Introduction
Primary brain and central nervous system (CNS) tumors have a mortality of 7% in patients <70 years old, and an estimated 270,000 non-malignant and 120,000 malignant tumors were reported between 2011 and 2015 in the USA (1,2). In 2017, ~80,000 new cases and 17,000 primary brain tumor-related mortalities were reported (3). Furthermore, one-third of the cases were diagnosed as malignant tumors (4). Glioma is a broad term that encompasses several types of malignant tumor derived from neuroepithelial cells, such as glial cells and other supporting cells of the CNS (1,5,6). Gliomas account for ~80% of primary brain tumors in adults and can be categorized as astrocytomas, oligodendrogliomas, ependymomas, mixed gliomas and other rare types, including brain stem and optic nerve gliomas (7). According to the World Health Organization criteria, gliomas vary histopathologically and may include benign ependymomas, as well as the most aggressive grade-IV glioblastoma (GBM) (5). Surgical removal is the primary therapy for intracranial tumors. However, tumors located in inoperable or sensitive regions of the brain pose clinical challenges (8). In addition, limited effectiveness of systemic chemotherapy can be attributed to the blood-brain barrier, which protects the brain from harmful compounds but also restricts the entry of chemotherapeutic drugs (9). The molecular mechanisms underlying the development of gliomas remain largely unknown. Therefore, elucidating the associated mechanisms may accelerate the development of targeted therapies for the disease.
The competing endogenous RNA (ceRNA) hypothesis, proposed by Salmena et al (10), has provided insight for RNA regulatory networks. According to this hypothesis, mRNA, long non-coding RNA (lncRNA), pseudogenes and other molecules can competitively bind to the same microRNA (miRNA/miR) response element and modulate miRNA function, forming an RNA regulatory network (10).
lncRNAs are RNAs >200 bp in length that do not encode proteins (11). Previous studies suggested that lncRNAs are a by-product of RNA transcription with biological functions that include modulation of the nervous system at the epigenetic, transcriptional and post-transcriptional levels (12–14). Additionally, lncRNAs serve as ceRNAs and participate in regulating gene expression and encoding miRNAs (10,15). Furthermore, lncRNAs were demonstrated to serve an important role in oncogenesis and tumor progression (16).
The present study constructed a ceRNA network in order to investigate and to identify potential biomarkers for glioma, and to determine the molecular mechanisms of glioma pathogenesis, using Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG). Additionally, the prognostic value of the identified mRNAs, miRNAs and lncRNAs was analyzed by survival analysis.
Discussion
Non-coding RNAs include several types of RNAs, and can be classified into lncRNAs and short non-coding RNAs. The latter can be further categorized into transfer RNAs, ribosomal RNAs, miRNAs, small interfering RNAs and snoRNAs (24). Previous studies demonstrated that lncRNAs played a key role in gene transcriptional control, epigenetic regulation and post-transcriptional modification (25,26). The present study analyzed RNA sequencing and clinical data from patients with gliomas acquired from TCGA in order to investigate the functions of lncRNAs in the ceRNA network. DElncRNAs, DEmRNAs and DEmiRNAs were identified using R packages and were used to construct a ceRNA network. The biological functions of the DEmRNAs were determined by GO and KEGG analyses. Further survival analysis was performed to investigate the association between DElncRNAs and OS.
miRNAs are involved in the regulation of a number of biological processes, including cell differentiation, proliferation and apoptosis, by binding with target mRNAs at the transcriptional or post-transcriptional levels (27). A previous study demonstrated that aberrantly expressed miRNAs were closely associated with several types of cancer (28). Moreover, growing evidence suggests that several miRNAs are implicated in the progression of glioma (29–31). hsa-mir-183 is frequently methylated and was correlated with poor outcome in hepatocellular carcinoma (32). Moreover, hsa-miR-183-5p was involved in cell cycle regulation via the MAPK and glioma signaling pathways (33). In addition, multivariative analysis revealed that low serum hsa-mir-137 levels were closely associated with high clinical grades and poor survival in patients with GBM (34). Moreover, hsa-miR-93-5p upregulation was correlated with poor prognosis in non-small cell lung cancer (35), while hsa-miR-93-3p was recognized as a diagnostic biomarker in triple negative breast cancer (36). Overexpression of hsa-mir-497 promoted the proliferation of glioma cells (37) and has been demonstrated to be an unfavorable prognostic biomarker in human glioma (38). In the present study, 6 DEmiRNAs, which may serve as promising prognostic biomarkers, were involved in the ceRNA network.
