AI Boosts Gene Editing for Epigenetic Therapies

Gene editing and artificial intelligence (AI) are two powerful discoveries that have the potential to revolutionize disease treatment. While gene editing can correct genetic defects, AI can analyze data better for improved decision-making. Using these technologies together would no doubt improve patient outcomes and potentially cure diseases.

Now a new AI program from the Grossman School of Medicine at NYU Langone-Health and the University of Toronto is taking an older gene editing technology and improving it to enhance and expedite the development of gene therapies – ones that can epigenetically modify the expression of genes in need of regulation.

Gene therapy (GT) is one of the fastest-growing sectors of medicine today, although most applications are primarily used in the research setting. One main reason for this lies in the risks associated with GTs. For instance, gene editing is a technique that directly alters a cell’s DNA code. If this type of GT isn’t delivered correctly, it could affect off-target cells in the body, causing harm.

Newer technologies like CRISPR-Cas9 have proven to be more targeted in their GT approach, but certain characteristics of these methods still pose a risk. For one, Cas9’s large protein size makes its delivery harder, lowering therapy efficacy. Plus, the immune system might recognize the protein, making it unsuitable for long-term treatment.

In the current study, the researchers sought to improve upon the use of Zinc finger (ZF) nucleases for gene editing. ZFs are one of the earliest and more commonly used tools for controlling and editing genes. They are also the smallest and less likely to trigger an immune response since they share a DNA-binding domain with many human transcription factors (TFs). Thus, they can easily latch onto TFs and draw them towards a specific gene that needs regulation.

However, designing ZFs for a particular task can be challenging. Researchers need to determine countless combinations to understand how each ZF interacts with others to achieve the desired genetic change. Here, the authors interfaced AI with ZF to overcome these obstacles with a new technology they call “ZFDesign”.

“Our program can identify the right grouping of zinc fingers for any modification, making this type of gene editing faster than ever before,” says first author David Ichikawa, PhD at NYU’s Grossman School of Medicine.

Ichikawa believes that using ZFs to edit genes could be safer than using CRISPR, which relies on bacterial proteins. He explained that the foreign proteins in CRISPR could trigger the immune system and cause harmful inflammation, whereas ZFs are human-derived and less likely to cause an adverse immune response. As well they are smaller and more versatile for gene therapy options.

The researchers also noted that the small size of ZFs provides more flexible options for gene therapy compared to CRISPR. This is because the compact nature of the tool enables multiple delivery methods to target the appropriate cells in patients.

Study author Marcus Noyes, PhD, assistant professor in the Department of Biochemistry and Molecular Pharmacology at NYU, added, “By speeding up zinc-finger design coupled with their smaller size, our system paves the way for using these proteins to control multiple genes at the same time. In the future, this approach may help correct diseases that have multiple genetic causes, such as heart disease, obesity, and many cases of autism.”

Dr. Noyes and his team used a tailored ZF-type to disrupt the gene coding sequence in human cells as a test for the computer’s AI design code. They also produced several ZFs that successfully reprogrammed TFs to bind near a specific gene sequence and increase or decrease its expression level, demonstrating the technology’s potential for epigenetic changes.

Because ZFs can be difficult to control and not always specific to a single gene, Dr. Noyes stresses that more research and testing are needed. The team plans to improve the AI program to develop more accurate zinc-finger groupings that only cause the desired edits.

Overall, the creation of the ZFDesign tool has generated exciting opportunities in systems and synthetic biology, allowing for the examination and alteration of gene regulation on a large-scale genome-wide level.

Source: Ichikawa, D.M., et al. A universal deep-learning model for zinc finger design enables transcription factor reprogramming. Nature Biotechnology, January 2023.

Reference: New Artificial Intelligence Tool Makes Speedy Gene Editing Possible. NYU Langone- Health / NYU Grossman School of Medicine. January 26, 2023.