Bioinformatic analysis of epigenetic effects, particularly in DNA methylation, following different interventions

Abstract

Epigenetics is defined as heritable changes in gene expression without a change in the underlying DNA sequence. In this thesis I concentrate on DNA methylation and the changes that occur in response to different conditions; more particularly, I develop methods to analyse methylation data and associated transcriptional and chromatin changes and apply
this to four different projects.
The first project focused on the effects of shRNA mediated DNMT1 depletion within immortalised human fibroblasts. Here we found four key classes of genes dependent on DNA methylation; protocadherins, genes involved in fat homeostasis, olfactory receptors and cancer testis antigens. In addition to an interplay with polycomb repressive complexes at certain loci. Within this project, I developed tools to examine complex loci and correlate
methylation with chromatin marks.
In the second project, we sought to carry out a similar experiment, but this time investigated the effects of UHRF1 depletion within the same cell line, as UHRF1 is known to
recruit DNMT1 to hemi-methylated DNA. Here we found depletion of UHRF1 caused demethylation and upregulation of endogenous retroviruses and a subsequent innate immune response. When the cells were rescued methylation did not recover but the innate
immune response and expression of retroviral elements was attenuated. However, rescued cells were hypersensitive to SETDB1 and KAP1 inhibition, implicating H3K9me3 in the UHRF1-mediated repression in absence of DNA methylation. UHRF1 cell lines which were mutated to affect the H3K9me3 binding domain could not repress endogenous retroviral expression, confirming the involvement of H3K9me3 here. Here, I aided in the analysis of
methylation array data for knockdown, rescue and mutant cell lines and developed a tool to analyse repeat elements covered by the Illumina Human Methylation 450k BeadChip and
MethylationEPIC arrays.
In the third project, we sought to investigate the effects of folic acid supplementation in the second and third trimester on the methylation of the offspring. Folate is a limiting factor of
one carbon metabolism and as a result, DNA synthesis and DNA methylation. Following intervention, cord blood was examined using the EPIC array and we discovered a folate sensitive differentially methylated region upstream of the imprint regulator ZFP57 and verified the change in an independent cohort and within in vitro models. In this project, I helped to develop statistical models with the initial and downstream bioinformatic analysis of methylation arrays and refined a tool for the investigation of target loci from methylation array data.
In project 4, we investigated the effects of mental illness on the methylation patterns of first year university students. We observed enrichment for genes involved in the immune response and the inflammatory skin condition psoriasis, with notable hypermethylation at the late cornified envelope gene cluster involved in skin cell differentiation. Results were confirmed via wet lab approaches and validated in part in an independent cohort, adding an immune component to the aetiology of depression. In this study, I aided with the initial and downstream bioinformatic analysis of methylation arrays, including taking advantage of their ability to score copy number variation.
Finally, in project 5, I formalised the tools I had used in project 1-4 into a complete pipeline called CandiMeth (available at www.bit.do/candimeth) which can be used by people with
little bioinformatics training to investigate DNA methylation at candidate genomic features. This pipeline is user-friendly, has no installation requirements and runs freely off the Galaxy framework (www.usegalaxy.org) to allow users to reproducibly quantify and visualise methylation differences among their samples and how these results correlate with different
genomic features, such as repetitive elements.
Overall, in this thesis I have developed novel approaches to analysing methylation data and applied these to a range of projects, culminating in the development of a user-friendly
methylation array analysis tool called CandiMeth.

Date of Award	Jul 2020
Original language	English
Sponsors	Department for the Economy
Supervisor	Shu-Dong Zhang (Supervisor), Kristina Pentieva (Supervisor) & Colum Walsh (Supervisor)