Genome-Wide Analysis of NOTCH1, ETS Family Factors, and RUNX1 Binding in Human T Lymphoblastic Leukemia Cells Reveals Distinct Regulatory Elements

Abstract 1277

NOTCH1 regulates gene expression by forming transcription activation complexes with the DNA-binding factor RBPJ and gain-of-function NOTCH1 mutations are common in human and murine T lymphoblastic leukemia/lymphoma (T-LL). Via ChIP-seq studies of T-LL cells with constitutive Notch activation, we previously showed that NOTCH1/RBPJ binding sites in T-LL genomes are highly enriched for motifs corresponding to Ets factors and Runx factors. In this study, we determined the relationship of NOTCH1, RBPJ, ETS1, GABPA and RUNX1 binding sites in human T-LL cells by performing ChIP-Seq for each of these factors, as well as the chromatin marks H3K4me1, H3K4me3, and H3K27me3, and aligning the resulting sequences to human genome reference hg19 using programs available through Cistrome. Peak calling was performed with MACS2, and motif analysis was performed using SeqPos, which relies on JASPAR, TRANSFAC, Protein Binding Microarray (PBM), Yeast-1-hybrid (y1h), and human protein-DNA interaction (hPDI) databases to find known motifs and can also perform de novo motif discovery. Our analysis showed even more pervasive overlap of NOTCH1/RBPJ binding with ETS1/GABPA and RUNX1 factor binding than was predicted by motif analysis, in part due to binding of Ets factors and RUNX1 to non-canonical sequences. Heat-map analysis with K-means clustering on NOTCH1 binding regions identified three major classes of RBPJ/NOTCH1: class 1, characterized by high NOTCH/RBPJ signals, binding of the cofactors ZNF143, ETS1 and GABPA, high H3K4me3 signals, localization to promoters, and binding motifs for ZNF143; class 2, characterized by low NOTCH/RBPJ signals, binding of the cofactors ETS1, GABPA and RUNX1, high H3K4me3 signals, and Ets factor and CREB binding motifs; and class 3, characterized by high NOTCH/RBPJ signals, binding of RUNX1 and ETS1 cofactors, high H3K4me1 signals, intergenic localization (consistent with enhancers), and motifs for RUNX factors, ETS factors, and RBPJ. Of note, the nearest binding sites to the most responsive NOTCH1 target genes (defined as >2 fold stimulation when NOTCH1 was activated following release of gamma-secretase inhibitor (GSI) blockade by drug washout) were preferentially associated with Class 3 sites. Furthermore, shRNA knockdown of Ets factors and RUNX1 in T-LL cell lines induced apoptosis and reduced cell proliferation, implicating these factors in maintenance of T-LL growth and survival. Combination of knockdown of either Ets factors or RUNX1 with GSI treatment resulted in more severe phenotype in terms of apoptosis and cell growth compared to the knockdown or GSI treatment alone. In summary, our studies represent a step forward towards genome-wide understanding of how Notch works in concerts with other transcription factors to regulate the transcriptome of T-LL cells.