Direct and indirect targets of the E2A-PBX1 leukemia-specific fusion protein

Abstract

E2A-PBX1 is expressed as a result of the t(1;19) chromosomal translocation in nearly 5% of cases of childhood acute lymphoblastic leukemia. The E2A-PBX1 chimeric transcription factor contains the N-terminal transactivation domain of E2A (TCF3) fused to the C-terminal DNA-binding homeodomain of PBX1. While there is no doubt of its oncogenic potential, the mechanisms of E2A-PBX1-mediated pre-B cell transformation and the nature of direct E2A-PBX1 target genes and pathways remain largely unknown. Herein we used chromatin immunoprecipitation assays (ChIP-chip) to identify direct targets of E2A-PBX1, and we used gene expression arrays of siRNA E2A-PBX1-silenced cells to evaluate changes in expression induced by the fusion protein. Combined ChIP-chip and expression data analysis gave rise to direct and functional targets of E2A-PBX1. Further we observe that the set of ChIP-chip identified E2A-PBX1 targets show a collective down-regulation trend in the E2A-PBX1 silenced samples compared to controls suggesting an activating role of this fusion transcription factor. Our data suggest that the expression of the E2A-PBX1 fusion gene interferes with key regulatory pathways and functions of hematopoietic biology. Among these are members of the WNT and apoptosis/cell cycle control pathways, and thus may comprise an essential driving force for the propagation and maintenance of the leukemic phenotype. These findings may also provide evidence of potentially attractive therapeutic targets.

Figure 1. E2A-PBX1 ChIP binding.

A. Peak promoter sequence analysis. Chromatin immunoprecipitation assays were performed to identify direct targets E2A-PBX1 using the human promoter array; HG18; (NCBI Build 36, Roche Nimblegen, Madison WI). A search was performed using the software WebMOTIFS and de novo motif finding a search for Transcription Factors that are likely to contact and regulate the binding sites identified by ChIP-chip. The 3 most significant transcription factor families (classes of transcription factors most likely to regulate the input sequences) are PBX, POU and STAT. B. Positional binding of E2A-PBX1. Among the 108 top hit ChIP genes, 76 genes have a single annotated transcription start site. The average intensity of ChIP pull-down is graphed along the promoter region in a smoothed plot. This histogram of the positions of maximal peak intensities (mean of 3 replicates) relative to the transcription start site (TSS) in target genes (Table S3). Most target genes show a maximum probe intensity around −500 bp upstream of the TSS. Probe regions with multiple TSS were excluded from this analysis. The promoter coverage area was complete for all genes between −2200 and +500 by the TSS (solid line) with some additional coverage for some genes outside of that region (dotted line).

Figure 2. Identified ChIP-chip targets on expression array in E2A-PBX1 silenced vs control cells.

A. ChIP-chip analysis data and expression analysis data upon E2A-PBX1 silencing were combined to get an overview of the direct and functional targets of E2A-PBX1. siRNA to E2A-PBX1 was employed to silence E2A-PBX1. The MA plot depicts the expression array changes upon silencing E2A-PBX1. Genes are ordered on the x-axis on the basis of their expression in untreated 697 cells. The Y-axis displays the change in expression upon E2A-PBX1 silencing. Those genes that are direct hits of E2A-PBX1 by chromatin immunoprecipitation assays (108 genes) were plotted against the expression array changes upon silencing and are shown in red color and marked with a blue dot; grey dots indicate all other genes assessed by the array. B. Here we compare the changes in expression of the 108 E2A-PBX1 direct targets to the expression change of all the other genes (non-direct E2A-PBX1 targets) upon E2A-PBX1 silencing. The set of identified ChIP-chip targets show collective down-regulation trend in the E2A-PBX1 silenced samples compared to controls (p<1.63e-06).

Figure 3. The 9p21 deletion is not always targeting CDKN2A.

Copy number alterations in a group of 22 t(1;19)⁺ E2A-PBX1 bone marrow samples are shown Figure 3A . Patient samples were subjected to hierarchical clustering analysis; we found the 9p21 deletion to exist in four of the 22 patients we have analyzed, with one patient exhibiting a homozygous deletion, white color refers to no changes, red to gain and blue to loss. Figure 3B . Fine map of deletions in the 9p21 region showing the deletion endpoints among 8 patients. The commonly deleted region among 10 patients (with 45%, including 7 patients in this figure and the other 3 others with complete arm loss) is the interferon gene cluster telomeric to the CDKN2A/B locus.

Figure 4. Combined KEGG Pathway analysis of direct and functional E2A-PBX1 targets.

The analysis was performed using Exploratory Gene Association Networks (EGAN) software tool. E2A-PBX1 direct (108 direct targets identified by ChIP-chip) and functional targets (122 significant differentially expressed genes between E2a-Pbx1 silenced samples and controls) and Pathways that might be regulated by them were visualized. Interfaces of direct and functional E2A-PBX1 targets are depicted. Magenta lines depict the connection of the genes to the direct target and/or functional target group; blue lines show the participation of the genes in KEGG pathways and brown lines show known interaction between genes connected.

Figure 5. Expanded combined analysis of direct and functional E2A-PBX1 targets.

The analysis was performed using the 102 direct targets identified by ChIP-chip analysis and the data set of the functional targets including the top 2000 genes up and down regulated after siRNA silencing of E2A-PBX1. Venn diagrams were generated and the interfaces of direct and functional E2A-PBX1 targets are depicted. 10 genes that were both direct targets and functionally down-regulated ( Figure 5A ) and 10 that were both direct targets and functionally up-regulated ( Figure 5B ) dependent on expression of E2A-PBX1 were identified. We are not able to address E2A-PBX1 as a functional repressor (for the 10 down regulated genes) or whether its depletion allowed accessibility to another transcriptional activator. Figure 5C Selected direct and functional E2A-PBX1 targets are highly expressed in primary E2A-PBX1 leukemias. Pediatric Cancer Genome Project. http://www.pediatriccancergenomeproject.org/site/accessed 14 Oct 2013. The expression of four genes UGT2B15, ASNS, WNT16 and HK2, that were among the direct and functional E2A-PBX1 targets in primary leukemia’s are shown. The expression of these genes was analyzed in Hyperdiploid leukemia’s (H), in E2A-PBX1 leukemia’s (E), TEL-AML1 (T) leukemia’s and BCR-ABL1 (B) leukemia’s are shown. The analysis of expression was performed using the gene expression tool and data available at St. Jude Children’s Research Hospital – (Washington University).