Expressions and functions of long non-coding RNAs in glioma

Abstract

Gliomas are the most common primary brain tumors in adults. Despite the development of multimodal and aggressive treatments that include surgical resection, radiotherapy and systemic chemotherapy in the past decades, patient outcomes remain unsatisfactory. To improve treatment efficacy, a better understanding of glioma pathogenesis at the molecular level is urgently needed.

Long non-coding RNAs (lncRNAs) are transcripts with length greater than 200 nucleotides and without protein-coding function. They are emerging as new regulators in cancer biology by acting as tumor suppressors or oncogenes. They may even be employed as biomarkers for cancer diagnosis and prognostication. However, the global expression patterns and functional roles of lncRNAs in the majority of cancer types, including glioma, remain largely unknown. The hypothesis of this project was that lncRNAs were aberrantly expressed in glioma, and may act as biomarkers or regulators in glioma.

With this hypothesis, the study set four objectives: 1) establish a bioinformatics mining method to obtain lncRNA expression profiles from the existing glioma microarray gene expression data; 2) explore the correlations between lncRNA expression signatures and clinical phenotypes of glioma (histological subtypes and malignancy grades); 3) explore the prognostic values of lncRNAs in glioma; and 4) characterize the functional roles of a representative lncRNA candidate-CRNDE, in regulating glioma growth in vitro and in vivo.

To achieve these goals, the study first developed a bioinformatics classification pipeline to identify the lncRNAs that were represented on the Affymetrix HG-U133 Plus 2.0 array, which is one of the most commonly used commercial microarrays in human cancer profiling. This identified 2448 probe sets (corresponding to 1970 lncRNA genes) that were represented on this array platform. By using this mining approach, lncRNA profiling was then performed in two large cohorts of public glioma microarray gene expression datasets (containing 268 and 157 clinical specimens, respectively). Sets of lncRNAs that were unique to different glioma histological subtypes and malignancy grades were identified and verified. Moreover, by analyzing the associations between lncRNA expression signatures and patient prognoses, the study identified a six-lncRNA signature (KIAA0495, PART1, MGC21881, MIAT, GAS5, PAR5) that was significantly associated with the overall survival of glioblastoma multiforme (GBM) patients. Further analysis revealed that the prognostic value of the lncRNA signature was independent of the known prognostic factors, i.e., age and O(6)-methylguanine-DNA methyltransferase (MGMT) promoter methylation status. The functional study for one representative lncRNA, CRNDE, showed that CRNDE knock-down dramatically inhibited glioma cell growth both in vitro and in vivo, and was associated with decreased BCL2/Bax ratio and induced cell apoptosis, suggesting the oncogenic role of CRNDE in glioma. This study established for the first time a readily available lncRNA profiling method, and provided an important platform for future study. The associations of lncRNA signatures with glioma clinical phenotypes and patient outcomes suggest that lncRNAs may play important roles in the pathogenesis of glioma, and that lncRNAs may act as potential biomarkers for glioma diagnosis and prognostication. The functional roles of CRNDE in regulating glioma growth indicate its potential of being employed as a therapeutic target.