Structure-guided discovery of highly efficient cytidine deaminases with sequence-context independence

Area of Science:

• Biochemistry
• Molecular Biology
• Structural Biology

Background:

• Cytosine base editors are crucial for gene editing but face limitations due to sequence context dependency and off-target effects.
• Developing novel base editors with improved specificity and broader applicability is essential for advancing gene therapy.

Purpose of the Study:

• To discover novel cytidine deaminases with enhanced editing efficiency, diverse editing windows, and reduced off-target effects.
• To leverage structural biology and machine learning to overcome current limitations in base editor technology.

Main Methods:

• Utilized AlphaFold2 for structural prediction of 1,483 cytidine deaminases.
• Employed partitional clustering for categorizing deaminases based on structural features.
• Experimentally characterized representative deaminases to assess editing efficiency and off-target effects.

Main Results:

• Identified several deaminases exhibiting high C-to-T conversion efficiency across various sequence contexts (AC/TC/CC/GC sites).
• Demonstrated the ability of these deaminases to introduce stop codons in mammalian genes without inducing double-strand breaks.
• Found that specific residue modifications at predicted DNA-interacting sites can significantly decrease off-target activity.

Conclusions:

• The discovery of novel deaminases expands the toolkit for precise gene editing.
• Structure-based design and machine learning hold promise for further optimizing base editors for therapeutic applications.
• These advancements pave the way for safer and more effective gene therapies.