Rational selection of ketopantoate hydroxymethyltransferase with low product feedback inhibition based on binding...

The microbial production of pantothenic acid (d-PA) is critically limited by feedback inhibition and low activity of the key enzyme ketopantoate hydroxymethyltransferase (KPHMT). To overcome this, rational enzyme mining based on computational prediction was established. Following sequence conservation analysis, molecular dynamics simulations, and binding free energy (ΔG) calculations, five representative native KPHMT enzymes were selected for experimental validation: EcKPHMT from Escherichia coli, CgKPHMT from Corynebacterium glutamicum, MpKPHMT from Mangrovibacter plantisponsor, EpKPHMT from Enterovibrio pacificus, and BsKPHMT from Bacillus subtilis. EpKPHMT and BsKPHMT exhibited 4.25- and 4.60-fold times that of EcKPHMT, respectively, with relieved pantoate feedback inhibition (IC₅₀: 14.07 and 19.86 mM vs. 1.08 mM) and virtually no inhibition by d-PA. The strain expressing BsKPHMT enhanced d-PA and pantoate titers by 74.30% (3.12 g/L) and 140.0% (0.84 g/L) in shake flasks, and further increased d-PA production by 55.4% in a 5 L bioreactor, over the control. This work established a predictive framework for mining superior enzymes based on in silico prediction, offering a valuable strategy for metabolic engineering of high-value chemicals.IMPORTANCEThe industrial-scale biosynthesis of pantothenic acid (d-PA) is often bottlenecked by the strict feedback inhibition of its key biosynthetic enzyme, ketopantoate hydroxymethyltransferase (KPHMT). This study describes a computational strategy for the mining and selection of naturally occurring KPHMTs with reduced feedback inhibition, providing superior genetic parts for metabolic applications. This approach provides a novel and rational framework for mining allostery-free enzymes, successfully delivering two highly efficient biocatalysts, BsKPHMT and EpKPHMT. These biocatalysts exhibit immediate potential for industrial applications, offering a direct solution to enhance the production of both pantoate and d-PA. Collectively, the integrated methodology demonstrates effective translation from fundamental discovery to practical application, presenting a generalizable model for overcoming similar metabolic bottlenecks.