Research On A Novel Enzyme Computational Strategy Based On Computational Enzyme Design And Molecular Dynamic Simulation

The application of chemical synthesis is indispensable in modern society;however,concerns about environmental hazards associated with its synthesis have led to a growing interest in more environmentally friendly and sustainable synthetic methods.The introduction of bio-based chemicals has opened new avenues for environmentally friendly synthesis.Enzymes play a crucial role as key catalysts in the production of bio-based chemicals,enabling efficient,selective,and environmentally friendly synthetic processes under relatively mild conditions.However,natural enzymes are often susceptible to environmental factors such as heat,p H,and other conditions,making their direct application in industrial production challenging.Therefore,the modification of natural enzymes becomes particularly important.Computational enzyme design provides a powerful tool for designing and optimizing the catalytic activity,selectivity,and stability of enzymes for specific reactions.In this project,computational methods are primarily employed to address two key issues:（1）developing universal,mechanism-based computational design strategies,and（2）achieving efficient design by modifying the reaction activity and selectivity of enzymes.This is focused on four representative bio-based chemical transformation processes:the synthesis of 4-hydroxyisophthalic acid,2-hydroxyterephthalic acid,adipic acid,and glucaric acid.The investigation involves a combination of experimental and multi-scale simulation methods.First,based on structural analysis,we screened a promising 5-carboxyvanillic acid decarboxylase as the most suitable enzyme for catalyzing carboxylated benzene ring small molecules.We experimentally assessed its specific activity towards different substrates,revealing a specific activity of 2.966 nmol mg^-1 min^-1 for catalyzing vanillic acid and 0.279nmol mg^-1 min^-1 for catalyzing p-hydroxybenzoic acid.It was found to be inactive towards m-hydroxybenzoic acid,and its ability to catalyze the carboxylation of p-hydroxybenzoic acid was discovered for the first time.We then established a comprehensive simulation index evaluation system based on computational design of three small molecule transition state structures combined with molecular dynamics simulations that corresponded well with experimental activities.Subsequently,we employed this high-throughput MD-assisted computational design strategy to modify enzyme activity for the synthesis of 4-hydroxyisophthalic acid from p-hydroxybenzoic acid.From a sequence space of 20⁷,we generated 273 designed sequences,from which 103 sequences were selected for MD simulation.Finally,13 focused experimental libraries were determined.The optimal M5 mutant（L47M）showed a threefold improvement in carboxylation activity,achieving a specific activity of 1.01 nmol mg^-1 min^-1at 30℃and 1.745 nmol mg^-1 min^-1 at 40℃,confirming the success of the design strategy.We further analyzed the reason why the 5-carboxyvanillic acid decarboxylase could not catalyze the carboxylation reaction of m-hydroxybenzoic acid.Consequently,we designed a novel hydrogen bond network to better accommodate the transition state of m-hydroxybenzoic acid.From a sequence space of 20⁵,we obtained 73 designed sequences,from which 52 sequences were selected for MD simulation.Finally,8 focused experimental libraries were determined.The yield of the M2 mutant（S91T/Y317F/L47N/R58W）can reach1200 ng/m L after 24 hours of reaction,with a specific activity of 0.0256 nmol mg^-1 min^-1.thereby modifying the enzyme’s substrate selectivity to achieve biocatalysis of a non-natural reaction.Furthermore,multiscale simulation methods were employed to design highly active mutations of enoate reductase for the synthesis of adipic acid.Due to the lack of a crystal structure,the catalytic mechanism of the enoate reductase ERBC is not well understood,making the computational design in this study particularly challenging.Firstly,we obtained a high-precision structure of the enoate reductase ERBC through structure prediction tools and bioinformatics analysis.Quantum mechanical cluster calculations were used to determine its catalytic mechanism and rate-limiting steps.Subsequently,utilizing a computational enzyme design approach based on transition state structures,we obtained 430low-energy sequences from a sequence space of 20⁷.Further screening using high-throughput MD simulations to eliminate false-positive designs resulted in 141 sequences.A focused library was finally generated with only 15 variants.Experimental validation confirmed that the best variant,M8（G27M/I374S）,achieved an in vivo enzyme activity of 2.75 m U/mg,a14.2-fold increase in catalytic activity compared to the wild type（0.19 m U/mg）.The production of adipic acid reached 0.9 g/L with a yield of 61%when adding 10 m M cis,cis-muconic acid,which was approximately 19 times higher compared to the wild type（3.2%）.Finally,this study shifted its focus to how multi-enzyme self-assembled systems at a larger scale affect enzyme activity.Previous experiments have verified that the self-assembly system of glucose oxidase（GOX）and catalase（CAT）constructed through supramolecular recognition betweenβ-cyclodextrin（β-CD）and adamantane（AD）can effectively enhance the production of gluconic acid.However,it is still unclear how theβ-CD/AD inclusion ratio affects the self-assembly process and the optimal enzyme ratio.Therefore,molecular dynamics simulations based on coarse-grained（CG）methods were used to simulate the self-assembly process between the two enzymes.Compared to low grafting conditions,under high grafting conditions,the dual enzymes achieved the self-assembly process within a reasonable simulation time scale.The results for different enzyme ratios also showed significant differences under different grafting conditions.Exploring the self-assembly process under various enzyme addition ratio conditions revealed varying degrees of enzyme self-aggregation not observed in experiments.This precisely elucidated that the higher the grafting ratio,the lower the enzyme activity in the self-assembly system.Only when the enzyme ratio is matched at 1:2 and the grafting ratio is at 2:1,can the two enzymes form an ordered supramolecular self-assembly structure.In other systems,GOX or CAT underwent self-aggregation,with CAT exhibiting a stronger tendency for self-aggregation.In summary,this paper employed a multiscale molecular simulation approach to achieve activity modification of carboxylases and enoate reductases,and studied the essence of enzyme activity changes at a microscopic level.Initially,starting with the catalytic activity of 5-carboxyvanillic acid decarboxylase,a universally applicable computational enzyme design workflow was developed based on atomic-scale catalytic mechanism elucidation and molecular-scale dynamic simulation methods.This design strategy was successfully applied to the activity modification of carboxylation and double bond reduction reactions,resulting in enhanced activity from p-hydroxybenzoic acid to 4-hydroxyisophthalic acid synthesis and the first biosynthesis from metahydroxybenzoic acid to 2-hydroxyterephthalic acid,as well as activity design of cis,cis-muconic acid to adipic acid with different catalytic mechanisms.Additionally,utilizing coarse-grained molecular dynamics simulations,the impact of proportions during the self-assembly process of the cascade reaction of GOX/CAT on enzyme activity was revealed.Different degrees of enzyme self-aggregation,not observable in experiments,were discovered and appropriate assembly proportions were determined.This multiscale enzyme self-assembly model can be used to optimize the design of the GOX/CAT system and further enhance enzyme activity.