Oncogenesis by E2A-PBX1 in ALL: RUNX and more

The mechanisms by which fusion transcription factors stimulate cancer pathogenesis are being elucidated with increasingly powerful molecular techniques. In this issue of Blood, Pi et al¹  performed a genome-wide identification of the targets of the E2A-PBX1 fusion protein associated with t(1;19), found in 4% of cases of childhood B-cell acute lymphocytic leukemia (ALL).² 

This translocation links E2A, a basic helix loop helix transcription factor that controls B-cell development to the PBX1 homeodomain protein. In the fusion, E2A transactivation domains that recruit p300³  and other coactivators are linked to the regions of PBX1 protein that mediate DNA binding to an AT-rich sequence.⁴  Early studies correlated the ability of the E2A-PBX1 fusion to bind to its cognate site and activate reporter genes with the oncogenic potential of the fusion protein in heterologous systems such as NIH 3T3 cells or transduction of the oncoprotein into murine marrow.⁵  More recent experiments in a more biologically relevant system showed that transduction of E2A-PBX1 into fetal liver progenitors blocked B-cell differentiation, whereas a point mutant in the PBX1 homeodomain was highly defective for this function.⁶  Prior chromatin precipitation-microarray studies identified E2A-PBX1 target genes, which were enriched for PBX1 DNA recognition motifs,⁷  and putative direct targets were identified, including WNT1 and WNT16. These studies supported a simple model that the chimeric transcription factor inappropriately activated target genes through the PBX1 DNA-binding motif of the fusion.

The genome-wide chromatin precipitation-sequencing (ChIP-Seq) study by Pi et al leads to a major refinement of this model. Expression of an epitope-tagged version of E2A or E2A-PBX1 in t(1;19)⁺ 697 cells followed by ChIP-Seq for E2A, the epitope-tagged E2A-PBX1 fusion, and PBX1 distinguished genes bound by E2A-PBX1 from genes bound by E2A. Analysis of the sequences of the genomic regions bound by E2A-PBX1 showed that PBX1 motifs were enriched 1.5-fold above background. Surprisingly, however, the most enriched DNA motif (>2.5-fold) was that of the RUNX1 transcription factor. RUNX1 is a transcriptional activator possessing a RUNT domain that complexes with DNA in combination with the cofactor CBFβ. RUNX1 is essential for normal hematopoietic development, is rearranged to create fusion transcription factors in acute myeloid leukemia (AML) and ALL, and is sporadically mutated in poor prognosis AML.⁸  In the Broad cell-line encyclopedia, RUNX1 levels are highest in T and B-ALL cells, and RUNX1 depletion inhibits growth of T-ALL lines,⁹  suggesting a role of the protein in leukemogenesis. Pi et al showed that RUNX1 and E2A-PBX1 form a complex in vivo, and RUNX1 recruited E2A-PBX1 to specific loci. Such regions displayed increased histone H3 lysine 27 acetylation, due to the recruitment of the p300 histone acetyltransferase through the E2A moiety. Knockdown of E2A-PBX1 by short hairpin RNAs (shRNAs) directed against E2A and PBX1 led to decreased expression of 210 genes, 46 of which were also suppressed by depletion of RUNX1. Among these were the previously identified WNT16 and notably RUNX1 itself, suggesting a positive feedback loop and oncogenic function. Accordingly, shRNA depletion of RUNX1 dramatically decreased the oncogenic activity of E2A-PBX1. Checking these 46 genes against the Broad Institute cancer dependency database (www.DepMap.org), CCDN3 encoding cyclin D3, RUNX1, and CLECL14 encoding a C-lectin are cell type–specific dependencies for t(1;19) 697 B-ALL cells. Furthermore, WNT16 can now be identified as a target activated via RUNX1. Collectively, these data indicate that E2A-PBX1 has a major effect on tumorigenesis through regulation of sets of genes directly bound by RUNX1. Which of these genes are most critical for transformation remains to be determined.

Although this work identifies a major oncogenic axis of E2A-PBX1, molecular and biological evidence still indicates that the full oncogenic function of E2A-PBX1 requires activation of PBX1 target genes. Pi et al identified >1200 E2A-PBX1 binding sites that were not bound by RUNX1 and >160 genes activated by E2A-PBX1 but not affected by RUNX1, including some identified as critical for 697 cell growth in DepMap, such as LEF1, whose expression has been associated with poorer outcome in B-cell ALL, IKZF1, encoding the Ikaros transcription factor, which is critical to B-cell development, and IL7R, encoding the receptor for interleukin-7 (IL-7), a critical B-cell growth factor. Furthermore, E2A-PBX1 activated, in a non-RUNX-dependent manner, BMI1 critical for E2A-PBX1–mediated transformation through its ability to repress the INK4A tumor suppressor locus,¹⁰  as well as ETV5, which is also important for transformation of B cells. Whether these are all direct E2A-PBX1 targets will require more detailed analysis of the ChIP-Seq data, but a targeted examination of these data shows that E2A-PBX1 but not RUNX1 binds the promoter region of IZKF1.

Structure-function studies from Pi et al and prior investigators indicated that the full transformation activity E2A-PBX1 requires the transactivation domains of E2A, responsible for recruiting p300,³  the DNA binding domain of PBX1, 5 amino acids C-terminal to the PBX1 homeodomain that allows interaction with various HOX homeodomain proteins, and a region N terminal to the PBX1 homeodomain that allows self-association of the protein. The requirement of the PBX1 DNA binding domain appears to be twofold: it directly binds and activates a specific set of genes and additionally mediates interaction with RUNX1, allowing recruitment to RUNX binding sites. Although an E2A-PBX1 construct completely devoid of the PBX1 DNA binding domain has no transformation activity and fails to activate activation through PBX sites⁵  or RUNX1 sites, Pi et al showed that a point mutant that prevents binding to PBX sites can stimulate growth and replating of murine hematopoietic progenitors, although at reduced efficiency. Thus, the full effects of the oncoprotein are mediated through the direct and indirect activation mechanisms (see figure).

Although t(1;19) ALL has a favorable 5-year prognosis of ∼90% survival, it is associated with an increased risk of central nervous system relapse, motivating the search for molecular mechanisms and targets. Along these lines, E2A-PBX1, collectively through both mechanisms, activated expression of genes implicated in the IL-12, IL-7, WNT, AKT, NOTCH, and neural development pathways, suggesting new avenues for therapeutic investigation.