Epigenome-wide association study of leukocyte telomere length

Abstract

Telomere length is associated with age-related diseases and is highly heritable. It is unclear, however, to what extent epigenetic modifications are associated with leukocyte telomere length (LTL). In this study, we conducted a large-scale epigenome-wide association study (EWAS) of LTL using seven large cohorts (n=5,713) - the Framingham Heart Study, the Jackson Heart Study, the Women's Health Initiative, the Bogalusa Heart Study, the Lothian Birth Cohorts of 1921 and 1936, and the Longitudinal Study of Aging Danish Twins. Our stratified analysis suggests that EWAS findings for women of African ancestry may be distinct from those of three other groups: males of African ancestry, and males and females of European ancestry. Using a meta-analysis framework, we identified DNA methylation (DNAm) levels at 823 CpG sites to be significantly associated (P<1E-7) with LTL after adjusting for age, sex, ethnicity, and imputed white blood cell counts. Functional enrichment analyses revealed that these CpG sites are near genes that play a role in circadian rhythm, blood coagulation, and wound healing. Weighted correlation network analysis identified four co-methylation modules associated with LTL, age, and blood cell counts. Overall, this study reveals highly significant relationships between two hallmarks of aging: telomere biology and epigenetic changes.

Keywords: DNA methylation; leukocyte telomere length; multi-ancestry.

Figure 1

Analysis flow chart.

Figure 2

Regional Manhattan plots and inter-CpG correlations for the top four genes identified in the global meta-analysis. (A) VARS; (B) MAN2A2; (C) C7orf41 (MTURN); (D) ERGIC1.

Figure 3

EWAS Manhattan plots of the fully adjusted LTL.

Figure 4

Scatter plots between the group-specific meta-Z scores. (A) European male vs African male; (B) European male vs European female; (C) European male vs African female; (D) African male vs European female; (E) African male vs African female; (F) African female vs European female; The black dots in the panels refer to the top 30 CpG sites detected by the global meta-analysis, whereas the grey dots indicate the remaining CpG sites. Pearson correlation coefficients (red font) reveal strong agreement (r=0.4) between males and females of European ancestry.

Figure 5

Hierarchical clustering of CpG sites by weighted gene co-expression network analysis (WGCNA). Each data point on the x-axis of the dendrogram refers to an individual CpG site. The color band ‘Consensus module’ displays co-methylated modules (clusters) in different colors. The other color bands highlight the degree of correlations between DNA methylation of CpG sites and traits of interest. Red represents a positive correlation, whereas blue represents a negative correlation.

Figure 6

Heat map of correlations between the co-methylated module representatives and LTL, the partially adjusted LTL, the fully adjusted LTL, age, and blood cell counts. The numbers in the cells refer to meta-Z scores and their corresponding p-values. Meta-Z scores were calculated based on biweight midcorrelations between DNAm and a trait of interest in the six strata. ¹Partially adjusted LTL for age, sex and ethnicity. ²Fully adjusted LTL for age, sex, ethnicity, CD4+ naïve, CD8+ naïve and exhausted cytotoxic T cell.