Mutation in Wilms' Tumor 1 Induces DNA Hypermethylation of PRC2 Targets, Blocks Myelomonocytic Differentiation, and Defines a Novel Subtype of AML Responsive to EZH2 Inhibition

Wilms’ Tumor 1 is a transcription factor found to be recurrently mutated (WT1mut) in 10% of normal karyotype acute myeloid leukemia (NK-AML), predominantly in young adults with intermediate-poor prognosis, often in association with FLT3-ITD and high white cell count. Mutations are usually heterozygous and consist of small insertions or deletions clustered around exons 7 and 9, which encode zinc-finger DNA binding domains. In general, mutations disrupt these DNA binding domains leading to a truncated protein that may act in a dominant negative fashion. Currently, the mechanism by which WT1mut contributes to leukemogenesis is unknown, and no lead drug targets linked to this mutation have been identified to date. Using a novel computational method based on Boolean implications that link the presence of a somatic mutation to CpG methylation on a site-by-site basis, we found that mutation in WT1 is strongly linked to DNA hypermethylation in AML patient samples. In order to validate these findings, we expressed mutant WT1 protein prematurely truncated at exon 7 in THP-1 AML cells (confirmed to be wildtype for WT1 at both alleles), and after 10 passages we measured DNA methylation by 450K bead-chip arrays. We found consistent upregulation of DNA methylation in mutant but not wildtype WT1-expressing cells when compared to parental THP-1 cells, validating WT1 mutation as an active driver of DNA hypermethylation. Additional methylome analysis of human hematopoietic stem and progenitor compartments (HSPC) including (HSC, MPP, L-MPP, CMP, and GMP) indicated that WT1mut induces predominantly de novo DNA methylation, as virtually all CpG sites induced by the mutant protein are unmethylated in normal HSPC. Strikingly, the pattern of methylation in both WT1mut patient samples and WT1mut-THP-1 cells was enriched for polycomb repressor complex 2 (PRC2) target genes (p<1.6E-87), implicating a role for this repressive chromatin-remodelling complex in WT1mut leukemogenesis. In keeping with this, gene expression analysis of WT1mut AMLs (but not other normal karyotype AMLs) showed marked repression of known hematopoietic PRC2 target genes (as defined by Chip-Seq), suggesting WT1mut may induce a differentiation block through deregulation and hypermethylation of PRC2 targets.

To explore this possibility, we expressed WT1mut in purified normal cord blood CD34+ HSPC using lentiviral transduction and performed in vitro liquid culture differentiation assays in IL-3, SCF, FLT3L, and GM-CSF. We found that WT1mut (but not wildtype or empty vector) induced a myelomonocytic differentiation block with fewer cells expressing CD11b, CD11c, and CD14. Separately, we also showed that WT1mut induced a differentiation block in a TF-1 cell model of erythroid differentiation. These findings suggest a role for WT1mut in perturbing myeloid differentiation in early HSPC.

To test the therapeutic implications of our findings, we asked whether inhibition of the major enzymatic histone trimethylase component of PRC2, EZH2, could reverse the differentiation block caused by WT1mut. Significantly, we found that treatment of primary WT1mut AML blasts with the selective EZH2 inhibitor GSK-126 induced upregulation of the mature myeloid markers CD11b, CD33, and CD14. In contrast, NK-AML without WT1mut or acute promyleocytic leukemia cells did not show a significant differentiation response.

Our results indicate that mutation in WT1 defines a novel subgroup of DNA hypermethyated AML with de novo hypermethylation of PRC2 target genes that may clinically respond to selective EZH2 inhibitors through differentiation. Importantly, our methods show that genome-wide analysis of mutation-specific DNA methylation patterns may have a future role in determining epigenetic therapies for personalized medicine.