Finding the ΔΔ G spot: Are predictors of binding affinity changes upon mutations in protein–protein interactions ready for it?

Predicting antibody affinity changes upon mutations by combining multiple predictors

Antibodies are proteins working in our immune system with high affinity and specificity for target antigens, making them excellent tools for both biotherapeutic and bioengineering applications. The prediction of antibody affinity changes upon mutations (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{\Delta \Delta {\mathrm{G}}}}_{\mathrm{binding}}$$\end{document}ΔΔGbinding) is important for antibody engineering. Numerous computational methods have been proposed based on different approaches including molecular mechanics and machine learning. However, the accuracy by each individual predictor is not enough for efficient antibody development. In this study, we develop a new prediction method by combining multiple predictors based on machine learning. Our method was tested on the SiPMAB database, evaluating the Pearson’s correlation coefficient between predicted and experimental \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{\Delta \Delta {\mathrm{G}}}}_{\mathrm{binding}}$$\end{document}ΔΔGbinding. Our method achieved higher accuracy (R = 0.69) than previous molecular mechanics or machine-learning based methods (R = 0.59) and the previous method using the average of multiple predictors (R = 0.64). Feature importance analysis indicated that the improved accuracy was obtained by combining predictors with different importance, which have different protocols for calculating energies and for generating mutant and unbound state structures. This study demonstrates that machine learning is a powerful framework for combining different approaches to predict antibody affinity changes.

Convergent structural features of respiratory syncytial virus neutralizing antibodies and plasticity of the site V epitope on prefusion F

Respiratory syncytial virus (RSV) is a global public health burden for which no licensed vaccine exists. To aid vaccine development via increased understanding of the protective antibody response to RSV prefusion glycoprotein F (PreF), we performed structural and functional studies using the human neutralizing antibody (nAb) RSB1. The crystal structure of PreF complexed with RSB1 reveals a conformational, pre-fusion specific site V epitope with a unique cross-protomer binding mechanism. We identify shared structural features between nAbs RSB1 and CR9501, elucidating for the first time how diverse germlines obtained from different subjects can develop convergent molecular mechanisms for recognition of the same PreF site of vulnerability. Importantly, RSB1-like nAbs were induced upon immunization with PreF in naturally-primed cattle. Together, this work reveals new details underlying the immunogenicity of site V and further supports PreF-based vaccine development efforts.

Respiratory syncytial virus (RSV) is a persistent, contagious seasonal pathogen and a serious public health threat. While infants are the most at-risk population, with infections potentially leading to bronchiolitis, adults, especially the elderly, are also burdened by RSV-induced respiratory infections. The only treatment currently available for RSV is passive immunization for high-risk infants. Thus, there is a critical need to develop a vaccine for the vast majority of the vulnerable population for which there is no preventative treatment. The RSV fusion protein in its prefusion form (PreF) is the target of the majority of naturally-induced neutralizing antibodies, and several clinical trials are currently evaluating PreF as a promising vaccine candidate. In this study, we solved the X-ray structure of PreF bound to the Fab fragment of a human neutralizing antibody. The structure reveals plasticity of the epitope, as well as a unique molecular signature for antibodies elicited towards this region of PreF. We also find that similar antibodies are induced upon immunization of naturally-primed cattle with a PreF vaccine antigen, suggesting that this epitope is highly immunogenic. These results will help us better understand the human immune response to RSV infection and vaccination, and guide future vaccine-design efforts.

Interaction of the maturation protein of the bacteriophage MS2 and the sex pilus of the Escherichia coli F plasmid

One promising strategy to combat antimicrobial resistance is to use bacteriophages that attach to the sex pili produced by transmissible antimicrobial resistance (AMR) plasmids, infect AMR bacteria and select for loss of the AMR plasmids, prolonging the life of existing antimicrobials. The maturation protein of the bacteriophage MS2 attaches to the pili produced by Incompatibility group F plasmid-containing bacteria. This interaction initiates delivery of the viral genetic material into the bacteria. Using protein-protein docking we constructed a model of the F pilus comprising a trimer of subunits binding to the maturation protein. Interactions between the maturation protein and the F pilus were investigated using molecular dynamics simulations. In silico alanine scanning and in silico single-point mutations were explored, with the longer term aim of increasing the affinity of the maturation protein to other Incompatibility group pili, without reducing the strength of binding to F pilin. We report our computational findings on which residues are required for the maturation protein and F pilin to interact, those which had no effect on the interaction and the mutations which led to a stronger interaction.

