Dynamic regulation of DNA methylation in human T-cell biology

To this end, we employed large-scale profiling techniques combining gene expression as well as genome-wide 5mC and 5hmC measurements to construct a time-series model of epigenetic regulation of differentiation. This revealed that early T-cell activation was accompanied by extensive genome-wide deposition of 5hmC which resulted in demethylation upon proliferation. Early DNA methylation remodelling through 5hmC was not only indicative of demethylation events during T-cell differentiation but also marked changes persisting longterm in memory T-cell subsets. These results suggest that priming of epigenetic landscapes in T-cells is initiated during early activation events, preceding any establishment of a stable lineage, which are then maintained throughout the cells lifespan. The regions undergoing remodelling were also highly enriched for genetic variants in autoimmune diseases which we show to be functional through disruption of protein binding. These variants could potentially disrupt gene-regulatory networks and the establishment of epigenetic priming, highlighting the complex interplay between genetic and epigenetic layers. In the course of this work, we discovered that a commonly used technique to study genome-wide DNA modifications, DNA immunoprecipitation (DIP)-seq, had a false discovery rate between 50-99% depending on the modification and cell type being assayed. This represented inherent technical errors related to the use of antibodies resulting in off-target binding of repetitive sequences lacking any DNA modifications. These sequences are common in mammalian genomes making robust detection of rare DNA modifications very difficult due to the high background signals. However, offtarget binding could easily be controlled for using a non-specific antibody control which greatly improved data quality and biological insight of the data. Although future studies are advised to use alternative methods where available, error correction is an acceptable alternative which will help fuel new discoveries through the removal of extensive background signals.

Taken together, this thesis shows how integrative use of high-resolution epigenomic data can be used to study complex biological systems over time as well as how these techniques can be systematically characterised to identify and correct errors resulting in improved detection.