Genome-wide analysis of DNA methylation in buccal cells: a study of monozygotic twins and mQTLs

Jenny Van Dongen,Gonneke Willemsen,Dorret I Boomsma,Rick Jansen,Erik A Ehli,Meike Bartels,Gareth E Davies,Bastiaan T Heijmans,Noah A Kallsen,Catharina E M Van Beijsterveldt,Charles E Breeze,Shanna A Peyton,Cornelis Kluft,Jouke J Hottenga

doi:10.1186/s13072-018-0225-x

Abstract

BackgroundDNA methylation arrays are widely used in epigenome-wide association studies and methylation quantitative trait locus (mQTL) studies. Here, we performed the first genome-wide analysis of monozygotic (MZ) twin correlations and mQTLs on data obtained with the Illumina MethylationEPIC BeadChip (EPIC array) and compared the performance of the EPIC array to the Illumina HumanMethylation450 BeadChip (HM450 array) for buccal-derived DNA.ResultsGood-quality EPIC data were obtained for 102 buccal-derived DNA samples from 49 MZ twin pairs (mean age = 7.5 years, range = 1–10). Differences between MZ twins in the cellular content of buccal swabs were a major driver for differences in their DNA methylation profiles, highlighting the importance to adjust for cellular composition in DNA methylation studies of buccal-derived DNA. After adjusting for cellular composition, the genome-wide mean correlation (r) between MZ twins was 0.21 for the EPIC array, and cis mQTL analysis in 84 twins identified 1,296,323 significant associations (FDR 5%), encompassing 33,749 methylation sites and 616,029 genetic variants. MZ twin correlations were slightly larger (p < 2.2 × 10−16) for novel EPIC probes (N = 383,066, mean r = 0.22) compared to probes that are also present on HM450 (N = 406,822, mean r = 0.20). In line with this observation, a larger percentage of novel EPIC probes was associated with genetic variants (novel EPIC probes with significant mQTL 4.7%, HM450 probes with mQTL 3.9%, p < 2.2 × 10−16). Methylation sites with a large MZ correlation and sites associated with mQTLs were most strongly enriched in epithelial cell DNase I hypersensitive sites (DHSs), enhancers, and histone mark H3K4me3.ConclusionsWe conclude that the contribution of familial factors to individual differences in DNA methylation and the effect of mQTLs are larger for novel EPIC probes, especially those within regulatory elements connected to active regions specific to the investigated tissue.

Highlights

DNA methylation arrays are widely used in epigenome-wide association studies and methylation quantitative trait locus studies
The genome-wide average MZ twin correlation of DNA methylation level obtained with the EPIC array was 0.21, and MZ twin correlations were slightly larger for the novel EPIC probes compared to the probes that are common to EPIC and HM450
When we analyzed the distribution of methylation sites with a large correlation between MZ twins (r > 0.5) across cell-type specific regulatory elements, we found the strongest enrichment in epithelial cell enhancers, DNase I hypersensitive site (DHS) and Histone H3 lysine trimethylation (H3K4me3); the histone mark associated with transcriptional start sites of actively transcribed genes

Summary

Introduction

DNA methylation arrays are widely used in epigenome-wide association studies and methylation quantitative trait locus (mQTL) studies. We performed the first genome-wide analysis of monozygotic (MZ) twin correlations and mQTLs on data obtained with the Illumina MethylationEPIC BeadChip (EPIC array) and compared the performance of the EPIC array to the Illumina HumanMethylation450 BeadChip (HM450 array) for buccal-derived DNA. Several validation studies of the EPIC array have been published that assessed the reproducibility of the EPIC array, compared the performance of the EPIC array to the HM450 array, or compared the performance of the EPIC array to wholegenome bisulfite sequencing (WGBS) [2,3,4,5] These studies have reported high correlations (r > 0.9, across all CpGs) between replicate samples on EPIC and between matched samples measured on HM450 and EPIC (r > 0.9, across all overlapping CpGs). No study has been published on EPIC data generated with DNA derived from buccal swabs, which may be used as a surrogate tissue in epigenome-wide association studies of human traits and in studies of genetic variants that influence DNA methylation

Methods

Results

Discussion

Conclusion