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

Navigating the Fragrance Space Using Graph Generative Models and Predicting Odors

We explore a suite of generative modeling techniques to efficiently navigate and explore the complex landscapes of odor and the broader chemical space. Unlike traditional approaches, we not only generate molecules but also predict the odor likeliness with a ROC AUC score of 0.97 and assign probable odor labels. We correlate odor likeliness with physicochemical features of molecules using machine learning techniques and leverage SHAP (SHapley Additive exPlanations) to demonstrate the interpretability of the function. The whole process involves four key stages: molecule generation, stringent sanitization checks for molecular validity, fragrance likeliness screening, and odor prediction of the generated molecules. By making our code and trained models publicly accessible, we aim to facilitate the broader adoption of our research across applications in fragrance discovery and olfactory research.

Comprehensive evaluation of the capacities of microbial cell factories

Systems metabolic engineering is facilitating the development of high-performing microbial cell factories for producing chemicals and materials. However, constructing an efficient microbial cell factory still requires exploring and selecting various host strains, as well as identifying the best-suited metabolic engineering strategies, which demand significant time, effort, and costs. Here, we comprehensively evaluate the capacities of various microbial cell factories and propose strategies for systems metabolic engineering steps, including host strain selection, metabolic pathway reconstruction, and metabolic flux optimization. We analyze the metabolic capacities of five representative industrial microorganisms as cell factories for the production of 235 different bio-based chemicals and suggest the most suitable host strain for the corresponding chemical production. To improve the innate metabolic capacity by constructing more efficient metabolic pathways, heterologous metabolic reactions, and cofactor exchanges are systematically analyzed. Additionally, we present metabolic engineering strategies, which include up- and down-regulation target reactions, for the improved production of chemicals. Altogether, this study will serve as a comprehensive resource for the systems metabolic engineering of microorganisms in the bio-based production of chemicals.

Protein Engineering of Substrate Specificity toward Nitrilases: Strategies and Challenges

Nitrilase is extensively applied across diverse sectors owing to its unique catalytic properties. Nevertheless, in industrial production, nitrilases often face issues such as low catalytic efficiency, limited substrate range, suboptimal selectivity, and side reaction products, which have garnered heightened attention. With the widespread recognition that the structure of enzymes has a direct impact on their catalytic properties, an increasing number of researchers are beginning to optimize the functional characteristics of nitrilases by modifying their structures, in order to meet specific industrial or biotechnology application needs. Particularly in the artificial intelligence era, the innovative application of computer-aided design in enzyme engineering offers remarkable opportunities to tailor nitrilases for the widespread production of high-value products. In this discussion, we will briefly examine the structural mechanism of nitrilase. An overview of the protein engineering strategies of substrate preference, regioselectivity and stereoselectivity are explored combined with some representative examples recently in terms of the substrate specificity of enzyme. The future research trends in this field are also prospected.

Codeine 3-O-demethylase catalyzed biotransformation of morphinan alkaloids in Escherichia coli: site directed mutagenesis of terminal residues improves enzyme expression, stability and biotransformation yield

The cultivation of opium poppy is the only commercially viable source of most morphinan alkaloids. Bioproduction of morphinan alkaloids in recombinant whole-cell systems provides a promising alternate source of these valuable compounds. The enzyme codeine 3-O-demethylase can transform morphinan alkaloids by O-demethylation and has been applied in single step biotransformation reactions or as part of larger biosynthetic cascade, however, the productivity for these reactions remains low and suboptimal enzyme properties could be improved. This mutagenesis study targeted non-conserved N-and C-terminal residues, which were replaced with the equivalent residues from enzyme thebaine 6-O-demethylase. Whole cell biotransformation performance was significantly improved in Escherichia coli expressing codeine 3-O-demethylase mutants, with a ~ 2.8-fold increase in the production of oripavine from thebaine and ~ 1.3-fold increase in the production of morphine from codeine. Statistical analysis of biotransformation yield, enzyme expression and stability, predicted using changes in Gibbs free energy (ΔΔG) with deep-learning-based model DDmut, suggested that altered enzyme stability and/or expression of soluble protein may contribute to the observed improvements in biotransformation. This approach could be beneficial for screening future codeine 3-O-demethylase mutations and for other enzymes.

