OBJECTIVES

This study aims to reveal regulatory network of lncRNAs-miRNAs-mRNAs in oral squamous cell carcinoma (OSCC) through gene expression data.

MATERIAL AND METHODS

Differentially expressed lncRNAs, miRNAs and mRNAs (cut-off: False discovery rate (FDR)<0.05 and |fold change|>1.5) were unveiled by package edgeR of R. Cox regression analysis was performed to screen prognostic factors in OSCC related with overall survival (OS) and relapse-free survival (RFS). Protein-protein interaction (PPI) network was constructed for differentially expressed mRNAs using BioGRID, HPRD and DIP. Key hub genes were identified from top 100 differentially expressed mRNAs ranked by betweenness centrality using recursive feature elimination. LncRNA-miRNA and miRNA-mRNA regulatory network were constructed and combined into ceRNAs regulatory network. Gene ontology biological terms and Kyoto Encyclopedia of Genes and Genomes pathways were identified using Fisher's exact test.

RESULTS

A total of 929 differentially expressed mRNAs, 23 differentially expressed lncRNAs and 29 differentially expressed miRNAs were identified. 59 mRNAs, 6 miRNAs (hsa-mir-133a-1, hsa-mir-1-2, hsa-mir-486, hsa-mir-135b, hsa-mir-196b, hsa-mir-193b) and 6 lncRNAs (C10orf91, C2orf48, SFTA1P, FLJ41941,PART1,TTTY14) were related with OS; and 52 mRNAs, 4 miRNAs (hsa-mir-133a-1, hsa-mir-135b, hsa-mir-196b, hsa-mir-193b) and 2 lncRNAs (PART1, TTTY14) were associated with RFS. A support vector machine (SVM) classifier containing 37 key hub genes was obtained. A ceRNA regulatory network containing 417 nodes and 696 edges was constructed. ECM-receptor interaction, cytokine-cytokine receptor interaction, focal adhesion, arachidonic acid metabolism, and p53 signaling pathway were significantly enriched in the network.

CONCLUSION

These findings uncover the pathogenesis of OSCC and might provide potential therapeutic targets.