Identification of key genes with clinical outcome in benign meningioma using bioinformatic analysis

Background: Meningioma is the second most common type of brain neoplasms.However,the underlying molecular mechanisms are still not clear,and the main treatment is mainly surgery plus radiotherapy. Material and method: To explore the key genes in benign meningioma,we downloaded microarray dataset GSE43290 from Gene Expression Omnibus(GEO) database.The differential genes (DEGs) between benign meningioma and normal meninges were identified by GEO2R.The gene ontology (GO) and Kyoto Encyclopedia of Gene and Genomes (KEGG) pathway were performed by the Database for Annotation,Visualization and Integrated Discovery (DAVID).The protein-protein interaction (PPI) network and module analysis were performed and visualized by the Search Tool for the Retrieval of Interacting Gene database (STRING) and Cytoscape.The hub genes were evaluated by the Cytohubba and further explored by MCODE plugin of Cytoscape and Enrichr.The relationship between hub genes and clinical factors were further explored by GSE16581 through R software. Result: A total of 358 DEGs were identified,including 15 upregulated genes and 343 downregulated genes.The main enriched functions were extracellular matrix organization 、 inflammatory response 、 cell adhesion 、 extracellular space and integrin binding.The main KEGG pathways were Malaria and focal adhesion.Among these DEGs,5 overlapping selected as correlated with age and tumor recurrence,which could be clinical therapeutic targets. Conclusion: This study indicates the key genes in benign meningioma which may help us understand the molecular mechanisms and provide the candidate therapeutic targets.


Microarray data
The GEO database (http://www.ncbi.nlm.nih.gov/geo) was a public functional genomics of high throughout gene expression data.The gene expression dataset (GSE43290) was downloaded from GEO database (Affymatrix GPL96 platform,Affymatrix Human Genome U133A Array).The GSE43290 contained 33 benign meningioma tissue samples and 4 normal meninges tissue samples.

Identification of DEGs
The DEGs between benign meningioma and normal meninges samples were detected by GEO2R (http://www.ncbi.nlm.nih.gov/geo/geo2r),and the adjusted P-value and |LogFold Change(FC)| were calculated.Probe sets without meaningful gene symbol and genes with more than one probes were eliminated.The |LogFC|≧2 and adjusted P-value<0.01 were considered statistically significant.

GO and KEGG pathway enrichment analysis of DEGs
The Database for Annotation,Visualization and Integrated Discovery (DAVID) was an online gene analysis database which integrated bioinformatic data、biological pathways and provides functional annotations genes [21,22] .The GO and KEGG pathway enrichment were performed by DAVID.The GO analyses which was a tool to annotate and analyze genes were classified into biological process(BP)、 cellular component(CC) and molecular function(MF) [23] .The KEGG pathway contained pathway maps for the molecular systems in both normal and perturbed states [24] .The gene counts ≧10 and FDR(adjusted P-value)<0.05 were considered statistically significant.

PPI network construction and module analysis
The Search Tool for the Retrieval of Interacting Gene database (STRING; http://string-db.org) was an online database to analyze interactions between proteins [25] .In this study,previous DEGs were mapped by STRING to evaluate the potential interactions of proteins with combined score >0.9.After that,PPI network was constructed and visualized by Cytoscape software(version 3.7.2) which was an open source bioinformatic software platform [26] .The plugin Molecular Complex Detection(MCODE,version 1.6) was used to find the intensively densely connected region based on topology [27] .The most significant network was detected by MCODE with following criteria:degree cutoff =2, node score cutoff = 0.2, k-core = 2, max depth = 100.

Identification and analysis of hub genes
The plugin CytoHubba which provided 11 topological analysis methods of Cytoscape was being used to rank node in previous PPI network [28] .In this study,the top 10 genes were ranked by Cytohubba with Degree、the maximal clique centrality (MCC), maximum neighborhood component (MNC) and edge percolated component (EPC).The overlapping genes were considered as hub genes and further explored the GO and KEGG pathways using Enrichr [29] .The adjusted P-value<0.01 was considered statistically significant.We further downloaded and analyzed GEO dataset(GSE16581) by R software(version 4.0.1),illuminating the relationship between the expression of hub genes and clinical factors,including age and recurrence.

Identification of DEGs
With the criteria |LogFC|≧2 and adjust P-value<0.01,a total of 358 DEGs were identified respectively,consisting of 15 upregulated genes and 343 downregulated genes between normal meninges and benign meningiomas,as showing in figure 1. Figure 1.Differential expression of data among two datasets. Note:The red points represent upregulated genes with |LogFC|≧2 and adjust P-value<0.01.The green points represent downregulated genes with |LogFC|≧2 and adjust P-value<0.01.The black points represent genes with no statistically significance.

