MIR502 Acts As a Tumor Suppressor Gene to Accelerate Apoptosis and Suppress Proliferation by Targeting Hippo Signaling Pathway in Ovarian Cancer

Background Ovarian cancer (OC) is the major cause of death among women due to the lack of early screening methods and complex pathological progression. Increasing evidence indicated that microRNAs were considered as gene expression regulators in tumors by interacting with mRNAs. Although researches regarding with OC and microRNAs are extensive, the vital role of MIR502 in OC still remains unclear. Methods We integrated two microRNA expression arrays from GEO to identify differentially expressed genes. Kaplan–Meier method was used to screen miRNAs that had influence on survival outcome. Upstream regulator of MIR502 was predicted by JASPAR and verified by ChIP-seq data. The LinkedOmics database was used to study correlated genes with MIR502 . Gene Set Enrichment Analysis (GSEA) was conducted to reveal the functional annotation of GO and KEGG pathway enrichment analysis by using the open access WebGestalt tool. We constructed PPI network by using the STRING to further explore core protein. Results We found that expression level of MIR502 was significantly down regulated in OC, which was related to poor overall survival outcome. NRF1 as the upstream regulator of MIR502 was predicted by JASPAR and verified by ChIP-seq data. In addition, anti-apoptosis and pro-proliferation genes attending in Hippo signaling pathway, including CCND1, MYC, FGF1 and GLI2 were negatively regulated by MIR502 in the analysis of GO and KEGG pathway enrichment results. PPI network further demonstrated that CCND1 and MYCN were at core position in the development of ovarian cancer. Conclusions MIR502 , which was regulated by NRF1, acted as a tumor suppressor gene to accelerate apoptosis and suppress proliferation by targeting Hippo signaling pathway in ovarian cancer. ,which


