Mechanisms of enhanced or impaired DNA target selectivity driven by protein dimerization

Abstract

Successful DNA transcription demands coordination between proteins that bind DNA while simultaneously binding to one another to form dimers or higher-order complexes. For proteins with numerous DNA targets throughout the genome, measurements that report on their dwell time or occupancy thus represent a convolution over a population interacting with specific DNA, nonspecific DNA, or protein partners on DNA. Dimerization is known to add contacts that can help a single protein to stably bind DNA. However, we show here that dimerization can also impair measured dwell times and occupancy on target sequences because the population redistributes across DNA. We combine mass-action kinetic models of pairwise reversible reactions between proteins and DNA with theory and spatial stochastic simulations to isolate the role of dimerization on observed DNA dwell times, occupancy, and spatial distribution of proteins on DNA. Three key themes emerge: (i) Protein-protein interactions, in addition to protein-DNA interactions, can localize a protein to DNA, and relative binding rates can thus widely tune dwell times. (ii) Dimensional reduction achieved through nonspecific binding and subsequent 1D diffusion controls the order-of-magnitude of enhancements despite nucleosome barriers. (iii) Dimerization enhances selectivity for locally clustered targets and often impairs binding to widely-spaced targets by sequestration. Compared with ChIP-seq data, our model explains how the distribution of the essential GAF protein throughout the genome is highly selective for clustered targets due to protein interactions. This model framework predicts when even weak dimerization can redistribute and stabilize proteins on DNA as a necessary part of transcription. SignificanceTranscription of DNA into RNA relies on a globally orchestrated but microscopically stochastic recruitment of multiple proteins to targeted DNA sequences. DNA binding is only a first step. Here we show how proteins with many targets throughout the genome can use protein-protein dimerization to not only help proteins bind DNA, but also to redistribute the population genome-wide, significantly enhancing or impairing dwell times and target occupancies dependent on target spacing. Using theory and simulations along with experimental comparison, outcomes depend on protein-to-target ratios, dimensional gains of 1D searchers, and relative binding rates. Our relatively simple formulas predict how these factors yield negligible or dramatic changes in target selection, with direct mapping to experimental measurements on essential transcription factors.