Autophagy involves in the evolution process of eukaryotic cells which are highly conservative, and functions in the degradation almost in all cells from yeast to mammals. Autophagy disorders are associated with many diseases. LncRNA as a novel autophagy-regulating molecule acts on gene expression at transcriptional and post-transcriptional levels. It acts as a scaffold for recruitment of chromatin modifiers in transcriptional regulation, which induces epigenetic silencing and participates in regulating histone modification, DNA methylation and chromosome remodeling, enhancing gene transcription, and regulating gene transcription by altering the subcellular localization of transcription factors. For example, lncRNA 2810403D21Rik/Mirf as a regulator of autophagy functions via the miR-26a-USP15 axis, to inhibit autophagy and induce myocardial injury by reducing intrinsic cardioprotective activity. While silencing lncRNA 2810403D21Rik/Mirf can reduce the H2O2-induced myocardial injury, which can be reversed under miR-26a or 3-MA knockout by AMO-26a to inhibit autophagy in NMCM. In the post-transcriptional level, lncRNA as a precursor of small RNA assists in mRNA stabilizing or degrading in a form of small RNA regulating mRNA processing and stability. As previous studies shown, lncRNA HOTAIRM1 functions on autophagy and tumor protein PML-RARA degradation during bone marrow cell differentiation arrest, and down-regulated HOTAIRM1 can inhibit the PML-RARA degradation induced by all-trans retinoic acid in acute promyelocytic leukemia (APL) cells and then block the differentiation from early promyelocytes to granulocytes.
Most studies on lncRNA focus on specific genes involved in autophagy, while autophagy-related lncRNA prognostic in LGG patients is a less studied field. We here established an autophagy-related lncRNA signature in LGG which was prognostic. In our research, co-expression of ATGs and lncRNAs was firstly analyzed to screen autophagy-related lncRNAs, followed by univariate, Lasso, and multivariate Cox analyses to identify 9 prognostic autophagy-related lncRNAs, including AL136964.1, ARHGEF26-AS1, PCED1B-AS1, AC104072.1, PRKCQ-AS1, LINC00957, AC125616.1, PSMB8-AS1, and AC087741.1. These 9 lncRNAs may be prognostic markers and therapeutic targets for LGG patients. Among the 9 genes, 5 of them (ARHGEF26-AS1, PCED1B-AS1, PRKCQ-AS1, LINC00957, and PSMB8-AS1) have been proven of cancer relevant. (1) ARHGEF26-AS1 (ARHGEF26 antisense RNA 1) as the antisense RNA of ARHGEF26 is reported to show a significant correlation with OS of colon cancer (P < 0.05). In addition, ARHGEF26-AS1 acts as a central gene in the ceRNA network to indicate the prognosis of intrahepatic cholangiocarcinoma and colon cancer. (2) PCED1B-AS1 presents profoundly up-regulated expression in clear cell renal cell carcinoma (ccRCC) tissues and cell lines, which indicates poor prognosis of patients. Besides, it can mediate ccRCC cell proliferative and migratory abilities as well as epithelial-mesenchymal transition (EMT) process, and promote the expression of ZEB1 by inhibiting miR-484. Recent studies have shown that PCED1B-AS1 induces immunosuppression and increases the expression and function of PD-Ls in liver cancer as a sponge for miR-194-5p. It also promotes the proliferation of transplanted tumor cell in nude mice, along with increased colony formation and tumor formation in vivo, and inhibited cell apoptosis. PCED1B-AS1 can bind HIF-1α mRNA 5'-UTR in a direct way and to enhance HIF-1α translation, resulting in the increase of HIF-1α proteins, thus advancing the Warburg effect and tumorigenesis of glioblastoma. In addition, PCED1B-AS1 can activate glioma cells to proliferate while inhibiting apoptosis with miR-194-5p/PCED1B axis. Other studies found that PCED1B-AS1 modulates macrophage apoptosis and autophagy in active tuberculosis via targeting miR-155. (3) LncRNA PRKCQ-AS1 is proven prognostic in colorectal cancer (CRC) patients as a risk factor. It is concentrated in the cytoplasm of CRC cells and reversely regulates