In the present study, 92 DEmRNAs were identified in the ceRNA network. The biological processes and pathways of the DEmRNAs were assessed by GO and KEGG pathway analyses. The most enriched GO biological processes included ‘protein binding’, ‘signal transducer activity’, ‘protein kinase activity’ and ‘regulation of cell cycle’. The pathway analysis revealed that several genes were involved in cancer-related pathways, such as ‘cell cycle’, ‘cellular senescence’, ‘Kaposi sarcoma-associated herpesvirus infection’, ‘relaxin signaling pathway’ and ‘human T-cell leukemia virus 1 infection’. Among the enriched pathways, ‘Kaposi sarcoma-associated herpesvirus infection’ has been reported to modulate the proliferation of glioma stem-like cells (39). Kaposi's sarcoma-associated herpesvirus exhibits neurotropism and its mRNA was detected in the plasma of patients with glioma in previous studies (39–41). Another previous study suggested that viral infection was a risk factor for glioma (42). The potential association between Kaposi's sarcoma-associated herpesvirus infection and glioma progression warrants further investigation. Among the DEmRNAs in the ceRNA network, tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein (YWHA)-H, E2F transcription factor (E2F) 5, E2F2, WEE1 G2 checkpoint kinase (WEE1), cell division cycle 25A (CDC25A) and checkpoint kinase 1 (CHEK1) were enriched in cell cycle-associated pathways. The present study demonstrated that YWHAH was significantly downregulated and associated with hsa-mir-424, hsa-mir-497 and hsa-mir-195 in the ceRNA network. A previous study demonstrated that YWHAH was involved in several cellular processes, including G1/S and G2/M cell cycle transition (43). Additionally, YWHAE, another YWHA isoform involved in cell cycle control and signal transduction (44,45), was associated with copy number aberrations in astrocytoma pathogenesis (46). E2F5 is a key transcription factor involved in cell cycle progression. In a previous study, E2F5 silencing inhibited the proliferation of GBM cells and induced cell cycle arrest (47). miR-218 inhibited the growth and metabolism of glioma cells by targeting E2F2 (48). WEE1, a regulator of the G2 checkpoint in GBM cells, was associated with malignancy and poor outcomes of patients with GBM (49). CDC25A and CHEK1 are significant regulators of the cell cycle and apoptosis in glioma cells (50).
The top 10 RNAs in the ceRNA network were identified using the cytohubba plugin. The network consisted of 6 miRNAs, including hsa-mir-93, hsa-mir-424, hsa-mir-497, hsa-mir-195, hsa-mir-182 and hsa-mir-96, and 4 lncRNAs, including AC092171.1, FAM87A, DLEU1 and LY86-AS1. LY86-AS1 and FAM87A were demonstrated to have a high degree of closeness with hsa-mir-424, hsa-mir-497 and hsa-mir-195. A previous study revealed that hsa-mir-497 demonstrated high sensitivity and specificity for the prediction of malignant astrocytomas (51). Moreover, overexpression of hsa-mir-497 promoted the proliferation of U87 glioma cells by targeting neuregulin receptor degradation protein 1 (37). Previous studies held conflicting views on the role of aberrantly expressed hsa-mir-195 in gliomas. For instance, hsa-mir-195 was demonstrated to inhibit the proliferation of human glioma cells by directly targeting cyclin D1 and cyclin E1 mRNA (52), and was positively correlated with the OS of patients with GBM (53). Additionally, Zhang et al (54) revealed that overexpression of hsa-mir-195 resulted in cell cycle arrest and significantly decreased invasion of GBM cell lines. By contrast, hsa-mir-195 promoted the proliferation and invasion of the human glioma cell line U87 via the transforming growth factor-β (TGF-β) signaling pathway (55). The present results are in agreement with those reported by Zhang et al (54) and suggested that hsa-mir-195 was upregulated in glioma tissues. The mechanisms underlying the interactions between hsa-mir-195, YWHAH and other potential mRNAs involved in cell cycle regulation require further investigation.
