Screening and identification of key genes between high-grade and low-grade serous ovarian carcinomas using integrated bioinformatics

doi:10.21203/rs.3.rs-4717976/v1

Download PDF

Research Article

Screening and identification of key genes between high-grade and low-grade serous ovarian carcinomas using integrated bioinformatics

https://doi.org/10.21203/rs.3.rs-4717976/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Epithelial ovarian cancer (EOC) is one of the most aggressive tumors in women. The most common pathological type of EOC is high-grade serous carcinoma (HGSC), which is often diagnosed at an advanced stage. Low-grade serous carcinoma (LGSC) is estimated to account for 10% of all serous carcinomas. Previous studies have demonstrated that molecular and clinical characteristics differences are apparent between these two subtypes of EOC. The objective of this study was to screen and identify key genes between HGSC and LGSC, and to explore potential molecular mechanisms in the pathogenesis of EOC. The microarray datasets GSE27651 and GSE14001, with a total of 23 LGSC tissue samples and 32 HGSC tissue samples, were obtained from the Gene Expression Omnibus (GEO). The differentially-expressed genes (DEGs) were selected out through the “affy” and “limma” package in R. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis were performed through the Database for Annotation, Visualization and Integrated Discovery (DAVID). The protein-protein interaction (PPI) analysis of DEGs was carried out through the Cytoscape software. Finally, survival analysis of some key geneswas conducted using the Kaplan Meier Plotter Online Tool. A total of 357 DEGs were found in HGSC, of which 181 were up regulated and 176 were down regulated. GO functional enrichment analysis showed that the DEGs were mainly associated with nucleus, cell proliferation and protein binding. KEGG pathway analysis showed that these genes were enriched in the PI3K-Akt signaling pathway, pathways in cancer, the p53 signaling pathway, cell cycle, microRNAs in cancer. Twelve hub genes (TOP2A, CDK1, CCNB1, MAD2L1, KIF11, CCNB2, TTK, AURKA, RACGAP1, BUB1, RRM2 and PBK) were screened out from PPI network. The mRNA expression of TOP2A, CDK1, CCNB1, MAD2L1, KIF11, CCNB2, TTK, AURKA, RACGAP1, BUB1, RRM2, and PBK were significantly increased in tumor tissues. The protein expression of TOP2A, CDK1, CCNB1, MAD2L1, KIF11, CCNB2, TTK, AURKA, RACGAP1 and PBK were distinctly higher in serous ovarian cancer tissues than non-serous ovarian cancer tissues detected by immunohistochemical staining. Survival analysis showed that TOP2A, CCNB1, KIF11, AURKA, and BUB1 were significantly associated with clinical survival outcome. In addition, there is a significant correlation between the expression levels of twelve hub-genes and immune cell infiltration in serous ovarian cancer. In summary, the present study identified DEGs and hub genes by two GEO datasets mining, which might offer new insights into the molecular mechanisms of these two subtypes of EOC and provide some prognostic biomarkers for the treatment of EOC.

high-grade serous ovarian cancer

genes expression profiles

microarray

pathways

Protein-protein interaction (PPI)

survival

Epithelial ovarian cancer (EOC) is the most aggressive gynecologic malignancy, with 22440 estimated new cancer cases and 14080 estimated new cancer deaths in 2017 in the United States[1]. Serous, mucinous, endometrioid and clear cell were the main histologic subtypes of ovarian cancer, of which serous epithelial carcinomas dominate the largest proportion[2]. The high-grade serous cancer (HGSC) is the single largest group of EOC, which is responsible for almost two-thirds of ovarian cancer deaths[3]. However, low-grade serous cancer (LGSC) takes only 5–10% proportion of EOC[4]. Patients with LGSC have a longer overall five-year survival, and tend to be chemo-resistant compared to HGSC[5]. It is accepted that these two types of EOC are two distinct tumor types with different clinical and genetic characteristics. Although many efforts have been made to explore new therapeutic approaches, most women with HGSC develop recurrence and chemoresistance[6]. At present, the underlying molecular mechanisms are still poorly understood, which greatly limits the treatment of HGSC. Like many solid malignances, HGSC frequently have a high proportion of chromosomal instability, such as gene copy number amplifications and deletions, which are associated with tumor grade and patient outcomes[7]. In addition, the alterations of oncogenes and tumor suppressor genes are also involved in the tumorigenesis of HGSC, such as HER-2/neu, c-myc, BRCA1 and BRCA2[8]. Identification of these key genes could provide advances in diagnosis and in therapeutic strategies. Thus, investigating key genes and understanding the molecular mechanism are urgent and necessary.

The pathogenesis of serous ovarian cancer at a molecular level has been explored for years. For example, Helland et al defined a pathway that may drive biological and clinical behavior of a distinct molecular subtype of HGSC[9]. Stronach et al declared that HDAC4 abrogated sensitivity to cisplatin through modulating STAT1 acetylation, phosphorylation, and nuclear translocation in ovarian cancer[10]. Zaid et al found that the FGFR4 is overexpression in HGSC and might be an indicator of poor prognosis[11]. However, most studies ignored to reveal key genes which regulate the occurrence and progress of serous ovarian cancer instead of concentrating on one certain molecular target. Nowadays, the microarray, a high throughput technique for large-scale gene expression analysis, makes it possible to investigate the molecular basis of human cancer on a genomic scale[12, 13]. In the past decades, the gene expression profiles of ovarian cancer using microarray technology have been investigated to better understand the ovarian tumorigenesis[14–16].

Given the limited samples and conflicting outcomes of individual microarray, we performed an integrated bioinformatics analysis to evaluate key genes between HGSC and LGSC. Two gene expression profiles microarrays (GSE27651 and GSE14001) were obtained from Gene Expression Omnibus database (GEO, http://www.ncbi.nlm.nih.gov/geo/) to screen out the DEGs between LGSC and HGSC. Then, gene ontology (GO) annotation and protein–protein interaction network (PPI) construction were applied to explore the function of DEGs. Finally, the hub genes were identified by the PPI network using the Cytoscape software. Its role in epithelial ovarian cancer were analyzed by studying its mRNA expression level, protein expression level, impact on survival, and immune cell infiltration. The present study aimed to compare the gene expression profiles between LGSC and HGSC at the molecular level and identify potential biomarkers of HGSC, providing a new approach for the clinical diagnosis and treatment of ovarian cancer patients.

