Bottom-up Construction of Orthogonal Regulation for Gene Circuits Using Transcription Factor-Promoter Pairs with Predictable Properties

The majority of well-described synthetic gene networks are constructed using the same transcription factors and so suffer from an inherent incompatibility. Although the number of transcription factors available to synthetic biology has expanded over recent years thanks to advances in modular zinc finger, CRISPR-Cas9 and TALE-based DNA binding proteins, generating and characterising unique synthetic transcription factor-promoter sets for every new network is time-consuming and impractical. Building on previous work where we introduced TAL Orthogonal Repressors (TALORs) for use in yeast, we show here that incorporating randomisation into TALOR assembly and coupling this with rapid promoter modification enables fast and efficient generation of valuable libraries of orthogonal transcription factor-promoter pairs. To do this we took a rational approach, first analysing a randomised 40-member constitutive promoter library using quantitative sequence-activity models  to identify base positions within a core promoter that correlate with promoter output. This information was then applied to create design criteria for TALOR randomisation reactions; allowing partial randomisation of DNA-binding sites whilst maintaining output levels of all regulated promoters. In tests for orthogonality with 100 engineered strains, 10 regulated library promoters were evaluated against their 10 corresponding library TALORs, showing repression only when promoters are paired with their cognate library member. Remarkably, orthogonality was seen despite sequence homology in the 20 bp binding site between some members being as high as 85%. To demonstrate the functionality of this library, TALORs and their promoters were implemented as orthogonal wires in a novel yeast genetic circuit. Together, the methodology for generating libraries of orthogonal transcription factor-promoter pairs and the use of modelling for predictable properties potentially enables routine construction of synthetic circuits and networks from de novo regulatory components.