Prime editing is often described as a molecular search-and-replace system. It couples a catalytically impaired Cas9 nickase to an engineered reverse transcriptase, and it is programmed by a prime editing guide RNA, or pegRNA, which both locates the genomic target and carries the template encoding the desired edit. Because the method writes new genetic information without making a double-strand break in the DNA and without requiring an external donor template, it has long been viewed as one of the most versatile tools in the genome editing repertoire, capable of installing all twelve possible base-to-base conversions as well as small insertions and deletions. Yet translating that versatility into animals, and ultimately patients, has proved difficult. The prime editor protein is far too large to fit inside a single adeno-associated virus vector, forcing researchers to rely on cumbersome dual-vector systems, while concerns about the long-term expression of an editor in a patient’s tissues have made viral delivery unattractive for clinical development.

Non-viral delivery offers a way around those concerns. Lipid nanoparticles, the same technology that carried mRNA vaccines into billions of arms, can deliver the prime editor in transient mRNA form together with its pegRNA, giving the editor a brief window of activity before it is degraded by the cell. The problem has been efficiency. Earlier attempts to deliver prime editing systemically with lipid nanoparticles produced editing levels far below what would be needed clinically, required repeated injections, or depended on doses of RNA so high that they exceeded what regulators would consider translatable. The fragile nature of long RNA molecules inside the bloodstream and inside cells is a central culprit: unmodified or lightly modified pegRNAs are rapidly chewed apart by nucleases, and even end modifications at the RNA termini, the standard protective measure borrowed from antisense oligonucleotide chemistry, leave long internal stretches of the guide exposed and vulnerable.

The Wuhan-led team’s solution was to extend chemical protection far beyond the ends of the guide RNA. Rather than modifying only the terminal nucleotides, the researchers densely incorporated modified RNA motifs throughout both the fixed regions of the pegRNA, such as the scaffold and the engineered extension motifs like evopreQ1, and the variable regions, including the spacer, the reverse transcriptase template, and the primer binding site. The chemical toolkit relied on well-established modifications from the RNA therapeutics field, principally 2′-O-methyl groups on the ribose sugar and phosphorothioate linkages in the phosphate backbone, both of which are known to shield RNA from nuclease degradation and, in the case of 2′-O-methyl groups, to subtly alter pairing geometry and affinity. The team describes the resulting molecules as super-end-modified RNAs, in which terminal protection and dense internal modification act in combination.

The optimization was systematic. In human cell experiments using electroporated prime editor mRNA, the researchers tested modification patterns in each structural element of the pegRNA in turn, measuring editing outcomes at endogenous loci such as EMX1, where the system installed a three-base deletion, and HEK3, where it installed a single-base insertion. They varied the density of 2′-O-methyl and phosphorothioate modifications at the 5′ end of the spacer, in the primer binding site, in the reverse transcriptase template, and in the extended scaffold motifs, and then combined the best-performing patterns. Analyses of editing byproducts showed that the dense modifications did not increase the proportion of indels or unwanted scaffold-insertion events relative to total editing, and cell viability assays indicated the heavily modified guides were not toxic to the treated cells, an important consideration since excessive chemical modification of guide RNAs has previously been associated with cellular stress.

The in vivo results were the centerpiece. Using lipid nanoparticles to co-deliver prime editor mRNA and the densely modified pegRNA into mice, the team achieved nearly 70 percent prime editing efficiency in the bulk liver, a level indicating that most hepatocytes in the organ had been edited. Perhaps more consequential for clinical translation, a single injection at a lipid nanoparticle dose the authors describe as clinically translatable was sufficient to suppress the expression of the target protein in vivo, and under those dose conditions the densely modified guide produced an approximately 80-fold increase in editing efficiency compared with conventional end-modified pegRNAs. Toxicity assessments, including serum liver enzyme measurements and analysis of editing byproducts, supported the tolerability of the approach, and off-target analyses at predicted sites suggested the chemical modifications did not worsen targeting fidelity.

The generality of the strategy proved equally important. The dense motif modification approach was not confined to prime editing. When the team applied the same modification logic to guide RNAs carrying the widely used MS2 RNA motif, which is exploited in RNA-guided recruitment platforms such as base editor systems that tether an editor protein to a Cas9 nickase through an RNA aptamer interaction, they saw broad improvements across multiple RNA sequences and split RNA-guided genome editing platforms, with base editing efficiencies rising by up to 11-fold. The researchers also combined the modified guides with mRNA encoding ancillary factors known to boost prime editing, such as MLH1-domain constructs, and observed further gains in vivo, indicating that RNA chemistry and protein engineering enhancements can be layered rather than traded off against one another.

The findings arrive at a moment when the field is converging on transient, non-viral delivery as the safest route to in vivo genome editing. Clinical programs using lipid nanoparticle delivery of CRISPR-Cas9 mRNA and guide RNA have already demonstrated that single-dose in vivo gene knockout is feasible in humans, and prime editing has shown curative promise in ex vivo settings such as engineered hematopoietic stem cells for sickle cell disease. What has lagged is in vivo precision editing, where the required correction must be installed in a large fraction of target tissue cells rather than simply disrupting a gene. By raising editing yields at clinically plausible doses, the dense modification strategy narrows the gap between what prime editing can do in a dish and what it can do in a body, and because the modifications are applied to the RNA guide rather than to the editor protein, they can in principle be adopted without redesigning the editing enzymes themselves.

