NUS Medicine's AI-Guided Gene Editing Breakthrough Enables Safer DNA Corrections

The Dawn of Smarter Gene Editing at NUS Medicine

Researchers at the National University of Singapore's Yong Loo Lin School of Medicine have unveiled a groundbreaking advancement in gene editing technology. Published just yesterday in the prestigious journal Advanced Science, their work introduces an AI-guided method to engineer more precise and safer base editors, compact tools designed to correct single-letter errors in DNA without the risks associated with traditional double-strand breaks. 128 76

Led by Assistant Professor Jungjoon K. Lee from the Department of Biochemistry and the Synthetic Biology for Clinical and Technological Innovation (SynCTI) group, the team addressed longstanding challenges in genome editing. Their innovation combines artificial intelligence-driven protein structure prediction with a novel multi-layered screening platform, resulting in base editors that are not only more efficient but also less toxic and less likely to cause unintended DNA damage.

This development holds immense promise for treating genetic diseases caused by point mutations, which account for approximately 90% of known pathogenic variants in humans. By making compact editors viable for delivery via adeno-associated viruses (AAVs)—the preferred vehicle for in vivo gene therapy—the NUS breakthrough could accelerate the path to clinical applications. 91

Understanding Base Editors: The Next Evolution in CRISPR Technology

Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR) technology, popularized since 2012, has transformed genetic research. Traditional CRISPR-Cas9 acts like molecular scissors, cutting DNA at specific locations to enable repairs. However, this double-strand break (DSB) process can lead to unwanted insertions, deletions, or rearrangements, posing safety risks for therapeutic use.

Base editors represent a leap forward. Developed by David Liu's lab in 2016, they fuse a CRISPR-Cas nickase (which makes a single-strand nick) with a deaminase enzyme to chemically convert one DNA base to another—C to T or A to G—without DSBs. This precision suits single-nucleotide variants (SNVs), responsible for diseases like sickle cell anemia, cystic fibrosis, and many rare disorders. Studies suggest base editors could potentially correct 62% of pathogenic SNVs in genes. 87

Despite their promise, challenges persist: larger editors exceed AAV packaging limits (4.7 kb), off-target edits, bystander edits (unintended changes nearby), and immunogenicity (immune reactions to foreign proteins). Compact editors like those derived from bacterial toxins offer hope but suffer from low efficiency and toxicity.

SsdAtox: A Compact Candidate from Bacterial Origins

The NUS team focused on SsdAtox, a cytosine deaminase toxin from Shewanella bacteria. At about two-thirds the size of standard editors like BE4max, SsdAtox fits AAVs easily. However, its natural form edits inefficiently and induces DSBs and cell death, limiting utility. 77

To overcome this, they employed a dual strategy. First, leveraging AlphaFold3—an AI model excelling in predicting protein-DNA interactions—they pinpointed residue K31 in the enzyme's active site entrance. Mutating it widened access for DNA substrates, boosting catalytic efficiency without compromising specificity. AlphaFold3's accuracy in multi-molecule modeling proved pivotal, marking one of the first direct applications in editing tool design. 99

Trinity-Screen: Revolutionizing Enzyme Optimization

Complementing AI design, the team invented Trinity-Screen, a bacterial evolution platform assessing three metrics simultaneously: editing efficiency, DSB avoidance, and cytotoxicity. Variants passing all layers underwent iterative selection, yielding synergistic mutations.

Tested in human HEK293T cells across 24 synthetic targets and 10 endogenous genes, top variants (e.g., SsdAtox-K31V/T) achieved:

• 11.8-fold higher editing efficiency vs. wild-type
• 50% fewer DSBs than prior mutants
• 10-fold reduced bacterial toxicity
• 31-fold BEPI score improvement

One variant cut DSBs by 37% compared to BE4max, with a tighter editing window for predictability. 76

Introducing BEPI: A Comprehensive Performance Metric

To benchmark fairly, NUS developed the Base Editor Performance Index (BEPI), weighting efficiency against DSB rates. Unlike efficiency-alone metrics, BEPI reveals trade-offs, positioning optimized SsdAtox as superior for therapeutic profiles.

Singapore's Rising Star in Biotech Innovation

Singapore positions itself as Asia's biotech hub, with over S$25 billion in Research, Innovation and Enterprise (RIE) 2025-2030 funding. NUS, a cornerstone, hosts SynBio Clinic and collaborates with A*STAR. The city-state's ecosystem—Biopolis, clinical trial efficiency, IP protection—attracts global players like Ring Therapeutics. 107 Read the full NUS announcement.

This breakthrough aligns with national goals for precision medicine, where gene therapies could address 2.4 million carriers of treatable SNVs regionally.

Real-World Impacts: Paving the Way for Genetic Disease Treatments

Over 7,000 rare diseases stem from SNVs amenable to base editing. Compact, low-immunogenicity editors like optimized SsdAtox enable AAV delivery to hard-to-reach tissues, potentially treating liver, muscle, and eye disorders. Reduced off-targets minimize cancer risks from DSBs. 87

"Our framework accelerates safer tool development," notes Asst Prof Lee. Experts praise its deimmunization via evolution, addressing AAV immunogenicity hurdles.

• Liver diseases: Hemophilia, alpha-1 antitrypsin deficiency
• Neuromuscular: Duchenne muscular dystrophy
• Ocular: Leber congenital amaurosis

Challenges Ahead and Ethical Considerations

While promising, hurdles remain: long-term off-targets, delivery specificity, and ethics. Singapore's Bioethics Advisory Committee guides germline editing bans, focusing somatic therapies. Clinical translation requires IND-enabling studies, with NUS's GMP facilities aiding.

Global trials like Vertex's CTX001 (base-edited sickle cell) validate feasibility, but compact editors expand scope.

Future Outlook: NUS Leading the SynBio Revolution

The NUS method generalizes to other editors, with plans for prime editing integration and in vivo testing. As Singapore invests S$800m in semiconductors and biotech, NUS eyes partnerships for commercialization.

This positions Singaporean universities at the forefront, fostering careers in synthetic biology—a field projected to grow 30% annually.Access the Advanced Science paper 128 .

Photo by Mufid Majnun on Unsplash

Cultivating Talent in Singapore's Higher Education Biotech Landscape

NUS's programmes like SynCTI train PhD students in AI-synbio fusion. With 40% of regional biotech jobs in Singapore, opportunities abound in research, clinical translation, and startups. The government's S$1b life sciences push underscores universities' role.