The difficulty begins with how conventional CRISPR editing works. The canonical Cas9 nuclease cuts both strands of DNA at a targeted site, and the cell’s repair machinery then patches the wound. That repair is precise only if researchers supply a matching template and the cell uses homology-directed repair, a pathway that operates mainly in dividing cells and is notoriously inefficient. Left to its default machinery, non-homologous end joining, the cell scrambles the junction, producing unpredictable insertions and deletions. At sites cut on both strands, graver outcomes can follow: large deletions, inversions, chromosome rearrangements, and the activation of DNA-damage responses that can select against successfully edited cells. For knocking out a gene, such collateral damage is tolerable. For writing a therapeutic sequence into a defined position, it is disqualifying. Many of the most valuable edits, moreover, are large by nature: whole exons, entire genes, or regulatory modules that can span thousands of bases. Large, precise insertions have therefore been the stubborn frontier of genome editing, achievable mainly with viral vectors, transposases, or site-specific recombinases, each of which carries its own cargo limits, targeting constraints, or safety concerns.

Prime editing, first demonstrated in 2019, offered a gentler alternative. Instead of cutting both strands, a prime editor pairs a nicking form of Cas9 with an engineered reverse transcriptase. Its guide RNA, the prime editing guide RNA or pegRNA, does double duty: it locates the genomic target and carries a short RNA template that the reverse transcriptase copies directly onto the nicked strand. Because the intact complementary strand then guides repair, the method can install substitutions, small insertions, and deletions without a double-strand break. Prime editing has since proved itself for point mutations and compact edits. Writing kilobases of DNA, however, means asking the reverse transcriptase to copy enormously long RNA templates, and the efficiency of that synthesis falls steeply as templates lengthen. Larger inserts have generally demanded workarounds — twin-prime-editing schemes, integrase-based platforms, or engineered transposases — each of which adds enzymes, extra steps, or sequence constraints that limit where and how well the method works.

DoPE’s central insight is to stop asking the editor to synthesize the insert and instead let the insert deliver itself. The method couples a PE2* prime editor with a pair of overhang-complementary prime editing guide RNAs — opegRNAs — and a double-stranded DNA donor whose ends carry short 3′ single-stranded overhangs. Each opegRNA directs the editor to one side of the intended insertion site and encodes an overhang sequence complementary to one end of the donor. As the editor nicks each flanking strand and extends the exposed 3′ ends, the genome itself acquires sticky ends that mirror the donor’s. The matching sequences then anneal like two halves of a zipper, tethering the donor DNA into the gap between the two nicked sites, after which the cell’s own ligation and repair activities seal both junctions and complete the insertion. Precision comes from that complementarity: the donor anneals only where its overhangs find matching genomic sequences, so integration is guided by design rather than left to chance. Because the cargo arrives as pre-made DNA rather than being reverse-transcribed base by base, its length is limited less by the editor’s synthetic capacity than by delivery, which is why the same chemistry accommodates everything from tiny fragments to sequences longer than ten kilobases.

In the new study, the team reports precise insertions reaching 12.5 kilobases, a scale few cut-free methods achieve without enlisting integrases or transposases. Strikingly, the sticky ends that make the system work are short: overhangs of roughly 30 nucleotides proved sufficient to support the full range of inserts, from small fragments to sequences exceeding 10 kilobases. That brevity has practical consequences. The overhang is essentially the only custom sequence the platform needs, so retargeting the system or changing the cargo means redesigning two short guide RNAs and the donor ends rather than re-engineering the editor itself. One opegRNA pair, one editor, and a donor of whatever size the experiment demands — the architecture stays constant whether the cargo is a few bases or an entire gene-sized module. That plug-and-play quality is precisely the behavior that earlier insertion platforms, with their fixed recognition sites and enzyme-specific requirements, have struggled to deliver.