Microarray data

Two microarrays of GSE14001 and GSE27651 were retrieved from GEO database. These two gene expression datasets were both based on Affymetrix GPL570 platform (Affymetrix Human Genome U133 Plus 2.0 Array). The microarray data for GSE14001 consisted of 10 HGSC tissue samples and 10 LGSC tissue samples. The microarray data for GSE27651 consisted of 22 HGSC tissue samples and 13 LGSC tissue samples.

Data processing and identification of DEGs

The raw data files of two datasets were download from GEO included CEL files. The probes annotation was obtained from GPL570 platform. Statistical software R (version 3.4.0, http://www.r-project.org/) and packages of Bioconductor (http://www.bioconductor.org/) were applied to analyze the DEGs between HGSC and LGSC samples. The method of Robust Multi-array Average was performed to normalize and to logarithmically convert raw expression profiles data[17]. We used the limma package (Linear Models for Microarray Analysis ) to select significant DEGs[18]. We used the empirical Bayes methods in the R package “sva” to adjust batch effects[19]. An adjusted P < 0.01 and a |logFC| ≥ 3 were defined as the threshold criteria. The clustering analysis of DEGs was performed by the ‘‘RColorBrewer’’, “gplots” packages in R.

Functional and pathway enrichment analysis of DEGs

The DAVID database (https://david.ncifcrf.gov/) is a web bioinformatics resource that could extract biological features of large gene lists[20]. The Gene Ontology (GO) project is a web tool for functional interpretation of genes, genes products, and sequences[21]. Kyoto Encyclopedia of Genes and Genomes (KEGG) is a comprehensive knowledge repository to provide higher order functional information of genes[22]. We used the DAVID database to perform the GO and KEGG pathways enrichment analysis of DEGs. The terms with P < 0.05 were considered as the significant enrichment.

Protein–protein interaction (PPI) network construction

The Search Tool for the Retrieval of Interacting Genes (STRING) database is a website tool that give evidence to the protein–protein biological interactions[23]. To better understand the direct and indirect interactions of key genes, the PPI network was constructed by the STRING database. A confidence score ≥ 0.7(high confidence score)was considered as significant. The PPI network of DEGs was reconstructed by the Cytoscape software. Cytoscape is an open source software that provides powerful function in integrating large data of genetic interactions, protein–DNA and protein–protein[24]. Finally, the CytoHubba46 plug-in in Cytoscape 3.5.1 was used to identify the hub genes of PPI, and the Molecular Complex Detection (MCODE) plug-in was used to select significant modules with degree cut-off = 10[25]. The protein expression of hub genes in cancer and normal tissues was also verified in the Human Protein Atlas (HPA) database (http://www.proteinatlas.org/)[26].

Survival analysis of twelve hub-genes

Kaplan–Meier plotter (www.kmplot.com) is an online tool that could provide patient survival analysis of 54675 genes in four types of cancer, including breast cancer, ovarian cancer, lung cancer and gastric cancer[27]. By using the median expression value as a cut-off, the patients were divided into two groups. Then we plotted the Kaplan-Meir overall survival (OS) curves in serous ovarian cancer patients. The hazard ratio (HR) with 95% confidence intervals and log rank test P value were also calculated.

TIMER Database Analysis

The Tumor IMmune Estimation Resource (TIMER) database (https://cistrome.shinyapps.io/timer/) is a reliable and comprehensive resource that allows the evaluation of the abundance of immune cell infiltration across diverse cancer types. In this study, the “Gene module” was used to evaluate the correlation between the expression level of immune-related genes, tumor purity, and the infiltration of immune cells including B cells, CD4 + T cells, CD8 + T cells, neutrophils, macrophages, and dendritic cells (DCs) in serous ovarian cancer. Meanwhile, the “Correlation module” was used to investigate the correlation between the expression level of immune-related genes and several gene markers of tumor-infiltrating immune cells.

Identification of DEGs

The raw CEL files were downloaded from the GEO, and then normalized with the “affy” package. After eliminating batch effects in the expression data, we used the limma package to select DEGs. Using P < 0.01 and |logFC| ≥ 3 criteria, 357 DEGs were screened out from two GEO datasets, including 181 up-regulation DEGs and 176 down-regulation DEGs in HGSC compared to LGSC. The top 10 up-regulated and top 10 down-regulated DEGs were listed in Table 1. In addition, the heat map of DEGs was shown as the Fig. 1. The hierarchical clustering analysis revealed a distinct separation in these two different types of ovarian cancer.

Table 1

Top ten up-regulation and top ten down-regulation DEGs (high-grade serous ovarian cancer versus low-grade serous ovarian cancer)
Probe set	Gene symbol	Log fold change	P value
Up-regulated
1565483_at	EGFR	3.29	1.78E-12
214677_x_at	IGLC1	3.19	1.91E-07
218542_at	CEP55	3.10	1.61E-17
203560_at	GGH	2.83	5.39E-13
201292_at	TOP2A	2.77	2.86E-12
203764_at	DLGAP5	2.77	5.43E-17
219787_s_at	ECT2	2.74	9.80E-15
204533_at	CXCL10	2.71	1.48E-09
219148_at	PBK	2.66	2.25E-13
242546_at	LINC01296	2.64	4.91E-11
Down-regulated
214218_s_at	XIST	-3.41	5.93E-11
229782_at	RMST	-3.39	3.47E-13
233249_at	LOC100507073	-2.85	6.36E-13
204014_at	DUSP4	-2.68	3.70E-08
229331_at	SPATA18	-2.61	2.69E-07
205765_at	CYP3A5	-2.58	3.42E-08
229245_at	PLEKHA6	-2.58	4.60E-14
240065_at	FAM81B	-2.50	7.43E-06
225996_at 227188_at	LONRF2 EVA1C	-2.50 -2.40	1.14E-08 9.98E-09

