Decoding cryptic defluorinases through a latent generative sequence landscape

Area of Science:

• Biochemistry and enzymology
• Bioinformatics and computational biology
• Environmental microbiology

Background:

• The carbon-fluorine bond's strength makes fluorinated natural products rare, despite the existence of natural defluorinase enzymes.
• Fluoroacetate dehalogenase is a well-characterized defluorinase, but the full scope of defluorination chemistry in nature is underexplored.
• Understanding and discovering defluorinases is crucial for both fundamental science and potential biotechnological applications.

Purpose of the Study:

• To develop and apply a machine learning framework, latent generative landscapes (LGLs), to explore the functional sequence space of the alpha/beta-hydrolase superfamily.
• To identify novel defluorinase enzymes within the biosphere.
• To experimentally validate and characterize newly discovered defluorinases.

Main Methods:

• Training and application of a machine learning-based framework (latent generative landscapes - LGLs) to predict enzyme function.
• Mapping the functional sequence space of the alpha/beta-hydrolase superfamily to identify potential defluorinases.
• Experimental validation including enzyme assays, thermal stability measurements (Tm), and enantioselectivity determination.

Main Results:

• Identification of 3014 putative defluorinases previously unannotated or misannotated.
• Experimental validation and reclassification of five novel defluorinases.
• The newly discovered defluorinases exhibit high thermal stability (Tm > 70 °C) and diverse catalytic efficiencies with conserved enantioselectivity.
• One novel enzyme (A0A4Z0BVY8) demonstrated 2.7-fold higher activity than the current state-of-the-art enzyme (Q6NAM1).

Conclusions:

• Latent generative landscapes (LGLs) modeling is a powerful strategy for uncovering cryptic carbon-fluorine bond chemistry in nature.
• This approach significantly expands the known repertoire of defluorinating biocatalysts.
• The findings pave the way for future discovery and engineering of novel defluorination enzymes for biotechnological purposes.