The same design turns out to be library-compatible, and that may prove its most consequential property. Because donors can be pooled, the researchers built collections from synthesized single-stranded oligonucleotides and used a single opegRNA pair to install a saturated library of mutations across a targeted region of EGFP, the gene for a green fluorescent reporter protein. The result was in situ saturation mutagenesis: the targeted stretch of the genome was rewritten with a comprehensive set of variants, resolvable at both amino-acid and single-nucleotide resolution, all generated inside cells in one experiment. Deep mutational scanning, the workhorse technique for measuring how thousands of protein variants behave, usually requires elaborate cloning campaigns to assemble variant libraries before they ever encounter a cell. DoPE compresses that workflow, writing the library directly into the genome in its native context, ready to be sorted and sequenced. For protein engineering, regulatory-element design, and systematic functional genomics, the method offers a route from sequence concept to cellular library without a cloning bottleneck.

The therapeutic proof of concept targeted PRKCSH, a gene whose loss-of-function mutations are linked to autosomal dominant polycystic liver disease, a condition in which fluid-filled cysts progressively enlarge the liver. Rather than correcting each patient’s mutation individually, the team used DoPE to replace mutant exons of PRKCSH — one exon at a time or two exons simultaneously — restoring the correct sequence wholesale. The strategy is mutation-agnostic: because whole exons are swapped for their healthy counterparts, any mutation lying within the replaced segment, whatever its chemical nature, is repaired by the same edit. That property addresses a persistent headache in gene therapy. Many disease genes harbor not one recurring mutation but a sprawl of rare variants scattered across the gene, and designing a bespoke editor for each is impractical. Exon-level replacement offers a single design that can cover many patients, and the study reports that distinct alleles were corrected uniformly in vitro, an early indication that the approach performs consistently across different mutant starting points.

Set against existing large-insertion tools, DoPE occupies a distinctive niche. Adeno-associated viral vectors can ferry genetic cargo but are size-restricted and integrate at random rather than chosen positions. Transposon systems move large fragments but with limited site specificity. Integrase-based platforms combine a prime editor with a serine integrase to install large sequences, and CRISPR-associated transposases target defined sites, but both recruit additional enzymes with their own sequence requirements and insertion preferences. DoPE’s parts list is minimal: one PE2* editor, two opegRNAs, and a synthetic DNA donor. No double-strand break occurs at any point, so the genotoxic hazards associated with cut-based editing — unpredictable indels at the junction, chromosomal scrambling, and DNA-damage signaling — are avoided by design rather than managed after the fact. And because targeting is written into the guide RNAs, any genomic site compatible with prime editing can in principle serve as a landing pad, keeping the method programmable in the same sense that CRISPR itself is.

The caveats are those that attend any new genome-editing platform. The results reported here were obtained in cultured cells, and performance in primary cells, tissues, and whole organisms — where delivering both a large editor and a double-stranded donor DNA is considerably harder — remains to be demonstrated. Efficiency and precision will need to be measured across many genomic contexts and cell types, and the platform’s byproduct profile, including any partial-edit intermediates or mis-annealed donors, will require systematic characterization. Off-target activity, a concern for every CRISPR-derived tool, will need dedicated assessment at scale. Independent replication across laboratories, as with any powerful new technique, will also shape how quickly the field adopts it. None of these open questions diminishes the conceptual advance; they simply mark the distance between an elegant chemistry demonstrated in vitro and a dependable research tool or therapy. The history of prime editing itself suggests a trajectory worth watching: the original system was a proof of principle in 2019 and has since been sharpened through successive rounds of protein and guide-RNA engineering.

