Identifying metal binding amino acids based on backbone geometries as a tool for metalloprotein engineering

Abstract

Metal cofactors within proteins perform a versatile set of essential cellular functions. In order to take advantage of the diverse functionality of metalloproteins, researchers have been working to design or modify metal binding sites in proteins to rationally tune the function or activity of the metal cofactor. This study has performed an analysis on the backbone atom geometries of metal-binding amino acids among 10 different metal binding sites within the entire protein data bank. A set of 13 geometric parameters (features) was identified that is capable of predicting the presence of a metal cofactor in the protein structure with overall accuracies of up to 97% given only the relative positions of their backbone atoms. The decision tree machine-learning algorithm used can quickly analyze an entire protein structure for the presence of sets of primary metal coordination spheres upon mutagenesis, independent of their original amino acid identities. The methodology was designed for application in the field of metalloprotein engineering. A cluster analysis using the data set was also performed and demonstrated that the features chosen are useful for identifying clusters of structurally similar metal-binding sites.

Structural Bioinformatics and Deep Learning of Metalloproteins: Recent Advances and Applications

“…Prediction of positions where metal ligands can be introduced, based on protein backbone coordinates, to design artificial MPs [50] Structural comparison of metal sites Sequence-based prediction of MPs using a NN trained with information derived from 3D structures [97] https://github.com/cerm-cirmmp/MBSDL (accessed on 5 July 2022)…”

“…Although it is not aimed at the identification of MBSs in apo-protein structures, another tool employing a random forest classifier was developed to analyze backbone protein structures to identify suitable positions to introduce metal-binding residues in order to engineer MBSs in proteins (i.e., to artificially design an MP given a protein scaffold of known 3D structure) [ 50 ]. In practice, the training set contains features that are based only on the coordinates of the backbone atoms, whereas all of the side chain atoms are removed.…”

“…The position of the Cβ is recalculated from the backbone coordinates to permit the computation of geometric parameters that are fed into the random forest classifiers. Neglecting the position of the side chain atoms makes it possible to predict the occurrence of potential MBSs independently of the identity of amino acids in the current structure, thereby opening the way towards creating engineered binding sites by introducing the appropriate side chains in the positions highlighted by the tool [ 50 ].…”

A Comprehensive Review of Computation-Based Metal-Binding Prediction Approaches at the Residue Level

“…Figure 3 reveals that 6 out of 32 methods adopt feature selection before training the model. These feature selection approaches include forward feature selection [ 79 , 102 ], experience-based [ 104 ], Boruta algorithm [ 57 , 81 ], minimum-redundancy maximum-relevancy [ 79 ], and mean decrease Gini index [ 78 ].…”

Machine Learning Approaches for Metalloproteins

“…In the past decades, there have been multiple attempts to understand metalloproteins using theoretical or computational methods, while many established databases and software documents have analyzed metal-binding processes [24][25][26]. For example, in 2000, Dudev and Lim [27] conducted ab initio and continuum dielectric calculations of the free energy change that occurs when a protein binds a metal ion in the presence of surrounding water molecules.…”