Structural Stabilization and Activity Enhancement of Glucoamylase via the Machine-Learning Technique and Immobilization

Abstract

Glucoamylases (GLL) hydrolyze starch to glucose syrup without yielding intermediate oligosaccharides, but their lack of stability under industrial conditions poses a major limiting factor. Using consensus- and ancestral-based machine-learning tools, a functional GLL with six mutations (GLL^{I73l/T130V/N212V/D238G/N327M/S332P}) was constructed that exhibited superior hydrolytic activity relative to the wild-type (WT-GLL). An oxidized multi-walled carbon nanotube (oMW-CNT) was used as a solid support to immobilize the WT-GLL with an immobilization capacity of 211.28 mg/g. The specific activity of mutant GLL-6M and GLL@oMW-CNTII was improved by 2.5-fold and 3.9-fold respectively, with both retaining 64.5% residual activity after incubation at 50 °C for 2 h compared to the WT-GLL with 42.6% activity. GLL and GLL-6M were however completely inactivated at 55 °C in 30 min while oMW-CNTII retained ∼43.1% activity. Our results demonstrate that employing a machine-learning approach for enzyme redesign and immobilization is a practicable alternative for improving enzyme performance and stability for industrial applications.