Integrating Regulatory Information in Eukaryotic Transcription

Abstract

Transcriptional regulation is a key determinant of cell differentiation during development. Changes in where and when a gene is expressed can lead to phenotypic variation among individuals, and ultimately to phenotypic divergence between species over evolutionary time. Spatiotemporal patterns of gene expression are primarily controlled by enhancers, and a given developmental gene can have multiple enhancers that direct portions of its expression in different tissues and at different times in embryogenesis. A major open question in gene regulation is how information from across a gene locus is combined to generate precise and robust patterns of expression. At developmental genes, more than one enhancer is often able to concurrently activate transcription in the same cells, and together these “shadow enhancers” drive non-additive patterns and levels of expression. How does the promoter integrate their inputs to determine the overall output of the gene? We tackled this question using a combination of mathematical modeling, controlled molecular biology and quantitative imaging in Drosophila embryos and mammalian cells. In this work, we challenged existing models of transcriptional regulation with new theory and experiments. We built a quantitative model to explore how combinatorial control might occur through regulators acting on different slow steps in transcription. We demonstrated that this can produce the same “computations” as existing thermodynamic models but allows for more flexibility in when transcription factors bind DNA and how they interact with one another. This may help explain some puzzling observations in eukaryotic transcription, including the rate of regulatory sequence evolution and the dynamic nature of transcription factor -DNA binding. We tested the predictions generated by this model using highly controlled reporter experiments. Using live imaging of nascent transcription driven by shadow enhancers and enhancer duplications in Drosophila embryos, we show support for competition between concurrently-active enhancers in activating transcription. However, the “computation” performed by the promoter to combine their inputs varies in space and time, and the relative positions of the enhancers influence this. Our results provide a nuanced view of how regulatory information is integrated at the locus level to produce precise patterns of gene expression.