(723c) Rational Design of Protein Devices: Predicting How Functionalization and Unnatural Amino Acid Mutations Affect Protein Stability Using Molecular Simulation and Experiment

Using proteins in new applications is a source of continual research because proteins can generate many desirable outcomes. However, using a protein outside of its native environment can compromise protein stability, activity, and bioavailability. Moreover, harnessing the power of a protein usually requires functionalization of the molecule or attachment to a solid substrate. Until recently, the location of the required alteration on the protein molecule could only be done non-specifically (e.g. histidine residues) or at a limited number of fixed sites (e.g. the N and C termini). However, state of the art methods used in our group allow the functionalization or attachment to occur at any residue along the amino acid chain. This unprecedented control is made possible by replacing one of the amino acids in the chain with an unnatural amino acid containing a specific chemical moiety not found in nature (e.g. azide and alkyne groups). The question thus becomes where to perform the mutation, and subsequent functionalization, to maintain protein activity. Answering this question is foundational to rapid development of cost-effective and reliable protein devices.

This work will explain how we are using an integrated experimental and theoretical approach to create heuristics for rational design of protein devices. The presentation will begin with a description of the types of protein devices involved and an explanation of how site-specific functionalization is done experimentally. Next, the coarse-grain and all-atom simulation methods used to determine how a mutation or functionalization will affect protein stability and activity are described including an explanation of a recently developed CHARMM-compatible model needed to accurately reproduce the biophysics of the unnatural amino acids used in this work. This is followed by a series of case studies showing how molecular simulation can be used to screen for the best candidate site for mutation on a given protein. Highlights of the results include: 1) an explanation of why mutation at certain sites destabilize T4 lysozyme while other do not and how simulation can be used to predict this, 2) how simulation can screen for the best sites to PEGylate and tether β-lactimase, and 3) how experimental and computational results feed into future selections to self-consistently optimize the prediction process. Taken as a whole, the results provide significant hope that rational design of protein-based devices is possible in the near future.