Functional determination by GO terms and KEGG pathways

To interpret the function of the DEGs, the gene lists were uploaded to the online software DAVID. The GO terms including cell component (CC), biological processes (BP) and molecular function (MF) ontologies were shown in Fig. 2. For the CC ontology, we found that most DEGs significantly enriched in nucleus and cytoplasm items, such as nucleus, nucleoplasm and cytosol. Some genes were associated with organelles in cytoplasm, such as centrosome, spindle and midbody (Fig. 2A). For the BP ontology, the majority GO terms were about cell proliferation items, such as cell division, G1/S transition of mitotic cell cycle. In addition, the other GO terms is significantly related to regulation activities of organism, including positive regulation of cell proliferation, positive regulation of GTPase activity (Fig. 2B). In the MF ontology, the binding function consisted a large proportion of GO categories, which involved protein binding, protein kinase binding, ATP binding and microtubule binding (Fig. 2C). As shown in the Table 2, the KEGG pathway enrichment analysis found five significantly pathways, including the PI3K-Akt signaling pathway, pathways in cancer, p53 signaling pathway, cell cycle, microRNAs in cancer.

Table 2

KEGG pathway enrichment of DEGs
Term	Count	P value
hsa04110: Cell cycle	12	1.97E-06
hsa04115: p53 signaling pathway	9	5.85E-06
hsa05200: Pathways in cancer	14	0.0053
hsa04151: PI3K-Akt signaling pathway	12	0.0133
hsa05206: MicroRNAs in cancer	10	0.0257
Count refers to the number of genes significantly enriched in this term

PPI network and hub genes

The protein-protein interaction network of DEGs was constructed through the STRING database. The 357 DEGs were uploaded to the STRING website to get PPI data. Then, the PPI data were analyzed by the Cytoscape software. The PPI which consisted of 167 nodes and 1794 edges shown in Fig. 3A. After the PPI network construction, the CytoHubba plug-in in Cytoscape was used to indentify hub genes. In PPI networks, 12 node protein, including TOP2A, CDK1, CCNB1, MAD2L1, KIF11, CCNB2, TTK, AURKA, RACGAP1, BUB1, RRM2, PBK had a strong association with other nodes (Degree ≥ 60). From the PPI network, a significant module including 51 nodes and 1206 edges was selected by the MCODE (Fig. 3B). The GO and KEGG enrichment analysis show these genes in the module were enriched in the mitotic cytokinesis, ATP binding and p53 signaling pathway (Table 3).

Table 3

GO and KEGG pathway enrichment analysis of the genes in module
Term	Description	Count	P value
GO:0000281	mitotic cytokinesis	5	7.37E-07
GO:0032467	positive regulation of cytokinesis	4	5.82E-05
GO:0007018	microtubule-based movement	5	6.19E-05
GO:0035556	intracellular signal transduction	4	0.044449
GO:0030496	midbody	6	5.04E-06
GO:0005654	nucleoplasm	14	9.49E-05
GO:0005737	cytoplasm	19	4.15E-04
GO:0005634	nucleus	16	0.004541
GO:0016020	membrane	9	0.004243
GO:0005524	ATP binding	19	8.45E-11
GO:0016887	ATPase activity	5	1.27E-04
GO:0003682	chromatin binding	5	0.008074
KEGG:hsa04110	Cell cycle	10	6.71E-12
KEGG:hsa04115	p53 signaling pathway	5	2.13E-05
Count refers to the number of genes significantly enriched in this term

Genetic Alterations of Twelve Hub-genes in Serious Ovarian Cancer Patients

The cBioPortal tool was used for the analysis of genetic alterations of twelve hub-genes from the TCGA PanCancer Atlas dataset. As a result, 5% TOP2A, 9% CDK1, 11% CCNB1, 6% MAD2L1, 5% KIF11, 5% CCNb2, 6% TTK, 14% AURKA, 5%ACGAP1, 4%BUB1, 10%RRM2 and 10% PBK were altered in ten types of genetic alterations, including infame mutation (unknown significance), missense mutation (unknown significance), splice mutation (unknown significance), truncating mutation (unknown significance), amplification, deep deletion, mRNA high, mRNA low, protein high and protein low in the queried TCGA serious ovarian cancer samples (Fig. 4A). The alteration frequency derived from mutations, copy-number alterations, mRNA expression data and protein expression data were shown in serous ovarian cancer (Fig. 4B). To explore whether these hub genes amplification had an influence on its mRNA and protein level, the results indicated that TTK and AURKA were amplified along with the significantly high mRNA and protein level from TCGA-OV cohort (Fig. 4C). The mutation types, number, and sites of TOP2A, CDK1, TTK and BUB1 genetic alterations were displayed in Fig. 4D.

The Expression Level of the Twelve Hub-genes in Ovarian Cancer Tissues with P53 Mutation

GO functional enrichment analysis showed that the DEGs was associated with the p53 signaling pathway. We conducted research and analysis on 199 ovarian cancer samples with P53 mutations and 19 ovarian cancer samples without P53 mutations. The results showed that the mRNA expression levels of AURKA (Fig. 5A), BUB1 (Fig. 5B), CCNB1 (Fig. 5C), CDK1 (Fig. 5D), MAD2L1 (Fig. 5E), PBK (Fig. 5F), TOP2A (Fig. 5G), and TTK (Fig. 5H) in P53 mutated ovarian cancer samples were higher than those in P53 non mutated ovarian cancer samples.

The Expression Level of the Twelve Hub-genes in Tumor Tissues

To better understand the differential expression, the CPTAC dataset was used to assess the twelve hub-genes mRNA expression level in large-scale mRNA data from the National Cancer Institute. As shown in Fig. 6A, the mRNA expression of TOP2A, CDK1, CCNB1, MAD2L1, KIF11, CCNB2, TTK, AURKA, RACGAP1, BUB1, RRM2, and PBK were significantly increased in tumor tissues. The GEPIA2 tool was also used to analyze the relationship between the twelve hub-genes expression and tumor pathological stage. Figure 6B showed stage-specific change of MAD2L1, RACGAP1, RRM2 and TTK in tumor tissues. On the other hand, the remaining hub genes were no clear association between the gene expression and patients’ stage.