Introduction
Ovarian cancer (OC) is a common gynecological malignancy with the highest mortality rate [1].Primary cytoreductive surgery following chemotherapy is the conventional therapy of OC .Tumor often occurs proliferation, invasion, and lymph node metastasis at the time of diagnosis due to the lack of typical symptoms at the early stage ,which leads to delay in initiating appropriate treatment and poor outcome in the meanwhile [2]. The pathogenesis of OC is complicated because it is regulated by a variety of oncogenes and tumor suppressor genes [3]. Currently, multiple ovarian cancer oncogenes have been elucidated, whereas relatively few studies have focused on antioncogenes, and the molecular mechanisms regulating the progression of OC still remains largely unclear. Therefore, it is of great significance to explore new molecular markers for regulating proliferation and apoptosis of ovarian cancer cells, and provide potential signaling pathway for clinical treatment.
microRNAs (miRNAs) are small RNA molecules with length of approximately 20 nucleotides, whose function is negatively regulating gene expression through binding target mRNAs on 3′-untranslated regions (3'-UTRs) in post-transcriptional level [4,5].A substantial amount of researches have confirmed that multiple miRNAs played pivotal role in the process of tumor development such as apoptosis, proliferation, invasion, migration and recurrence by overexpression or low expression [6,7]. In particular, various miRNAs have been shown to serve different roles in ovarian cancer. However, the regulatory mechanism and target genes of miRNAs are still in their infancy, and the relationship between miRNAs and tumor, especially ovarian cancer, is not fully understood. Nowadays, the effect of MIR502 in cancer has been researched widely. The results from our study indicated that MIR502 produced a marked effect on suppressing ovarian cancer proliferation.
Nuclear respiratory factor 1(NRF-1)is an important transcription factor in human genome. It was estimated that 6% of human promoter region genes contained NRF-1 response element in a systematic bioinformatics study [8]. NRF-1,also known as a-pal, was originally identified as a mitochondrial gene involved in the regulation of energy conduction [9].NRF1 encodes a protein, which has the function of homologous dimer and transcription factor, activates the expression of some key metabolic genes regulating cell growth [10].
Hippo signaling pathway is known as a crucial role in regulating cell proliferation, regeneration and controlling organ growth [11]. This pathway is composed by a large number of proteins. It has the function of controlling cell fate not only in the process of development and differentiation, but also in the pathological process including cancer [12]. The main Hippo core kinase cascades are Yes-associated transcriptional regulator (YAP) and transcriptional coactivator with the PDZ-binding motif (TAZ). Observations have demonstrated that there was a strong relationship between Yap activation and cancer. In many tumors, including brain ,lung, breast, pancreatic, liver, colon, skin and ovarian cancer, YAP and TAZ promote cell proliferation and anti-apoptosis in cooperation with transcription factors via translocating into the nucleus to regulate many well-known oncogenes [13][14][15][16][17]. A study about podocytes found that YAP overexpression lead CCND1 significantly up-regulated, which confirmed CCND1 as a downstream target gene of YAP [18]. Previous research of gastric tumor has identified MYC as a key downstream molecular target of YAP. The positive correlation between MYC and YAP in human gastric cancers also supported regulation of MYC by YAP, which was an important molecular mediator of gastric tumorigenesis [19]. Shan Xu has verified that YAP promoted VEGFA by targeting GLI2 in renal cancer [20].Some studies have shown that FGF promoted Hippo/Yap signal transduction in the proliferation and differentiation process of lens epithelial cell, and FGF induced nuclear Yap expression played an important role in promoting lens epithelial cell proliferation [21]. Accordingly, it has been reported that YAP, acting as an oncogene, was associated with poor prognosis in ovarian cancer [22][23][24].
The big data generated by high-throughput research is generally characterized by large amount, a wide range of data types, deep value mining and fast processing response. The big data provides opportunities for the discovery of tumor molecular targets, but also brings great challenges to the full mining, integration and utilization of it. It's certainly needed to investigate the complex genetic mechanisms by applying appropriate statistical method [25]. In this study, we found that the expression level of MIR502 is significantly down regulated in ovarian cancer by using bioinformatics analysis of two public databases. We also analyzed the association of MIR502 expression with clinical overall survival (OS) outcome, and correlated pathways were explored to improve prognostic and therapeutic value in preventing ovarian cancer progression. Gene Ontology (GO) and Kyoto Encyclopedia of Gene and Genomes (KEGG) pathway analysis showed that Hippo signaling pathway was correlated with MIR502. And transcription factor NRF1 was predicted as the upstream regulator of MIR502.The authors believe that it may provide more effective and scientific guidance to clinicians for the early diagnosis of patients with ovarian cancer, along with the individual treatment and improve prognosis to the patients and achieve benefit ultimately.

Access of public database
The microRNA expression datasets used in this study (GEO: GSE83693 and GSE119055) were acquired from the National Center for Biotechnology Information (NCBI) Gene expression (http://www.ncbi.nlm.nih.gov/geo/) .

Analysis Of Public Database
GEO2R (http://www.ncbi.nlm.nih.gov/geo/geo2r/) is an analysis tool which is used to compare two sets of data coming from the GEO database. We used GEO2R to screen miRNAs of differential expression between normal tissue and ovarian cancer tissue in the GSE83693 and GSE119055 dataset. We selected genes whose |log2FC (fold change)| > 2 and adjust p-value < 0.05 as differential expression genes.

Survival Analysis
According to the lower quartile expression level of MIR502, OC patients were divided into high expression group and low expression group. The overall survival was analyzed by using Kaplan-Meier plotter (http://kmplot.com/analysis/index.php?p=background). The hazard ratio with 95% confidence intervals and log rank P value were calculated and displayed.

Gene Correlation Expression Analysis
The LinkedOmics database (http://www.linkedomics.org/admin.php) contains 32 TCGA cancer-associated multidimensional datasets, including ovarian cancer. The website was used to study correlation between MIR502 and interested genes expression in the TCGA OC cohort. Results were analyzed statistically using Pearson's correlation coefficient.

Prediction And Verify Of Transcription Factors
We developed JASPAR (http://jaspar.genereg.net) to predict the transcription factors of CLCN5, and Cistrome Data Browser (http://cistrome.org/db/), which provided resource of human cis-regulatory information obtained from the chromatin analysis from ChIP-seq, DNase-seq and ATAC-seq was used to verify the prediction results.

