Machine learning-guided malate dehydrogenase engineering for improved production of L-malic acid in Aspergillus niger

Abstract

L-malic acid (MA) is an important four-carbon platform compound, with wide applications in food, pharmaceuticals, and cosmetics. Malate dehydrogenase (MDH), which also catalyzes the reduction of oxaloacetate (OAA) as the reaction is reversible, is the key enzyme for microbial production of L-MA. Here, machine-learning-guided protein engineering was applied to AnMDH from Aspergillus niger, resulting in the best variant G212A/G234 V with a remarkable 11-fold improvement of enzyme activity. Then, the efficient L-MA-producing strain A. niger RG020 was constructed by introducing AnMDH^G212A/G234V into A. niger RG018. Shake-flask and bioreactor studies showed that A. niger RG020 had an increased L-MA production rate and yield compared to the parental strain RG018. In a 5-L batch bioreactor, the fermentation period was shortened from 146 to 104 h, with the productivity of MA increasing from 0.93 to 1.42 g/L/h. Our work provides practical guidance for improving the productivity of industrial strains.