Pythia: Non-random DNA repair allows predictable CRISPR/Cas9 integration and gene editing

Area of Science:

• Molecular Biology
• Genomics
• Bioengineering

Background:

• CRISPR-based genome engineering offers significant potential for scientific research and therapeutic development.
• Efficient and controlled DNA integration and editing remain challenging in various cellular environments due to limitations in managing the DNA repair process.

Purpose of the Study:

• To develop predictable methods for controlling DNA repair at the genome-cargo interface for enhanced CRISPR applications.
• To facilitate the integration of large DNA sequences and precise small edits in diverse cell types and organisms.

Main Methods:

• Utilized deep learning models to predict DNA repair outcomes based on sequence context.
• Designed and implemented triplet base-pair repeat repair arms (trimologies) to guide microhomology-mediated end joining.
• Employed oligonucleotide templates for precise single or double nucleotide edits.

Main Results:

• Demonstrated predictable DNA repair at the genome-cargo interface using deep learning predictions and sequence-specific rules.
• Successfully integrated large DNA cargo (>2 kb) and performed precise small edits in over 30 loci across human cells and in vivo models.
• Achieved germline-transmissible transgene integration in Xenopus and endogenous tagging in differentiated mouse brain cells, showcasing broad applicability.

Conclusions:

• Developed a novel strategy using predicted repair arms for efficient and controlled large DNA integration and small nucleotide edits.
• The Pythia design tool (pythia-editing.org) offers a user-friendly platform to optimize custom genome engineering strategies.
• This approach significantly expands the utility of CRISPR technology for experimental and therapeutic applications across various cell types and organisms.