Epigenomics and transcriptomics association study of blood pressure and incident diagnosis of hypertension in twins

Abstract

Hypertension is the most frequent health-related condition worldwide and is a primary risk factor for renal and cardiovascular diseases. However, the underlying molecular mechanisms are still poorly understood. To uncover these mechanisms, multi-omics studies have significant potential, but such studies are challenged by genetic and environmental confounding - an issue that can be effectively reduced by studying intra-pair differences in twins. Here, we coupled data on hypertension diagnoses from the nationwide Danish Patient Registry to a study population of 740 twins for whom genome-wide DNA methylation and gene expression data were available together with measurements of systolic and diastolic blood pressure. We investigated five phenotypes: incident hypertension cases, systolic blood pressure, diastolic blood pressure, hypertension (140/90 mmHg), and hypertension (130/80 mmHg). Statistical analyses were performed using Cox (incident cases) or linear (remaining) regression analyses at both the individual-level and twin pair-level. Significant genes (p < 0.05) at both levels and in both types of biological data were investigated by bioinformatic analyses, including gene set enrichment analysis and interaction network analysis. Overall, most of the identified pathways related to the immune system, particularly inflammation, and biology of vascular smooth muscle cell. Of specific genes, lysine methyltransferase 2 A (KMT2A) was found to be central for incident hypertension, ataxia-telangiectasia mutated (ATM) for systolic blood pressure, and beta-actin (ACTB) for diastolic blood pressure. Noteworthy, lysine methyltransferase 2A (KMT2A) was also identified in the systolic and diastolic blood pressure analyses. Here, we present novel biomarkers for hypertension. This study design is surprisingly rare in the field of hypertension. We identified biological pathways related to vascular smooth muscle cells and the immune system, particular inflammation, to be associated with hypertension and blood pressure. Of specific genes, we identified KMT2A (lysine methyltransferase 2A) to be central for blood pressure and hypertension development. ACTB beta-actin, ATM ataxiatelangiectasia mutated, BP blood pressure, EWAS epigenome-wide association studies, KMT2A lysine methyltransferase 2A, LMER linear mixed effect regression, LR linear regression, TWAS transcriptome-wide association studies.

Keywords: Hypertension; Immune system; KMT2A gene; Multi-Omics Association Study; Twins.

We identified biological pathways related to vascular smooth muscle cells and the immune system, particular inflammation, to be associated with hypertension and blood pressure. Of specific genes, we identified KMT2A (lysine methyltransferase 2A) to be central for blood pressure and hypertension development. ACTB beta-actin, ATM ataxiatelangiectasia mutated, BP blood pressure, EWAS epigenome-wide association studies, KMT2A lysine methyltransferase 2A, LMER linear mixed effect regression, LR linear regression, TWAS transcriptome-wide association studies.

Fig. 1

Study Design. We conducted gene set enrichment analyses (GSEA) with the aim to explore the biological pathways related to the five phenotypes in which the identified CpGs and probes take part. Before GSEA, the CpGs found in the EWAS with an FDR < 0.05, and the probes found in the TWAS with an FDR < 0.05 were identified in each statistical analysis. Next, for each of the statistical analyses, these CpG sites and probes were annotated to their gene names. As some different CpGs and probes might be annotated to the same gene, and as some CpGs might be annotated to multiple genes or different aliases, the unique genes were identified for the result of each statistical analysis. This resulted in 20 lists of gene names, i.e., five phenotypes times two analyses (individual-level and twin pair-level) times two kinds of molecular data. Afterwards, for each phenotype, the overlap in gene names of CpGs (EWAS) with FDR < 0.05 and the gene names of the probes (TWAS) with FDR < 0.05 were explored within each statistical analysis, for example the overlap in gene names between EWAS and TWAS in the incident hypertension Cox regression analysis at the individual-level corresponding to 1a in Fig. 1. This resulted in ten lists of EWAS-TWAS overlapping gene names (five phenotypes times two statistical models (an individual-level analysis and a twin pair-level analysis)). Next, for these ten lists, the overlap in EWAS-TWAS overlapping gene names from the individual-level analysis and, and EWAS-TWAS overlapping gene names from the twin pair analysis was found for each phenotype, corresponding to b in Fig. 1. This resulted in five overlaps of gene names identified in the individual-level analysis and confirmed in the twin pair level analysis, i.e., the genes confirmed when reducing the potential confounding due to shared genetic factors and early-life environment. Such genes must be considered the most relevant genes when exploring the association of molecular markers to hypertension. These five overlaps of confirmed genes (i.e. five phenotypes) were used for GSEA in the GSEA database with application of the Kyoto Encyclopedia of Genes and Genomes [41]. Moreover, genes present in some, or all, of the five overlaps of confirmed genes were explored. Finally, Interaction Networks were also conducted for the five overlaps of confirmed genes by application of StringApp in Cytoscape [42]. These steps were similarly done using a cut off p value < 0.05 instead of an FDR < 0.05 to explore the overlapping genes with p value < 0.05. Notes: BP is measured in mmHg. BP blood pressure, EWAS epigenome-wide association studies, LMER linear mixed effect regression, LR linear regression, TWAS transcriptome-wide association studies

Fig. 2

Interaction networks for Systolic Blood Pressure. Notes: green: the most connected genes, blue: the second-most connected gene, red: the third-most connected gene. Only genes having minimum 3 connections are displayed

Fig. 3

Interaction networks for Diastolic Blood Pressure. Notes: green: the most connected genes, blue: the second-most connected gene, red: the third-most connected gene. Only genes having minimum 3 connections are displayed

Fig. 4

Interaction networks for Incident Hypertension. Notes: green: the most connected genes, blue: the second-most connected gene, red: the third-most connected gene. Only genes connected to networks larger than 3 are displayed