Cell-Free Protein Synthesis as a Method to Rapidly Screen Machine Learning-Generated Protease Variants.

Area of Science:

• Biochemistry
• Molecular Biology
• Protein Engineering

Background:

• Machine learning (ML) models require extensive, high-quality data for protein structure prediction, engineering, and design.
• Traditional protein purification for data generation is time-consuming and resource-intensive, hindering ML scalability.
• Developing efficient methods for generating functional protein data is crucial for advancing ML in protein science.

Purpose of the Study:

• To present cell-free protein synthesis (CFPS) as a rapid and efficient tool for screening protein variants within ML workflows.
• To demonstrate the application of CFPS in optimizing protein function by improving the kinetic properties of a protease.
• To validate CFPS as a viable method for exploring protein fitness landscapes.

Main Methods:

• Utilized cell-free protein synthesis for rapid generation and activity assessment of diverse protein variants.
• Employed a strategy of initial random variant screening followed by targeted variant selection to efficiently explore the protein fitness landscape.
• Quantified kinetic properties of protease variants to identify improvements.

Main Results:

• Successfully screened 48 random protein variants and 32 targeted variants using CFPS.
• Identified several protease variants exhibiting enhanced kinetic properties.
• Demonstrated the speed and effectiveness of CFPS in identifying beneficial protein modifications.

Conclusions:

• Cell-free protein synthesis is a powerful and efficient platform for generating screening data to train and refine ML models in protein engineering.
• CFPS accelerates the discovery of protein variants with improved functional characteristics, such as enhanced enzyme kinetics.
• This methodology streamlines the ML-driven protein design cycle, reducing time and resource requirements.