Biology Seminar - Lacramioara Bintu

Work over the last few decades has identified ~1600 human transcription factors (TFs) involved in gene regulation and epigenetic memory and are at the core of many biological processes from development to immunity. While a few TFs have been studied in detail, we are currently missing a systematic understanding of how they work mechanistically: what are protein sequences that make good transcriptional activators or repressors and how do they enact their effects on gene expression? To answer these questions in a systematic manner, we have recently developed a high-throughput method for measuring the effect of 10,000-100,000 protein sequences in human cells. Using this approach, we were able to pinpoint the exact location and strength of hundreds of new activation and repressive domains across human TFs. By measuring mutants for all of them, we showed that the sequences necessary for activation consist of hydrophobic amino acids interspersed with either acidic, proline, serine, or glutamine residues. In order to understand the effects of particular activators once bound to DNA, we use single-molecule footprinting in live cells to measure the simultaneous occupancy of TFs, nucleosomes, and transcriptional machinery. By using these data in conjunction with a statistical mechanics approach, we show that activators effectively cooperate with each other by lowering nucleosome occupancy at nearby sites via recruitment of a chromatin remodeler. We also provide the first comprehensive categorization of repression domains based on the sequence necessary for function: in addition to hundreds of KRAB domains that recruit the major corepressor KAP1 and create heterochromatin at their target gene, we discovered that hundreds of TFs contain SUMOylation motifs and zinc fingers in the sequence necessary for repression. Finally, we discovered a new class of bifunctional effector domains that can dynamically switch from activation to repression over time, and are dissecting the parameters that govern this behavior.