Protein-coding genes have long been recognized as conventional cancer biomarkers. Nevertheless, lncRNAs, which play a major role in the regulation of gene expression, have become the focus of predictive biomarker investigation due to their ability to better reflect tumor prognosis (11). Accumulating evidence suggests that dysregulation of lncRNA expression is associated with carcinogenesis and progression of glioma. Therefore, lncRNAs may serve as potential biomarkers for prognosis (56–58). For instance, colorectal neoplasia differentially expressed (CRNDE), the most upregulated lncRNA in glioma (59), facilitated cancer cell proliferation, migration and invasion by downregulating miR-384 expression. CRNDE knockdown combined with miR-384 overexpression significantly attenuated tumor progression in vivo (57). lncRNA activated by TGF-β (lncRNA-ATB) served as a ceRNA and a sponge for miR-200a, promoting cell proliferation and invasion in glioma. Upregulation of lncRNA-ATB was associated with worse OS in patients with glioma (60).
In the present study, 92 DElncRNAs were involved in the ceRNA network. Kaplan-Meier curve analysis indicated that 36 DElncRNAs were associated with OS. Therefore, these lncRNAs may serve as potential prognostic biomarkers in patients with glioma. Tian et al demonstrated that LINC00320, a tumor-suppressive lncRNA, was downregulated in glioma tissues and inhibited the proliferation of glioma cells by repressing the Wnt/β-catenin signaling pathway in vitro and in vivo (26). Moreover, Mills et al (61) demonstrated that full-length LINC00320 was only expressed in human brain tissue, suggesting that it may be a highly specific biomarker for glioma. Zhu et al (62) have demonstrated that the expression level of HAS2-AS1 was closely associated with lymph node metastasis and hypoxic tumor status in patients with oral squamous cell carcinoma. Notably, the expression level of HAS2-AS1 was negatively associated with OS in the present study. DLEU1 was identified to act as a sponge for miR-490 and contributed to the development of endometrial cancer (63). Furthermore, DLEU1 has been widely accepted as an oncogenic lncRNA and its aberrant upregulation was associated with the outcome of various types of cancer, including gastric (64), pancreatic (65) and colorectal cancer (66). Terashima et al (67) demonstrated that MEG8 was required for the epithelial-mesenchymal transition in lung and pancreatic cancer cells. DBH-AS1 was demonstrated to promote cell proliferation and survival through the MAPK signaling pathway in hepatocellular carcinoma cells (68). MIR600HG independently correlated with the outcome of patients with pancreatic cancer (69). In the present study, low MEG8 and DBH-AS1 expression, as well as high MIR600HG expression, were associated with an improved outcome. Therefore, further investigation is required to validate whether these lncRNAs may serve as potential predictive biomarkers for glioma.
However, there were some limitations in the present study. Indeed, the present study was designed as a bioinformatics analysis based on a small sample size, which also lacked experimental verification. Further experiment validation in larger samples is required to verify the present findings.
In conclusion, the present study constructed a specific ceRNA network consisting of lncRNAs, miRNAs and mRNAs, and the mechanisms of ceRNA regulation were further analyzed. Survival analysis indicated that lncRNAs may serve as potential biomarkers for predicting the outcomes of patients with glioma. To the best of the authors' knowledge, the present study was the first to attempt to construct a specific ceRNA network and to investigate the interactions between miRNAs and lncRNAs in glioma based on TCGA data.