Quantification of protein-protein interactions from molecular simulations

Background.

Molecular modeling and simulations of proteins and their complexes give insight at a time and space resolution that is often not attainable experimentally. Methods to quantify the affinity of small molecules to protein hosts based on the free energy of binding are currently quite well established. However, for protein-protein interactions (PPIs) it is difficult to give robust and reliable predictions of the binding affinity (Siebenmorgen, 2020; Guo, 2022). Quantification of PPIs is important to judge novel designs, determine the correct poses, and to elaborate the function of enzymes and biologics. While relative binding affinity approaches that describe the effect of one or a small number of mutations are relatively accessible, a prediction of the full binding affinity remains highly challenging. These challenges originate from the sampling issue, which hampers a correct treatment of entropy.
Specificity in PPIs is essentially the difference in binding affinity for one preferred partner over another. For strongly interacting proteins, unspecific binding events are short-lived and readily distinguished from the relevant interaction. However, for transient interactions, e.g. to facilitate the electron transfer between two enzymes, the physiological binding affinity is smaller, and specificity is the effect of a much more subtle difference in affinity. Accordingly, distinguishing nonspecific binding from a productive (transient) binding mode of a protein-protein complex from computer models requires a very high accuracy in the estimate of the binding affinity.

Aims and Methods.

In this project, we will study efficient means to quantify the strength of PPIs with the aim to predict binding affinities and to predict relevant binding poses of transient complexes. The most robust methods for small molecule binding rely on so-called alchemical approaches, in which two states (bound and unbound) are connected via unphysical pathways, e.g. involving the annihilation of the small molecule from the active site. Due to the size of proteins, such methods are not feasible for PPIs. Elaborate pathways methods which describe the physical binding process have been described (Gumbart, 2013; Perthold, 2017; Suh, 2019), but do not seem very broadly applicable, probably due to lack of sufficient sampling and hysteresis effects. Faster approaches to identify the optimal binding poses have also been suggested, which may offer a way to describe transient complexes (Perthold, 2019). Effective binding modes for e.g. electron transfer may be studied using alternative pathway methods (Laurent, 2019), and give insight into the origins of specificity in interactions between enzymes. This project brings together the complete toolbox of molecular simulations, including alchemical free-energy calculations, enhanced sampling and screening of many putative complex structures. Experimental validation of predictions may be performed in collaboration with other doctoral candidates in the program.

REFERENCES
1. Gumbart, J. C., B. Roux and C. Chipot (2013) Efficient determination of protein-protein standard binding free energies from first principles. J Chem Theory Comput 9. doi: 10.1021/ct400273t
2. Guo, Z. and R. Yamaguchi (2022) Machine learning methods for protein-protein binding affinity prediction in protein design. Front Bioinform 2, 1065703. doi: 10.3389/fbinf.2022.1065703
3. Laurent, C., E. Breslmayr, D. Tunega, R. Ludwig and C. Oostenbrink (2019) Interaction between cellobiose dehydrogenase and lytic polysaccharide monooxygenase. Biochemistry 58, 1226-1235. doi: 10.1021/acs.biochem.8b01178
4. Perthold, J. W. and C. Oostenbrink (2017) Simulation of reversible protein-protein binding and calculation of binding free energies using perturbed distance restraints. Journal of Chemical Theory and Computation 13, 5697-5708. doi: 10.1021/acs.jctc.7b00706
5. Perthold, J. W. and C. Oostenbrink (2019) GroScore: Accurate scoring of protein-protein binding poses using explicit-solvent free-energy calculations. J Chem Inf Model 59, 5074-5085. doi: 10.1021/acs.jcim.9b00687
6. Siebenmorgen, T. and M. Zacharias (2020) Computational prediction of protein-protein binding affinities. Wiley Interdisciplinary Reviews-Computational Molecular Science 10, e1448. doi: 10.1002/wcms.1448
7. Suh, D., S. Jo, W. Jiang, C. Chipot and B. Roux (2019) String method for protein-protein binding free-energy calculations. J Chem Theory Comput 15, 5829-5844. doi: 10.1021/acs.jctc.9b00499