Efficient Computational Quantification of Protein-Protein Interactions
Abstract
It has been proven difficult to obtain experimental structures of protein-protein complexes. Computational methods like protein-protein docking attempt to overcome the mismatch between the number of available complex structures and single protein structures by the prediction of binding interfaces. However, the binding free-energy estimates given by the scoring algorithms used in such approaches show only poor correlation with experimentally determined binding strengths. Molecular dynamics (MD) simulations are a premier computational technique which allows for the atomistic modeling of the interactions, structures and motions of (bio-)molecular systems. Very recently, we calculated the binding free energy of two small proteins, namely Ubiquitin and the very flexible Ubiquitin-binding domain of the human DNA Polymerase ι (UBM2), using an MD simulation-based approach. Our results were in very good agreement with experimentally determined values (the mean unsigned error to experimentally determined values was 2.5 kJ/mol or lower) and the statistical errors of the calculations were also mostly in the order of thermal noise. In the proposed project, we aim to develop more efficient approaches that can be readily used on a wide variety of protein-protein complexes. In particular, the project addresses three aims. The first aim is the generation of reference data on the calculation of biomolecular binding affinity using the previously described approach for validation and subsequent optimization. The second aim is the optimization of the simulation method to efficiently score a high number of possible protein docking poses for similarity to the canonical complex structure. Preliminary analyses suggest that we can reduce the overall simulation time by two to three orders of magnitude, which with current computational resources makes a more high-throughput approach feasible, while simultaneously retaining sufficient accuracy to provide binding affinity estimates that can be compared to experimental values. In a more independent part of the project we will focus on a specific aspect of the binding affinity calculation, namely the contribution of conformational preferences of biomolecules. To avoid adverse effects upon administering e.g. mouse-derived antibodies for therapeutic purposes, the framework regions of a mouse antibody are being mutated to become (more) human-like. Such mutations do not affect the antibody-antigen interface directly, but are often seen to negatively influence the binding affinity due to altered conformational preferences of the antibody. As a third aim, we will develop a method to predict the binding free-energy change upon antibody framework mutation based on the conformational preferences of the antibody molecules.
keywords Free Energy Calculation Computational Thermodynamics Molecular Simulation Scoring Functions Statistical Mechanics
Publikationen
Project staff
Jan Walther Perthold
Dipl.-Ing. Jan Walther Perthold
jan.perthold@boku.ac.at
Project Leader
01.07.2018 - 30.06.2021