Super-enhancer activation of a repressor: EVI1 in AML

This thesis explores EVI1 biology in AML: its upstream regulation, essential interaction partners and downstream targets.

In Chapter 2, we reviewed the mechanisms of aberrant gene activation by defective regulatory elements in cancer, taking EVI1 transcription by translocated enhancers as a paradigm for enhancer dysregulation. We also discuss ways to identify “novel” hijacked- or neo-enhancers as well as previously unrecognized transforming genes. Enhancer translocations, such as in acute myeloid leukemia (AML) with 3q26 aberrations, are strong drivers of oncogenesis and create potential vulnerabilities. In the review we also hint towards the idea that differential activation of an oncogene by “neo-enhancers’ may provide possibilities to pharmacologically intervene with altered transcription.

In the next two chapters, cofactor dependencies of enhancers activating EVI1 in AML were explored using inhibitors and degraders for P300 and CBP, two important histone acetyltransferases. In chapter 3, we demonstrate that the MYC-BENC SE sensitizes EVI1 to P300/CBP degradation in primary AML samples with a translocation t(3;8)(q26;q24), and in a K562 EVI1-eGFP t(3;8) model. The MYC-BENC enhancer is one of a subset of regions which lose accessibility upon degradation of P300/CBP. In a sample of an AML patient with a translocation t(3;8)(q26;q24), all enhancers which can translocate to EVI1 in the distinct variant 3q26 rearranged AMLs, exhibit dependency on P300/CBP. In contrast, none of the elements in the native MECOM locus are P300/CBP dependent. This indicates that the regulatory elements in the locus of MECOM itself are regulated differently from the ones that are hijacked. This is in line with our observation that in CD34+ donor derived HSPCs, EVI1 is not sensitive to P300/CBP degradation. Footprinting analysis suggests a role for GATA factors in mediating sensitivity to P300/CBP. Chapter 4 further explores P300/CBP dependency by studying individual domain inhibitors for the acetyltransferase- or bromodomain of P300 and CBP. We found that combining both inhibitors leads to a synergistic reduction of EVI1 transcription driven by MYC-BENC. We found that a subset of distal CREs (including MYC-BENC) loses P300 binding, and this is accompanied by loss of accessibility and H3K27ac. Again, a role for GATA factors is predicted with footprinting analyses. Preliminary experiments show the ability of both GATA1 and GATA2 to downregulate EVI1 levels. We propose experiments in which we further investigate the role of GATA factors and address the question whether they indeed play a role in p300/CBP driven activation and whether their binding to the genome is acetylation-dependent.

In Chapter 5 we investigated what the important interaction partners are of EVI1 in 3q26 rearranged AML. CTBP binds a 5-amino acid protein motif in EVI1 with the sequence PLDLS. These 5 amino acids are essential for EVI1 to cause leukemia using in vitro models, indicating that this interaction is essential for leukemic transformation by EVI1. Using the knowledge of this binding site, overexpression of PLDLS in the was studied. We found that overexpression of a PLDLS-repeat construct outcompetes EVI1-CTBP1/2 binding and abrogates leukemia cell growth in vitro and in vivo. Our next step will be to develop a cell-penetrating peptidomimetic “PLDLS compound”, that can replace the expression construct that we used and is able to interfere EVI1 to CTBP interaction in primary AML cells.

In chapter 6, we carried out experiments to uncover the downstream targets of EVI1 using an auxin-inducible degron model. The key advantage of a degron model is the rapid depletion of protein levels, which enabled us to probe the cells where EVI1 had been depleted within 1.5 hrs. After depletion of EVI1, we found only upregulated genes, suggesting EVI1 acts as a repressor. Among the upregulated genes was CEBPA, a master regulator of myeloid differentiation. This gene is repressed by EVI1, probably via a long-distant enhancer of CEBPA located 42kb upstream (+42kb) of the gene. After depletion of EVI1, the CEBPA +42kb enhancer gains accessibility and shows increased H3K27Ac levels. In addition, CEBPA target genes are upregulated over prolonged EVI1 depletion; and CEBPA binding motifs are predicted to gain binding. They are also enriched in regions which gain occupancy of P300, RUNX1 and H3K27ac. This suggests that after EVI1 is depleted, CEBPA takes over and drives myeloid development. Indeed, CEBPA overexpression in presence of EVI1 is able to induce differentiation. Furthermore, when the +42kb CEBPA enhancer is deleted (-/+), differentiation is not induced upon EVI1 knock-out. We also show that overexpression of 4xPLDLS (Chapter 5) leads to a increase of CEBPA levels, indicating CEBPA repression is partially CTBP-dependent. CEBPA expression will be an excellent read-out to further unravel the mechanism of repression by EVI1 and will help to identify molecules able to interfere with EVI1 function in 3q26 rearranged AMLs.