Engineering of Sensory Proteins with New Ligand-Binding Capacities

Abstract

Biosensors based on whole cell living bacteria or their isolated components mostly use the capacity of natural protein classes to “sense,” in other words to interact with, chemical ligands. Frequently deployed sensor protein classes consist of transcription regulators, two-component sensory proteins, methyl-accepting chemoreceptors and periplasmic binding proteins. All classes have been linked to some form of transducer and sensor output upon binding of the ligand, such as by de novo synthesis of reporter proteins, through Föster resonance energy transfer, from cell accumulation, from chemotaxis, or in dye-encapsulating competitive liposome assays. Despite the partially successful deployment of such biosensors, one of their current limitations is the covered spectrum of chemical ligands, which in most cases reflects the cognate ligand of the used sensory protein. Here we will summarize some of the past and recent efforts to obtain new ligand-binding specificities in these sensory protein classes. Most strategies have followed combinations of random mutagenesis, selection and screening methodologies, and protein-structure guided predictions. They have led to some success of obtaining new specificities, but which not unexpectedly lay relatively close to the original ligand. Computational structure-function predictions have become more important to reduce the mutagenesis efforts to smaller library sizes, but their success has been limited so far to those sensory proteins that do not undergo major conformational changes upon ligand-binding. In particular for the class of periplasmic binding proteins, new ligand-binding specificities have been notoriously difficult to predict, likely as a result of the current limitations in dynamic structure predictions. It is to be expected that future advances in computational algorithms will facilitate this process, which will make it more straightforward to obtain sensory proteins with targeted ligand-binding properties, enabling plug-and-play biosensor design.