The Protein Expression of Twelve Hub-genes in Serous Ovarian Cancer tissues vs. Non-serous Ovarian Cancer tissues

To assess the protein expression of twelve hub-genes in serous ovarian cancer tissues vs. non-serous ovarian cancer tissues, we performed immunohistochemical analysis. The expression levels of ten proteins, including TOP2A, CDK1, CCNB1, MAD2L1, KIF11, CCNB2, TTK, AURKA, RACGAP1 and PBK were distinctly higher in serous ovarian cancer tissues than non-serous ovarian cancer tissues (Fig. 7).

The Prognostic Value of Twelve Hub-genes in Patients with Serous Ovarian Cancer

The prognostic value of twelve hub-genes were assessed using an online tool of KM plotter (http://www.kmplot.com). The survival curves were calculated according to the gene expression levels. Among those hub-genes, our results showed that high expression of TOP2A (HR 1.32 95%CIs [1.13–1.55], P = 0.00063), CCNB1 (HR 1.24 95%CIs [1.05–1.47], P = 0.01), KIF11 (HR 1.23 95%CIs [1.04–1.45], P = 0.017), AURKA (HR 1.47 95%CIs [1.23–1.75], P = 2.2e-05), and BUB1 (HR 1.18 95%CIs [1.02–1.38], P = 0.03) were associated with worse overall survival (Fig. 8A-E). We also found increased expression of MAD2L1 (HR 0.79 95%CIs [0.62–1.01], P = 0.055) was not significantly associated with worse overall survival (Fig. 8F).

Correlation of Twelve Hub-Genes with Tumor Purity and Immune Cell Infiltration in Patients with Serous Ovarian Cancer

Since the functional annotation analysis revealed that the twelve hub-genes participated in the process of the immune response, next, the correlation between the expression of twelve hub-genes and immune cell infiltration in the TIMER database was further analyzed. Interestingly, high expression levels of twelve hub-genes were found to be associated with high immune cell infiltration in serous ovarian cancer. A positive correlation between TOP2A expression and the infiltration of macrophage (Cor = 0.12, p = 8.67e − 03) and purity (Cor = 0.123, p = 6.83e − 03) were observed, while the TOP2A expression was negatively associated with the infiltration of CD8 + T cells (Cor = -0.121, p = 7.90e − 03). There is no significant correlation between the expression level of TOP2A and the infiltration level of B cells, CD4 + T cells, neutrophil and dendritic cells (Fig. 9A). The change of KIF11 expression level is the same as that of TOP2A (Fig. 9E). As shown in Fig. 9B, the expression level of CDK1 is positively correlated with tumor purity (Cor = 0.152, p = 1.63e − 02), but not significantly correlated with the infiltration level of B cells, CD4 + T cells, CD8 + T cells, macrophage, neutrophil and dendritic cells. Figure 9C showed that the expression level of CCNB1 is positively correlated with the infiltration level of macrophage (Cor = 0.119, p = 8.80e − 03) and neutrophil (Cor = 0.125, p = 6.01e − 03), but not significantly correlated with the infiltration level of B cells, CD4 + T cells, CD8 + T cells, dendritic cells and tumor purity. The expression level of MAD2L1 is positively correlated with the infiltration level of macrophage (Cor = 0.18, p = 7.00e − 05), neutrophil (Cor = 0.238, p = 1.38e − 07) and dendritic cells (Cor = 0.17, p = 1.75e − 04), but not significantly correlated with the infiltration level of B cells, CD4 + T cells, CD8 + T cells and tumor purity (Fig. 9D).The change of CCNB2 and TTK expression level is the same as that of MAD2L1 (Fig. 9F, G). Figure 9H showed that the expression level of AURKA is positively correlated with the infiltration level of CD4 + T cells (Cor = 0.118, p = 9.90e − 03), macrophage (Cor = 0.225, p = 6.06e − 07) and neutrophil (Cor = 0.12, p = 8.49e − 03), but not significantly correlated with the infiltration level of B cells, CD8 + T cells, dendritic cells and tumor purity. The expression level of RACGAP1 and RRM2 are positively correlated with the infiltration level of CD4 + T cells, macrophage, neutrophil and dendritic cells, but not significantly correlated with the infiltration level of B cells and CD8 + T cells. In addition, the expression level of RACGAP is also positively correlated with tumor purity, while the expression level of RRM2 is not significantly correlated with tumor purity (Fig. 9I, K). Figure 9J showed that the expression level of BUB1 is positively correlated with the infiltration level of macrophage (Cor = 0.128, p = 4.96e − 03) and tumor purity (Cor = 0.09 p = 4.78e − 02), while the BUB1 expression level was negatively associated with the infiltration of CD8 + T cells (C or = -0.121, p = 7.90e − 03). but not significantly correlated with the infiltration level of B cells, CD4 + T cells, neutrophil and dendritic cells. The expression level of PBK is positively correlated with the infiltration level of macrophage (Cor = 0.131, p = 4.14e − 03), but not significantly correlated with the infiltration level of B cells, CD4 + T cells, CD8 + T cells, macrophage, dendritic cells and tumor purity (Fig. 9L).

In recent years, there has been a growing clinical and genetic evidence to support a two-pathway model of ovarian cancer[28, 29]. From the evidence, the LGSC derived from a stepwise fashion from benign serous cystadenoma to LGSC. However, HGSC seemed to be in the absence of recognizable precursor lesion. From the molecular genetics perspective, an increasing number of studies have found that LGSC have a high frequency mutation of KRAS, BRAF, PTEN, CTNNB1, whereas the HGSC is associated strongly with p53 mutations[30, 31]. In view of clinical features, women with HGSC usually present at advanced stages, have a sensitive responsive to chemotherapy, have a high incidence of recurrence and decreased survival[32, 33]. However, many younger women with LGSC tend to be resistant to chemotherapy, but have an improved survival[34]. Thus, it is worth emphasizing that HGSC and LGSC are two distinctly different tumor types rather than different grades of the same neoplasm. Based on clinical and molecular features of ovarian cancer, it is generally accepted that ovarian carcinomas could be divided into Type I tumor which are slowly developing in a stepwise fashion; and Type II tumor which are rapidly growing, high-grade tumors[31, 35, 36]. Thus, it is necessary to compare the different molecular mechanisms about these two different tumor types and find potential candidate markers in a prognostic, predictive, or therapeutic sense.

