Researchers in South Korea Develop ‘OpenABE,’ Overcoming Limitations of AI-Designed Gene Editors

Researchers in South Korea have developed a new technology that overcomes the limitations of next-generation gene editors designed with artificial intelligence (AI), substantially improving their efficiency and safety.

Sungkyunkwan University said on the 31st that a research team led by Dae-sik Kim, professor in the Department of Medicine, jointly developed the next-generation adenine base editor ‘OpenABE’ with Yong-seop Kim, professor at the University of Ulsan College of Medicine, and Jae-hyun Park, professor at Sungkyunkwan University School of Medicine. The study is significant because it overcame the limitations of early AI-designed gene editors using structure-based protein engineering.

Adenine base editors (ABEs) are an advanced biotechnology that corrects an incorrect adenine (A) in a DNA sequence to the normal guanine (G). Research using AI to rapidly design new gene editors that do not exist in nature has recently attracted attention, but early models had low editing efficiency and safety concerns, including off-target effects.

The joint research team precisely analyzed the three-dimensional structure of the gene editor using ‘AlphaFold’ predictions. This enabled the researchers to identify a key region that helps the editor firmly hold onto target DNA without losing it. The team introduced customized mutations into this region and refined the structure by attaching a specialized tail structure derived from the highest-performing existing gene editor.

As a result, the newly developed ‘OpenABE 1.1’ and ‘OpenABE 1.2’ were found to have up to 36 times greater editing activity than existing AI gene editors. Their performance was comparable to that of ‘ABE8e,’ which is currently used as a standard technology. The editors also achieved high precision in correcting only the desired genes by dramatically reducing off-target effects around target genes.

In particular, the study demonstrated that the technology can precisely edit not only DNA in the cell nucleus but also mitochondrial DNA, which has been difficult to edit because of its structural characteristics. The research team also packaged the gene editor in engineered virus-like particles (eVLPs) and delivered it into cells, achieving the safety of selectively editing only the genes requiring treatment.

Dae-sik Kim said, “This is an innovative study in which human scientists overcame the limitations of AI-designed gene editors through structure-based protein engineering,” adding, “We expect it to accelerate the development of treatments for genetic diseases by opening a path to safely treating the causes of genetic disorders in the cell nucleus and mitochondria.”

Meanwhile, the research team recently reported that another next-generation gene editor designed by AI, ‘OpenCRISPR-1,’ reduced off-target mutations by up to 553-fold compared with existing Cas9 while maintaining editing efficiency. The finding demonstrated the potential for expansion into a next-generation gene therapy platform with high efficiency and precision.