Computational Redesign of the Substrate Binding Pocket of Glutamate Dehydrogenase for Efficient Synthesis of Noncanonical l-Amino Acids

Noncanonical amino acids are attractive molecules in synthetic chemistry as important building blocks for pharmaceuticals and agrochemicals. Asymmetric reductive amination of prochiral keto acids catalyzed by glutamate dehydrogenase (GluDH) is an efficient process for the production of these molecules. However, the stringent substrate specificity and the lack of an engineering strategy hampered its wide application in substrate-oriented chiral carbon–nitrogen bond formation. To address this issue, first, the substrate recognition mechanism of GluDH was investigated using CdGluDH (an NADH-dependent GluDH from Clostridium difficile) and four molecules including the natural and three non-natural substrates as models through computational simulation and experimental verification. Second, a computational engineering strategy based on reducing the pocket steric hindrance and tuning enzyme–substrate electrostatic interactions and/or X–H···π interactions was developed to systematically redesign the binding pocket with mutations of 10 sites to accept distinct substrates. As a result, tailor-made variants were created to synthesize the corresponding high-value products, l-norvaline, l-phosphinothricin, and l-homophenylalanine, with specific activities improved 2.3-fold, 916.2-fold, and from no activity to 66.64 U/mg, respectively. Finally, reductive amination efficiency of the variants was tested by a simulated industrial catalytic reaction, demonstrating that the computational enzyme redesign strategy is promising in enhancing the production of noncanonical amino acids using GluDHs.