AlphaFold3-guided optimization of a photoactivatable endonuclease for top-down genome engineering.

Area of Science:

• Biotechnology
• Synthetic Biology
• Structural Biology

Background:

• Artificial intelligence, particularly protein structure prediction, is revolutionizing enzyme engineering.
• Photoactivatable restriction enzymes offer precise control over genome engineering.
• MagMboI, a split-enzyme, utilizes light-inducible dimerization to control nuclease activity.

Purpose of the Study:

• To enhance the MagMboI photoactivatable restriction enzyme using AI-guided structural insights.
• To investigate the structural basis of MagMboI-DNA interaction for improved enzyme design.
• To develop advanced light-controlled genome engineering tools.

Main Methods:

• Utilized AlphaFold3 for modeling the MagMboI-DNA complex structure.
• Employed a split-protein strategy with light-inducible dimerization modules.
• Introduced and tested a redesigned enzyme variant (MagMboI-plus) in Saccharomyces cerevisiae.

Main Results:

• AlphaFold3 provided structural insights into MagMboI's interaction with its 5'-GATC-3' DNA target.
• An alternative split site was identified, increasing interface area and complex stability in MagMboI-plus.
• MagMboI-plus showed increased in vivo DNA cleavage activity but also more genomic rearrangements upon light activation.

Conclusions:

• AI-based structure prediction accelerates functional improvements in engineered enzymes.
• AlphaFold3 facilitates the rational design of novel genome engineering tools.
• The redesigned MagMboI-plus variant demonstrates potential and highlights considerations for light-controlled genome editing applications.