From Precise Editing Without Double-Strand Cuts to Kilobase-Scale Rewriting: The Molecular Evolution of Prime Editing Technology
Background
Conventional gene correction technologies have relied on methods that cut the DNA double helix. CRISPR-Cas9 effectively targets genes for cutting but can cause unintended random mutations as a side effect during the cellular repair process. Homology-Directed Repair (HDR), which introduces foreign genetic material, is also inefficient in non-dividing cells.
To overcome these side effects and limitations, Prime Editing (PE) was developed. PE operates using a Cas9 nickase enzyme, which cuts only one strand of DNA, fused with a reverse transcriptase (RT). It incorporates a Prime Editing Guide RNA (pegRNA) carrying correction information to guide precise editing.
However, the initial PE system showed irregular efficiency in cells due to its complex components. Its correction range was narrow, and it failed to demonstrate significant results in inserting large gene fragments. As a result, both academia and industry have focused on molecular improvements and the development of new design tools to enhance practicality.
Key Findings
The leap in PE technology began with molecular engineering of its core components. Early PE systems had limited target ranges and low RT binding efficiency. Researchers focused on introducing Cas9 variants with expanded Protospacer Adjacent Motif (PAM) recognition and improving RT thermal stability. As a result, accessibility was broadened, and correction efficiency increased by more than tenfold.
The most notable achievement is the emergence of large-genome insertion technologies that expanded the editing range from dozens of base pairs to several kilobases (kb). TwinPE, a representative platform, uses two pegRNAs to first generate recombination enzyme attachment sequences at the target genomic location. Subsequently, PASTE or PrimeRoot, specialized for plant correction, combine with recombination enzymes such as Bxb1 to stably introduce target genes. Using this method, gene fragments over 36 kb, including therapeutic genes, can be inserted safely and seamlessly.
The proliferation of AI design tools has further accelerated optimization. pegRNA performance varies significantly depending on target sequence length, making optimization cumbersome. Recently, deep learning-based prediction models such as PRIDICT and DeepPE have been used to accurately predict pegRNA correction efficiency across cell types. AI models analyze tens of thousands of combinations in just a few minutes, drastically reducing experimental failure rates.
Implications and Outlook
Improvements in genome correction performance are expected to bring real change in disease treatment and plant biotechnology. PE has demonstrated excellent safety in correcting single-base mutations in rare diseases. Research is also gaining momentum in developing climate-resilient crops by inserting beneficial traits. It is increasingly being positioned as a technology that can precisely place genes at desired locations, unlike traditional insertion methods.
Nevertheless, challenges remain before industrial adoption. PE complexes are large in molecular size, making it difficult to overcome delivery barriers into cells. There are spatial constraints in fully loading genes into delivery vehicles such as Adeno-Associated Virus (AAV) or Lipid Nanoparticles (LNP). Additionally, it is difficult to completely prevent off-target editing at non-specific sites, requiring long-term safety validation.
Prime editing has rapidly emerged as a versatile and precise genome engineering technology capable of introducing targeted substitutions, insertions, and deletions without requiring double-strand breaks or donor DNA templates. Since its initial development, substantial progress has been achieved through iterative innovations in core components, including Cas protein variants, reverse transcriptase engineering, and prime editing guide RNA (pegRNA) design. These advancements have significantly improved editing efficiency, fidelity, and target accessibility across diverse biological systems. In parallel, the development of large-fragment editing platforms such as TwinPE, PASTE, and PrimeRoot has expanded the scope of prime editing from small sequence modifications to kilobase-scale genome rewriting. Furthermore, the integration of artificial intelligence-driven design tools has transformed pegRNA optimisation and system selection, enabling more predictable and efficient editing outcomes. This review provides a comprehensive overview of recent molecular innovations, large-fragment editing strategies, and computational approaches shaping the evolution of prime editing. Collectively, these developments position prime editing as a highly programmable and scalable platform with broad applications in medicine, agriculture, and synthetic biology.
This research can be directly applied to the treatment of genetic disorders such as Duchenne Muscular Dystrophy (DMD), which are caused by large gene deletions or mutations. With traditional gene scissors, it was practically impossible to correctly insert dystrophin gene fragments, which can be several kilobases in size, into patient cells. However, using the improved large-volume prime editing technology, it is now possible to insert normal gene sequences precisely at the target gene location without the risk of random insertion-induced cancer. This will be a key factor in accelerating the commercialization of gene therapies that replace entire disease-causing genes with normal ones, going beyond single-base corrections.