Bazzoli 2014

Andrea Bazzoli, Jimmy Budiardjo, John Karanicolas

Protein biosensors produce a detectable signal in response to binding to a target compound. In traditional protein biosensor design, the conformational change undergone by a protein upon binding of the target is coupled to activation of a reporter molecule (typically a fluorophore or an enzyme) fused to the protein for easy readout. A major limitation of these methods is the need to start from a protein that is known to bind the target. To overcome this problem, we recently devised a technique—called chemical rescue of structure—in which any protein of measurable activity can be deactivated by a cavity-creating mutation and then reactivated by addition of a compound that matches the deleted “constellation” of atoms both structurally and chemically; by being activated contingent on the presence of the compound, the mutant protein is de facto a sensor for that compound. We recently demonstrated the technique by designing a β-glycosidase into a sensor for indole through a single WG mutation. While they served as a proof-of- principle for the technique, single-residue cavity-creating mutations have little applicability, as the number of compounds rescuing from them is very small (unlikely more than 43, the number of such mutations). More interesting in practice is the possibility to combine two or more contiguous cavity-forming mutations; in addition to an exponential growth in the number of potentially rescuing compounds, another advantage of these combined mutations should be an increased rescue power, as the rescuing compounds will tend to be larger and therefore to bind more strongly to the cavity. We present here the computational design of a sensor for tryptamine working through rescue from a combined WG, Kx mutation (x {A, G, I, L, M, S, T, V}). The sensor will be built from a fusion between an “input” protein, containing the contiguous W– K residue pair, and an “output” luciferase, acting as a reporter of tryptamine binding. Adopting the fusion construct allowed us to decouple rescue from signaling, so that we could search for tryptamine-like W–K constellations regardless of whether the input protein had a detectable function. Experimental testing of candidate sensors is underway.