The role of the epigenome in pathogenesis and therapy of acute myeloid leukemia

Aberrant epigenetic patterning of the genome is emerging as a hallmark of acute myeloid leukemia, a malignancy characterized by a relative paucity of genetic lesions. AML can be classified into disease subtypes based on DNA methylation signatures. The most informative methylcytosine residues driving epigenetic clustering of AML patients are located in enhancer elements, positioned outside of CpG islands. Hence perturbation of enhancer function through aberrant cytosine methylation may play a key role in encoding the unique phenotypes and clinical outcomes of subsets of AML patients. We predicted that epigenetic signatures could be used as molecular blueprints that point towards causative mechanisms that drive leukemia phenotypes. Along these lines a subset of AMLs that share a particular DNA hypermethylation signature exhibit somatic mutations of the IDH1 and IDH2 genes, which generate the oncometabolite 2-hydroxyglutarate, which can inhibit aKG dependent dioxygenases including the TET family of enzymes that convert 5-methylcytosine into 5-hydroxymethlcytosine. Indeed patients with TET2 mutations share the aberrant DNA methylation profile of IDH mutant AML patients. We found that WT1, a transcriptional regulator often mutated in AML interacts with TET2. Notably, somatic mutations in IDH1, IDH2, TET2 and WT1 are almost universally mutually exclusive of each other in AML. Patients with all four of these mutations exhibit reduced global levels of 5hmC and in genome wide profiling studies display overlapping 5hmC distribution profiles. Collectively these data define a functional axis whereby WT1 mutation defines a subset of the effects emanating from TET2 loss of function, which in turn mediates a subset of the actions of 2HG generated by mutant IDH1 and IDH2. Although loss of TET2 does not cause AML in mice, leukemia does occur when TET2 deficient mice are crossed with FLT3ITD transgenic animals. Methylome analysis revealed that TET2 or FLT3-ITD alone had little effects on cytosine methylation alone. However both together resulted in a massive hypermethylation signature distinct from those affected by either mutation alone. Most of this perturbation consisted of DNA hypermethylation. Among the top methylated genes was GATA2, which was silenced in the double but not single mutant animals, as well as human TET2 + FLT3ITD AML patients. Restoration of GATA2 rescued the TET2-FLT3ITD leukemia phenotype, demonstrating that loss of GATA2 plays a critical role in the biological effect of the combined mutations. The inherent plasticity of the epigenome could potentially endow populations of AML cells with epigenetic heterogeneity and hence greater opportunities to encode chemotherapy resistant phenotypes. We developed an approach to identify epigenetic “alleles” that could be assessed to determine AML “epi-clonality”. Indeed we observed that epigenetic allele burden is linked to unfavorable clinical outcome in AML, independent of somatic mutations or other biomarkers. Genetic and epigenetic allelic diversity did not track together and appeared to be distinct biological phenomenon. AMLs could be classified into those with high epiallele burden with low somatic mutations, and a second group with low epiallele burden and higher somatic mutation burden. The presence of epialleles was accompanied by transcriptional deregulation as shown by single cell RNA-seq profiling. The presence of epigenetic allele diversity may thus enable AMLs to sample different transcriptional states.