Improving the Cβ stereoselectivity of L-threonine aldolase for the preparation of L-threo-3,4-dihydroxyphenylserine, a powerful anti-Parkinson's disease drug

L-threo-DOPS (L-threo-3,4-dihydroxyphenylserine) is a powerful anti-Parkinson's disease drug approved by FDA in 2014. The industrial production of optical L-threo-DOPS relied on optical resolution after chemical reactions, which did not fit into the atom-economy principle. L-Threonine aldolase (L-TA) could catalyze the synthesis of L-threo-DOPS from glycine and 3,4-dihydroxybenzaldehyde in one step, which is a promising alternative route. However, the substrate 3,4-dihydroxybenzaldehyde bearing an electron-rich catechol group is inactive for aldol condensation reaction, resulting in poor Cβ diastereoselectivity and low yields in enzymatic synthesis. In this work, virtual protein mutagenesis revealed the importance of His128 of L-TA to its binding with L-threo-DOPS intermediate. Further, the site-saturation mutagenesis of His128 provided a high-diastereoselectivity mutant H128N with 92.9 % de in the synthesis of L-threo-DOPS. With the whole cells harboring H128N, L-threo-DOPS was synthesized in a great de value of 95.4 % and a yield of 1.6 g/L in 1 h, which was significantly superior to the best-ever enzyme (55.4 % de). Based on the mutagenesis result and in silico docking, it was suggested that the enlarged volume and active hydrogen donor of the side chain of the amino acid 128 were disadvantageous, while proper hydrogen receptors could enhance substrate binding, leading to improved conversion.