Machine Learning Reveals Signatures of Promiscuous Microbial Amidases for Micropollutant Biotransformations

Area of Science:

• Environmental Science
• Biochemistry
• Microbiology

Background:

• Organic micropollutants (pharmaceuticals, pesticides) contaminate the environment, posing toxic risks.
• Human urine is a significant source of micropollutants, containing microbial communities capable of biotransformation.
• Amidase signature (AS) enzymes are known for broad activity, but their role in urine micropollutant transformation is unexplored.

Purpose of the Study:

• To investigate the potential of urine-derived microbial AS enzymes in micropollutant biotransformation.
• To characterize AS enzyme-substrate specificity and identify key features for biotransformation.
• To inform the development of strategies for micropollutant removal and benign-by-design chemicals.

Main Methods:

• Tested 17 diverse compounds against 40 AS enzyme homologues from urine microbial isolates.
• Assessed enzyme activity and substrate specificity.
• Employed a machine learning model to identify critical chemical and amino acid features.

Main Results:

• 16 of 40 AS enzymes showed activity on at least one tested micropollutant.
• Enzymes exhibited diverse substrate specificities, with some active on up to nine compounds.
• Identified molecular weight and charge as key chemical features, and specific amino acids near the active site as important protein features.

Conclusions:

• Urine microbial AS enzymes represent an underutilized resource for micropollutant biotransformation.
• Insights into substrate and enzyme features can guide the design of novel environmental remediation technologies.
• This research provides a foundation for developing targeted approaches to mitigate micropollutant pollution.