Promoter reliability in modular transcriptional networks

Abstract

Synthetic biologists engineer systems with desired properties from simple and well-characterized biological parts. Among the most popular and versatile parts are tunable promoters and the transcription factors (TFs) that regulate them. Individual TFs can transduce physical or chemical signals to regulate gene expression; networks of TFs regulating each other's expression can filter signals, reduce noise, store memories, and oscillate. However, the biochemical parameters that describe TF-promoter interactions are often context dependent, making it challenging to build systems that reliably achieve specific outcomes. Here, we explore this problem using plasmid-borne transcriptional networks in Escherichia coli. We demonstrate that the expression properties of a positive-feedback module quantitatively and qualitatively change when this module is embedded within the context of a larger network, where the original TF is used to drive new outputs. A mathematical model suggests this might be due in part to the sequestration of the TF by additional copies of its cognate promoter. The parameters describing TF-promoter interactions (the Hill coefficient and half-saturation constant) can vary depending on promoter copy number. This problem is acute for plasmid-borne systems where promoter concentrations exceed the TF-promoter equilibrium constant. In this regime, we advocate the use of operator buffers: passive multimeric stretches of TF-binding sites that insulate promoter properties from context. If such buffers are included in a standard host chassis, promoters once characterized can be reliably integrated into larger networks.