Preview

PDF, English Download (13MB) | Terms of use

Abstract

Acute myeloid leukemia (AML) is a hematological cancer characterised by a block in differentiation and accelerated proliferation of myeloid progenitor cells. Epigenetic regulators are among the most frequent targets for mutations and structural variations in AML, and the disruption of these genes can result in profound epigenetic heterogeneity between and within tumors. Deletion 5q [del(5q)] is the most common copy number alteration (CNA) in older AML patients and is associated with poor clinical outcome and therapy resistance, however the mechanisms linking del(5q) to leukemic development and progression are not understood. I began this thesis with an analysis of DNA methylation profiles from 477 elderly AML patients using a DNA methylome deconvolution approach. Here I discovered that del(5q) AML constitutes an epigenetically distinct subgroup characterized by a unique signature of DNA hypermethylation. In an attempt to pinpoint the epigenetic disturbance leading this signature to arise, I investigated the 5q Minimally Deleted Region (MDR) for potential epigenetic regulators, and identified the H3K9me1/2 demethylase KDM3B as a promising target. Precise mapping of the MDR, together with differential transcriptional, protein and mutational analysis of 5q genes strengthened the argument that KDM3B is the most likely candidate for haploinsufficiency in del(5q) AML. I further linked the del(5q) methylation signature to dysregulation of other H3K9me1/2 regulators, and consistent overexpression of the de novo DNA methyltransferase and leukemic stem cell marker, DNMT3B. Moreover, I discovered that del(5q) and MECOM -overexpressing leukemias share a common DNA methylation signature, which in both subgroups coincides with increased expression of DNMT3B. These findings suggest that del(5q) AML deserves to be reappraised as an epigenetically-defined subgroup, which I suggest may be driven by haploinsufficiency of KDM3B. Reduction in protein levels of KDM3B should result in an increase in H3K9me1/2. In addition, I hypothesized that haploinsufficiency of this enzyme may result in an imbalanced removal of H3K9me1/2, such that variable patterns of these histone marks may arise between cells. Such cell-to-cell epigenomic heterogeneity could provide a powerful driving force for leukemic progression by allowing selection of favorable phenotypes throughout cancer evolution and in response to therapy. To study this phenomenon, I developed a heterogeneity metric called epiCHAOS (epigenetic/Chromatin Heterogeneity Assessment Of Single cells). EpiCHAOS is the first tool that enables quantitative comparisons of epigenetic heterogeneity between single-cell groups/clusters within a biological sample. I validated epiCHAOS in silico and demonstrated its functionality by applying the metric to a range of biological datasets from developmental systems, cancers, and aging to investigate both genome-wide and region-specific differences in epigenetic heterogeneity. Finally, to investigate the epigenetic consequences of KDM3B disruption in AML, I analyzed single-cell assay for transposase-accessible chromatin with sequencing (ATAC-seq) data generated from KDM3B-heterozygous OCI-AML3 cell lines, which were established to mimic haploinsufficiency of the enzyme. Heterozygous deletion of KDM3B resulted in the expected global chromatin compaction as well as epigenetic heterogeneity at H3K9me1/2- associated regions. This thesis provides two important contributions for the research community. First, my findings shed light on the mechanisms driving one of the most aggressive forms of AML, which until now has not been studied from an epigenetic perspective, and where KDM3B has received very little attention as a putative target gene. Secondly, I provide the first computational strategy for quantitative single-cell analysis of epigenomic heterogeneity, which should offer a useful tool for biologists, especially those interested in stemness, plasticity and mechanisms of therapy resistance in cancer.