Binding of Transcription Factors Adapts to Resolve Information-Energy Tradeoff

We examine the binding of transcription factors to DNA in terms of an information transfer problem. The input of the noisy channel is the biophysical signal of a factor bound to a DNA site, and the output is a distribution of probable DNA sequences at this site. This task involves an inherent tradeoff between the information gain and the energetics of the binding interaction—high binding energies provide higher information gain but hinder the dynamics of the system as factors are bound too tightly. We show that adaptation of the binding interaction towards increasing information transfer under a general energy constraint implies that the information gain per specific binding energy at each base-pair is maximized. We analyze hundreds of prokaryote and eukaryote transcription factors from various organisms to evaluate the discrimination energies. We find that, in accordance with our theoretical argument, binding energies nearly maximize the information gain per energy. This work suggests the adaptation of information gain as a generic design principle of molecular recognition systems.