As many as 59% of the transcription factors in Escherichia coli regulate the transcription rate of their own gene. This suggests that auto-regulation has one or more important functions. Here, one possible function is studied. Often the transcription rate of an auto-regulator is also controlled by additional transcription factors. In these cases, the way the expression of the auto-regulator responds to changes in the concentrations of the \input" regulators (the response function) is obviously aected by the auto-regulation. We suggest that, conversely, auto-regulation may be used to optimize this response function. To test this hypothesis, we use an evolutionary algorithm and a chemical{physical model of transcription regulation to design model cis-regulatory constructs with predened response functions. In these simulations, auto-regulation can evolve if this provides a functional benet. When selecting for a series of elementary response functions|Boolean logic gates and linear responses|the cis-regulatory regions resulting from the simulations indeed often exploit auto-regulation. Surprisingly, the resulting constructs use auto-activation rather than auto-repression. Several design principles show up repeatedly in the simulation results. They demonstrate how auto-activation can be used to generate sharp, switch-like activation and repression circuits and how linearly decreasing response functions can be obtained. Auto-repression, on the other hand, resulted only when a high response speed or a suppression of intrinsic noise was also selected for. The results suggest that, while auto-repression may primarily be valuable to improve the dynamical properties of regulatory circuits, auto-activation is likely to evolve even when selection acts on the shape of response function only.

Additional Metadata
Reviewer N. de Keijzer
Persistent URL dx.doi.org/10.1371/journal.pcbi.1000813
Journal PLoS Comput. Biol.
Citation
Hermsen, R, Ursem, B, & ten Wolde, P.R. (2010). Combinatorial gene regulation using auto-regulation. PLoS Comput. Biol., 6(6, Article number: 1000813), 1–13. doi:10.1371/journal.pcbi.1000813