DNA methylation regulator-mediated modification patterns and tumor microenvironment characterization in glioma

Abstract

Growing evidences indicate DNA methylation plays a crucial regulatory role in inflammation, innate immunity, and immunotherapy. However, the overall landscape of various DNA methylation regulatory genes and their relationship with the infiltration of immune cells into the tumor microenvironment (TME) as well as the response to immunotherapy in gliomas is still not clear. Therefore, we comprehensively analyzed the correlation between DNA methylation regulator patterns, infiltration of immune cell-types, and tumor immune response status in gather glioma cohorts. Furthermore, we calculated the DNA methylation score (DMS) for individual glioma samples, then evaluated the relationship between DMS, clinicopathological characteristics, and overall survival (OS) in patients with gliomas. Our results showed three distinct DNA methylation regulator patterns among the glioma patients which correlated with three distinct tumor immune response phenotypes, namely, immune-inflamed, immune-excluded, and immune desert. We then calculated DMS for individual glioma samples based on the expression of DNA methylation-related gene clusters. Furthermore, DMS, tumor mutation burden (TMB), programmed death 1 (PD-1) expression, immune cell infiltration status in the TME, and Tumor Immune Dysfunction and Exclusion (TIDE) scores were associated with survival outcomes and clinical responses to immune checkpoint blockade therapy. We also validated the predictive value of DMS in two independent immunotherapy cohorts. In conclusion, our results demonstrated that three DNA methylation regulator patterns that correlated with three tumor immune response phenotypes. Moreover, we demonstrated that DMS was an independent predictive biomarker that correlated with survival outcomes of glioma patients and their responses to immunotherapy therapeutic regimens.

Keywords: DNA methylation; glioma; immune phenotypes; immunotherapy; tumor mutation burden.

Figure 1

Multi-omics landscape of the DNA methylation regulators in glioma. (A) The summary of 20 DNA methylation regulators and their molecular functions in mediating the dynamic reversible process of DNA methylation. (B) Mutation frequency of the 20 DNA methylation regulators based on TCGA glioma dataset. Each column represents a single glioma samples. (C) The CNV frequency of DNA methylation regulators based on the TCGA glioma dataset. Note: gain, red; loss, blue. (D) PPI network of the 20 DNA methylation regulators. Size of the node denotes the number of proteins. (E) Circos plots illustrating the chromosomal locations of the CNV alternations in 20 DNA methylation regulatory genes. (F) Boxplot shows the expression levels of the 20 DNA methylation regulators in patients with LGG and GBM bases on the gather glioma cohort. Note: ns P > 0.05; ^*P < 0.05; ^**P < 0.01; ^***P < 0.001.

Figure 2

Characterization of distinct DNA methylation modification patterns in the gathered glioma cohort. (A) Consensus clustering matrix of the gather glioma cohort for k = 3. (B) Unsupervised clustering of 20 DNA methylation regulators in the gather glioma cohort. The glioma samples were annotated according to the DNA methylation regulator patterns, glioma grades, 1p19q codeletion status, and IDH status. (C) PCA confirmed three distinct patterns based on the expression of the 20 DNA methylation regulator in 2228 glioma samples. (D) Kaplan-Meier survival curve analysis showed the OS of glioma samples belonging to the three DNA methylation regulator patterns based on gather glioma cohorts. (E, F) GSVA analysis shows relatively enriched hallmark gene sets among the three patterns.

Figure 3

Different clinical and transcriptome characteristics of the three DNA methylation regulator patterns in the gather glioma cohort. (A) Heatmap of several immune signatures for the three DNA methylation regulator patterns in the gather glioma cohort. (B) Box-plots show the expression levels of few stroma-activated related genes in the three DNA methylation regulator patterns based on the gather glioma cohort. (C) Box-plots show the proportions of several immune cells types in the three patterns based on the gather glioma cohort. (D) Box-plots show the TMB for the three patterns in the TCGA dataset. (E) Functional annotation of the DNA methylation related genes between three patterns in the gather glioma cohort. Note: ns P > 0.05; ^*P < 0.05; ^**P < 0.01; ^***P < 0.001.

Figure 4

Construction of DMS in the gather glioma cohort. (A) Unsupervised clustering of the overlapping DNA methylation-related genes in the gather glioma cohorts. (B) Survival analysis of glioma patients belonging to the three DNA methylation-related gene clusters; P < 0.001. (C) The proportion of immune cell types in the glioma and the transcriptome traits in the three DNA methylation-related gene clusters. (D) The differences in the expression of genes related to the activated stromal pathways including EMT1, EMT2, EMT3, and pan-F-TBRS between three DNA methylation-related gene clusters. (E–G) Box-plots shows the DMS for DNA methylation regulator patterns (E), gene clusters (F), different glioma grades groups (G), P < 0.001. Note: ns P > 0.05; ^*P < 0.05; ^**P < 0.01; ^***P < 0.001.

Figure 5

Survival characteristics of glioma patients based on DMS and the relationship between DMS and tumor somatic mutation. (A–C) Survival analyses with the OS rates for the low DMS and high DMS groups among all glioma (A), GBM (B) and LGG (C) samples patients based on gather glioma cohorts, P < 0.001. (D) Sankey diagram shows the association between DNA methylation regulator patterns, glioma grades, DNA methylation-related gene cluster, and DMS groups. (E) Differences in DMS between patients with high and low TMB, P < 0.001. (F) Scatter plot shows the relationship between DMS and TMB in the glioma samples (R = 0.56; P < 0.001). (G–H) Waterfall plot shows tumor somatic mutations in high (G) and low (H) DMS subgroups. Note: ns P > 0.05; ^*P < 0.05; ^**P < 0.01; ^***P < 0.001.

Figure 6

The relationship between DMS and response to anti-PD-1/L1 immunotherapy. (A–F) Distribution of TIDE scores between high and low DMS subgroups in the gather glioma cohorts, as well as CGGA1, CGGA2, GSE180474, GSE16011 and TCGA datasets, respectively. (G–I) Survival analyses with the OS (G), clinical response to anti PD-1 immunotherapy (H), proportion of patients responding to PD-1 blockade immunotherapy (I), differences in PD-L1 expression (J) based on the low and high DMS subgroups in the IMvigor210 cohort. (K–N) Survival analyses with the OS (K), clinical response to anti PD-1 immunotherapy (L), proportion of patients responding to PD-1 blockade immunotherapy (M), differences in PD-L1 expression (N) based on the low and high DMS subgroups in the GSE78220 cohort. Note: ns P > 0.05; ^*P < 0.05; ^**P < 0.01; ^***P < 0.001.