簡單6基于cbioportal等數(shù)據(jù)庫聯(lián)合出圖發(fā)文

1、Shen et al. Journal of Ovarian Research019 0578 1(2019) 12:110Identification of key biomarkers associated with development and prognosis in patients with ovarian carcinoma: evidence from bioinformatic analysisJiayu Shen1, Shuqian Yu2, Xiwen Sun3, Meichen Yin1, Jing Fei1 and Jianwei Zhou1*AbstractBac

2、kground: Ovarian cancer (OC) is the deadliest cause in the gynecological malignes. Most OC patients arediagnosed in advanced stages with less than 40% of women cured. However, the possible mechanism underlying tumorigenesis and candidate biomarkers remain to be further elucidated.Results: Gene expre

3、ssion profiles of GSE18520, GSE54388, and GSE27651 were available from Gene ExpressionOmnibus (GEO) database with a total of 91 OC samples and 22 normal ovarian (OV) tissues. Three hundred forty nine differentially expressed genes (DEGs) were screened between OC tissues and OV tissues via GEO2R and

4、online Venn software, followed by KEGG pathway and gene ontology (GO) enrichment analysis. The enriched functionsand pathways of these DEGs contain male gonad development, cellular response to transforgrowth factorbeta stimulus, positive regulation of transcription from RNA polymerase II promoter, c

5、alcium independent cell cell adhesion via plasma membrane cell adhesion molecules, extracellular matrix organization, pathways in cancer, cell cycle, cell adhesion molecules, PI3K AKT signaling pathway, and progesterone mediated oocyte maturation. The protein protein network (PPI) was established an

6、d module analysis was carried out using STRING and Cytoscape.Next, with PPI network analyzed by four topological methods in Cytohubba plugin of Cytoscape, 6 overlap genes (DTL, DLGAP5, KIF15, NUSAP1, RRM2, and TOP2A) were eventually selected. GEPIA and Oncomine wereimplemented for validating te expr

7、ession and all the six hub genes weighly expressed in OC specimenscompared to normal OV tissues. Furthermore, 5 of 6 gexcept for DTL were associated with worse prognosisusing Kaplan Meier plotter online tool and 3 of 6 genes were significantly related to clinical stages, including RRM2, DTL, and KIF

8、15. Additionally, cBioPortal showed that TOP2A and RRM2 were the targets of cancer drugs in patients with OC, indicating the other four genes may also be potential drug targets.: Six hub genes (DTL, DLGAP5, KIF15, NUSAP1, RRM2, and TOP2A) present promising predictive value for the development and pr

9、ognosis of OC and may be used as candidate targets for diagnosis and treatment of OC.Keywords: Ovarian cancer, Differentially expressed genes, Bioinformatic analysis* Correspondence: 2195045Jiayu Shen and Shuqian Yu contributed equally to this work.1Department of Gynecology, The second affiliated ho

10、spital ofUniversity School of Medicine, No88, Jie Hangzhou, Zhengjiang 310002, PeoplesRoadShangcheng District,Full list of author information is available at the end of the article© The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4

11、.0International License (), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were. The Creative Commons Public Domain Dedi

12、cation waiver() applies to the dataavailable in this article, unless otherwise stated.RESEARCHOpen AccessShen et al. Journal of Ovarian Research(2019) 12:110Page 2 of 13IntroductionOvarian cancer (OC) is the leading cause of death in gynecological malign es. There were 22,530 new diagnoses in the Un

13、ited States in 2019. Due to lack of representative symptoms and sensitive diagnostic methods, more than 70% of patients are diagnosed with advanced disease (FIGO III or IV, The International Federation of Gynecology and Obestetrics) as defined by the sp of disease outside the pelvis. The standard tr

14、eatment remains appropriate surgical staging and debulking surgery, followed by platinum-based system- atic chemotherapy. Despite that standard treatment anddrugs, CBio Cancer Genomics Portal (cBioPortal) was used and showed that TOP2A and RRM2 were the tar- gets of cancer drugs in patients with OC,

15、 indicating the other four genes may also be potential drug targets. In, this bioinformatic study provides some promising biomarkers associated with development and prognosis in patients of OC.Materials and methodsMicroarray dataGEO () functions as apublic functional genomics database of high throug

