Engineered Probiotics Enabling Specific Sensing of Neurochemicals

Microbes have evolved diverse sensor systems that detect metabolites of great medical importance. These sensors have the potential to be utilized in probiotic microbes to provide diagnostic information and deliver therapeutics with temporal and geographical precision. However, microbial sensors found in nature often have promiscuity to several structurally similar aromatic amino acids, common neurotransmitters, or neuromodulators, limiting their practical applications. Due to differences in associated diseases and functions, specific sensing of these chemicals is critical for probiotic sensor applications. Despite advances in protein engineering for specific ligand-protein interaction, however, engineering ligand-specific sense-and-respond systems remains challenging, especially when the target ligands are structurally similar and ligand-protein binding should control downstream functions such as gene expression. This is mainly due to the challenge in coupling subtle protein conformational changes caused by binding of similar ligands with differential DNA interactions.

We first characterized three promiscuous sensors that recognize aromatic metabolites associated with various metabolic and neurological disorders. Protein engineering often requires extensive structural knowledge of the proteins or massive library sizes. In contrast, to improve ligand selectivity, we rationally engineered the responsible regulators by identifying and individually mutagenizing specific amino acids. We show that our simple and generalizable method of protein engineering is effective and time-efficient, requires small library sizes with only a basic understanding of the protein structure, and enables changes in ligand-protein binding specificity while maintaining protein-DNA interaction and thus downstream gene expression control. Importantly, these sensors are the first ligand-specific sensors for phenylalanine, tyrosine, and phenylethylamine. In addition, our computational and experimental analyses provide novel insights into the uncharacterized regulator structure for the first time, suggesting residues that can be mutagenized to generate novel ligand-specific sensors for additional aromatic neurotransmitters and neuromodulators. The novel ligand-selective sensors generated in this work will have diverse biomedical applications.