Rigorous molecular characterization of biological systems has uncovered a variety of gene variations underlying normal and disease states and a remarkable complexity in the forms of RNA transcripts that exist. A recent concept, competitive endogenous RNA, suggests that some non-coding RNAs can bind to miRNAs to modulate their role in gene expression. Here, we used several platforms, integrating mRNA, non-coding RNAs and protein data to generate an RNA-protein network that may be dysregulated in human glioblastoma multiforme (GBM). Publicly available microarray data for mRNA and miRNA were used to identify differentially expressed miRNAs and mRNAs in GBM relative to non-neoplastic tissue samples. Target miRNAs were further selected based on their prognostic significance, and the intersection of their target gene set with the differentially expressed gene set in Venn diagrams. Two miRNAs, miR-637 and miR-196a-5p, were associated with poor and better prognosis, respectively, in GBM patients. Non-coding RNAs, ENSG00000203739/ENSG00000271646 and TPTEP1, were predicted to be miRNA target genes for miR-637 and miR-196a-5p and positively correlated with the selected mRNA, CYBRD1 and RUFY2. A local protein interaction network was constructed using these two mRNAs. Predictions based on the ENSG00000203739/ENSG00000271646-miR-637-CYBRD1 and TPTEP1-miR-196a-5p-RUFY2 regulation axes indicated that the two proteins may act as an oncogene and tumor suppressor, respectively, in the development of GBM. These results highlight competitive endogenous RNA networks as alternative molecular therapeutic targets in the treatment of the disease.
Keywords: GBM; lncRNA; mRNA; molecular datasets; network.
Copyright © 2020 Zheng, Su, Kong, Lin, Liu, Wang and Wang.