The effect of mutations on binding interactions between the SARS-CoV-2 receptor binding domain and neutralizing antibodies B38 and CB6

Abstract

SARS-CoV-2 is the pathogen responsible for COVID-19 that has claimed over six million lives as of July 2022. The severity of COVID-19 motivates a need to understand how it could evolve to escape potential treatments and to find ways to strengthen existing treatments. Here, we used the molecular modeling methods MD + FoldX and PyRosetta to study the SARS-CoV-2 spike receptor binding domain (S-RBD) bound to two neutralizing antibodies, B38 and CB6 and generated lists of antibody escape and antibody strengthening mutations. Our resulting watchlist contains potential antibody escape mutations against B38/CB6 and consists of 211/186 mutations across 35/22 S-RBD sites. Some of these mutations have been identified in previous studies as being significant in human populations (e.g., N501Y). The list of potential antibody strengthening mutations that are predicted to improve binding of B38/CB6 to S-RBD consists of 116/45 mutations across 29/13 sites. These mutations could be used to improve the therapeutic value of these antibodies.

Exploring the ability of the MD+FoldX method to predict SARS-CoV-2 antibody escape mutations using large-scale data

“…In particular, the yeast-DMS approach excludes mutants with significant misfolding 33 . Conversely, predictive models often do not calculate antigen folding stability; when they do, their accuracy can be variable, posing additional challenges to accurately predicting escape mutations [19][20][21][22][23] . 3) The presence of multimeric structures, such as trimers in spike proteins, could introduce cooperative binding effects that could affect binding strength and escape.…”

“…Third, while some methods provide fast results, their accuracy may be compromised, and more accurate methods may be computationally intensive and time-consuming. Current methods range from sequence-based 17,18 to structure-based [19][20][21][22] , including hybrid models 23 . Structure-based approaches are particularly important since protein structure profoundly influences their biophysical properties.…”

“…S. Huang et al proposed a hybrid structure/sequence-based model to predict similar high-risk immune-escaping hotspots on the RBD averaging results from 145 non-specified Abs 19 . Barnes et al used physics-based molecular dynamics (MD) in combination with FoldX to develop a watchlist for pandemic surveillance for the SARS-CoV-2 RBD in complex with Abs B38 and CB6 21 . These studies offer valuable insights, however, there are some reasons to expand on this body of work.…”

“…Second, some do not employ conformational sampling of the experimental structures for their predictions 19,20 . Third, some are limited in their focus on a few specific systems 21,22 . Finally, some present average effects rather than antibody-specific data; this is important since Abs targeting structurally similar epitopes may elicit different responses to the same mutation 19 .…”

“…Among the array of possible computational approaches, the MD+FoldX method stands out as a promising avenue for predicting Ab-escape mutations, as evidenced by our previous comparison of eight methods across various protein-protein systems 27 . The MD+FoldX approach was developed by our lab in 2016 and has since been applied to various studies 21,27–30 . It is a 3-D structure-based approach that integrates MD simulations with FoldX estimations.…”