16、hout gene expression data, chips and microarrays 2. Three gene expression profiles in OC and normal OV tissues were downloaded from GEO, that is GSE18520 3, GSE54388 4 and GSE27651 5. Microarray data ofcurrent novel therapies (such as-angiogenesis agentsand PARP inhibitors) do improve patients outco

17、me and reduce the mortality, the five-year survival is still low (about 47%) due to frequent relapse and drug1. Therefore, it is of vital importance and urgency to better understand the mechanism underlying tumorigen- esis in OC and develop new strategies for early diagno- sis, disease monitoring, a

18、nd prognosis evaluation.It is well-known that tumorigenesis is a heterogeneous disease characterized of various gene aberrations, so does ovarian carcinoma. However, the underlying mech- anisms of OC development have not been fully under- stood. Over the past decades, an array of high- throughput te

19、chnologies for measuring RNA intermedi- ates and epigenetic markers, such as DNA methylation and histone modifications, are widely available. With the continuously rapid development of microarray technol- ogy and bioinformatics analysis, genetic alterations at genome level have been widely dug out t

20、o identify the differentially expressed genes (DEGs) and functional pathways related to tumorigenesis and prognosis. In the present study, microarray data of three gene expression profiles were downloaded and differentially expressed genes (DEGs) were identified between OC and normal ovarian (OV) ti

21、ssues, followed by further assessment using Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway. Furthermore, protein-protein interaction (PPI) network of DEGs was established and Cytohubba plugin of Cytoscape was ap- plied for identifying some core genes. Moreover, the ex-

22、these three datasets were all onof GPL570 Plat-forms, HG-U133 Plus 2 Affymetrix Human Genome U133 Plus 2.0 Array. The GSE18520 dataset contained 53 high grade OC samples and 10 normal OV samples. GSE54388 contained 16 high grade OC samples and 6 normal OV samples while GSE27651 contained 22 high gra

23、de OC specimen and 6 normal OV specimen.Identification of DEGsGEO2R()isregarded as an interactive online tool designed to com- pare two or more datasets in a GEO series for the pur- pose of DEGs identification across experimental conditions. The DEGs between OC tissues and normal OV tissues were ide

24、ntified using GEO2R with the threshold of |logFC| > 2 and P value < 0.05 which were considered of statistically significance. For the next step, the online Venn software was applied to detect the inter- section DEGs among three datasets.Functional enrichment analysisThe GO datasets and KEGG pa

25、thway enrichment were used to analyze DEGs at the functional level with DA- VID (The Database for Annotation, Visualization and Integrated Discovery,6.8) 68. DAVID is a compensive database of func-pression of the overlapgenes between OC andtional annotation tools for connecting functional terms with

26、 gene lists using a clustering algorithm. In order to elucidate the functional profiles of the DEGs, we used DAVID to obtain the enriched biological process (BP), cellular component (CC), molecular function (MF) and KEGG pathway. P < 0.05 was considered statistically significant.normal OV tissues

27、 were validated using Gene Expression Profiling Interactive Analysis (GEPIA) and Oncomine. Taken together, six DEGs were selected for further ana- lysis, namely DTL, DLGAP5, KIF15, NUSAP1, RRM2,and TOP2A. Then, the Kaplan Meier plotter online tool was used to assess the prognostic value of these cor

28、e genes, showing that 5 genes, except for DTL, were corre- lated with worse survival. Three of those six hub genes were found to be significantly differentiated in various clinical stages, including RRM2, DTL, and KIF15. In addition, to explore relationships between genes andPPI network construction

29、 and module analysisSTRING (Search Tool for the Retrieval of InteractingGenes, version 11.0) online databasewas used to predict the PPI network which may furtherShen et al. Journal of Ovarian Research(2019) 12:110Page 3 of 13explain the mechanisms of the occurrence and progres- sion of diseases 9. B

30、y using STRING database, PPI net- work of DEGs was analyzed and an interaction with a combined score > 0.4 was recognized as statistical signifi- cance. The plug-in MCODE (Molecular Complex Detec- tion) app of Cytoscape (an public bioinformatics software, version 3.7.1) is constructed for cluster

31、ing a network based on topology to determine intensively con- nected regions 10, 11. The PPI network was plotted with the application of Cytoscape and the most significant module in the PPI network was narrowed down using MCODE with the following criteria: degree cutoff =10, node score cutoff = 0.2,