Deep learning for NAD/NADP cofactor prediction and engineering using transformer attention analysis in enzymes

Understanding and manipulating the cofactor preferences of NAD(P)-dependent oxidoreductases, the most widely distributed enzyme group in nature, is increasingly crucial in bioengineering. However, large-scale identification of the cofactor preferences and the design of mutants to switch cofactor specificity remain as complex tasks. Here, we introduce DISCODE (Deep learning-based Iterative pipeline to analyze Specificity of COfactors and to Design Enzyme), a novel transformer-based deep learning model to predict NAD(P) cofactor preferences. For model training, a total of 7,132 NAD(P)-dependent enzyme sequences were collected. Leveraging whole-length sequence information, DISCODE classifies the cofactor preferences of NAD(P)-dependent oxidoreductase protein sequences without structural or taxonomic limitation. The model showed 97.4% and 97.3% of accuracy and F1 score, respectively. A notable feature of DISCODE is the interpretability of its transformer layers. Analysis of attention layers in the model enables identification of several residues that showed significantly higher attention weights. They were well aligned with structurally important residues that closely interact with NAD(P), facilitating the identification of key residues for determining cofactor specificities. These key residues showed high consistency with verified cofactor switching mutants. Integrated into an enzyme design pipeline, DISCODE coupled with attention analysis, enables a fully automated approach to redesign cofactor specificity.

The INSIGHT platform: Enhancing NAD(P)-dependent specificity prediction for co-factor specificity engineering

Enzyme specificity towards cofactors like NAD(P)H is crucial for applications in bioremediation and eco-friendly chemical synthesis. Despite their role in converting pollutants and creating sustainable products, predicting enzyme specificity faces challenges due to sparse data and inadequate models. To bridge this gap, we developed the cutting-edge INSIGHT platform to enhance the prediction of coenzyme specificity in NAD(P)-dependent enzymes. INSIGHT integrates extensive data from principal bioinformatics resources, concentrating on both NADH and NADPH specificities, and utilizes advanced protein language models to refine the predictions. This integration not only strengthens computational predictions but also meets the practical demands of high-throughput screening and optimization. Experimental validation confirms INSIGHT's effectiveness, boosting our ability to engineer enzymes for efficient, sustainable industrial and environmental processes. This work advances the practical use of computational tools in enzyme research, addressing industrial needs and offering scalable solutions for environmental challenges.

Improvement of Heterologous Soluble Expression of L-amino Acid Oxidase Using Logistic Regression

Successful implementation of enzymes in practical application hinges on the development of efficient mass production techniques. However, in a heterologous expression system, the protein is often unable to fold correctly and, thus, forms inclusion bodies, resulting in the loss of its original activity. In this study, we present a new and more accurate model for predicting amino acids associated with an increased L-amino acid oxidase (LAO) solubility. Expressing LAO from Rhizoctonia solani in Escherichia coli and combining random mutagenesis and statistical logistic regression, we modified 108 amino acid residues by substituting hydrophobic amino acids with serine and hydrophilic amino acids with alanine. Our results indicated that specific mutations in Euclidean distance, glycine, methionine, and secondary structure increased LAO expression. Furthermore, repeated mutations were performed for LAO based on logistic regression models. The mutated LAO displayed a significantly increased solubility, with the 6-point and 58-point mutants showing a 2.64- and 4.22-fold increase, respectively, compared with WT-LAO. Ultimately, using recombinant LAO in the biotransformation of α-keto acids indicates its great potential as a biocatalyst in industrial production.

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

NAD(P)H-dependent oxidoreductases are crucial biocatalysts for synthesizing chiral compounds. Yet, the industrial implementation of enzymatic redox reactions is often hampered by an insufficient supply of expensive nicotinamide cofactors. Here, a cofactor self-sufficient whole-cell biocatalyst was developed for the enzymatic asymmetric reduction of 2-oxo-4-[(hydroxy)(-methyl)phosphinyl] butyric acid (PPO) to L-phosphinothricin (L-PPT). The endogenous NADP+ pool was significantly enhanced by regulating Preiss-Handler pathway toward NAD(H) synthesis and, in the meantime, introducing NAD kinase to phosphorylate NAD(H) toward NADP+. The intracellular NADP(H) concentration displayed a 2.97-fold increase with the strategy compared with the wild-type strain. Furthermore, a recombinant multi-enzyme cascade biocatalytic system was constructed based on the Escherichia coli chassis. In order to balance multi-enzyme co-expression levels, the strategy of modulating rate-limiting enzyme PmGluDH by RBS strengths regulation successfully increased the catalytic efficiency of PPO conversion. Finally, the cofactor self-sufficient whole-cell biocatalyst effectively converted 300 mM PPO to L-PPT in 2 h without the need to add exogenous cofactors, resulting in a 2.3-fold increase in PPO conversion (%) from 43% to 100%, with a high space-time yield of 706.2 g L-1 d-1 and 99.9% ee. Overall, this work demonstrates a technological example for constructing a cofactor self-sufficient system for NADPH-dependent redox biocatalysis.