If the method’s cell-culture performance carries forward, its implications are broad. Kilobase-scale, DSB-free, site-specific insertion without recombinases would let researchers install entire genes, swap promoters, or build synthetic regulatory circuits at defined loci, and would give gene therapy a candidate strategy for the many disorders caused by scattered mutations across large genes. The library mode, meanwhile, turns the genome itself into the substrate for high-throughput experimentation, potentially accelerating everything from enzyme evolution to the dissection of noncoding DNA elements. Genome editing began as a scalpel — a way to cut a chosen sequence and let the cell cope with the consequences. Prime editing recast it as a pencil, correcting individual letters without breaking the strand. DoPE pushes the metaphor further still: less a pencil than a compositor’s hand, setting whole paragraphs of DNA onto the genome’s page, one designed insert at a time.

CRISPR systems do not generally cut DNA at arbitrary locations. Their molecular targeting depends on two interacting components: a guide sequence that pairs with the chosen DNA site and a nearby short motif known as a protospacer adjacent motif, or PAM. The PAM acts as a molecular permission signal, allowing a CRISPR nuclease to bind and activate only when the correct neighboring sequence is present. Cas12a, a nuclease widely used for genome editing and nucleic-acid detection, has distinctive advantages, including its ability to recognize a PAM positioned differently from the one used by the more familiar Cas9 enzyme. But those advantages come with a limitation: commonly used Cas12a proteins generally require relatively long or restrictive PAM sequences. A desired genetic target may therefore be perfectly suited to the guide RNA yet remain inaccessible because the adjacent DNA does not satisfy the enzyme’s PAM rule.

Prime editing adds another layer of precision to this problem. Rather than relying on a conventional double-strand break followed by the cell’s repair machinery, prime editing typically uses a modified Cas protein fused to a reverse transcriptase. The Cas component nicks one DNA strand, while a specialized prime-editing guide RNA both directs the complex to the target and carries a template encoding the intended change. The reverse transcriptase copies that template into the DNA, creating a modified strand that the cell can incorporate into the genome. In principle, this arrangement allows researchers to install single-base substitutions, short insertions and deletions while avoiding the potentially disruptive double-strand breaks associated with standard CRISPR cutting. Yet prime editing is only as versatile as the nuclease that delivers it: if the Cas enzyme cannot recognize a nearby PAM, the edit may be impossible or require a less favorable target site.

To loosen that constraint, the team examined and engineered variants related to four Cas12a enzymes: LbCas12a, AsCas12a, MbCas12a and FnCas12a. Their goal was to create PAM-flexible proteins that remained active inside mammalian cells rather than merely showing altered biochemical behavior in a test tube. Among the variants tested, flexiAsCas12a—derived from AsCas12a—proved the most effective in the reported experiments. The researchers found that the variant could extend Cas12a’s recognized PAM repertoire to include sequences described as NATN, NCCN and GTCN. In this notation, N represents any nucleotide, while the specified letters impose only partial constraints on the four-base motif. Compared with a highly restrictive PAM requirement, these patterns make more sites across the genome potentially addressable by the nuclease.

The significance of the expanded recognition rules is not simply that three additional sequence patterns have been added to a catalog. PAM availability is distributed unevenly across DNA, and the position of a PAM relative to a disease-associated mutation, regulatory element or coding sequence can determine whether an edit is practical. A target may require the guide and editing machinery to approach from a particular direction, or the desired alteration may fall within a limited distance from the nick introduced by the nuclease. More permissive PAM recognition increases the odds that a usable site exists in the right genomic context. It could also reduce the need to redesign an experiment around a nearby, imperfectly positioned target. However, broader recognition must be balanced against the risk of unintended activity: an enzyme that accepts more PAMs may encounter more possible sites, making rigorous off-target testing essential before therapeutic applications can be considered.

The study also connects PAM flexibility to a less conventional form of prime editing. The researchers used currently available Cas12a variants with relaxed PAM recognition—impLbCas12a, flexiAsCas12a and enAsCas12a—to develop circular RNA-guided split prime editors. In a split editor, the molecular machinery is divided into separate components rather than delivered as one large protein or complex. This strategy can help address the size limitations that complicate the delivery of genome-editing systems, particularly with viral vectors, whose cargo capacity is restricted. Circular RNAs are RNA molecules whose ends are joined, a structure that can offer greater resistance to degradation than conventional linear RNA. By combining circular RNA guidance with split prime-editing architecture, the researchers sought to create systems that could function at targets carrying non-canonical PAMs—sequences that would not ordinarily be accepted by standard Cas12a tools.