32、 k-core = 2, max depth = 100.datasets contained 349 genes as shown in the Venn dia- gram (Fig. 1d).Enrichment analysis for DEGsTo elucidate the biological functions of the overlap DEGs, we performed functional annotation and pathway enrichment analysis via DAVID online tool. Results indi- cated that

33、 the overlapDEGs in biological process of GO enrichment were markedly associated with male gonad development, cellular response to transfor growth factor beta (TGF) stimulus, positive regulation of transcription from RNA polymerase II (RNAP II) pro- moter, calcium independent cell-cell adhesion via

34、plasma membrane cell adhesion molecules, and extracellular matrix organization (Fig. 2a). As for molecular function of GO enrichment, DEGs were remarkably related to cal- cium ion binding, transcriptional activator activity (RNAP II core promoter proximal region sequence specific bind- ing), transcr

35、iptional factor activity (RNAP II distal enhan- cer sequence specific binding), heparin binding, and microtubule binding (Fig. 2b). In addition to cellular com- ponent, the overlapDEGs were particularly enriched in extracellular region, extracellular space, proteinaceous extracellular matrix, midbod

36、y, and extracellular matrix (Fig. 2c). Besides, signaling pathway analysis of KEGG demonstrated that those DEGs played pivotal roles in pathways in cancer, cell cycle, cell adhesion molecules, PI3K-AKT signaling pathway, and progesterone mediated oocyte maturation (Fig. 2d).Hub genes selection and a

37、nalysisThe plug-in Cytohubba of Cytoscape is an APP provided with 11 topological analysis methods for ranking nodes in a PPI network by their network features 12. In the present study, the top 20 hub genes were ranked accord-ing to theal clique centrality (MCC),umneighborhood component (MNC), Degree

38、 and edge per-colated component (EPC). The overlaphub genes intop 20 by these four topological methods were selectedfor further bioinformatics analysis using t NIA App of Cytoscape which contains a compeMA-ensivesets of datasets from GEO, BioGRID, Pathway Commons and I2D, as well as organism specifi

39、c functional genom- ics datasets 13. Meanwhile, the biological process of hub genes was also visualized using BiNGO (Biological Networks Gene Oncology tool, version 3.0.3) plugin of Cytoscape 14. Furthermore, GEPIA (PPI network construction and significant module identificationSTRING database was us

40、ed to predict the potential rela-cer- Oncomine (plied for validating t is a web-based toolwebsite and online) weredatabase both ap-tionships among these overlapDEGs at proteine expression 15, 16. GEPIA to provide key interactive andlevels with combined score > 0.4. The establishment of the PPI ne

41、twork was constructed via Cytoscape software, including 270 nodes and 1169 edges (Fig. 3a). Addition- ally, the most important PPI network modules were ob- tained using MCODE, consisted of 37 nodes and 640 edges (Fig. 3b). To further identify the hub genes, we ap- plied Cytohubba plugin of Cytoscape

42、 for ranking the top 20 nodes in the above PPI network according to four topological analysis methods, including MCC, MNC, De-customizable functions based on TCGA (The Cancer Genome Atlas) and GTEx (Genotype-Tissue Expression) data. The overall survival of hub genes was analyzed using Kaplan Meier-p

43、lotter online tool which is com- monly applied for assessing the effect of genes on sur- vival based on EGA, TCGA database and GEO 17. CBioPortal was used for exploring genetic alterations of hub genes and relationships between genes and drugs 18.gree, and EPC (Table 1). A total of 6 overlaphubgenes

44、 were determined for further analysis, namely DTL, DLGAP5, KIF15, NUSAP1, RRM2, and TOP2A.ResultIdentification of DEGs in ovarian cancerVia GEO2R online tools, DEGs in three datasets (1273 DEGs in GSE18520, 910 in GSE54388, and 905 inGSE27651, respectively) were extracted after gene ex- pression pro

45、file data processing and standardization with the cutoff standard of P value < 0.05 and |logFC| > 2Re-analysis of the six selected genesGEPIA and Oncomine tool were used to further validate the expression of these 6 genes. Both the databases verified that the expression of DTL, DLGAP5, KIF15,

46、NUSAP1, RRM2, and TOP2A presented significant dissimilarities in OC samples and normal OV samples (Fig. 4). Via Kaplan Meier plotter online tool, we found that five of the six(Fig. 1a-c). The overlapDEGs among these threeShen et al. Journal of Ovarian Research(2019) 12:110Page 4 of 13Fig. 1 a c Volc