Network Organization of Antibody Interactions in Sequence and Structure Space: the RADARS Model

Adaptive immunity in vertebrates is a complex self-organizing network of molecular interactions. While deep sequencing of the immune-receptor repertoire may reveal clonal relationships, functional interpretation of such data is hampered by the inherent limitations of converting sequence to structure to function. In this paper, a novel model of antibody interaction space and network, termed radial adjustment of system resolution, RAdial ADjustment of System Resolution (RADARS), is proposed. The model is based on the radial growth of interaction affinity of antibodies towards an infinity of directions in structure space, each direction corresponding to particular shapes of antigen epitopes. Levels of interaction affinity appear as free energy shells of the system, where hierarchical B-cell development and differentiation takes place. Equilibrium in this immunological thermodynamic system can be described by a power law distribution of antibody-free energies with an ideal network degree exponent of phi square, representing a scale-free fractal network of antibody interactions. Plasma cells are network hubs, memory B cells are nodes with intermediate degrees, and B1 cells function as nodes with minimal degree. Overall, the RADARS model implies that a finite number of antibody structures can interact with an infinite number of antigens by immunologically controlled adjustment of interaction energy distribution. Understanding quantitative network properties of the system should help the organization of sequence-derived predicted structural data.

Implementation of adaptive integration method for free energy calculations in molecular systems

Estimating free energy differences by computer simulation is useful for a wide variety of applications such as virtual screening for drug design and for understanding how amino acid mutations modify protein interactions. However, calculating free energy differences remains challenging and often requires extensive trial and error and very long simulation times in order to achieve converged results. Here, we present an implementation of the adaptive integration method (AIM). We tested our implementation on two molecular systems and compared results from AIM to those from a suite of other methods. The model systems tested here include calculating the solvation free energy of methane, and the free energy of mutating the peptide GAG to GVG. We show that AIM is more efficient than other tested methods for these systems, that is, AIM results converge to a higher level of accuracy and precision for a given simulation time.

The Role of Conformational Dynamics and Allostery in Modulating Protein Evolution

Advances in sequencing techniques and statistical methods have made it possible not only to predict sequences of ancestral proteins but also to identify thousands of mutations in the human exome, some of which are disease associated. These developments have motivated numerous theories and raised many questions regarding the fundamental principles behind protein evolution, which have been traditionally investigated horizontally using the tip of the phylogenetic tree through comparative studies of extant proteins within a family. In this article, we review a vertical comparison of the modern and resurrected ancestral proteins. We focus mainly on the dynamical properties responsible for a protein's ability to adapt new functions in response to environmental changes. Using the Dynamic Flexibility Index and the Dynamic Coupling Index to quantify the relative flexibility and dynamic coupling at a site-specific, single-amino-acid level, we provide evidence that the migration of hinges, which are often functionally critical rigid sites, is a mechanism through which proteins can rapidly evolve. Additionally, we show that disease-associated mutations in proteins often result in flexibility changes even at positions distal from mutational sites, particularly in the modulation of active site dynamics.

Predicting changes in protein stability caused by mutation using sequence-and structure-based methods in a CAGI5 blind challenge

Predicting the impact of mutations on proteins remains an important problem. As part of the CAGI5 frataxin challenge, we evaluate the accuracy with which Provean, FoldX, and ELASPIC can predict changes in the Gibbs free energy of a protein using a limited data set of eight mutations. We find that different methods have distinct strengths and limitations, with no method being strictly superior to other methods on all metrics. ELASPIC achieves the highest accuracy while also providing a web interface which simplifies the evaluation and analysis of mutations. FoldX is slightly less accurate than ELASPIC but is easier to run locally, as it does not depend on external tools or datasets. Provean achieves reasonable results while being computational less expensive than the other methods and not requiring a structure of the protein. In addition to methods submitted to the CAGI5 community experiment, and with the aim to inform about other methods with high accuracy, we also evaluate predictions made by Rosetta's ddg_monomer protocol, Rosetta's cartesian_ddg protocol, and thermodynamic integration calculations using Amber package. ELASPIC still achieves the highest accuracy, while Rosetta's catesian_ddg protocol appears to perform best in capturing the overall trend in the data.