The experiments validated the functionality of these circular RNA-guided split editors on non-canonical PAM sequences, according to the study. That result is important because it demonstrates more than a theoretical change in target recognition: the engineered enzymes could be incorporated into a working prime-editing system in mammalian cells. The work does not, however, establish that every newly recognized target will be edited with equal efficiency, nor does it show that the approach is ready for clinical use. Genome editing performance depends on many variables, including chromatin structure, guide-RNA design, the exact DNA change being attempted, cellular repair pathways and the concentration and duration of the editing components. The study’s central achievement is therefore an expansion of the addressable target space, rather than a universal solution to the technical and safety challenges facing prime editing.

Cas12a’s biology may make these variants attractive for applications beyond prime editing as well. Cas12a enzymes are already used in nucleic-acid detection because their activity can produce detectable signals after target recognition. In genome manipulation, their guide-RNA processing properties and targeting geometry differ from those of Cas9, providing alternative ways to design multiplexed or directionally constrained editing experiments. A broader PAM range could allow researchers to choose among more Cas12a configurations according to the genomic site and desired outcome. The new variants may also be useful when a Cas9-based tool has unfavorable off-target behavior or cannot reach a sequence in the required orientation. Whether flexiAsCas12a or related enzymes outperform established systems will depend on head-to-head measurements of editing yield, product purity, unintended edits, cellular toxicity and delivery efficiency.

The authors describe flexiAsCas12a as a new addition to the expanding collection of Cas12a PAM variants, with the broader objective of making more genomic sequences accessible to precision editing. The work was conducted by researchers affiliated with the HUN-REN Research Centre for Natural Sciences, the University of Szeged, the Biological Research Centre, Eötvös Loránd University, Semmelweis University and Hungarian biotechnology organizations. The article was made available as an early peer-reviewed, accepted version carrying a permanent DOI, with the publisher noting that it may later be replaced by a final version of record. As the field moves toward increasingly programmable genome modification, the practical importance of the advance will be determined by how reliably these flexible nucleases edit difficult targets while preserving the selectivity that makes prime editing appealing. For now, the study offers a molecular workaround to one of CRISPR’s most persistent limitations: the short sequence beside a target that can decide whether the target is editable at all.

Suppressor tRNAs serve as critical tools in the field of gene editing by enabling cells to read through premature stop codons, thus restoring the production of functional proteins. However, optimizing these molecules for therapeutic applications has remained a formidable challenge due to the complex interplay between tRNA structure, cellular machinery, and the repair mechanisms that govern genomic stability. The latest research addresses these hurdles by systematically introducing mutations directly into sup-tRNAs to enhance their performance post-prime editing.

The team began their investigation by selecting three candidate human sup-tRNAs—tRNA-Arg-CCT-4-1, tRNA-Tyr-GTA-2-1, and tRNA-Leu-TAA-4-1—alongside a mouse ortholog, tRNA-Leu-TAA-2-1. They engineered comprehensive lentiviral libraries that incorporated every conceivable single-nucleotide substitution, single-base deletion, and paired-base modification within the tRNA sequences. This high-throughput approach enabled a granular exploration of structure-function relationships within the sup-tRNAs, facilitating identification of mutations that could amplify their suppressive activity.

Upon transducing these lentiviral libraries into a reporter cell line, which allowed precise quantification of sup-tRNA function, the researchers observed that most mutations diminished tRNA performance. Deletions, in particular, were largely detrimental, underscoring the structural sensitivity of tRNAs to nucleotide loss. However, a notable subset of single-nucleotide variants and paired-base substitutions led to measurable improvements in sup-tRNA efficacy. This highlights the delicate balance between preserving tRNA integrity and introducing beneficial alterations to optimize function.