Acquire overexpression genes of ovarian cancer from the Cancer genome atlas database
The Gene Expression Profiling Interactive Analysis (GEPIA) website (http://gepia.cancer-pku.cn) can provide varied functions based on TCGA data, including gene expression, gene correlation analysis, survival analysis and so on. GEPIA was used to find overexpression genes in ovarian cancer. P < 0.05 was considered statistically significant.

Go And Kegg Pathway Analysis
Gene Set Enrichment Analysis (GSEA) was conducted to reveal the functional annotation of GO and KEGG pathway enrichment analysis by using the open access WebGestalt tool (http://www.webgestalt.org). GO analysis included biological process (BP), cellular component (CC) and molecular functions (MF). Results with the false discovery rate (FDR) ≤ 0. 05 were considered noteworthy.
Target genes prediction of MIR502 miRWalk(http://zmf.umm.uni-heidelberg.de/apps/zmf/mirwalk/micrornapredictedtarget.html) was applied to forecast the genes targeted by MIR502. In total, five servers with DIANA-mT, miRanda, miRWalk, PICTAR5 and Targetscan were used. Only those genes projected by all of five servers were selected as target genes.

Protein-protein Interaction Network Construction
Protein-protein interaction (PPI) network was constructed based on the overlaps genes that appeared both in predicted genes in miRWalk and overexpression genes in GEPIA by using the Search Tool for the Retrieval of Interacting Genes (STRING, version 11.0, https://string-db.org/) database.

Statistical Analysis
Statistical analysis was performed by using Prism software (GraphPad, CA, USA). Statistical significance of differences between and among groups was assessed using the t-test. Significant differences are indicated as follows: * P < 0.05; * * P < 0.01; * * * P < 0.001.

The expression level of MIR502 was lower in ovarian cancer tissue comparing with normal ovary tissue
In order to explore the difference of microRNAs expression in human ovarian cancer tissue, we obtained two microarray gene profiling datasets (GSE83693 and GSE119055) from public GEO datasets of NCBI. The detailed information about two datasets were showed in Table 1. After analyzing the expression of microRNAs, we screened out 39 and 25 differentially expressed genes (DEGs) from GSE83693 and GSE119055 datasets respectively, which were showed in volcano plots (Fig. 1A,1B). Seven common DEGs were screened out with Bioinformatics and Evolutionary Genimocs (http://bioinformatics.psb.ugent.be/webtools/Venn/) (Fig. 1C) and listed on Fig. 1D.

Expression of MIR502 affected the overall survival in OC patients
After analyzing overall survival in OC patients by using Kaplan-Meier plotter, we found that only the different expression of MIR502 (p < 0.01) and MIR532(p = 0.013) affected the overall survival outcome among seven microRNAs ( Fig. 2A-2G). The researches about the interaction between MIR532 and ovarian cancer have made headway, but relatively few studies demonstrated the mechanism of MIR502 in OC, so our object of the study was MIR502.For obvious view, we presented box-plot to show the expression of MIR502 in each database (Fig. 2H,2I).

Correlated genes with MIR502 in ovarian cancer
The volcano plot showed the positive and negative correlation genes with MIR502 (Fig. 3A). The top 50 significant gene sets with positive and negative correlation to MIR502 were shown in the heat map (Fig. 3B,3C). This result demonstrated a widespread influence of MIR502 on the transcriptome.

MIR502 was closely related to CLCN5
CLCN5 showed the strongest positive correlation with MIR502 in Fig. 3B and Fig. 4B (Pearson-correlation = 0.6512, P < 0.01). For further exploration, we found MIR502 hosting in the third intron of the CLCN5 gene by searching NCBI (Fig. 4A). The expression level of CLCN5 in OC was lower than that in normal ovary tissue (Fig. 4C), what was consistent with MIR502. JASPAR (http://jaspar.genereg.net/) database were used to analyze and predict the transcription factors that potentially regulated the expression of CLCN5. By matching the 2000 bp region of nucleotide sequence upstream of the promoter of CLCN5 gene, we found transcription factor NRF1 was the highest matched ( Fig. 4D).