In this study, gene expression data of 23 LGSC tissue samples and 32 HGSC tissue samples were retrieved from two GEO datasets. We identified 357 DEGs between LGSC tissue samples and HGSC tissue samples and GO and pathway enrichment analysis were further performed. First, in CC ontology, most of the DEGs were significantly enriched in nuclear-related and cell mitosis items including nuclear, centrosome, spindle and microtubule cytoskeleton[37]. These cell biological processes were mainly associated with genomic stability and cancer etiology[38]. Many studies suggested that the spindle assembly are playing a role in the multi-step process of acquired paclitaxel resistance in ovarian cancer[39, 40]. Then, for the GO category of BP, our results showed that the most significant items were the cell division items including mitotic nuclear division, G1/S transition of mitotic cell cycle. Admittedly, the majority of patients with HGSC will develop chemoresistance through multiple complex mechanisms. Some study found defects in mitotic checkpoints were associated with paclitaxel resistance in ovarian cancer cell lines[40, 41]. In MF ontology, the binding-related items were the most significant items such as protein binding, protein kinase binding and ubiquitin protein ligase binding. These data suggested that these DEGs might participate in the binding functions among nucleus, protein, histone and extracellular matrix, which influence tumor microenvironment and signal transduction in cancer cells[42]. Obviously, protein binding and kinase binding are involved in the multiple signaling pathways in cancer. A large number of kinase and phosphatases were found in ovarian cancer tissue, which were considered to have taken part in biological process of ovarian cancer[43]. In the present study, pathway analysis found five significantly enriched pathways including p53 signaling pathway, PI3K-Akt signaling pathway[44]. Hayano et al reported a novel intact p53 pathway subtype of HGSC, which provided a new insight into the pathogenesis of HGSC with an intact p53 pathway[45]. In addition, the deregulation of the PI3K/Akt pathway is an important genomic change in ovarian cancer[46]. Tanaka et al reported that high levels of pAkt were correlated with decreased progression free survival and overall survival in ovarian cancer, thus Akt might be a potential molecular target[47].

This study also constructed the protein-protein interaction network of DEGs and screened out top 12 hub genes. Among these hub genes, TOP2A was the highest degree of connectivity. The TOP2A gene encode the enzyme topoisomerase IIa (topo IIa), which is responsible for resolving topological problems during the DNA metabolism[48]. TOP2A is located close to HER2 on chromosome 17q21 and the clinical value of TOP2A in breast cancer have been studied for years[49, 50]. It reported that TOP2A gene copy number or protein overexpression could predict the treatment response of pegylated liposomal doxorubicin in platinum resistant in EOC[51]. So far, the carcinogenic effect of TOP2A for ovarian cancer have not explored fully. The second hub gene CDK1 is a mitotic cyclin-dependent kinases (CDKs), which is a key factor for G2/M phase transition[52]. CDK1 might have the potential to regulate resistance and suppression of CDK1 could reverse paclitaxel resistance in ovarian cancer cells[53]. Taken together, these results showed that these hub genes might be associated with the clinicopathological features of these two types of ovarian cancer.

In our research, three hub genes were found to have significantly increased mRNA expression levels in ovarian cancer tissues with P53 mutation and tumor tissues, significantly increased protein expression levels in serous ovarian cancer tissues, and its high expression levels were significantly correlated with worse survival outcome, including TOP2A, CCNB1 and AURKA. Cyclins refer to proteins that differ in their levels to activate specific cyclin-dependent kinases (CDKs) required for progression in the cell cycle[54]. Cyclin B1 (CCNB1) has a pivotal role in regulating and forming a complex with CDK1 to promote the transition from the G2 phase of cell cycle to mitosis[55]. Increasing evidence demonstrates that the over-expression of CCNB1 is observed in certain number of human cancers including colorectal cancer[55], breast cancer[56], and ovarian cancer[57]. Yiping Zou et al reported that the mRNA expression level of CCNB1 is upregulated in various tumor tissues, including hepatocellular carcinoma (HCC). The high expression of CCNB1 gene is associated with poor prognosis in HCC patients[58]. Hui Zhang et al demonstrated that CCNB1 silencing activates the p53 signaling pathway and consequently inhibits cell proliferation and promotes cell senescence in pancreatic cancer[59]. As showed in our study, we found that the mRNA expression level and protein expression level of CCNB1 were significantly increased in tumor tissues, and the high expression of CCNB1 was positively correlated with adverse survival outcomes in ovarian cancer patients. Aurora kinases belong to serine/threonine kinases which share a highly conserved catalytic domain containing auto-phosphorylating sites[60]. This family contains three members: Aurora A (AURKA), Aurora B (AURKB), and Aurora C (AURKC). Both AURKA and AURKB play essential roles in regulating cell division during mitosis while AURKC has a unique physiological role in spermatogenesis[61]. Relatively less information is available for the roles of AURKC in cancer. AURKA and AURKB have been found to function as oncogenes to promote tumorigenesis in multiple types of cancer including solid tumors and hematological malignancies[62]. Apart from playing a role in mitosis, an increasing number of studies have suggested that AURKA, when abnormally expressed, could be an oncogene involved in tumorigenesis[63]. Gene amplification, transcriptional activation and inhibition of protein degradation could contribute to the elevated levels of AURKA expression in cancer tissues. AURKA promotes tumorigenesis by participating in the cancer cell proliferation, epithelial-mesenchymal transition (EMT), metastasis, apoptosis, and self-renewal of cancer stem cells[64]. Our research results were consistent with these findings. The mRNA and protein expression levels of AURKA were significantly increased in tumor tissue and ovarian cancer tissue, and the high expression of AURKA could predict adverse prognostic outcomes for ovarian cancer patients.