Beyond optimizing sup-tRNA activity, the team aimed to devise mutations that could circumvent intrinsic cellular mismatch repair (MMR) pathways. By identifying silent mutations that evade MMR detection, they sought to enhance the durability and effectiveness of prime editing outcomes. This strategy also addresses the issue of prime editor rebinding to the repaired locus, which can impede editing precision. Through these refined mutations, the researchers demonstrated enhanced genomic editing fidelity by reducing unwanted cellular responses.

The effect of beneficial mutations discovered in the tRNA-Leu-TAA-4-1 variant was further corroborated on paralogous tRNA family members, including tRNA-Leu-TAA-1-1, -2-1, and -3-1. This cross-applicability suggests a potential for broad utility of these optimized tRNAs in various genomic contexts and across species. Such versatility articulates the promise for these engineered sup-tRNAs to be tailored for a wide spectrum of therapeutic targets.

Prime editing itself is a relatively nascent genome editing technique that leverages a fusion of a catalytically impaired Cas9 and a reverse transcriptase. It allows highly specific nucleotide modifications without introducing double-strand breaks, thereby reducing off-target effects and enhancing cellular safety profiles. Integrating sup-tRNAs into this framework further expands the capability to correct nonsense mutations—key culprits in many genetic diseases.

The implications of this research extend far beyond the bench. By developing sup-tRNAs that can be seamlessly integrated into prime editing workflows and resist cellular repair obstacles, therapeutic gene correction approaches become more feasible and efficient. This could transform treatment paradigms for a host of disorders caused by premature stop codons, such as cystic fibrosis, Duchenne muscular dystrophy, and various inherited retinal diseases.

Importantly, the systematic mutational approach taken by the researchers provides a blueprint for future engineering of noncoding RNAs in gene therapy applications. Rather than relying solely on alterations in the anticodon region, this study underscores the value of probing and optimizing additional structural elements within tRNAs to unlock enhanced functionality.

This work also contributes vital insights into the mechanistic underpinnings of tRNA performance within the complex cellular milieu. Understanding the nuanced influence of nucleotide substitutions and deletions on tRNA stability, folding, and interaction with ribosomes and editing complexes is critical for designing next-generation gene editing tools.

As prime editing technologies continue their rapid evolution, the introduction of sup-tRNAs with improved activity and MMR evasion capabilities could mitigate current limitations, such as partial editing efficiency and undesired genomic outcomes. The integration of these enhanced molecular components stands to elevate the precision and durability of gene correction protocols in clinical settings.

Looking forward, further investigation into the long-term stability and immunogenicity of these optimized sup-tRNAs in vivo will be essential before broader therapeutic implementation. Nonetheless, this pioneering study marks a monumental advance in the genetic toolkit available to researchers and clinicians alike.

The convergence of synthetic biology, RNA engineering, and prime editing at the heart of this research exemplifies the future of personalized medicine—where tailored molecular interventions can correct the very root causes of genetic diseases with remarkable accuracy and minimal side effects. The findings illuminate a promising pathway toward realizing the full therapeutic potential of genome editing.

Subject of Research:
Gene editing enhancement through engineered suppressor tRNAs integrated with prime editing technology.

Article Title:
Prime editing-installed suppressor tRNAs for disease-agnostic genome editing.

Article References:
Pierce, S.E., Erwood, S., Oye, K. et al. Prime editing-installed suppressor tRNAs for disease-agnostic genome editing. Nature (2025). https://doi.org/10.1038/s41586-025-09732-2

DOI:
https://doi.org/10.1038/s41586-025-09732-2

Keywords:
Prime editing, suppressor tRNAs, genome editing, nucleotide substitutions, mismatch repair evasion, lentiviral libraries, genetic therapy, noncoding RNA engineering