Excavating novel diagnostic and prognostic long non-coding RNAs (lncRNAs) for head and neck squamous cell carcinoma: an integrated bioinformatics analysis of competing endogenous RNAs (ceRNAs) and gene co-expression networks

Abstract

Long non-coding RNAs (lncRNAs) have been demonstrated to fine-tune gene regulations that govern a broad spectrum of oncogenic processes. Nonetheless, our understanding of the roles of lncRNAs and their interactions with miRNAs and mRNAs in HNSCC is still highly rudimentary. Here, we present a comprehensive bioinformatics analysis in which competing endogenous RNA (ceRNA) network construction and weighted gene co-expression network analysis (WGCNA) were combined to explore novel diagnostic and prognostic lncRNAs for HNSCC. Differentially expressed mRNAs (DEGs), miRNAs (DEMs) and lncRNAs (DELs) were identified based on the RNA sequencing data and clinical data retrieved from TCGA database. LncRNA-regulated ceRNA networks were constructed based on the interactive RNA pairs predicted by miRDB, miRcode and TargetScan. WGCNA was conducted to identify lncRNAs that were significantly correlated with patient overall survival (OS) and HNSCC tumor. RT-qPCR was employed to validate the expression of lncRNAs in HNSCC cell lines and patient sera. A ceRNA network consisting of 90 DEGs, 7 DEMs and 67 DELs associated with clinical traits was established. WGCNA and Kaplan-Meier survival analysis revealed that 5 DELs (MIR4435-2HG, CASC9, LINC01980, STARD4-AS1 and MIR99AHG) were significantly correlated with OS of HNSCC patients, whereas DEL PART1 was most significantly correlated with the HNSCC tumor. The in silico predicted expression patterns of PART1, LINC01980 and MIR4435-2HG were further validated in HNSCC cell lines and patient sera. Collectively, the present study provided novel insights into the lncRNA-regulated ceRNA networks in HNSCC and identified novel lncRNAs that harbor diagnostic and prognostic potentials for HNSCC.