N6-methyladenosine (m6A)-connected lncRNAs are linked to survival and immune infiltration in glioma patients

Abstract

Background: The connection between m6A-assiociateed lncRNAs and prognosis has been demonstrated in multiple types of tumors. However, potential roles of m6A-assiociateed lncRNAs in glioma is still rare.

Methods: We implemented consensus cluster analysis to group the downloaded samples into two subtypes. The least absolute shrinkage and selection operator (LASSO) analysis was used to create a risk model. Additionally, the conjunction between m6A-related lncRNAs and immune cells infiltration was explored by conducting the R package. Ultimately, we inspected the underlying downstream pathways of the two subtypes by performing Gene Set Enrichment Analysis (GSEA). The expression level of m6A-connected lncRNAs in glioma were examined by conducting in vitro experiments.

Results: We ascertained two subtypes of glioma in line with the consensus clustering of m6A-associated lncRNAs. We confirmed that age, grade, and IDH are related to the two subtypes. Additionally, the immune cells infiltration and immune checkpoint molecules of the two clusters were discussed. A risk signature including AL359643.3, AL445524.1, AL162231.2, AL117332.1, AP001486.2, POLR2J4, AC120036.4, LINC00641, LINC00900, CRNDE, and AL158212.3, was identified using the Cox regression and LASSO analyses. We also verified the prognostic value and discussed the immune cells infiltration and immune checkpoint molecules of the risk signature. In vitro experiments verified that the m6A-associated lncRNAs was abnormally expressed in glioma.

Conclusion: We elaborated the significant role of m6A-connected lncRNAs in glioma prognosis and immune infiltration and suggest that these key regulators may serve as underlying therapeutic targets to build up the efficacy of glioma immunotherapy.

Keywords: Glioma; LncRNA; M6A regulators; immune infiltration; prognosis.

Figure 1. Data analysis flowchart

(A) Data sources. (B) m6A genes and their related lncRNAs. (C) Consensus Cluster. (D) Immune checkpoint and infiltration. (E) LASSO Model. (F) Train and Test groups. (G) Prognostic risk scores and qPCR.

Figure 2. The expression traits and interactions of m6A-associated genes in glioma

(A) Waterfall blot displayed the mutation of 23 m6A-related genes. (B) Heatmap exhibited the overall expression of 14 m6A-conencted genes in LGG and GBM from TCGA datasets. (C) The connection between m6A-related genes (red) and the related IncRNAs (yellow). (D) Univariate Cox regression analysis for the features of 14 m6A-associated lncRNAs. (E) Heatmap showed the overall expression of 14 m6A-connected lncRNAs in LGG and GBM from TCGA datasets. (F) Differential expression of the m6A-connected lncRNAs in LGG and GBM was displayed by boxplot. P<0.05 (*), P<0.01 (**), and P<0.001 (***).

Figure 3. Conjunction between the m6A methylated lncRNAs and prognostic and clinical characteristics of glioma

(A) Consensus clustering model with CDF for k = 2–9 (k represents cluster count). (B) For k = 2–9, relative change in area under the CDF curve. (C) Glioma TCGA cohort was categorized as two clusters with k = 2. (D) The overall survival of glioma patients in cluster 1 and cluster 2 was figured out by Kaplan–Meier analysis. (E) Heatmap presented the correlation of cluster1 and cluster 2 with clinical features; P<0.001 (***).

Figure 4. Immune cell infiltration and TME score in cluster 1 and cluster 2 in TCGA

(A) Infiltration levels of 22 immune cell types in the two clusters in TCGA. (B–D) The stromal (B), immune (C), estimate (D) scores in the two clusters in TCGA.

Figure 5. ICPGs differential expression in cluster 1 and cluster 2 in TCGA dataset

(A) Expression levels of 47 ICPGs in two clusters in TCGA. (B–G) The expression levels of CD28 (B), CD80 (C), CD86 (D), CD274 (E), CTLA4 (F), and PDCD1 (G) in the two cluster in TCGA.

Figure 6. Establishment and verification of prognostic signatures of m6A-connected lncRNAs in TCGA

(A,B) LASSO regression analysis for quantifying the minimum criteria. (C,D) Heatmap revealed the allocation of eleven prognostic signatures in train (C) and test (D) subtypes. (E,F) Kaplan-Meier analysis of OS for glioma patients in line with the risk score in train (E) and test (F) subtypes. (G,H) Distribution of risk score, OS, and OS status of the eleven prognostic biomarkers in train (G) and test (H) subtypes. (I,J) ROC curves reflecting the predictive ability of the risk score in train (I) and test (J) subtypes.

Figure 7. Prognostic risk scores associated with clinical characteristics and immune score in TCGA train sutype

(A) Heatmap and clinical characteristics of high-risk and low-risk cohorts; P<0.001 (***). (B–I) The risk scores distributed by gender (B), age (C), grade (D), IDH (E), MGMT (F), 1p/19q (G), cluster l/2 (H), and immune-score (I).

Figure 8. Correlations between risk score and infiltration abundances of 15 immune cell types

(A–O) T cells CD8 (A), T cells CD4 memory activated (B), T cells follicular helper (C), T cells gamma delta (D), Macrophages M0 (E), Macrophages M1 (F), Macrophages M2 (G), Neutrophils (H), T cells regulatory (Tregs) (I), Mast cells activated (J), Dendritic cells activated (K), Eosinophils (L), T cells CD4 memory resting (M), NK cells activated (N), and Monocytes (O).

Figure 9. Validation of the expression levels of m6A-associated lncRNAs between normal brain tissues, LGG, GBM tissues by qRT-PCR

(A–D) The expression levels of AL117332.1 (A), AL359643.3 (B), AL445524.1 (C), CRNDE (D) were much higher in the LGG and GBM tissues when compared to normal brain tissues. (E,F) The expression levels of LINC00641 (E) and AP001486.2 (F) in normal brain tissues were much higher than in the LGG and GBM tissues.