Advancing Enzyme’s Stability and Catalytic Efficiency through Synergy of Force-Field Calculations, Evolutionary Analysis, and Machine Learning

Abstract

Thermostability is an essential requirement for the use of enzymes in the bioindustry. Here, we compare different protein stabilization strategies using a challenging target, a stable haloalkane dehalogenase DhaA115. We observe better performance of automated stabilization platforms FireProt and PROSS in designing multiple-point mutations over the introduction of disulfide bonds and strengthening the intra- and the inter-domain contacts by in silico saturation mutagenesis. We reveal that the performance of automated stabilization platforms was still compromised due to the introduction of some destabilizing mutations. Notably, we show that their prediction accuracy can be improved by applying manual curation or machine learning for the removal of potentially destabilizing mutations, yielding highly stable haloalkane dehalogenases with enhanced catalytic properties. A comparison of crystallographic structures revealed that current stabilization rounds were not accompanied by large backbone re-arrangements previously observed during the engineering stability of DhaA115. Stabilization was achieved by improving local contacts including protein–water interactions. Our study provides guidance for further improvement of automated structure-based computational tools for protein stabilization.

Computer-assisted stabilization of fibroblast growth factor FGF-18

“…It shows the importance of iteration of calculations to address the changes introduced by the initial mutations. With each iteration, fewer mutations and lower stabilizing effects can be expected, as was recently shown on the stabilization of haloalkane dehalogenase [32] .…”

Machine Learning-Guided Protein Engineering

“…MutCompute predictions are orthogonal to force-field calculations and phylogenetic analyses conducted with the popular automated web tools PROSS and FireProt, and these predictions can be exploited to remove destabilizing substitutions from multiple-point mutants designed using those tools. This was recently demonstrated by the successful stabilization of haloalkane dehalogenase DhaA115 (PDB ID 6SP5 ).…”

Electrostatics as a Guiding Principle in Understanding and Designing Enzymes

“…Nowadays, machine learning techniques have been also incorporated in the process of protein design. − In this case, an algorithm obtained after training on a large data set can predict the properties of a protein based on a set of descriptors. For example, the three-dimensional structure of the protein can be nowadays predicted from its sequence, using algorithms trained on already existing structures. , For a catalytic process, the selection of a particular protein scaffold to integrate an adequate active site can be performed using a deep-learning based approach, while mutations introduced to improve catalytic efficiency can be also selected integrating machine learning methods . The machine learning designing process does not require an explicit knowledge of how a particular residue affects the properties of the biocatalyst but a prediction of the consequences of its mutation.…”

“… 7 , 8 For a catalytic process, the selection of a particular protein scaffold to integrate an adequate active site can be performed using a deep-learning based approach, 9 while mutations introduced to improve catalytic efficiency can be also selected integrating machine learning methods. 10 The machine learning designing process does not require an explicit knowledge of how a particular residue affects the properties of the biocatalyst but a prediction of the consequences of its mutation. The rational contribution to this designing process is in the selection of the descriptors used by the algorithm in its prediction.…”