Molecular Mechanism Of CDCP1 Promoting Proneural-mesenchymal Transformation In Primary Glioblastoma

Background and objectiveGlioblastoma（GBM）accounts for 45.2%of primary malignant tumors of central nervous system.Although glioblastoma patients have been treated with STUPP regimen,the quality of life and prognosis of GBM patients are still very poor,with an average survival time of 15 months.In the 2016 WHO classification,the classification of tumors in the central nervous system（CNS）puts forward the advantages of integrating histology and molecular classification in the diagnosis of brain tumors,which makes the hierarchical classification of patients more accurate.According to the research of The Cancer Genome Atlas（TCGA）team,GBM can be divided into four specific molecular subtypes,including Mesenchymal,MES),Classical,CL),Neural,NL)and pro-neural（PN）.CUB domain-containing protein 1（CDCP1）is a transmembrane protein that transmits signals through other receptors（such as EGFR,HER2 and β1 integrins）,metabolic mediators（such as acyl coenzyme A synthetase）and key intracellular signal transduction factors（including Src,PKCδ,Akt and FAK）.The expression of CDCP1 protein is related to the low Overall survival,OS)of patients with breast cancer,colorectal cancer,renal cancer,lung cancer,ovarian cancer and pancreatic cancer.CDCP1 plays an important role in other tumors,but no relevant research has been found in glioma.Does CDCP1 play a similar role in promoting cancer in glioma and glioblastoma as other tumors?What molecular mechanism does CDCP1 affect the prognosis of glioblastoma patients?In order to explore these scientific problems,we carried out this research.Materials and methodsR language（version 4.0.1）was used for statistical analysis of data.Use "limma"package and "beeswarm" package to analyze two or more sets of classification data.The survival package was used for Kaplan-Meier analysis of glioma patients.Cox regression analysis was used to analyze the risk of prognosis and draw the forest map of independent prognosis.Transwell assays and Western Blotting verified the function of CDCP1 in GBM.The gene ontology（GO）database and Kyoto encyclopedia（KEGG）of genes and genomes database were used to analyze the pathway of CDCP1 in GBM.According to the first 2%of CytoNCA parameters and MCODE in Cytoscape plug-in,the key genes related to CDCP1 in protein-protein interaction network（PPI network）were screened out.Immunohistochemistry verified the correlation between CDCP1,CD44 and ITGAM.Use "pROC" package to draw the subject-operation characteristic curves of multiple gene models in gene matrix.Use"rms" package and "survival" package to draw nomogram.P<0.05（bilateral）indicates that the difference is statistically significant.ResultsBy mining TCGA,CGGA and GEO databases,it was found that the expression of CDCP1 in glioblastoma was significantly higher than that in normal brain tissue,and increased with the grade of glioma（P<0.05）.Further prognostic risk assessment indicated that the higher the expression of CDCP1,the lower the overall survival rate.Besides,excluding age,sex and tumor grade,it can independently influence the prognosis of patients（P<0.05）.In vitro experiments verified that overexpression of CDCP1 promoted GBM invasion and migration.Ten key genes with high correlation were screened by protein protein interaction network（PPI network）and Cytoscape software,among which CD44 and STAT3 were one of MES-GBM markers,and the other eight key genes were related to immune infiltration.GO function analysis and KEGG pathway analysis indicated that EMT and immune infiltration related functions were mainly enriched.According to the TCGA team’s research on four types of GBM in TCGA database,it is found that CDCP1 is highly expressed in MES-GBM.According to correlation analysis,CDCP1 is positively correlated with MES-GBM and negatively correlated with PN-GBM.In vitro experiments verified that CDCP1 was positively correlated with CD44 and ITGAM.Risk prediction models of these three genes were established.the receiver-operating characteristic curve（ROC）results indicated that the areas under the curves of CDCP1,CD44,ITGAM and their interactions were 0.558,0.642,0.528 and 0.692,respectively.Compared with the risk model with low expression,the prognosis of patients with high expression was significantly worse.The reliability of the model is verified by nomogram.ConclusionThe expression of CDCP1 is up-regulated in glioma,and the high expression of CDCP1 is positively correlated with the poor prognosis of glioma patients.CDCP1 promotes the proneural-mesenchymal transition（PMT）of primary GBM by interacting with CD44 and ITGAM.The prognostic risk model composed of these three genes is an important index for diagnosis and prognosis prediction of glioma patients.