Logistic Regression-Guided Identification of Cofactor Specificity-Contributing Residues in Enzyme with Sequence Datasets Partitioned by Catalytic Properties

Abstract

Changing the substrate/cofactor specificity of an enzyme requires multiple mutations at spatially adjacent positions around the substrate pocket. However, this is challenging when solely based on crystal structure information because enzymes undergo dynamic conformational changes during the reaction process. Herein, we proposed a method for estimating the contribution of each amino acid residue to substrate specificity by deploying a phylogenetic analysis with logistic regression. Since this method can estimate the candidate amino acids for mutation by ranking, it is readable and can be used in protein engineering. We demonstrated our concept using redox cofactor conversion of the Escherichia coli malic enzyme as a model, which still lacks crystal structure elucidation. The use of logistic regression with amino acid sequences classified by cofactor specificity showed that the NADP+-dependent malic enzyme completely switched cofactor specificity to NAD+ dependence without the need for a practical screening step. The model showed that surrounding residues made a greater contribution to cofactor specificity than those in the interior of the substrate pocket. These residues might be difficult to identify from crystal structure observations. We show that a highly accurate and inferential machine learning model was obtained using amino acid sequences of structurally homologous and functionally distinct enzymes as input data.

Curse and Blessing of Non‐Proteinogenic Parts in Computational Enzyme Engineering

“…Specifically for protein structures featuring a Rossman fold, a deep learning algorithm called Rossman‐toolbox predicts the preference between NADH and NADPH, [212] and Cofactory, also optimized for Rossman folds, predicts a mutant's cofactor specificity and binding affinity for NAD, NADP, FAD, and SAM [212] . A purely sequence‐based ML algorithm for ranking mutations that may cause a switch between NAD and NADP specificity is also available, although the creation of a new optimized model for each protein is required [262] . Another promising, although computationally expensive, approach is to use ML to predict energies, by training them on MD or QM/MM trajectories of a protein of interest.…”

“…[212] A purely sequencebased ML algorithm for ranking mutations that may cause a switch between NAD and NADP specificity is also available, although the creation of a new optimized model for each protein is required. [262] Another promising, although computationally expensive, approach is to use ML to predict energies, by training them on MD or QM/MM trajectories of a protein of interest. At the expense of throughput, such algorithms may lead to more robust predictions than simply using these simulations for rational design.…”

Design of a cofactor self‐sufficient whole‐cell biocatalyst for enzymatic asymmetric reduction via engineered metabolic pathways and multi‐enzyme cascade

“…[16] Although this strategy can improve the rate of NAD(P)H regeneration, it is still necessary to add a certain stoichiometric oxidized cofactor exogenously, which increases the process costs and complicates the workup upon reaction completion. Altering cofactor preference by enzyme modification based on structural biotechnology, [17][18][19][20][21] such as turning NADPH-dependent to NADH-dependent, can reduce redox cofactor costs. However, changing the cofactor specificity from the NADPH preference to the NADH preference through protein engineering is difficult to achieve.…”

Protein property prediction based on local environment by 3D equivariant convolutional neural networks

“…Protein engineering is widely employed in both academia and industry to enhance the desired properties of proteins, such as modifying enzyme activities [1], improving antibody binding specificity [2], and increasing stability or yield for productions purposes [3, 4]. In practice, this process typically starts with a natural protein, followed by mutagenesis to generate a set of candidates (mutants), and screening these mutants to identify the ones with desired properties [5].…”