Related Experiment Video
Updated: Aug 5, 2026
11:13
Mass Spectrometry-Guided Genome Mining as a Tool to Uncover Novel Natural Products
Published on: March 12, 2020
Machine learning-assisted discovery of AdMysD for enhanced porphyra-334 biosynthesis
Longfei Yuan1, Shuting Feng2, Sheng Xie3
1State Key Laboratory of Ocean Sensing & Ocean College, Zhejiang University, Zhoushan 316021, China; Hainan Institute, Zhejiang University, Sanya 572025, China.
Bioorganic Chemistry
|August 1, 2026
Summary
Researchers discovered a new enzyme, AdMysD, that significantly enhances the production of UV-protective compounds called mycosporine-like amino acids (MAAs). This finding offers a powerful biocatalytic tool for efficient MAA synthesis.
Area of Science:
- Biochemistry
- Enzyme Engineering
- Metabolic Engineering
Background:
- Mycosporine-like amino acids (MAAs) are vital secondary metabolites offering UV protection and antioxidant benefits.
- The enzyme MysD is crucial for MAA biosynthesis, catalyzing a key step that shifts UV absorption to the UVA range.
- MysD is a rate-limiting enzyme, and limited characterized enzymes hinder discovery and application.
Purpose of the Study:
- To develop a data-driven framework for discovering novel MysD enzymes.
- To identify and characterize new enzymes with enhanced catalytic efficiency for MAA production.
- To explore the substrate promiscuity of newly discovered MysD enzymes.
Main Methods:
- Integrated screening framework combining Sequence Similarity Networks (SSN), deep representation learning (UniRep), and Positive-Unlabeled Bagging (PU Bagging).
- Computational pipeline to prioritize potential MysD candidates from a large dataset of unannotated homologues.
- Heterologous expression and biochemical characterization of prioritized enzyme candidates.
Main Results:
- The screening framework successfully reduced 951 unannotated homologues to 42 prioritized candidates.
- Discovery of AdMysD from Aphanothece hegewaldii with a 3-fold higher catalytic efficiency for porphyra-334 production compared to NlMysD.
- AdMysD demonstrated substrate promiscuity, accepting L-threonine, L-serine, L-alanine, and L-cysteine.
Conclusions:
- The developed PU-learning-based strategy is effective for identifying rare enzyme families with limited functional data.
- AdMysD represents a significant advancement in biocatalysis for efficient MAA production.
- This study provides a valuable biocatalytic tool and a novel approach for enzyme discovery.
