CRISPR gets a power boost from AI-designed ‘molecular scissors

Harnessing AI for Synthetic CRISPR Proteins

The research, titled Structure and evolution-guided design of minimal RNA-guided nucleases, was conducted by a team including scientists from the Innovative Genomics Institute and the California Institute for Quantitative Bioscience, both at the University of California, Berkeley, and collaborators at other institutions. The team was led by University of California, Berkeley biochemist Jennifer Doudna, who won the 2020 Nobel Prize for her CRISPR-related work.

Soeren Lienkamp, a molecular biologist at the University of Zurich in Switzerland who was not involved in the research, noted that the paper marries two transformative fields: AI-guided design and enzymes called RNA-guided nucleases, which can cut DNA and RNA strands. Much like CRISPR democratized the ability to edit DNA at will, AI-based protein design promises to allow anyone to create totally novel properties in the protein space, Lienkamp added.

Overcoming Evolutionary Constraints

CRISPR systems rely on the machinery that bacteria use to defend themselves against viruses. These systems use a “guide RNA” to direct a nuclease—acting like molecular scissors—to a target DNA sequence, where it can snip out material to enable scientists to edit, delete, or add genetic information. While tools like Cas9 and Cas12 are powerful, the process is complex. The activity of these enzymes depends on coordinated RNA and DNA recognition, activation, and cleavage by distinct conformational states. Because of this, seemingly small changes can disrupt enzyme activity, making the design of synthetic versions a difficult challenge for protein design methods.

Previous attempts to use sequence-based biological language models often produced versions of proteins that closely resembled the reference sequences used to train them. To address this, Petr Skopintsev and colleagues utilized a strategy built around ESM Inverse Folding (ESM-IF1). Instead of allowing the model to drift toward sequences close to training references, the team introduced evolution-informed residue constraints designed to keep key functional elements in place while still permitting large sequence divergence.

The Development of SynTnpBs

The researchers focused their starting point on TnpB, a minimal CRISPR-Cas12-like nuclease. By applying their AI-based framework, they designed new variants they call SynTnpBs. These synthetic nucleases are engineered to remain RNA-guided and catalytically active despite being non-natural. This minimal nuclease context was particularly important because multi-domain architectures can be fragile, and the team sought to avoid the common pitfalls where small changes destroy activity.

The study highlights AI’s ability to expand the CRISPR toolbox to include RNA-guided nucleases with novel properties beyond those found in nature. The scientists wrote that the results establish a strategy for creating non-natural RNA-guided nucleases and conformationally active nucleic acid binders, enlarging the designable protein space. By using structure-guided protein design, the team successfully yielded genome-editing proteins with substantially different sequences while preserving function.

Future Implications for Genome Editing

Advancements in CRISPR-Cas9 systems have revolutionized genetic engineering, yet current enzymes still produce off-target effects. Researchers have been exploring ways to extend the capabilities of these systems for more efficient gene editing, and the ability to redesign nucleases to be more specific is a major goal. By establishing an approach to create non-natural nucleases, this research expands the scope for protein engineering to improve existing CRISPR-Cas9 tools.

Such synthetic CRISPR systems could one day power discoveries in fields from medicine to agriculture. As the work shows, the ability to generate functional equivalents with novel sequence features, rather than near replicas of known enzymes, is a significant step forward for the field of biotechnology.