Title: Single-molecule states link transcription factor affinity, chromatin occupancy, and gene expression in human cells.

Abstract: 

The binding of multiple transcription factors (TFs) to genomic enhancers drives gene expression in mammalian cells. However, the quantitative, causal relationship between changes in DNA sequence and changes in gene expression remains largely unsolved, making it difficult to engineer novel regulatory elements or to predict the impact of genetic variation in non-coding DNA. Even the molecular mechanisms behind simple observations, such as cooperativity in gene expression output with increasing numbers of TF binding sites, have been controversial and incomplete. Starting with this question, we applied single-molecule footprinting (SMF; Krebs 2017) to measure the simultaneous occupancy of TFs, nucleosomes, and other regulatory proteins on engineered enhancer-promoter constructs for both a synthetic TF and an endogenous TF involved in type I interferon response. We found that TFs cooperatively bind within chromatin by recruiting cofactors that destabilize nucleosomes. Once bound, TF occupancy linearly determines gene expression, enabling the decomposition of TF ‘strength’ into separable binding and activation terms. Finally, we develop thermodynamic and kinetic models that quantitatively predict enhancer binding microstates and gene expression dynamics.

The predictive power of these simple physical models is at odds with observations that in vitro TF binding measurements often poorly predict where and how TFs bind in cells. Motivated by these discrepancies, we paired high-throughput in vitro physical measurements with SMF in cells to dissect how KLF1, a human TF critical for the fetal-to-adult hemoglobin switch, interacts with DNA. Sequence context surrounding a motif drives over 40-fold variation in affinity in a manner consistent with distal flanking regions tuning TF search parameters. These context preferences are captured by a linear energy model and are validated in vivo both with SMF and a deep learning model trained on chromatin immunoprecipitation data, BPNet (Avsec 2021). We found that motif recognition, rather than the prevailing assumption of time in the bound state, drives variation in KLF1 affinity. With a novel strategy to measure TF dissociation from known motifs in cells, we identified minutes-long TF residence times consistent with in vitro measurements. Taken together, these studies provide a template for the quantitative dissection of multiple distinct contributors to gene expression, including TF binding affinity and kinetics, concentration, activation domains, binding site configurations, and recruitment of chromatin regulators.

Zoom: Please contact leyrec@stanford.edu for the zoom link.