Mathematical Biology Seminar

Jason Gertz
Department of Oncological Sciences, University of Utah

Genomic features that influence cell type-specific transcription factor binding

Wednesday, October 29, 2014, at 3:05pm
LCB 219

While many possible binding sites are present in a genome, most transcription factors bind a very small subset of these sites. To study how transcription factors choose their binding sites, we focused on estrogen receptor α (ER), a transcription factor that drives proliferation in both breast epithelial and endometrial cells. Because ER's DNA binding is induced by estradiol, studying cells that are starved for estradiol provides the opportunity to measure the genomic context that a transcription factor encounters just prior to binding. With an integrative functional genomics approach that included ChIP-seq, RNA-seq, DNase-seq and DNA methylation data in breast and endometrial cells, we found that ER has two distinct modes of binding. One set of ER binding sites, which we term "shared", is bound in multiple cell types, depends on very high affinity estrogen response elements (EREs) and appears to be agnostic to chromatin accessibility, being most often found in chromatin that is closed prior to estradiol treatment. A second set of ER binding sites, which we term "cell-specific", differs between cell types, is characterized by lack of high affinity EREs and is accompanied by cell type-specific open chromatin and differential DNA methylation. The co-occupancy of interacting transcription factors occurs at many of the cell-specific ER binding sites, but is infrequently present at shared sites. Thermodynamic models based on genome sequence and chromatin accessibility support a tethering mechanism in which ER does not directly bind DNA at one-third of cell-specific sites. The distinct properties of ER binding sites provide an elegant and flexible genomic encoding scheme for creating estrogen responsive enhancers. The integration of multiple functional genomics datasets allowed us to uncover the mechanisms of transcription factor binding site selection and quantitatively model how ER chooses its binding sites.