Challenges remain before such a system could reach patients, including extending the approach beyond the liver, where lipid nanoparticles naturally accumulate, to other tissues, scaling the manufacture of long, densely modified pegRNAs under clinical quality standards, and confirming long-term safety in larger animal models. The authors note that the methodology builds on their earlier work developing rapid methods for generating long chemically modified pegRNAs, and patent applications on the pegRNA modifications have been filed through Wuhan University. Still, the study offers a concrete, generalizable recipe: protect the guide RNA densely, deliver the editor transiently, and let chemistry do much of the work that delivery vehicles alone could not. If the efficiency gains hold across tissues and disease targets, the modest RNA molecule at the heart of prime editing may prove to be the lever that finally lifts the technology from laboratory promise toward therapeutic reality.

Subject of Research: Engineered prime editing guide RNAs with densely modified RNA motifs for robust in vivo genome editing

Article Title: Dense RNA motif modifications enable robust in vivo prime editing and enhance efficiencies of diverse editing systems

Article References: Lei, X., Chen, D., Zhang, K., Liu, X., Chen, Q., Zhang, Y., Ji, R., Zhu, J., Zhang, Q., Zhang, Y., & Yin, H. (2026). Dense RNA motif modifications enable robust in vivo prime editing and enhance efficiencies of diverse editing systems. Nature Biomedical Engineering. https://doi.org/10.1038/s41551-026-01787-4

Image Credits: AI Generated

DOI: 10.1038/s41551-026-01787-4

Keywords: prime editing, pegRNA, chemical modification, lipid nanoparticles, in vivo genome editing, base editing, RNA therapeutics, gene therapy, 2'-O-methyl, phosphorothioate, mouse liver, non-viral delivery

A team led by Zhu et al. addressed this challenge by engineering a novel cytosine base editor termed transformer base editor (tBE), which was optimized to robustly inhibit UDG activity within cells. This improved mechanism safeguards the uracil intermediates generated during C-to-T conversion, thus enhancing editing persistence and effectiveness. The tBE was subsequently packaged into VLPs to facilitate precise in vivo delivery.

In their rigorous animal models, tBE-VLPs demonstrated remarkable editing efficiencies. A single systemic injection in mice led to average editing rates of 46.0% at the mouse Pcsk9 locus and 64.2% at the Hpd locus within the liver. Furthermore, targeted delivery to the retinal pigment epithelium achieved an average 24.2% editing at the Vegfa gene, underscoring the platform’s versatility across diverse tissue types.

These editing rates translated to meaningful therapeutic outcomes in mouse disease models. Pcsk9 editing is known to reduce cholesterol levels, while Hpd editing can alleviate symptoms of hereditary tyrosinemia. Vegfa editing in the eye is associated with modulating vascular growth, relevant to diseases such as age-related macular degeneration. The study confirmed that tBE-VLP treatment produced beneficial physiological effects aligned with gene correction.

A key advantage of this approach is its superior specificity. Off-target editing events were undetectable both in vitro and in vivo, addressing a major concern associated with base editor delivery methods. Comparison with adeno-associated virus (AAV) vectors and lipid nanoparticle (LNP) mRNA delivery highlighted the enhanced precision of tBE-VLPs, underscoring their safety profile for potential translational applications.

Mechanistically, the enhancement comes from the improved recruitment of uracil DNA glycosylase inhibitor proteins within the tBE editing complex. By preventing excision of uracil bases, the tBE maintains base conversions until DNA replication or repair solidifies the desired C-to-T substitutions. This insight into enzymatic inhibition marks a critical step forward for base editing technology.

Beyond demonstrating robust editing in challenging tissues like the retina and liver, the study’s findings open avenues for broadening the therapeutic landscape of base editing. The modularity of VLPs combined with the refined inhibitor design promises scalable, transient, and safe genome editing interventions without viral genome integration risks.

Zhu and colleagues’ work represents a milestone in genome engineering, establishing transformer base editor VLPs as an efficient and accurate in vivo cytosine base editing system. As the field advances, this platform may catalyze novel treatments for genetic diseases with precision previously unachievable using traditional delivery methods.

With exciting potential for human therapeutic development, tBE-VLPs stand poised to revolutionize the future of gene editing by marrying efficacy, specificity, and safety in a virus-inspired delivery system.

Subject of Research: In vivo cytosine base editing with virus-like particles and uracil DNA glycosylase inhibition

Article Title: Efficient in vivo cytosine base editing using virus-like particles with uracil DNA glycosylase inhibition

Article References:
Zhu, J., Ding, L., Liu, KM. et al. Efficient in vivo cytosine base editing using virus-like particles with uracil DNA glycosylase inhibition. Nat Biotechnol (2026). https://doi.org/10.1038/s41587-026-03227-9

Image Credits: AI Generated

DOI: https://doi.org/10.1038/s41587-026-03227-9