Nrf1 Acted As Transcription Factor Of Clcn5
Match score and binding site of NRF1 were showed in Fig. 5A. Expression of NRF1 is positively correlated with CLCN5(Pearson-correlation = 0.33, P < 0.01) (Fig. 5B). We used Cistrome Data Browser (http://cistrome.org/db) database to analyze the ChIP-seq data of tumor cells, we found that NRF1 emerged DNA binding peak in the promoter region of CLCN5 (Fig. 5C). It is further confirmed that NRF1 bound and regulated CLCN5 expression as a transcription factor.

GO and KEGG pathway analysis of correlated genes with MIR502 in ovarian cancer
GO term analysis was given a broad overview by using Go Slim (Fig. 6). Results indicated that these genes could be categorized into several important biological processes, including biological regulation, metabolic process, membrane, nucleus, protein binding and ion binding. Significant GO terms were performed more detailedly by GSEA, showing that genes correlated with MIR502 were located mainly in protein localization to endoplasmic reticulum (GO:0070972) and translational initiation (GO:0006413)for BP, ribosome(GO:0005840) and tertiary granule(GO:0070820) for CC, and structural constituent of ribosome(GO:0003735) and pattern recognition receptor activity(GO:0038187)for MF ( Table 2). The KEGG pathway analysis showed that correlated genes enriched in various pathways (Figuer7A), including ribosome, allograft rejection pathways, systemic lupus erythematosus and so on. It should be noted that Hippo signaling pathway also appeared in the enrichment results. The detailed signaling pathway diagram is showed in Fig. 7B. The correlated genes with MIR502 is marked red. The significant enrichment results were showed in Fig. 8. Our study showed that six popular genes with the function of anti-apoptosis and pro-proliferation participated in the Hippo signaling pathway, including CCND1, FGF1, MYC, GLI2, AFP and AXIN2 (Fig. 7B). The LinkedOmics database was used to confirm the correlation between six genes and MIR502.The results indicated that MIR502 regulated CCND1(Pearson-correlation=-0.2092,p < 0.01), FGF1(Pearson-correlation=-0.1955, p < 0.01), MYC(Pearson-correlation=-0.1448, p < 0.05) and GLI2(Pearson-correlation=-0.1395, p < 0.05) negatively (Fig. 9).

CCND1 and MYCN were at core position in PPI network
860 common genes were selected as predicted target genes of MIR502 (Fig. 10A). A total of 44 genes were selected in the overlaps area of 860 predicted target genes and 1501 overexpression genes in GEPIA of ovarian cancer (Fig. 10B). PPI network revealed CCND1 and MYCN at core position (Fig. 10C). NRAS, PMAIP1 and MYBL2 showed interaction relationship with both CCND1 and MYCN.