GO and KEGG pathway enrichment of DEGs
The GO function and KEGG pathway analysis of DEGs were performed by DAVID.GO function analysis showed that changes in biological process (BP) were mainly enriched in extracellular matrix organization、inflammatory response、cell adhesion、regulation of blood pressure.Changes in cellular components (CC) were mainly enriched in extracellular space、extracellular region、axonextracellular matrix.Changes in molecular function (MF) were mainly enriched in integrin binding.The KEGG pathways showed that DEGs mainly enriched in Malaria、Focal adhesion (Table 1，Figure 2

PPI network construction and module analysis
The PPI network of EDGs were constructed by STRING and visualized by Cytoscape.A total of 341 nodes and 324 edges were involved in PPI network( Figure   3).The most significant module was constructed by MCODE plugin( Figure  4),including 14 nodes and 91 edges.

Hub genes selection and analysis
To further explore hub genes,the Cytohubba plugin were being used to rank the top 10 nodes in the above PPI network with four topological methods,including Degree、 MCC、MNC and EPC (Table 2).A total of 5 overlapping genes were selected as hub genes and further explored the GO and KEGG pathways by Enrichr( Table 3).The clinical outcomes of hub genes were showed in figure 5 and 6.
GPR183 CCL19 IL6 P2RY14 APLNR PPBP S1PR1 Note:Bold genes were the overlapping hub genes in the top 10 ranked genes with four methods in the Cytohubba.MNC,maximum neighbourhood component. MCC,maximal clique centrality.EPC,edge percolated component.Degree,node connect degree.

4.Discussion
The growth pattern of meningioma can't be predicted precisely and need long-term follow-up [30] .Complete resection may not be applied to every single type of meningiomas,but tumor location is strongly associated with different Simpson grade removal and there could be many serious functional impairment,such vascular insults or the 9 th through 12 th nerve deficit in ventral foramen magnum meningiomas [31][32][33] .Radiotherapy is regarded as supplementary post-operative treatment but should be applied discreetly,because of radiation resistance or the potential malignant transformation [17,34] .Because of lacking potential therapeutic targets,we can use few ways to control meningioma.Thus,it's important to explore the molecular mechanisms of benign meningioma and find the potential chemotherapeutic targets.
Nowadays,microarray data and high-throughout sequence which detect millions of genes play an important role in finding potential targets for cancer diagnosis and treatment.In present study,microarray data GSE43290 downloaded from GEO database was further analyzed to identify the key genes which might contribute to the occurrence and development of benign meningioma.A total of 358 DEGs were identified respectively,consisting of 15 upregulated genes and 343 downregulated genes.GO enrichment analysis showed that DEGs were mainly enriched in extracellular matrix organization、inflammatory response、 cell adhesion、 extracellular space and integrin binding.The significant KEGG pathway of DEGs were malaria and cell adhesion.At last,the selected hub genes were AGT、CXCL8、CXCL2、CXCL12 and CXCR4.
AGT,also called angiotensinogen,was knowned as precursor of the renin-angiotesin system.The human AGT gene ,which coded for a 485-amino acid protein,consisted of five extrons and four introns.Mature AGT protein contained 452 amino acid residues,which consisted of des(Ang I) AGT and angiotensin I(Ang I) ,an inactive decapeptide which is converted into Ang II [35] .Local Ang II in tumor environment(TME) produced by hypoxia-lactate-chymase-dependent mechanism was involved in immune escape and mediating radioresistance in tumor cells [36][37][38] .And augmented AGT and angiotesin-converting enzyme(ACE) released from apoptotic endothelial cells played a vital role in turning to Ang II to accelerate the progress of vascular remodeling [39] .It was demonstrated that human AGT inhibits endothelial cell proliferation and angiogenesis in vivo,and prevents tumor sinusoids from remodeling and anterialization,thus inhibits tumor progression [40,41] .Sun et al [42] also reported that high glucose promoted tumor proliferation by suppressing AGT expression.In this study,the downregulation of AGT represented that AGT expression decreased and the role of delaying tumor angiogenesis was weakened.
CXCL8， also known as chemokine (C-X-C motif) ligand 8,plays a very important role in tumor growth、invasion and metastases in an autocrine and paracrine maner [43] .Some researches demonstrated that CXCL8 was associated with many types of tumors,such as glioblastoma 、 lung ademocarcinoma、 hepatocellular carcinoma and colon cancer [44][45][46] . Luo et al [44] demonstrated that high gene expression levels of CXCL8 and VEGF was correlating with recurrent glioblastoma.Daniel et al [47] showed that IL-8 was strongly associated with glioma formation and progression mediated by activator protein-1(AP-1) and nuclear transcription factor-kappa(NF-кB) site. Liu et al [45] showed that CXCL8 was an independent unfavorable factor with recurrence free survival and overall survival in patients suffering from adenocarcinoma,and the human Dachshund homologue 1(DACH1) was able to antagonize CXCL8 through AP-1and NF-кB.Shen et al [46] reported that CXCL8 may induce tumor proliferation、invasiveness through induction of epithelial-mesenchymal transition via the phosphatidylinositol 3-kinase (PI3K)/protein kinase B(AKT)/nuclear factor-кB(NF-кB) pathway.Researches demonstrated that human meningioma cells secreted specifically the CXC chemokine-8 [48,49] .In our study,the differential expression of CXCL8 between two datasets suggested it as a important target.
CXCL2,also known as chemokine (C-X-C motif) ligand 2,was a member of chemokine subfamily.Some researches demonstrated that CXCL2 was overexpressed in breast、hepatic、colon and bladder cancer [50][51][52][53] .Typically,Ding et al [51] reported that CXCL2 was downregulated in hepatocellular carcinoma.Overexpression of CXCL2 significantly emasculated HCC cell proliferation,and the overall survival rates of high CXCL2 expression group were significantly higher than group with low CXCL2 expression.In this study,we manifested that CXCL2 was downregulated in benign meningiomas other than normal meninges,which may not inhibit meningioma proliferation.
CXCL12,also known as chemokine (C-X-C motif) ligand 12 or sromal cell-derived factor-1(SDF-1),also was a member of chemokine subfamily,widely expressed in many tissues [54,55] .CXCL12 was secreted by meningeal cells as a chemotactic factor attracting external grnual cells and played important role in embryogenesis,including vascluarization、neurodevelopment,and maintained hemostasis in mature brain system [56,57] . CXCL12 regulated tumor cell adhesion with laminin、fibrinogen and endothelial cell,and removal of CXCL12 abolished the chemotactic effect entirely [55,57] .Du R et al [58] and Santiago B et al [59] reported that under hypoxic conditions,tumor cells or fioblasts upregulated SDF-1 miRNA to facilitate angiogenesis.Some researches [60,61] demonstrated that hypoxia could also significantly stimulate CXCR4 expression.CXCR4,also known as chemokine (C-X-C motif) ligand receptor 4,was a G protein-coupled receptor activating by CXCL12 [62] .Zhang et al [63] reported that the expression of CXCR4 was positively correlated with CXCL12 expression,and the chemotactic effect of CXCL12 was modulated by controlling CXCR4 expression.Moreover Hartmann et al [64] reported that CXCL12-induced integrin activation and CXCR4 co-operated in mediating adhesion and survival signals from the TME.CXCL12/CXCR4 axis was also involved in blood vessel growth and remodeling [65] ,and Hiratsuka et al [66] demonsrated that inhibition of both VEGF1 and CXCR4 signaling showed greater effects on tumor vascular density、growth and metastasis than inhibition of VEGF1 alone.So we could know that CXCL12/CXCR4 axis plays a crucial role in tumor formation、 proliferation and metastasis,thus causing tumor formation and invasion [49,64,67,68] . Mo et al [69] demonstrated that CXCL12/CXCR4 promoted neurofibromatosis 1-associated malignant peripheral nerve sheath tumor(MPNST) proliferation,which illustrated that high CXCR4 expression was correlated with poor survival.Zeng et al [70] reported that blockage of CXCL12/CXCR4 axis by plerixafor prolonged survival of tumor-bearing mice by immunosuppression in the TME.Cai et al [60] reported that plasminogen kringle 5 reduced nuclear HIF-1α accumulation and impaired HIF-1α DNA-binding and transactivation function,thus suppressing the expression of CXCL12/CXCR4 to inhibit the neovascularization and tumor growth.
In those hub genes,the expressions of CXCL2 and CXCL8 were correlated with age,which was divided into two groups with the limits of 60 years old.With more genes and more pathways involving in tumorgenesis,it could be explained why elder people were frequently found with meningoma.Meanwhile these two hub genes were correlated with tumor recurrence.But we need further evidence to prove the effects in cell and tissue level.

