Engineering an Allosteric Transcription Factor to Respond to New Ligands

Bacterial transcription factors (TFs) that respond to small molecules (e.g. LacI/GalR, LysR, TetR and AraC) constitute large families of allosteric proteins. Bacterial allosteric TFs act as switches that transduce small molecule binding into a transcriptional actuation, regulating the expression of metabolic operons, transporter genes and other genes. This switch-like behavior makes bacterial allosteric TFs a cornerstone in synthetic biology applications. Expanding allosteric TFs beyond naturally-occurring TF-inducer pairs would greatly increase their utility as sensors for valuable chemicals and orthogonal switches for engineering higher order synthetic circuits. However, redesigning an allosteric TF to bind to a new inducer is challenging. Mutations in the ligand-binding site often disrupt allosteric communication with the DNA-binding domain and it is difficult to incorporate allostery into the design considerations because the amino acids involved in allosteric signal propagation are generally not known.

Using the classical lac repressor as a model system, we describe a general scheme for redesigning allosteric proteins to bind to new inducer molecules without disrupting allosteric communication. We use computational protein design to generate design candidates, synthesize DNA encoding tens of thousands of designs using chip-DNA, and apply a high-throughput screen to enrich for designs that are allosterically functional and specifically respond to the target inducer molecule. We have redesigned lacI to respond to four new non-metabolizable  (in E. coli) inducers: gentiobiose, fucose, lactitol and sucralose. The induction response of the designer lacIs toward their respective new inducers is comparable to WT lacI response to IPTG. The crystal structure of a sucralose-bound LacI variant illustrates how the binding pocket reconfigures for sucralose, a ligand substantially larger than IPTG, especially through stabilizing interactions with the highly nucleophilic chlorines. We compared the amino acid substitution profiles of laboratory-evolved lacI variants with 14,000 lacI paralogs. We find that laboratory-evolved lacI variants have chosen mutational paths not sampled by natural lacI family members.

Allowing designer sensors to be constructed, this method opens the door to accessing nature’s chemical diversity for generating new biosynthetic pathways that serve human needs. Our findings also provide evidence for the allosteric effect of distal mutations, which could lead to fundamental insight into the molecular mechanism of allostery and design rules for engineering allosteric proteins.