In conclusion, this study aimed to identify DEGs between two different pathology types of serous ovarian cancer and find the potential predictor. In the present study, a total of 357 DEGs were selected and TOP2A, CCNB1 and AURKA might be prognostic biomarkers of HGSC. However, the absence of experimental verification is the major drawback of our study. Thus, further experiments are required to confirm our results obtained from bioinformatics analysis.

Authors’ contributions

Conceived and conducted the study: LWZ and JJF; analyzed the data: ZLP, WGS,WYW,LTF,JL,RHH,QYW,YW,SHZ,WJS,BW ; wrote the manuscript: LWZ and ZLP, all authors read and approved the final version.

Funding

This work was supported by The Second Hospital of Anhui Medical University, National Natural Science Foundation of China(grant numbers 2022GQFY09).

Availability of data and materials

The dataset used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Ethics approval and consent to participate

Patient data were obtained from Gene Expression Omnibus (GEO) and Database for Annotation, Visualization and Integrated Discovery (DAVID),two publicly open database resource ,Informed patient consent was not required

Competing interests

The authors declare that there is no confict of interest regarding the publication of this paper.

Conflicts of interest: The authors declare no competing financial interests.

R.L. Siegel, K.D. Miller, and A. Jemal, Cancer Statistics, 2017. CA Cancer J Clin, 2017. 67(1): p. 7-30.
T. Bonome, J.Y. Lee, D.C. Park, M. Radonovich, C. Pise-Masison, J. Brady, et al., Expression profiling of serous low malignant potential, low-grade, and high-grade tumors of the ovary. Cancer Res, 2005. 65(22): p. 10602-12.
D.D.L. Bowtell, The genesis and evolution of high-grade serous ovarian cancer. Nat Rev Cancer, 2010. 10(11): p. 803-8.
R.J. Kurman, and M. Shih Ie, The Dualistic Model of Ovarian Carcinogenesis: Revisited, Revised, and Expanded. Am J Pathol, 2016. 186(4): p. 733-47.
R.N. Grisham, G. Iyer, K. Garg, D. Delair, D.M. Hyman, Q. Zhou, et al., BRAF mutation is associated with early stage disease and improved outcome in patients with low-grade serous ovarian cancer. Cancer, 2013. 119(3): p. 548-554.
G.C. Jayson, E.C. Kohn, H.C. Kitchener, and J.A. Ledermann, Ovarian cancer. Lancet, 2014. 384(9951): p. 1376-88.
B.T. Hennessy, R.L. Coleman, and M. Markman, Ovarian cancer. Lancet, 2009. 374(9698): p. 1371-82.
B. Aunoble, R. Sanches, E. Didier, and Y.J. Bignon, Major oncogenes and tumor suppressor genes involved in epithelial ovarian cancer (review). Int J Oncol, 2000. 16(3): p. 567-76.
A. Helland, M.S. Anglesio, J. George, P.A. Cowin, C.N. Johnstone, C.M. House, et al., Deregulation of MYCN, LIN28B and LET7 in a molecular subtype of aggressive high-grade serous ovarian cancers. PLoS One, 2011. 6(4): p. e18064.
E.A. Stronach, A. Alfraidi, N. Rama, C. Datler, J.B. Studd, R. Agarwal, et al., HDAC4-regulated STAT1 activation mediates platinum resistance in ovarian cancer. Cancer Res, 2011. 71(13): p. 4412-22.
T.M. Zaid, T.L. Yeung, M.S. Thompson, C.S. Leung, T. Harding, N.N. Co, et al., Identification of FGFR4 as a potential therapeutic target for advanced-stage, high-grade serous ovarian cancer. Clin Cancer Res, 2013. 19(4): p. 809-20.
A. Sturn, J. Quackenbush, and Z. Trajanoski, Genesis: cluster analysis of microarray data. Bioinformatics, 2002. 18(1): p. 207-8.
O.P. Kallioniemi, U. Wagner, J. Kononen, and G. Sauter, Tissue microarray technology for high-throughput molecular profiling of cancer. Hum Mol Genet, 2001. 10(7): p. 657-62.
A.D. Santin, F.H. Zhan, S. Bellone, M. Palmieri, S. Cane, E. Bignotti, et al., Gene expression profiles in primary ovarian serous papillary tumors and normal ovarian epithelium: identification of candidate molecular markers for ovarian cancer diagnosis and therapy. Int J Cancer, 2004. 112(1): p. 14-25.
C.D. Hough, C.A. Sherman-Baust, E.S. Pizer, F.J. Montz, D.D. Im, N.B. Rosenshein, et al., Large-scale serial analysis of gene expression reveals genes differentially expressed in ovarian cancer. Cancer Res, 2000. 60(22): p. 6281-7.
J.B. Welsh, P.P. Zarrinkar, L.M. Sapinoso, S.G. Kern, C.A. Behling, B.J. Monk, et al., Analysis of gene expression profiles in normal and neoplastic ovarian tissue samples identifies candidate molecular markers of epithelial ovarian cancer. Proc Natl Acad Sci U S A, 2001. 98(3): p. 1176-81.
S. Hochreiter, D.A. Clevert, and K. Obermayer, A new summarization method for Affymetrix probe level data. Bioinformatics, 2006. 22(8): p. 943-9.
G.K. Smyth, Linear models and empirical bayes methods for assessing differential expression in microarray experiments. Stat Appl Genet Mol Biol, 2004. 3: p. Article3.
W.E. Johnson, C. Li, and A. Rabinovic, Adjusting batch effects in microarray expression data using empirical Bayes methods. Biostatistics, 2007. 8(1): p. 118-27.