5.Conclusion
The present study intended to investigate the DEGs which involved in the mechanism of tumor generation、 proliferation and invasion.A total of 358 DEGs and 5 hub genes were identified and two hub genes,CXCL8 and CXCL2,were found to be correlated with tumor recurrence.Thus we have more methods to handle those meningiomas which can't be resected completely in the long run. But more studies are needed to illustrate the clinical significance of these genes in meningioma.

Abbreviations:
GEO ： Gene Expression Omnibus DEG ： differential expressed genes GO ： gene ontology KEGG ：Kyoto Encyclopedia of Gene and Genomes DAVID ：Database for Anotation,Visualization and Integreted Discovery PPI ：protein-protein interaction STRING ：Search Tool for the Retrieval of Interacting Gene database CNS ：central nervous system MRI ：magnetic resonance imaging TME ：tumor microenvironment AGT ：angiotensinogen CXCL8 ：chemokine (C-X-C motif) ligand 8 CXCL2 ：chemokine (C-X-C motif) ligand 2 CXCL12 ：chemokine (C-X-C motif) ligand 12 SDF-1 ：stromal cell-derived factor-1 CXCR4 ：chemokine (C-X-C motif) ligand receptor 4 MPNST ：malignant peripheral nerve sheath tumor Declarations: Ethics approval and consent to participate:Not applicable. Consent for publication:All the authors have read and approved the final version of manuscript and give consent. Availability of data and materials:Not applicable. Competing interests:The authors declare that they have no competing interests. Funding:Not applicable. Authors' contributions:MC conceived the conception,MC and CS searched the literature, MC drafted the manuscript with help of JZ. YHW corrected the details and references. All authors read and approved the final manuscript.