DNA methylation regulator-mediated modification patterns and risk of intracranial aneurysm: a multi-omics and epigenome-wide association study integrating machine learning, Mendelian randomization, eQTL and mQTL data

Abstract

Background: Intracranial aneurysms (IAs) pose a significant and intricate challenge. Elucidating the interplay between DNA methylation and IA pathogenesis is paramount to identify potential biomarkers and therapeutic interventions.

Methods: We employed a comprehensive bioinformatics investigation of DNA methylation in IA, utilizing a transcriptomics-based methodology that encompassed 100 machine learning algorithms, genome-wide association studies (GWAS), Mendelian randomization (MR), and summary-data-based Mendelian randomization (SMR). Our sophisticated analytical strategy allowed for a systematic assessment of differentially methylated genes and their implications on the onset, progression, and rupture of IA.

Results: We identified DNA methylation-related genes (MRGs) and associated molecular pathways, and the MR and SMR analyses provided evidence for potential causal links between the observed DNA methylation events and IA predisposition.

Conclusion: These insights not only augment our understanding of the molecular underpinnings of IA but also underscore potential novel biomarkers and therapeutic avenues. Although our study faces inherent limitations and hurdles, it represents a groundbreaking initiative in deciphering the intricate relationship between genetic, epigenetic, and environmental factors implicated in IA pathogenesis.

Keywords: DNA methylation regulator; Genome-wide association studies; Intracranial aneurysms; Machine learning; Mendelian randomization; Multi-omics.

Fig. 1

Genetic variations and expression of MRGs in IA tissue samples. A Localization of 19 MRGs on the 23 chromosome. B PPI analysis of 19 MRGs. C At the tissue level, principal component analysis separates normal (grey), RIA (yellow) and UIA samples (blue). Heat map (D) and box plot (E) showing 19 MRGs differentially expressed in normal, RIA and UIA tissues. *p < 0.05; **p < 0.01; ***p < 0.001; ns no statistical significance

Fig. 2

Genetic variations and expression of MRGs in IA blood samples. A PCA for combined expression profile before and after ComBat. B At the blood level, principal component analysis separates normal (grey), RIA (yellow) and UIA samples (blue). Heat map (C) and box plot (D) showing 19 MRGs differentially expressed in normal, RIA and UIA blood. *p < 0.05; **p < 0.01; ***p < 0.001; ns no statistical significance

Fig. 3

Consensus clustering results in IA cohorts. A Consensus clustering matrix of k = 2 as the optimal cluster number. B CDF curves of the consensus score from k = 2–9. C PCA principal component analysis of two clusters. Each subgroup was distinguished by different colors. Heat map (D) and box plot (E) showing 19 MRGs differentially expressed in A and B cluster. *p < 0.05; **p < 0.01; ***p < 0.001; ns no statistical significance

Fig. 4

Biological function in two clusters. A Heatmap of matrix of KEGG enrichment scores using GSVA algorithm. B Heatmap of matrix of GO enrichment scores using GSVA algorithm. C Volcano map showed DEGs between two patterns. D KEGG enrichment analysis of DEGs. *p < 0.05; **p < 0.01; ***p < 0.001; ns no statistical significance

Fig. 5

Single-cell RNA-sequencing analysis identifies Aneurysm and Health cell marker genes. A T-SNE plots show cells from Aneurysm and Health samples. B The cell types identified by marker genes. C The proportion Plot The cell types identified by marker genes. D, E The UMAP and boxplot plots represent 7 cell clusters from Aneurysm and Health samples

Fig. 6

The role of DNA methylation in IA and normal cells. A Difference in activation of HALLMARK pathways between IA and normal cells. B The enrichment scores of different HALLMARK signal pathways in normal and IA cells of each Intracranial Aneurysm sample. C Difference in expression of DNA methylation-related genes between IA and normal cells. D The expression of DNA methylation-related genes in normal and IA cells of each Intracranial aneurysm sample

Fig. 7

Construction and testing of the Methylation-related genes (MRGs) riskscore. A The AUC value of 100 machine-learning algorithm combinations in the three testing cohorts. ROC curves (B), decision curves (C) and calibration curves (D) were used in MRGs to identify RIA and uIA in the GSE122897 cohort, GSE36791 + GSE159610 cohort and GSE13353 + GSE54083 + GSE75436 cohort. E Diagnostic value of 17 diagnostic genes in MRGs

Fig. 8

Identification of pathways that the 17 risk genes involved in. A Gene-pathway correlation heatmap; B Enrichment score heatmap for key pathways. *p < 0.05; **p < 0.01; ***p < 0.001; ns no statistical significance

Fig. 9

The difference in immune infiltration among patients in RIA and UIA. A Heat map showing differences in immune infiltrating cells between RIA and UIA. B Heat map showing molecular differences in immunomodulators between RIA and UIA. C, D The box plot illustrated the absolute abundance scores of the 16 immune cells and 13 immune function components in UIA and RIA. *p < 0.05; **p < 0.01; ***p < 0.001; ns no statistical significance

Fig. 10

A, B Manhattan plot shows GWAS results. The y-axis indicates the Z-score for each gene tested on all autosomal and single nucleotide polymorphism weight sets. The x-axis indicates the chromosomal position corresponding to the gene, and the black line indicates the threshold of significance. C–E Pleiotropic association of DNMT3A with SAH/UIA (C, D) and MBD2 with UIA (E) using genome-wide cis-eQTLs. Top plot, grey dots represent the – log10 (p values) for SNPs from the GWAS of SAH/UIA, with solid rhombuses indicating that the probes pass HEIDI test. Middle plot, eQTL results. Bottom plot, location of genes tagged by the probes. Effect estimates IA, SAH, and UIA (F, N). Investigation of the association of a genetically determined unit increase in exposure with the risk of IA/SAH/UIA using inverse-variance weighted, MR Egger, and weighted median estimates. F, I, L Scatter plots of individual SNP effects and estimates from different MR techniques for the effect of DNA methylation related-genes on IA/SAH/UIA. G, J, M, Funnel plots of DNA methylation related-genes on IA/SAH/UIA. H, K, N Leave-one-out analysis plots for DNA methylation related-genes on IA/SAH/UIA. eQTL expression quantitative trait loci, GWAS genome–wide association studies, HEIDI heterogeneity in dependent instruments, SMR summary data–based Mendelian randomization, SNP single nucleotide polymorphism, SAH subarachnoid hemorrhage, IA Intracranial aneurysm