Discussion
The aim of our study was to distinguish miRNAs that were obviously different expression in OC comparing with normal tissue, and to improve ovarian cancer patients' clinical survival outcome by exploring the mechanism attending in particular pathway. We selected MIR502 as our interested miRNA after screening miRNAs through strict selection process. Our survival analysis showed that MIR502 conferred a protective phenotype to OC patients in which higher expression of MIR502 predicted longer overall survival. MIR502 located in the third intron of the CLCN5 gene, and it showed strong positive correlation with CLCN5 in ovarian cancer, we predicted NRF1 as a transcription factor of CLCN5, and ChIP-seq data of various tumor cells verified the binding peak between NRF1 and CLCN5. We demonstrated NRF1 as transcription factor of CLCN5 regulated the expression of MIR502 indirectly, which expounded the upstream regulatory mechanism of MIR502 deeply.
In order to explore downstream regulatory mechanism of MIR502 in the process of ovarian cancer, we further predicted and analyzed genes correlated with MIR502. We identified a set of biological functions and related signaling pathways which MIR502 might regulate in ovarian cancer. Furthermore, GSEA annotation analysis results showed that MIR502 regulated anti-apoptosis and pro-proliferation genes such as CCND1, FGF1, MYC, GLI2 negatively in Hippo signaling pathway. All the results demonstrated strongly that the expression of MIR502 was down regulated in OC, which increased the expression level of oncogene CCND1, FGF1, MYC and GLI2, they served an important function of anti-apoptosis and promote development of OC. PPI network also suggested that CCND1 and MYCN were both target genes regulated by MIR502, and they were at the center position interaction with other proteins.
CCND1,also known as cyclin D1, is a member of cell cycle family protein [26]. CCND1 regulates cell cycle progression by promoting the cell cycle transition from the G1 to S phase [27][28][29]. The abnormal expression of CCND1 promotes cell proliferation by regulating cell cycle. [30]. Previous researches have demonstrated that CCND1, identified as a proto-oncogene, served as an essential role in the development of many kinds of tumors, including lung adenocarcinoma, glioma and renal cell cancer [31][32][33]. In addition, some studies have shown that overexpression of CCND1 promoted tumor cell invasion and metastasis in breast cancer gastric cancer, leading to a poor prognosis [34,35]. Compared with normal tissues, the expression of CCND1 were obviously higher in bladder cancer, reproductive system tumors, gastric cancer and lung cancer, which were correlated with the pathological type and clinical stage of the tumor [36][37][38] .CCND1 expression was closely related with cell proliferation ability and apoptosis in epithelial ovarian cancer cells. A study of epithelial ovarian cancer observed that overexpression of CCND1 leads to stronger cell growth ability and fewer apoptosis [39]. In our study, MIR502 is down regulated in ovarian cancer, and CCND1 was negatively correlated with MIR502, which means the expression of CCND1 is overexpressed in OC. In addition, PPI network showed that CCND1 played a core function in interacting with other proteins, further verified the important role of CCND1 in regulating progress of OC. The development of OC may be slowed down by up regulating MIR502, which decreases expression of CCND1 and restrains cell cycle.
The MYC family proto-oncogenes is comprised of c-MYC, MYCN and MYCL [40].c-MYC as an oncogene in numerous cancer cells, plays an important role in a myriad of biological processes including cell growth, cell cycle progression and proliferation [41,42] by cooperating with YAP and activating a great number of target genes [43]. In fact, amplification of c-MYC has been reported in ovarian cancer [44].And previous studies showed that higher levels of c-MYC expression led faster recurrence and worse overall survival rate of patients with high grade serous ovarian cancer, and was related to cisplatin resistance in ovarian cancer cells. Silencing of c-MYC inhibited cell growth of cisplatin-resistant ovarian cancer. Thus, c-MYC targeted therapy is a potential treatment for ovarian cancer patients with high expression of c-MYC, including those who are resistant to cisplatin. It means that c-MYC may act as a new biomarker and therapy target of chemotherapy response. Another member of the MYC family, MYCN, controls the basic process of embryonic development. MYCN signaling disorder leads to a variety of tumors, including neuroblastoma, medulloblastoma, rhabdomyosarcoma, Wilms tumor, prostate cancer and lung cancer. In neuroblastoma, genetic aberration of MYCN amplification is the most related to poor prognosis and failure of therapy. MYCN targeted therapy has been proposed as a new strategy for cancer treatment, and a lot of efforts have been made to develop direct and indirect MYCN inhibitors with potential clinical applications [45].
FGF1 belongs to fibroblast growth factors (FGFs) family, whose function is regulating many cellular processes including cell proliferation, differentiation and survival as an oncogene [46][47][48].FGF1 is associated with tumor development, as it is upregulated in various cancers, including breast cancer, gliomas and ovarian cancer. The expression of FGF-1 has strong relationship with severity of prognosis and chemoresistance of tumor [49][50][51][52]. FGF1 has been considered as a potential prognostic marker for OC [53]. Compared with other family members, FGF1 genetic variation has the most significant correlation with the increased risk of ovarian cancer [54]. In addition, FGF1 expression is also an important determinant of survival and response to platinum chemotherapy. Therefore, the regulation of FGF1 by different mechanisms may play an important role in the development of ovarian cancer [55].Our study suggested that MIR502 played counter-regulation expression effect on FGF1, MIR502 low level of expression increases FGF1expression in ovarian cancer, which may lead OC development and platinum chemotherapy resistance.
GLI family zinc finger proteins mediate Sonic hedgehog (Shh) signaling and they exist in embryonic tumor cells as effective oncogenes. The proteins encoded by GLI2 belong to the C2H2-type zinc finger protein subclass of the GLI family. Researchers have found that the expression of GLI2 was regulated by Yap/TAZ, which activated the downstream regulatory factors of Shh signaling and promote proliferation [56]. A large number of evidences implicate that GLI2 regulates the key link of cell cycle. Nagao et al. reported that silencing the expression of GLI2 made the cell cycle stop in G1 phase, which prevented the growth of osteosarcoma [57]. Similar mechanisms have been reported in human vascular smooth muscle cells [58] and myofibroblasts [59]. Same cases occurred in cervical cancer that overexpression of GLI2 increased proliferation. All researches demonstrated that GLI2 promoted cell proliferation and exerted a tumor-promoting role in cancer. In our study, GLI2 as a downstream molecular of Hippo signaling pathway, highly expressed due to the negatively regulated by MIR502, resulting in accelerating the pathological process of ovarian cancer. And GLI2 may be targeted as a novel therapeutic strategy in the future.
In summary, we have discovered that MIR502 expression in ovarian cancer was lower than that in normal tissue, which means MIR502 acting as a significant tumor-suppressor in ovarian cancer and through K-M analysis, MIR502 expression is connected to ovarian cancer patient overall survival outcomes. Additionally, analysis showed that the expression of MIR502 was regulated by NRF1 and further exerted apoptosis and inhibiting proliferation by regulating genes downstream of the Hippo signaling pathway including, CCND1, FGF1, MYC and GLI2. In our study, we propose novel mechanisms between MIR502 and ovarian cancer that have not been elucidated previously. The immediate application of our findings is that MIR502 can be used as prognostic tools in ovarian cancer. The better result is that our research on MIR502 in ovarian cancer will promote the more extensive research on the molecular mechanism of MIR502 and provide reference for improving the clinical treatment of ovarian cancer.

