AI‑Engineered Synthetic Nuclease Boosts CRISPR Editing Efficiency

A synthetic RNA-guided nuclease engineered using AI and structural biology showed high editing efficiency, revealing a strategy to improve CRISPR‑based genome editing.

Image credit:© iStock.com, Panuwach

Advancements in CRISPR‑Cas9 systems have enabled precise gene modifications, transforming genetic engineering. However, off‑target activity remains a challenge, prompting researchers to seek ways to sharpen the specificity and efficiency of these tools.

One approach involves redesigning nucleases—the molecular scissors that cleave DNA during editing. These enzymes are guided by RNA sequences that home in on target sites, but their multi‑domain architecture makes rational engineering difficult without compromising function.

Researchers led by University of California, Berkeley biochemist Jennifer Doudna, a 2020 Nobel laureate for CRISPR work, employed an AI‑driven inverse protein‑folding model to craft a synthetic nuclease with reduced off‑target effects. The newly designed enzyme performed as well as or better than its natural counterpart. Their framework, reported in Science, opens a path to generate non‑natural nucleases and expand protein‑engineering possibilities for CRISPR‑Cas9 tools.

To build this nuclease, Doudna’s team used an in‑silico pipeline that reverses engineers protein sequences from a desired three‑dimensional structure. They applied the method to TnpB, a CRISPR‑Cas12a‑like nuclease family. Phylogenetic analysis indicated that the redesigned TnpB variants diverged from the wild‑type sequence, yet the computational design introduced mutations that could affect guide‑RNA and DNA binding.

The scientists countered potential disruptions by layering evolutionary constraints onto the AI model, limiting mutations in critical amino‑acid residues. The resulting sequences retained fewer changes in key binding regions, suggesting higher compatibility and likely increased activity.

Full‑length TnpB variant proteins were expressed in bacterial cells, and the most active candidates were selected for further testing in plant and animal cells. Gene‑editing assays pinpointed a synthetic TnpB variant (SynTnpB) that delivered the highest editing efficiency while minimizing off‑target cleavage.

Structural investigation using cryo‑electron microscopy revealed how SynTnpB achieves its improved performance. Comparison of the wild‑type and engineered proteins showed new electrostatic and hydrogen‑bond interactions at the guide RNA–DNA interface, a feature attributed to the AI‑based, evolution‑informed design.

These findings demonstrate that integrating artificial intelligence with evolutionary insights can produce superior RNA‑guided nucleases, advancing the precision of CRISPR genome editing.

Source link