(61c) Redesigning Known Proteins to Detect Insulin and IL-6 across a Four Order of Magnitude Concentration Range

Certain features of biosensors represent an uncommon challenge to rational protein engineering methods. Most such methods solely focus on making the best possible protein: the one with the strongest affinity or the highest catalytic efficiency. However, a well-designed biosensor must be able to detect its analyte of interest across a range of concentrations. If the sensor only uses very high affinity proteins, it will become saturated at low analyte concentrations and unable to provide meaningful information at higher concentrations.

As part of an NSF EPSCoR project, we are working with a multijurisdictional team to design in-line, continuous, insulin and interleukin-6 (IL-6) detecting biosensors. The sensors must be able to detect the analytes across concentrations ranging from 10^-9 to 10^-5 molar. To design the protein recognition elements of the sensors, we began with a literature review that identified a total of 13 high affinity proteins and peptides that bind either insulin or IL-6. The structure of each was predicted using RosettaFold, followed by the prediction of each complex using Haddock. The results were compared to literature information about the protein structures and complexes when that information was available to validate the workflow. Next, molecular mechanics calculations using CHARMM and Rosetta and molecular dynamics calculations using VMD were performed to determine the hotspot residues in each complex that were most important to binding. Finally, mutations to non-hotspot residues were identified which are predicted to slightly destabilize the hotspot interactions. It is our hypothesis that binding variants with a range of affinities by identifying mutations around the critical residues.

This presentation will discuss the findings from each step of our workflow, as well as provide a detailed overview of the final designs selected for experimental validation.