W. Huang da, B.T. Sherman, and R.A. Lempicki, Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat Protoc, 2009. 4(1): p. 44-57.
M.A. Harris, J. Clark, A. Ireland, J. Lomax, M. Ashburner, R. Foulger, et al., The Gene Ontology (GO) database and informatics resource. Nucleic Acids Res, 2004. 32(Database issue): p. D258-61.
M. Kanehisa, and S. Goto, KEGG: kyoto encyclopedia of genes and genomes. Nucleic Acids Res, 2000. 28(1): p. 27-30.
A. Franceschini, D. Szklarczyk, S. Frankild, M. Kuhn, M. Simonovic, A. Roth, et al., STRING v9.1: protein-protein interaction networks, with increased coverage and integration. Nucleic Acids Res, 2013. 41(Database issue): p. D808-15.
P. Shannon, A. Markiel, O. Ozier, N.S. Baliga, J.T. Wang, D. Ramage, et al., Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res, 2003. 13(11): p. 2498-504.
G.D. Bader, and C.W. Hogue, An automated method for finding molecular complexes in large protein interaction networks. BMC Bioinformatics, 2003. 4: p. 2.
M. Uhlén, L. Fagerberg, B.M. Hallström, C. Lindskog, P. Oksvold, A. Mardinoglu, et al., Proteomics. Tissue-based map of the human proteome. Science, 2015. 347(6220): p. 1260419.
B. Gyorffy, A. Lánczky, and Z. Szállási, Implementing an online tool for genome-wide validation of survival-associated biomarkers in ovarian-cancer using microarray data from 1287 patients. Endocr Relat Cancer, 2012. 19(2): p. 197-208.
C.J. O'Neill, M.T. Deavers, A. Malpica, H. Foster, and W.G. McCluggage, An immunohistochemical comparison between low-grade and high-grade ovarian serous carcinomas: significantly higher expression of p53, MIB1, BCL2, HER-2/neu, and C-KIT in high-grade neoplasms. Am J Surg Pathol, 2005. 29(8): p. 1034-41.
C.N. Landen, M.J. Birrer, and A.K. Sood, Early events in the pathogenesis of epithelial ovarian cancer. J Clin Oncol, 2008. 26(6): p. 995-1005.
K. Levanon, C. Crum, and R. Drapkin, New insights into the pathogenesis of serous ovarian cancer and its clinical impact. J Clin Oncol, 2008. 26(32): p. 5284-93.
R. Dehari, R.J. Kurman, S. Logani, and I.M. Shih, The development of high-grade serous carcinoma from atypical proliferative (borderline) serous tumors and low-grade micropapillary serous carcinoma: a morphologic and molecular genetic analysis. Am J Surg Pathol, 2007. 31(7): p. 1007-12.
S.C. Plaxe, Epidemiology of low-grade serous ovarian cancer. Am J Obstet Gynecol, 2008. 198(4): p. 459.e1-8; discussion 459.e8-9.
A. Gockley, A. Melamed, A.J. Bregar, J.T. Clemmer, M. Birrer, J.O. Schorge, et al., Outcomes of Women With High-Grade and Low-Grade Advanced-Stage Serous Epithelial Ovarian Cancer. Obstet Gynecol, 2017. 129(3): p. 439-447.
K.M. Schmeler, and D.M. Gershenson, Low-grade serous ovarian cancer: a unique disease. Curr Oncol Rep, 2008. 10(6): p. 519-23.
K.R. Cho, and I.M. Shih, Ovarian cancer. Annu Rev Pathol, 2009. 4: p. 287-313.
R.J. Kurman, Origin and molecular pathogenesis of ovarian high-grade serous carcinoma. Ann Oncol, 2013. 24 Suppl 10: p. x16-21.
J. Wu, and A. Akhmanova, Microtubule-Organizing Centers. Annu Rev Cell Dev Biol, 2017. 33: p. 51-75.
M. Chen, R. Linstra, and M. van Vugt, Genomic instability, inflammatory signaling and response to cancer immunotherapy. Biochim Biophys Acta Rev Cancer, 2022. 1877(1): p. 188661.
A.A. Ahmed, A.D. Mills, A.E.K. Ibrahim, J. Temple, C. Blenkiron, M. Vias, et al., The extracellular matrix protein TGFBI induces microtubule stabilization and sensitizes ovarian cancers to paclitaxel. Cancer Cell, 2007. 12(6): p. 514-27.
B. McGrogan, S. Phelan, P. Fitzpatrick, A. Maguire, M. Prencipe, D. Brennan, et al., Spindle assembly checkpoint protein expression correlates with cellular proliferation and shorter time to recurrence in ovarian cancer. Hum Pathol, 2014. 45(7): p. 1509-19.
X. Hao, Z.G. Zhou, S.M. Ye, T. Zhou, Y.P. Lu, D. Ma, et al., Effect of Mad2 on paclitaxel-induced cell death in ovarian cancer cells. J Huazhong Univ Sci Technolog Med Sci, 2010. 30(5): p. 620-5.
S. AlMusawi, M. Ahmed, and A.S. Nateri, Understanding cell-cell communication and signaling in the colorectal cancer microenvironment. Clin Transl Med, 2021. 11(2): p. e308.
J.D. Wulfkuhle, J.A. Aquino, V.S. Calvert, D.A. Fishman, G. Coukos, L.A. Liotta, et al., Signal pathway profiling of ovarian cancer from human tissue specimens using reverse-phase protein microarrays. Proteomics, 2003. 3(11): p. 2085-90.
M. Kessler, C. Fotopoulou, and T. Meyer, The molecular fingerprint of high grade serous ovarian cancer reflects its fallopian tube origin. Int J Mol Sci, 2013. 14(4): p. 6571-96.
T. Hayano, Y. Yokota, K. Hosomichi, H. Nakaoka, K. Yoshihara, S. Adachi, et al., Molecular characterization of an intact p53 pathway subtype in high-grade serous ovarian cancer. PLoS One, 2014. 9(12): p. e114491.
A. Astanehe, D. Arenillas, W.W. Wasserman, P.C. Leung, S.E. Dunn, B.R. Davies, et al., Mechanisms underlying p53 regulation of PIK3CA transcription in ovarian surface epithelium and in ovarian cancer. J Cell Sci, 2008. 121(Pt 5): p. 664-74.
