Abstract
Methylation of DNA molecules is a fundamental mechanism for regulating gene function that is prevalent in all vertebrates. Although, the role of DNA methylation is well recognised in the disease context, the knowledge of variable DNA methylation patterns in healthy individuals is very limited. To our knowledge, there was no study available investigating inter-individual DNA methylation variation in healthy individuals in a homogenous cell population. Identification of epigenetic variation in a normal population is an important step towards understanding phenotypic variation, altered disease susceptibility and drug response. Therefore, we sought to generate reference methylomes from neutrophils of 11 healthy individuals and then identify and document inter-individual variation in DNA methylation between them.
We adopted the technique of Reduced Representation Bisulphite Sequencing (RRBS) to profile methylation pattern of the individuals. We optimised, improved and streamlined the process of RRBS library preparation. Next, a robust bioinformatic pipeline was established to analyse large-scale epigenomic data as part of the project, which could be efficiently applied to other large-scale epigenetic analysis. Several novel tools and new strategies were developed during the course of the study to facilitate genome-scale DNA methylation analysis. From the 11 individuals, 344 million sequenced reads were obtained and successfully aligned with the reference genome using our methylation pipeline.
From the high quality methylation information available from the fragments, we were able to detect 12851 autosomal Variably Methylated Fragments (iVMFs) associated with 6353 protein coding genes. Almost 51% of the variably methylated fragments were found to be present in gene bodies, i.e., exons, introns and exon-intron boundary and overall 64% of them were located within CpG islands and CpG island shores. Integrating our results with recently released ENCODE data revealed that variably methylated fragments strongly overlap with active regions of the genome. It was observed that the variably methylated fragments strongly overlap with RNA polymerase II, CTCF binding sites and showed enrichment of active histone marks, DNase hypersensitive regions and strong enhancers.
Further, upon ranking, we found 83 genes associated with 10 or more variably methylated fragments. Functional analysis of top candidate genes indicated that they disproportionately involve in neuron function, memory formation, and RNA metabolic processes. Pathway analysis indicated their involvement in melanogenesis, calcium signalling, neoplasms and carcinogenesis. Examples of the novel variably methylated genes include FBRSL1 (involved in would healing and a potential marker for alcohol-induced liver injury), STY10 (involved in immune dependent memory performance and circadian timing system), DLGAP2 (in organisation of synapse and neuronal cell signalling, implicated in autism spectrum disorders), CDH4 (a cell adhesion protein and putative tumour suppressor gene), SALL3 (regulation of DNA methylation, embryonic development and congenital disorders), TERT (involved in maintenance of telomere length).
This study provides the first genome-wide, base-pair resolution DNA methylation profiles in a homogenous, human cell type to document inter-individual variation in DNA methylation. The results from this work will serve as a resource for future studies aiming to understand the nature and mechanism of altered phenotypic traits and disease susceptibility due to variable DNA methylation in normal individuals.