The ability to sense small molecules has important and wide-ranging applications for probing biological processes, for biological engineering and for medical diagnostics. This R21 project seeks to demonstrate feasibility of a computational design platform to engineer protein-based (thus genetically encodable) small molecule biosensors. To achieve this goal, we will develop and assess methods to engineer small molecule binding sites into heterodimeric protein-protein interfaces such that the protein-protein interaction becomes dependent on the small molecule. Each of the two protein partners will be linked to a fragment of a split reporter. A functional sensor then detects the presence of the small molecule by complementation of the reporter. In this fashion, the sensor output is in principle modular: different reporter fragments can be attached to the small molecule sensor components and tested for either detection (via for example split-GFP) or actuation (split-enzymes or gene expression). Our novel contribution will be to develop a design approach that uses a four-step process to transplant small molecule binding sites from liganded protein structures into protein-protein interfaces: (1) Given a small molecule target, we define geometric constraints for a binding "motif" (generally parts of protein side chains) that coordinates the target in a liganded protein monomer structure. (2) We search >600 heterodimeric protein-protein interfaces ("scaffolds") for those that could accommodate the target and its binding motif. (3) We computationally remodel and design the scaffold interface around the target-binding site to stabilize the motif residues in their binding conformation, using robotics-inspired approaches we recently developed. (4) We predict libraries of sensor sequences and test these for dimerization only in the presence of the small molecule using in vitro binding assays and in bacteria using the biosensor output. Computer-aided design of modular small-molecule induced dimerization would be a first. Our experimental plan seeks to ameliorate the risk inherent in this design-driven project in several ways: (i) Moderate target binding affinity is intended to be pre-programmed into the interface using a motif- directed approach that has recently led to successful design of new enzymes. (ii) Computation is complemented with experimental testing of not only the top predictions, but libraries of sequences. (iii) Testing these libraries is facilitated by the biosensor platform, where the intended output itself will be used to screen, select and improve sensors. While small-molecule induced protein dimerization, such as in the rapamycin system, exists in nature and has been reengineered, these systems are limited to a few molecules that can be sensed. Successful completion of this project would demonstrate a platform technology that could greatly broaden application of small-molecule sensing and actuation by providing a route to modular biosensors for many targets.
This research develops new technologies to engineer sensors made out of biological components that allow real-time detection of molecules in living cells and organisms. Such new biosensors have many practical applications in biomedical research and biological engineering, and will also help to advance our understanding of fundamental cellular processes by monitoring molecules in health and disease states. The software and technology developed in this project will be widely available to researchers to improve biosensor design.