Introduction of Our Project

Alginate is an acidic linear polysaccharide widely found in brown algae, composed of β-D-mannuronic acid (M) and α-L-guluronic acid (G) linked by 1,4-glycosidic bonds in alternating or consecutive arrangements to form polymers.

Alginate oligosaccharides (AOSs), obtained by the cleavage of alginate with alginate lyase, possess diverse biological activities, including antioxidant, anti-inflammatory, antitumor, and plant growth-promoting effects. Their activity is closely related to degree of polymerization (DP), with low-molecular-weight AOSs (DP 2~4) showing particularly strong physiological effects.

Therefore, increasing the proportion of tetrasaccharides (DP4) in alginate lyase products is valuable for understanding substrate recognition and for the targeted preparation of high-value oligosaccharides.

Idea of Our Design

Instead of just relying on trial-and-error, we combined semi-rational design — using structural biology to pick key sites for saturation mutagenesis, and machine learning — training models on experimental data to predict which mutations are most likely to boost DP4 production.

Together, this strategy lets us redesign PyAly to make more tetrasaccharides, faster and smarter.

Approach & Key Achievements

1 Understand the enzyme’s structure: We analyzed PyAly’s 3D structure and found four key amino acids (R143, R159, F170, K172) that control how sugars fit into the enzyme.

2 Make single mutations: By changing these amino acids one by one, we tested which variants increased DP4 production.

3 Use Machine Learning to predict the best combos: We trained machine learning models on our experimental data to suggest the most promising double mutations.

4 Build and test these variants: We created these new enzymes and measured their sugar outputs with HPLC.

Eventually, we identified mutations (like at R159 and K172) that significantly increased DP4 yield, built a machine learning model that predicts good mutation combinations, and developed a workflow that combines wet-lab experiments with computational predictions, making enzyme engineering more efficient and powerful.