Structure and evolution-guided design of minimal RNA-guided nucleases

Area of Science:

• Protein engineering
• Synthetic biology
• Genomics

Background:

• Designing novel RNA-guided nucleases with enhanced properties is crucial for advancing programmable genome editing.
• Generating diverse, enzymatically robust multi-domain proteins presents a significant challenge in protein engineering.

Purpose of the Study:

• To develop an AI-driven strategy for designing active, divergent variants of TnpB, a minimal CRISPR-Cas12-like nuclease.
• To expand the capabilities of programmable genome editing by creating nucleases with properties not constrained by natural evolution.

Main Methods:

• Employed an artificial intelligence strategy combining structure-guided inverse protein folding and evolution-informed residue constraints.
• Generated and functionally screened a library of AI-designed TnpB variants using high-throughput methods.
• Determined the structure of a highly active variant using cryo-electron microscopy (Cryo-EM).

Main Results:

• AI-generated variants of TnpB demonstrated retained or enhanced nuclease activity compared to wild-type in bacterial, plant, and human cells.
• The most divergent active variant exhibited new stabilizing contacts at RNA/DNA interfaces across different conformational states.
• Cryo-EM analysis revealed insights into the structural basis of enhanced activity and stability.

Conclusions:

• The study establishes a powerful strategy for designing non-natural RNA-guided nucleases with tailored properties.
• This AI-driven approach significantly enlarges the designable protein space for nucleic acid binders and editing tools.
• The findings demonstrate the potential for creating novel biological tools beyond natural evolutionary constraints.