Conclusion
Our study suggested that MIR502 might be modulated by NRF1 and play function as a potential tumor suppressor by regulating Hippo signaling pathway downstream anti-apoptosis and pro-proliferation genes, therefore provided a novel candidate for developing MIR502-based therapeutic strategies.

Ethics approval and consent to participate
Not applicable.

Consent for publication
Not applicable.

Availability of data and materials
The microRNA expression datasets used in this study (GEO: GSE83693 and GSE119055) were acquired from the National Center for Biotechnology Information (NCBI) Gene expression (http://www.ncbi.nlm.nih.gov/geo/) .

Competing interests
The authors declare that they have no competing interests. Funding Not applicable.  The expression level of MIR502 is lower in ovarian cancer tissue comparing with normal ovary tissue. A-B Volcano plots of detectable genome-wide miRNA profiles in ovarian cancer tissue and normal ovarian tissue samples from GSE83693 and GSE119055, respectively. Green and red plots represent aberrantly expressed miRNAs with P<0.05 and |log2(FC)|>2. Green plots indicate down-regulated genes, red plots indicate upregulated genes and grey plots indicate normally expressed miRNAs. C Venn diagram of GSE83693 and GSE119055D Detailed information of seven common different expression miRNAs were listed.       MIR502 regulated CCND1, FGF1, MYC and GLI2.Correlation of the expression levels of MIR502 and Hippo signaling pathway downstream genes, including CCND1, FGF1, MYC, GLI2, AFP and AXIN2.Data were analyzed using Pearson's R correlation. *, P < 0.05; **, P < 0.01; ***, P < 0.001. Figure 10 CCND1 and MYCN were at core position in PPI network. A Venn diagram of predicted target genes of MIR502 by using miRanda, miRWalk, PICTAR5, Targetscan and DIANAmT, 860 common genes were selected. B Venn diagram of 860 common predicted target genes and 1501 overexpression genes in ovarian cancer obtained from GEPIA,44 common genes were selected as hub genes. C The protein-protein interaction networks of 44 hub genes of MIR502 in ovarian cancer. Nodes represent gene-encoded proteins. Connections between nodes represent the relationship between proteins. A bolder line implies a higher confidence level.