Improved Bst DNA Polymerase Variants Derived via a Machine Learning Approach

Area of Science:

• Biochemistry
• Molecular Biology
• Enzyme Engineering

Background:

• DNA polymerase I from Geobacillus stearothermophilus (Bst DNAP) is crucial for isothermal amplification due to its strand displacement activity.
• Enhanced enzyme robustness is needed for diagnostic applications, particularly for high-temperature reactions that increase speed.

Purpose of the Study:

• To improve the stability and thermotolerance of Bst DNAP for advanced diagnostic applications.
• To explore the use of machine learning in predicting beneficial enzyme mutations.

Main Methods:

• A fusion domain from the actin-binding protein villin was appended to Bst DNAP to enhance stability and purification.
• A machine learning algorithm was developed to predict amino acid substitutions based on their microenvironment.
• Predicted sequence substitutions were introduced into Bst DNAP to assess their impact on thermotolerance.

Main Results:

• Enzyme variants with significantly improved thermotolerance were identified through machine learning predictions.
• Combined mutations resulted in additive thermostability, increasing denaturation temperatures by up to 2.5 °C compared to the parental enzyme.
• The thermostabilized enzyme facilitated faster loop-mediated isothermal amplification assays at 73 °C, a temperature inactivating other Bst DNAP versions.

Conclusions:

• Machine learning can effectively identify mutations for enzyme thermostabilization.
• The engineered Bst DNAP exhibits enhanced stability, enabling faster and higher-temperature isothermal amplification assays.
• This work represents a novel application of machine learning for enzyme engineering in molecular diagnostics.