Thermodynamic and Structural Investigations of Protein Mutations Using λ-Dynamics

Abstract

The λ-dynamics (λD) alchemical free energy method is a powerful tool to investigate the impact that an amino acid change has on the stability of a protein or a protein complex. This method simultaneously calculates thermodynamic relative free energy changes (ΔΔGs) due to protein perturbations and generates simulation trajectories that allow the specific origin of the thermodynamic change to be identified. Historically, alchemical investigations of protein mutations have encountered difficulties when the perturbation involved buried small-to-large changes, changes to backbone flexibility, and/or changes in charge. However, such perturbations frequently arise when studying disease-linked protein mutations. In this thesis, λD was used to model mutations in the 20S proteasome, the insulin receptor, and the RNA exosome complex, demonstrating its ability to model each kind of challenging protein mutations listed above. Using multisite λ-dynamics (MSλD), a series of thermal-sensitive mutants in the proteasome subunit, Pup2 (C76R, T113M, and L204Q) were investigated. The thermodynamic results revealed a large ΔΔGbind of 5 kcal/mol accompanying the C76R mutation. Complimentary stabilization by the T113M and L204Q variants was observed at lower temperatures (30 °C) but disappeared at higher temperatures (50 °C). The insulin ValA3Leu mutation gives the clinically observed insulin Wakayama a 140-500-fold worse binding affinity for the insulin receptor. This loss of binding, along with the binding trends of six other insulin A3 variants, were successfully captured with λD. Structural investigation revealed that the substantially worse impact from LeuA3 stems from a steric clash with Leu’s second Cδ. With λD, a series of challenging EXOSC3 mutations were analyzed, including the neurodegenerative disorder-linked D132A, A139P, G191C, and G191D and the variant of uncertain significance (VUS) G135R. All of these variants were found to destabilize EXOSC3 folding while A139P, G191D, and G135R were also found to negatively impact binding affinity. Structural analysis identified a clear rationale for the thermodynamic impact of all EXOSC3 variants. This study concludes that the VUS G135R is likely pathogenic. This work demonstrates the utility of λD in investigations including extremely challenging protein mutations in macromolecular complexes and applies this tool to explain disease-linked missense mutations and to predict the significance of a VUS mutation.