Free energy perturbation calculations of mutation effects on SARS-CoV-2 RBD::ACE2 binding affinity

Abstract

The strength of binding between human angiotensin converting enzyme 2 (ACE2) and the receptor binding domain (RBD) of viral spike protein plays a role in the transmissibility of the SARS-CoV-2 virus. In this study we focus on a subset of RBD mutations that have been frequently observed in sequenced samples from infected individuals and probe binding affinity changes to ACE2 using surface plasmon resonance (SPR) measurements and free energy perturbation (FEP) calculations. We find that FEP performance is significantly better than that of other computational approaches examined here, in part due to its ability to account for protein structure relaxation resulting from the mutation of interfacial residues. Moreover, analysis of FEP trajectories offers physical insights not available from other methods. Notably, FEP calculations successfully predict the observed cooperative stabilization of binding by the Q498R N501Y double mutant present in the Omicron variant and offer a physical explanation for the underlying mechanism. Our results furthermore suggest a strategy as to how to effectively deploy FEP methods in the optimization of neutralizing antibodies.

Comparative Study of the Mutations Observed in the SARS-CoV-2 RBD Variants of Concern and Their Impact on the Interaction with the ACE2 Protein

“…A variety of computational studies have been conducted and continue to be performed, comparing the affinities of SARS-CoV-2 variants of concern with the ACE2 receptor. Several methods, such as MM-PBSA or MM-GBSA, ,− FEP, and neural network models, were applied in this research. These computational methods have made it possible to compare various variants of concern and predict future mutations, as seen in several neural network models …”

Biophysical principles predict fitness of SARS-CoV-2 variants

“…Their study accurately determined the binding free energy of the RBD for four prevalent variants and the wild type, when complexed with ACE2 or antibodies S2E12 and H11-D4. In parallel, Sergeeva et al [25] offered an effective strategy to determine the impact of interfacial mutations on the binding affinities between RBD and ACE2, using free energy perturbation. Williams et al [26] constructed a multi-layer neural network, using biophysical parameters as inputs to predict binding affinities of SARS-CoV-2 antibodies with various VoCs.…”

Surfaces: A software to quantify and visualize interactions within and between proteins and ligands

“…To demonstrate the reliability of Surfaces, we compare its results with a highly curated dataset of 23 SARS-CoV-2 Spike Receptor Binding Domain (RBD)/ACE2 binding DDG values measured by Surface Plasmon Resonance (Sergeeva et al, 2022) and a compare these to large number of computational methods. Specifically: FEP + 100ns MD simulations From Sergeeva et al (Sergeeva et al, 2022), methods based on machine-learning: Mutabind2 (Zhang et al, 2020), mCSM-PPI (Rodrigues et al, 2019), SAAMBE-3D (Pahari et al, 2020); based on statistical potentials: BeAtMusic (Dehouck et al, 2013); and force field related scoring functions: FoldX (Schymkowitz et al, 2005), Rosetta flex ddG (Barlow et al, 2018). Surfaces shows a Pearson's correlation coefficient (PCC) of 0.556 with the experimental data, while FEP+100 ns MD obtains a PCC of 0.598.…”

“…Molecular dynamics (MD) simulations are at the forefront of such efforts, but such methods are computationally expensive and often also difficult to implement and thus remain impractical for high-throughput applications or broad adoption. Among these methods, we can highlight free-energy perturbation (FEP) methods that aim at predicting the DDG of binding for different mutations relative to wild-type or the binding DG for biomolecular complexes (Lavigne et al, 2000;Sergeeva et al, 2022;Zhu et al, 2022;McCarrick and Kollman, 1999) and gRINN (Serçinoglu and Ozbek, 2018), a tool based on MD simulations to breakdown the per-residue energetic contribution of molecular interactions.…”

“…Representation of the .CSV output and of the visual output colored from red (unfavorable) to blue (favorable) showing the net value of interactions mapped into the surface of the protein, as well as the interactions involving the residues of interest highlighted with dash lines. (E) Comparative results of Surfaces with the calculations performed using FEP + 100 ns MD simulations for the interaction of 23 mutants of the Spike protein and the receptor ACE2 (Sergeeva et al, 2022), and with the pairwise Interaction Energies calculated with gRINN (Serçinoglu and Ozbek, 2018) for a MD simulation of the Spike protein Receptor-Binding Domain of the Delta variant and the receptor ACE2 (Cheng et al, 2022).…”