrix (ECM) organization, collagen binding, and ECM constituent conferring tensile strength. They could perform the molecular function of ECM structural constituents, collagen binding, glycosaminoglycan binding and integrin binding ( Figure 4D ). In the KEGG pathway analysis of these genes, we found that they were significantly enriched in the following pathways: PI3K-Akt signaling pathway, focal adhesion, protein digestion and absorption and ECM-receptor interaction ( Figure 4E ).

HitPredict website is a database of quality assessed protein-protein interactions (26). We searched for the proteins interacting with EPYC. The results showed that phospholipase Cg2 (PLCG2) and CRK proto-oncogene, also named adaptor protein (CRK, or p38, or CRKII), were two proteins interacting with EPYC with high interaction score ( Figure 5A ). PLCG2 is a transmembrane signaling enzyme which catalyzes the conversion of PIP2 into second messengers IP3 and DAG by utilizing calcium catalysis. They initiate intracellular calcium flux and activate protein kinase C, respectively. PLCG2 is highly expressed in cells of hematopoietic origin and is responsible for regulating immune responses, platelet adhesion and spreading (26). CRK is a member of adapter protein families, and it has been implicated in many signal transduction events due to its SH2 and SH3 domains (27). SH2 interacts with phosphotyrosine, which results in CRK recruiting cytoplasmic proteins around tyrosine kinase (28). CRK has been shown to play a role in cancer progression and malignant behavior (29). CrkII is an alternatively spliced isoform of CRK, and could interact and be phosphorylated by insulin-like growth factor (IGF) receptor. The phospholated Crk may be related with the interference of IGF-1 regulatory pathway (30). The predicted interaction to these 2 proteins means EPYC maybe play important roles in regulating signal transduction and further experimental verification is needed.

It is a key step in the occurrence of EMT that SNAI2 as a transcription inhibitor directly down-regulates the expression of CDH1, and the TGFβ signaling pathway can regulate the expression of SNAI2 through SMAD pathway and the non-SMAD pathway (31). The PI3K/AKT pathway is one of the non-SMAD signaling pathways for TGF-β to induce epithelial-mesenchymal transition (EMT), which can regulate the metastasis of cancer cells (32).

Using the TCGA public database, we performed Gene Set Enrichment Analysis (GSEA) analysis. Gene enrichment analysis showed that EPYC might activate EMT, TGF-β and PI3K/AKT signals pathways ( Supplementary Figure 1 ). Correlation analysis revealed that EPYC was significantly correlated with the expression of multiple TGF-β signaling pathway molecules (TGFB1, TGFBI, TGFB3, TGFBR1, TGFBR2, TGFBR3) ( Supplementary Figure 2 ). After siRNA-EPYC interference, the invasion and metastasis ability of ovarian cancer cells was weakened, while the expression of SNAI2 was down-regulated and the expression of CDH1 is increased ( Figure 5B ).

Therefore, we speculated that EPYC in ovarian cancer might regulate the expression of SNAI2 by activating the TGF-β/PI3K/AKT signaling pathway, and promoted the occurrence of EMT and invasion and metastasis of ovarian cancer.

Using the TCGA database, we found out that EPYC gene expression was associated with late FIGO staging of ovarian cancer, without correlation with age, pathological grade, or death ( Figure 6A and Table 2 ). The expression of EPYC and overall survival time of OC is not correlated using TCGA database, maybe because limited patients were included. However, when using the online tool Kaplan-Meier Plotter website, which includes 1435 OC patients, we found that EPYC expression was associated with poor prognosis of serous ovarian cancer, with cutoff value 23 used in analysis ( Figure 6B ).