miR-1287-5p expression, which indicates adverse outcomes. Recent studies have also shown that PRKCQ-AS1, as an immune-related lncRNA, provides a reliable prognostic indicator for glioblastoma patients. (4) LINC00957 is significantly related to poor OS of CRC patients upon increased expression (P < 0.05), and inhibiting LINC00957 expression can reverse 5-FU resistance via P-gP down regulation. (5) PSMB8-AS1 is involved in pancreatic cancer progression with miR-382-3p/STAT1/PD-L1 axis, while knockout of PSMB8-AS1 leads to weakened cell proliferation, migration, invasion and EMT, while increased cell apoptosis. In addition, PSMB8-AS1 as a ceRNA for miR-22-3p regulates DDIT4 in glioblastoma. Contrarily, over expression of DDIT4 could antagonize the effects of PSMB8-AS1 on proliferation, apoptosis and radioresistance of glioblastoma cells. Other studies showed that the activation of PSMB8-AS1 by transcription factor ELK1 advances cell proliferation of glioma by regulating miR-574-5p/RAB10, while silencing PSMB8-AS1 inhibits the proliferation. For the other 4 autophagy-related lncRNAs (AL136964.1, AC104072.1, AC125616.1 and AC087741.1), there is no report on their prognostic role in cancer. It is necessary to conduct more research to identify the mechanism behind the effect of these lncRNAs on the prognosis of LGG individuals via autophagy.
The lncRNA signature we identified could be used to effectively predict the survival of patients suffering from LGG. According to the expression of the 9 signature lncRNAs, each LGG patient was conferred a Riskscore, based on which two risk cohorts were generated (cutoff: median Riskscore). As analyzed, LGG samples with high Riskscores had poorer survival compared to those with low Riskscores (P < 0.05). ROC analysis verified the prognostic accuracy of the lncRNA signature in LGG patients. Besides, the signature-based Riskscore was identified independently prognostic in multivariate analysis (P < 0.05), and presented more reliable prognostic performance relative to other traditional clinical indicators. Nomogram, C-index, calibration curve and DCA curve are effective and reliable clinical tools for survival prediction of cancer patients. Here, we established a robust nomogram involving multiple clinical variables (age, gender, diagnoses, grade, radiation_therapy, seizure_history, type, IDH1_mutation_status) and the signature-based Riskscore which helps improve the prognosis prediction of LGG patients. Patients of a higher age ( > = 40 years) or higher tumor grade generally show poor cancer prognosis, which is consistent with our results. Moreover, the calibration curve displayed a similar result between the actual and predicted 3-year and 5-year survival rates based on the nomogram. GSEA results suggested that the 9 autophagy-related lncRNAs may negatively regulate autophagy and cancer-related signaling pathways, and play a vital part in the pathogenesis of LGG.
However, there remain some limitations that need to be concerned. Firstly, data analyzed here were from TCGA, GEO, and CGGA databases, which requires a further validation in other prospective cohorts to ensure the robustness of the signature we identified. Secondly, the potential molecular correlation between autophagy-related lncRNAs and autophagy need further study. Finally, the role and mechanism behind these autophagy-related lncRNAs in LGG call for a further investigation.
In conculsion, the prognostic signature involving autophagy-related lncRNAs we identified here was accurate in survival prediction of LGG sufferers, which show great clinical application potential in individualized prognosis and treatment. Remarkable associations were revealed of the 9 signature lncRNAs with survival of LGG cases, and a high performance of the signature-based Riskscore in distinguishing patients at different risks was uncovered, suggesting the independent role of the signature. The autophagy-related lncRNAs and their signature, therefore, are potential molecular biomarkers and therapeutic targets of LGG.