Metagenomi, Inc., a precision genetic medicines company committed to developing curative therapeutics for patients using its proprietary gene editing toolbox, today presented a poster titled “Unlocking the Therapeutic Potential of Compact SMART Nucleases through Ancestral Reconstruction, Structure-Guided Engineering, and Generative AI” at the tenth meeting on Genome Engineering: CRISPR Frontiers at Cold Spring Harbor Laboratory, hosted in Cold Spring Harbor, NY from August 27-31, 2024. The work leverages the company’s metagenomics-derived gene editing tools and a variety of AI methods to rapidly develop novel programmable CRISPR nuclease and base editing systems.
“Gene editing is a revolutionary technology and novel gene editing tools are needed to address limitations with current systems, including the need for small systems that are compatible with currently available delivery technologies,” said Brian C. Thomas, CEO and founder of Metagenomi. “While there is no shortage of naturally occurring proteins that we can discover using our metagenomics platform, AI is one of several approaches we leverage to generate novel systems capable of highly efficient in vivo gene editing. We believe our metagenomics-based, AI-enhanced approach will allow us to achieve our goal of having the most effective gene editing tools to address any genetic disease anywhere in the human genome.”
In today’s poster presentation, Metagenomi demonstrated that metagenomics-informed synthetic sequence generation, including ancestral sequence reconstruction and generative AI techniques, can produce novel nucleases and base editors. These AI-based gene editing system variants show high editing efficiency in mammalian cells and include compact base editors that are less than 1,000 amino acids in length. Other key points from the presentation include:
- Both the SMART nucleases and base editors fit well within the carrying capacity of viral vector-based delivery systems.
- Compact SMART nucleases are promising for genome editing due to their small size but are extremely rare in nature, significantly limiting engineering efforts based on similar proteins.
- De novo, AI-generated, synthetic nucleases expand on natural systems and demonstrate robust activity in mammalian cells.
- Rational engineering further improved activity based on a newly solved structure of the enzyme and achieved saturating levels of editing in human cells (Ocampo, Rodrigo Fregoso, et al. “DNA targeting by compact cas9d and its resurrected ancestor.” bioRxiv, https://doi.org/10.1101/2024.04.08.588528).
“While other groups use generative AI to focus on creating systems very similar to SpCas9, our work highlights how Metagenomi is uniquely positioned to combine the incredible diversity from our metagenomics database of over 11 billion proteins with our understanding of both SMART biochemistry and generative AI to create gene editing tools that open up new capabilities and therapeutic applications,” said Christopher Brown, Head of Discovery at Metagenomi. “Combining metagenomics and generative AI is in its early stages, but our poster at today's CRISPR Frontiers Conference suggests the potentially significant value we seek to realize by training generative AI models with our vast collection of proteins recovered from the natural environment. Large datasets such as ours allow generative AI algorithms to find novel solutions with even greater precision and speed, and are valuable resources for accelerating drug development.”