Y. Tanaka, Y. Terai, A. Tanabe, H. Sasaki, T. Sekijima, S. Fujiwara, et al., Prognostic effect of epidermal growth factor receptor gene mutations and the aberrant phosphorylation of Akt and ERK in ovarian cancer. Cancer Biol Ther, 2011. 11(1): p. 50-7.
E. Arriola, C. Marchio, D.S. Tan, S.C. Drury, M.B. Lambros, R. Natrajan, et al., Genomic analysis of the HER2/TOP2A amplicon in breast cancer and breast cancer cell lines. Lab Invest, 2008. 88(5): p. 491-503.
K.V. Nielsen, S. Müller, S. Møller, A. Schønau, E. Balslev, A.S. Knoop, et al., Aberrations of ERBB2 and TOP2A genes in breast cancer. Mol Oncol, 2010. 4(2): p. 161-8.
K. Vang Nielsen, B. Ejlertsen, S. Møller, J. Trøst Jørgensen, A. Knoop, H. Knudsen, et al., The value of TOP2A gene copy number variation as a biomarker in breast cancer: Update of DBCG trial 89D. Acta Oncol, 2008. 47(4): p. 725-34.
J. Erriquez, P. Becco, M. Olivero, R. Ponzone, F. Maggiorotto, A. Ferrero, et al., TOP2A gene copy gain predicts response of epithelial ovarian cancers to pegylated liposomal doxorubicin: TOP2A as marker of response to PLD in ovarian cancer. Gynecol Oncol, 2015. 138(3): p. 627-33.
Q.H. Xi, M.H. Huang, Y.Y. Wang, J.X. Zhong, R. Liu, G.Q. Xu, et al., The expression of CDK1 is associated with proliferation and can be a prognostic factor in epithelial ovarian cancer. Tumour Biol, 2015. 36(7): p. 4939-48.
T. Bae, K.Y. Weon, J.W. Lee, K.H. Eum, S. Kim, and J.W. Choi, Restoration of paclitaxel resistance by CDK1 intervention in drug-resistant ovarian cancer. Carcinogenesis, 2015. 36(12): p. 1561-71.
J.U. Shin, C.H. Lee, K.T. Lee, J.K. Lee, K.H. Lee, K.M. Kim, et al., Prognostic significance of ATM and cyclin B1 in pancreatic neuroendocrine tumor. Tumour Biol, 2012. 33(5): p. 1645-51.
Y. Fang, H. Yu, X. Liang, J. Xu, and X. Cai, Chk1-induced CCNB1 overexpression promotes cell proliferation and tumor growth in human colorectal cancer. Cancer Biol Ther, 2014. 15(9): p. 1268-79.
S.P. Wang, S.Q. Wu, S.H. Huang, Y.X. Tang, L.Q. Meng, F. Liu, et al., FDI-6 inhibits the expression and function of FOXM1 to sensitize BRCA-proficient triple-negative breast cancer cells to Olaparib by regulating cell cycle progression and DNA damage repair. Cell Death Dis, 2021. 12(12): p. 1138.
X. Yang, S. Zhou, C. Yang, C. Cao, M. He, and S. Zi, CCNB1, Negatively Regulated by miR-559, Promotes the Proliferation, Migration, and Invasion of Ovarian Carcinoma Cells. Mol Biotechnol, 2022. 64(9): p. 958-969.
Y. Zou, S. Ruan, L. Jin, Z. Chen, H. Han, Y. Zhang, et al., CDK1, CCNB1, and CCNB2 are Prognostic Biomarkers and Correlated with Immune Infiltration in Hepatocellular Carcinoma. Med Sci Monit, 2020. 26: p. e925289.
H. Zhang, X. Zhang, X. Li, W.B. Meng, Z.T. Bai, S.Z. Rui, et al., Effect of CCNB1 silencing on cell cycle, senescence, and apoptosis through the p53 signaling pathway in pancreatic cancer. J Cell Physiol, 2018. 234(1): p. 619-631.
R. Du, C. Huang, K. Liu, X. Li and Z. Dong, Targeting AURKA in Cancer: molecular mechanisms and opportunities for Cancer therapy. Mol Cancer, 2021. 20(1): p. 15.
S.K. Heo, E.K. Noh, Y.K. Jeong, L.J. Ju, J.Y. Sung, H.M. Yu, et al., Radotinib inhibits mitosis entry in acute myeloid leukemia cells via suppression of Aurora kinase A expression. Tumour Biol, 2019. 41(5): p. 1010428319848612.
E.O. Dos Santos, T.C. Carneiro-Lobo, M.N. Aoki, E. Levantini, and D.S. Bassères, Aurora kinase targeting in lung cancer reduces KRAS-induced transformation. Mol Cancer, 2016. 15: p. 12.
K. Sasai, W. Treekitkarnmongkol, K. Kai, H. Katayama, and S. Sen, Functional Significance of Aurora Kinases-p53 Protein Family Interactions in Cancer. Front Oncol, 2016. 6: p. 247.
L. Yang, Q. Zhou, X. Chen, L. Su, B. Liu, and H. Zhang, Activation of the FAK/PI3K pathway is crucial for AURKA-induced epithelial-mesenchymal transition in laryngeal cancer. Oncol Rep, 2016. 36(2): p. 819-26.

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Screening and identification of key genes between high-grade and low-grade serous ovarian carcinomas using integrated bioinformatics

Status:

Version 1

Abstract

Figures

INTRODUCTION

Materials and methods

Microarray data

Data processing and identification of DEGs

Functional and pathway enrichment analysis of DEGs

Protein–protein interaction (PPI) network construction

Survival analysis of twelve hub-genes

TIMER Database Analysis

Results

Identification of DEGs

Functional determination by GO terms and KEGG pathways

PPI network and hub genes

Genetic Alterations of Twelve Hub-genes in Serious Ovarian Cancer Patients

The Expression Level of the Twelve Hub-genes in Ovarian Cancer Tissues with P53 Mutation

The Expression Level of the Twelve Hub-genes in Tumor Tissues

The Prognostic Value of Twelve Hub-genes in Patients with Serous Ovarian Cancer

Discussion

Declarations

References

Additional Declarations

Status:

Version 1