AbstractsBiology & Animal Science

DNA sequence and chromatin landscape regulate genomic binding of the glucocorticoid receptor

by Stephan Raphael Starick

Institution: Freie Universität Berlin
Department: FB Biologie, Chemie, Pharmazie
Degree: PhD
Year: 2015
Record ID: 1112244
Full text PDF: http://edocs.fu-berlin.de/diss/receive/FUDISS_thesis_000000099220


Organisms have evolved cell types for different functions. Transcriptional regulation of subsets of genes by binding of transcription factors (TFs) to DNA sequences within regulatory regions is essential to manifest cellular phenotypes. Examination of genomic binding sites of the glucocorticoid receptor (GR), a hormone inducible TF, showed that only a small percentage of many possible GR binding sequences in the genome is actually GR-bound. Furthermore, only a fraction of the GR-bound regions appeared to contain the classical GR consensus sequence, indicating that this sequence is neither necessary nor sufficient to explain GR binding. This raises two questions: First, what features of the chromatin landscape discriminate bound from unbound GR binding sequences and second, what sequences, other than the classical binding sequence, can recruit GR to the genome? Density and organization of DNA-enwound nucleosomes divides the genome’s chromatin landscape into TF-accessible “open“ and “closed“ regions. Our GR chro- matin immunoprecipitation (ChIP) showed a predominant GR-binding to “open“ chromatin, with one distinctive feature: A specific depletion of GR-binding at pro- moter regions of the “open“ chromatin universe. This depletion could be explained to some extend by a reduced frequency of GR’s canonical motif-matching sequences in these regions. However, sequence composition of promoter regions explains only part of the depletion and our Hierarchical Bayes Modeling of GR-binding, using chromatin marks as model features, indicated additional candidate mechanisms. For example, we found typical promoter marks (e.g. H3K9ac) to be negatively correlated with GR-binding in the “open“ chromatin universe. These acetylation marks are set by histone acetyltransferases that can also post-translationally modify GR and attenuate its interaction with DNA (Kino & Chrousos 2011) thereby providing a possible explanation for the promoter-proximal depletion observed. As we found that a large fraction of GR-bound regions lack a canonical GR binding sequence, we asked what sequences recruit GR to individual loci? To identify these sequences at high resolution, we used ChIP-Exo, taking advantage of an exonuclease trimming of ChIP fragments to the protein:DNA cross-linking point which protects the DNA from further digestion. We exploited this signal by determining footprint profiles of TF binding at single base pair resolution, using ExoProfiler, a computational motif-based pipeline. Comparison of our and the few public available ChIP-Exo datasets revealed: footprints are protein and recognition-sequence-specific signatures of TF binding sites that allow to distinguish direct and indirect (tethering to other DNA-bound proteins) GR binding and captures information about TFs other than the one directly targeted by the antibody. We show that the absence of classical recognition sequences can be explained, in part, by direct GR binding to degenerate canonical GR binding sequences. Furthermore, my study identified a new mode of DNA binding,…