Gene regulation by RUNX1 in the absence of consensus sequences

Runt-related transcription factor 1 (RUNX1) is a transcription factor that has an important role in haematopoietic cell development and function and is frequently disrupted in leukaemia. RUNX1 is commonly described as a sequence-specific DNA binding factor which recognises the consensus sequence TG(T/C)GGT in the promoter and enhancer regions of its target genes to affect changes in gene expression. However, the advent of techniques to study DNA-protein interactions on a genome-wide scale has provided the opportunity to re-assess RUNX1 localisation and function, and this analysis suggests that the classical model of RUNX1 function is incomplete. By fully understanding the mechanisms by which RUNX1 maintains its target gene expression profiles under normal cellular conditions, insights into disrupted function can be gained and interventions can be developed. Analysis of publicly available RUNX1 ChIP-Seq data determined that the majority of RUNX1 binding in haematopoietic cells occurs outside of gene promoter regions, in intergenic or intronic regions. Furthermore, approximately one fifth of all RUNX1-DNA binding sites on a genome-wide scale were not associated with a canonical consensus sequence, and this was particularly prevalent in promoter regions, with almost half of RUNX1 binding in promoter regions occurring in the absence of consensus sequences. This suggests that recruitment of RUNX1 to gene targets occurs through multiple mechanisms and raises the possibility that it may function differently depending on its location and mode of recruitment. Similar results were obtained for localisation of the RUNX1 fusion protein RUNX1-ETO, the product of a common chromosomal translocation in leukaemia. This data set was used to investigate the different modes of binding and action of RUNX1 with the aim of establishing whether binding in the absence of consensus sequences constitutes a novel mechanism of RUNX1 binding and if genes regulated in this way respond differently to RUNX1 disruption. This study identified biological pathways enriched for genes with RUNX1 binding in their promoters, and in which only one type of binding (in the presence or absence of a consensus sequence) occurred. Clusters of functionally related RUNX1-bound genes were selected from two of these pathways; the RhoA and HMGB1 signalling pathways and used as models to investigate regulation by RUNX1. The RhoA signalling pathway genes, ARHGAP1, ARHGAP4 and ARHGAP12, all bound RUNX1 at their promoters, in association with a consensus sequence. However, while ARHGAP4 and ARHGAP12 responded to RUNX1 in reporter assays, ARHGAP1 did not, suggesting that a consensus sequence alone is not sufficient for a promoter to respond to RUNX1. In contrast, a group of histone acetyl transferase genes from the HMGB1 signalling pathway bound RUNX1 at their promoters in the absence of a consensus sequence. Both the KAT6B and KAT2B promoters were activated by RUNX1 in the absence of a consensus sequence, suggesting a mechanism whereby RUNX1 recruitment to DNA does not require the canonical consequence sequence and may rely on recruitment by additional transcription factors or distal chromatin elements. Interestingly, evidence presented here indicates that while RUNX1-ETO inhibits RUNX1 activity at classically regulated promoters which contain RUNX consensus sequences, it cannot inhibit RUNX1 activity in the absence of canonical consensus sequences. This suggests that genes regulated by RUNX1 through different transcriptional mechanisms may respond differently to disruption of RUNX1 in diseases such as leukaemia. The majority of RUNX1 binding occurs outside of promoters, with one potential explanation for intergenic or intronic RUNX1 binding being that it functions at gene enhancers. Data presented here identified two such potential enhancers of the HAT1 gene. While HAT1 bound RUNX1 at its promoter, although in the absence of a consensus sequence, it does not respond directly to RUNX1 in reporter assays. Potential enhancers were identified approximately 31 kb upstream and 190 kb downstream of the promoter. While interactions with these regions were confirmed, neither were found to behave as an enhancer in reporter assays. However, the data presented is consistent with the +190 kb region representing a promoter-promoter interaction which may be related to the formation of large interconnected transcriptional complexes. Such promoter-promoter interactions could at least partly explain the prevalence of RUNX1 binding at promoters in the absence of a consensus sequence. This study has expanded our understanding of the mechanisms by which RUNX1 affects transcription of its target genes and has characterised distinct modes of operation at candidate genes: either directly through a consensus sequence; directly in the absence of a consensus sequence; or distally through an interacting regulatory element.