Computational Redesign of an Ancestral Xylose Isomerase: Tuning the Substrate Preference and Thermostability for Biomass Valorization

Xylose isomerase is a promising biocatalyst for lignocellulose valorization, but natural enzymes are limited by a preference for either d-xylose or d-glucose. Here, we integrated ancestral sequence reconstruction with deep learning methods to identify ASR285, an enzyme active toward both d-glucose and d-xylose. To enhance its potential for practical applications, we designed a flexible lid by truncating the α3-helix and reducing steric hindrance through the W140F mutation. Guided by computational analysis, the ASR285-M2 mutant (ASR285-Δhelix9/W140F/S147A/W189Y) was engineered, resulting in a 7.85-fold increase in the catalytic activity toward d-glucose while preserving the native d-xylose isomerization capacity. It also exhibited an approximately 2-fold longer half-life, indicating improved thermostability. In real lignocellulosic hydrolysates, ASR285-M2 achieved a 9-fold higher d-fructose yield than ASR285, enabling simultaneous production of d-fructose and d-xylulose. This semirational strategy successfully optimized both substrate preference and stability, providing a practical biocatalyst for biomass valorization.