47、ano plot of DEGs between OC tissues and normal OV tissues in each dataset. Red dots: significantly up regulated genes in OC; Green dots: significantly down regulated genes in OC; Blue dots: non differentially expressed genes. P < 0.05 and |logFC| > 2 were considered asstatistically significant

48、. d Venn diagram of overlap349 DEGs from GSE18520, GSE54388 and GSE27651 datasetsselected genes, except for DTL, were correlated with worse survival (Fig. 5). The training dataset (TCGA-OV) was ap- plied to validate the correlations between the six hub genes and clinical stages. Three of those six h

49、ub genes were found to be negatively related to clinical stages, including RRM2,182 patients. Among the six genes, DTL and RRM2 altered most (both were 7%) and the main type was amplification (Fig. 8a). Besides, regarding with the relationship between ancer drugs and hub genes, TOP2A and RRM2 wereth

50、e targets of cancer drugs in patients with OC (Fig. 8b).DTL, and KIF15 (Fig. 6). A protein/gene interaction net-DiscussionOvarian cancer remains the deadliest cause among ma-work for the six genes and their products with 20 proteins/ genes was generated via GeneMANIA plugin of Cytoscape, including R

51、RM1, TPX2, GLRX, MKI67, AURKA, HMMR, CENPF, ZWINT, RRM2B, ASPM, KIF11, CDK1, CCNA2, NCAPG, MELK, FOXM1, KIAA0101, BUB1B, KIF20A, andCENPE (Fig. 7a). The ranking order based on the score ran- ging from high to low was NUSAP1, DTL, KIF15, DLGAP, RRM2, and TOP2A. Besides, the biological process analysi

52、s of the hub genes was visualized in Fig. 7b, indicating the top 5 involved BPs includes chromosome segregation, posi- tive regulation of nuclear division, positive regulation of mi- tosis, DNA replication, and chromosome condensation. For genetic alteration, six hub genes were altered in 39 (21%) o

53、flignes of the female reproductive system. Long-termsurvival of OC patients is still unsatisfactory as a resultof late diagnosis, recurrence and drug. Earlydiagnosis plays a crucial role in the prevention and prog- nosis of cancers, including ovarian carcinoma. Cancer gen 125 (CA125) have been most

54、widely used in diagnosis and monitoring in OC patients 19, 20. Any- how, not all the OC patients present with abnormalCA125 level. Though humanprotein 4 (HE4)is also approved by the US Food and Drug Administra- tion (FDA) for monitoring, HE4 is not applied in routineShen et al. Journal of Ovarian Re

55、search(2019) 12:110Page 5 of 13Fig. 2 GO and KEGG analysis of the overlapDEGs in OC. a Biological process. b Cellular component. c Molecular function. d KEGG pathway.All of the enrichment pathways were generated using the ggplot2 package in R languageFig. 3 Common DEGs PPI network construction and m

56、odule analysis. a A total of 270 DEGs were visualized in the DEGs PPI network complex: the nodes represent proteins, the edges represent the interaction of the proteins. b Module analysis using MCODE: degree cutoff =10, node score cutoff = 0.2, k core = 2, max depth = 100Shen et al. Journal of Ovari

57、an Research(2019) 12:110Page 6 of 13Table 1 Hub genes for highly differentiated expressed genes ranked in Cytohubba plugin of Cytoscaperelated to calcium ion binding, transcriptional activator activity, transcriptional factor activity, heparin binding, and microtubule binding. In terms of cellular c

58、ompo- nent, these DEGs were particularly enriched in extracel-CatelogyRank methods in cytoHubbaMCCMNCDegreeEPClularregion,extracellularspace,proteinaceousGene symbol top 20DTL RRM2 DLGAP5 KIF15 NUSAP1 KIF20A PBKTTK CCNB1 BUB1B BUB1 NCAPG KIF11 CDK1 RAD51AP1 KIAA0101 TOP2A CENPF CCNB2CDC20CDK1 CCNB1 TOP2A NCAPG RRM2 DLGAP5 NUSAP1 KIF20A TTK UBE2C BUB1B BUB1 KIF11 DTL MELK KIF15 PBK CENPF CCNB2CDC20CDK1 CCNB1 RRM2 DLGAP5 TOP2A NCAPG HMMR NUSAP1 KIF20A TTK UBE2C BUB1B BUB1 CDC20 KIF11