David Liu advances disease-agnostic gene editing at ESGCT

By Sophie Lindqvist

Prime-editing-mediated readthrough produced about 7% of wild-type enzyme activity in a Hurler syndrome mouse model, a preclinical result that David Liu, PhD, said could “rescue the disease.”

In a plenary at the European Society of Gene & Cell Therapy (ESGCT) 2026 meeting in Hamburg, Liu used the experiment to support a disease-agnostic strategy: one treatment for mutations sharing a premature stop codon. The mouse result, clinical reach and economic case remain separate stages of evidence.

Before Liu, 2018 Kavli Award winner Virginijus Siksnys, PhD, cast himself jokingly as the warm-up. Liu had addressed ESGCT in Barcelona in 2019. He said progress comes from combining scientific methods with regulatory, manufacturing and economic advances.

Liu cited more than 200,000 known mutations, about 8,000 genetic diseases and roughly 400 million families. He did not derive the family estimate or distinguish mutations reachable by current editors, leaving gaps for portfolio sizing.

A decade ago, Liu’s group published the first landmark base-editing report; prime editing followed three years later. Both use a CRISPR-Cas chassis and can make single-base substitutions. Beam Therapeutics and Prime Medicine, both co-founded by Liu, carry them into commercial development.

How are AI tools changing editor design?

A Nature Biotechnology paper published a few months ago combined two optimization steps. In 2023, phage-assisted continuous evolution, or PACE, produced PE6 editors with higher activity but lower stability and expression. ProteinMPNN then redesigned up to 25% of reverse-transcriptase amino acids through inverse protein folding.

Liu said PE8 variants showed stronger expression and editing efficiency across several cell types. The account gave no sample sizes, numerical gains, comparators or safety assays and disclosed no study funding, leaving reproducibility for peer review.

Another recent paper presented OptiPrime, a free machine-learning tool for predicting prime-editing outcomes. It addresses a guide-RNA bottleneck: variable regions can alter efficiency, and one desired edit may have hundreds of possible pegRNAs. OptiPrime returns a ranked list for copying into a gene-synthesis order form.

That gives teams a design-to-procurement route. The source gave no hit rate, benchmark set, runtime or cost comparison, so the ranking remains a candidate filter, not a substitute for experiments.

What clinical evidence supports the platform?

Liu counted nine precision-editing examples, estimated at one person-century, that rescued animal-model disease phenotypes: progeria, sickle cell disease, cystic fibrosis, phenylketonuria and spinal muscular atrophy. Several programs reached IND-enabled studies with encouraging clinical results, but he did not tabulate sample sizes or effect sizes.

Clinical milestones gave portfolio teams more concrete signals:

• In March 2025, Beam Therapeutics reported the first clinical correction of a human pathogenic mutation in an alpha-1 antitrypsin deficiency subject, BEAM-302. The source gave no patient count, dose, follow-up or biochemical endpoint.
• Around the same time, attention focused on Baby KJ, treated with bespoke base editing; no outcome data appeared in the account.
• In April 2025, 57-year-old Nebraska resident Tracy Atteberry became the first ex vivo prime-editing patient dosed. Her edited stem cells aimed to repair a two-base deletion. Jennifer Gori, PhD, and colleagues reported the case in The New England Journal of Medicine in late 2025.
• In October, investigators dosed the first patient in an in vivo prime-editing trial for Wilson disease, PM577a.

These entries move the platform beyond animals, but the source gave no control data, manufacturing yield, adverse-event profile or durability.

Liu said U.S., U.K. and other regulators generally were not blocking development and described more permissive policies for trial initiation and BLAs. The day before his lecture, the FDA issued draft guidance titled Individualized Human Genome Editing Products: Considerations for BLA Submission and Postapproval Management.

The source did not explain comparability or manufacturing changes across patients, leaving standardization and development planning unanswered.

Can one editing strategy cover many diseases?

Liu’s proposal targets the roughly 30% of genetic diseases with at least one premature stop codon. Suppressor tRNAs can read through those codons and restore full-length protein production. “A tRNA approach enables far fewer drugs to target a larger number of mutations/diseases,” Liu said.

The genome has 418 tRNA genes but only 61 coding triplets, indicating redundancy; many people lack at least one tRNA gene, Liu said. PERT—prime editing-mediated readthrough of premature termination codons—converts a dispensable tRNA into a permanent suppressor, aiming at a one-time treatment for diseases sharing a stop codon.

Led by postdoc Sarah Pierce, PhD, the team used prime editing to convert each of the 418 human tRNAs into a suppressor candidate, engineering anticodons for stop codons CUA, UCA and UUA. They selected a leucine tRNA because it had the highest suppression activity, then optimized it.

To limit cellular burden, Liu said the therapy would not overexpress the optimized suppressor and would make about 40% of the replaced tRNA. In the Hurler mouse model, PERT produced the reported 7% wild-type enzyme activity. Beyond that readout, the source gave no dose-response, toxicity, normal-tissue suppression or manufacturing data.

Liu called the work early and hopeful, but said regulatory streamlining and clinical de-risking must lower risk and cost. Without those gains, most genetic diseases remain too rare for industry, making a shared product economically different from a mutation-specific one.

Who is supporting the rare-disease model?

The Broad Institute, Boston Children’s Hospital and The Jackson Laboratory established the nonprofit Center for Therapeutic Genetics. Liu said he hopes the model can be copied to connect rare-disease patients. Apertura, Viralgen, Beam Therapeutics and Prime Medicine provide initial commercial support.

The source did not disclose grant sizes, governance rights, intellectual-property terms, milestones or a work plan, so research leaders cannot yet compare the collaboration’s independence and commercial alignment.

The next portfolio test is whether PERT can move from one mouse-model readout to evidence that one treatment suppresses multiple stop-codon diseases, while clinical and regulatory work continues lowering the risk and cost Liu identified as barriers.