Prime Editing Explained: A CRISPR Alternative for Precise Gene Therapy

Prime editing has been successfully translated to therapeutic applications using two major delivery modalities. For ex vivo editing, mRNA delivery via electroporation achieves robust editing in patient-derived cells. A sickle cell disease study demonstrated 40% editing efficiency in mice after transplanting edited HSCs, with reduced sickling phenotype. For in vivo applications, dual AAV delivery systems deliver prime editing components to specific organs. A PCSK9 mutation study achieved ~40% liver editing efficiency using split PE2 Max, resulting in corresponding LDL reduction. Twin prime editing extends capabilities for larger modifications by performing prime editing on both DNA strands simultaneously, enabling installation of recombination sites for gene-sized insertions. The choice between delivery methods depends on whether the target is accessible ex vivo or requires in vivo modification. These examples demonstrate prime editing's translational potential for treating genetic diseases through precise, programmable genome modification.

Base editing rapidly advanced to clinical trials: lipid nanoparticle delivery of adenine base editors achieved 90% PCSK9 reduction in primates; multiple trials now enroll patients for sickle cell disease, T-cell leukemia, and familial hypercholesterolemia. The first positive clinical outcome came when triple-base-edited CAR-T cells achieved complete remission in a 13-year-old SMA patient. Prime editing optimization focused on improving efficiency and reducing off-target effects. Research revealed that DNA mismatch repair pathway components significantly impact editing outcomes, leading to strategies of adding silent mutations to evade repair mechanisms. Three key improvements emerged: mismatch repair evasion, improved protein architecture (PX), and epegRNAs with RNA pseudoknots resisting degradation. Phage-assisted continuous evolution (PACE) accelerated protein evolution by linking prime editing activity to phage propagation, enabling hundreds of generations in days rather than years. Research revealed that optimal editors depend on edit complexity: variants evolved for one-base insertions excel at that task but fail on longer edits, and vice versa. This understanding enabled development of compact variants (PE6A, PE6B) offering PE-Max performance at half the size for dual-AVG delivery. Further optimization produced PE6C and PE6D, which excel at structured templates but cause indels with unstructured ones. Template secondary structure (folding energy > -23 kcal/mol) now guides variant selection. Advanced applications demonstrate broad therapeutic potential: ex vivo prime editing of sickle cell disease achieved 42% wild-type beta-globin expression in transplanted mice. In vivo delivery via split AAVs and lipid nanoparticles achieved therapeutic editing in brain, liver, and heart tissues. Twin prime editing enables simultaneous installation of both DNA strands, facilitating large-scale gene insertions including recombinase landing sites for site-specific gene insertion.

CRISPR delivery strategies are categorized into ex vivo and in vivo approaches. Ex vivo editing involves extracting cells, editing them outside the body, and reintroducing them, which is more straightforward and well-understood. In vivo delivery requires injecting CRISPR components directly into the human body, facing challenges including immune system survival, tissue targeting, and achieving sufficient editing efficiency. Two primary in vivo platforms are lipid nanoparticles (LNPs) and adeno-associated viruses (AAVs). LNPs have been successfully used for siRNA delivery and can carry larger genetic payloads, with preclinical studies showing 50-75% editing efficiency in liver cells. AAVs offer tissue-specific targeting through different serotypes but face challenges with pre-existing immunity and limited cargo capacity. CRISPR has diverse therapeutic applications including transthyretin amyloidosis (single point mutation correction), hepatitis B virus (clearing latent viral DNA), and inborn errors of metabolism. Technical challenges include delivery efficiency, achieving sufficient editing percentages, and ensuring both gene copies are edited. Safety concerns include off-target effects and immunogenicity of bacterial Cas9 protein.

Effective gene therapy requires efficient delivery of prime editing components to target tissues. Traditional approaches using two adeno-associated viruses (AAVs) are effective but extremely costly ($2-3 million per patient). Alternative delivery systems under development include lipid nanoparticles (successfully used in the Pfizer-BioNTech COVID-19 vaccine), extracellular vesicles (~100 nm particles naturally produced by all cells, ~10^15 per liter in human serum where Cas9 nuclease has been successfully introduced), and virus-like particles for direct protein delivery. These innovations aim to make prime editing therapies more affordable and accessible for clinical applications in treating Duchenne muscular dystrophy and potentially thousands of other inherited diseases caused by single nucleotide mutations.

Gene editing requires three components: Cas9 protein, guide RNA, and donor DNA. Delivery can be as DNA plasmids (cheaper, longer expression, more toxic) or as pre-assembled RNP complexes (more efficient, less toxic, transient expression). Four main delivery methods exist: (1) Liposome delivery packages tools into lipid vesicles that mimic cell membranes, entering via fusion or endocytosis; (2) Polymer transfection uses positively charged polymers to neutralize nucleic acid charges; (3) Gold nanoparticles protect cargo from immune detection and enable in vivo applications; (4) Electroporation uses electrical pulses to create temporary membrane pores, with nucleofection enabling direct nuclear delivery. Viral vectors are modified viruses that deliver genetic material without replicating. Three main types exist: (1) Adenovirus (36 kb capacity, broad tropism, transient expression, high immunogenicity); (2) AAV (4.7 kb capacity, tissue-specific tropism, stable expression, low immunogenicity); (3) Lentivirus (8 kb capacity, infects non-dividing cells, integrates into genome). AAV vectors are produced by co-transfecting HEK293 cells with three plasmids containing capsid genes, transgene, and replication helper genes. Clinical applications include Zolgensma (AAV9 for spinal muscular atrophy) and Luxturna (AAV2 for inherited retinal dystrophy).

Effective prime editing optimization follows an iterative workflow using systematic screening and progressive refinement. Begin with workhorse cell lines (HEK293T or N2a) that are easy to transfect and have the target edit already installed. Clone initial pegRNA libraries using Golden Gate or Gibson cloning, screening 9-15 variants per round. First-pass experiments may test PE2 or PE4 while optimizing only PBS and RT template lengths, or simultaneously screen PBS, RT templates, and 2-3 nicking guides. Use 96-well plates for high-throughput screening. After initial screening, select top-performing pegRNAs and iterate by adding improvements: nicking guides, silent mutations, and mlh1dn. Each round builds upon previous results, progressively improving editing efficiency. Machine learning tools from other labs can predict optimal pegRNAs for new edits. Analysis uses Crispresso 2 with discard_indel_reads=True and appropriate quantification windows covering both nick sites. Essential controls include untreated cells, positive controls at established loci, and side-by-side comparisons to account for transfection variability.

Prime editing is a search-and-replace genome editing technique using Cas9 nickase fused to reverse transcriptase and a pegRNA that encodes both the target site and intended edit. The pegRNA consists of four key components: a spacer sequence targeting the genome, a primer binding site (PBS) for reverse transcription initiation, a reverse transcriptase template (RT) carrying the edit, and a scaffold for Cas9 binding. Successful pegRNA design requires careful optimization of each component: spacer selection should target unique genomic regions near appropriate PAM sequences (NGG for Cas9, TTV for Cas12A) while avoiding repetitive sequences and high GC content; RT template length typically ranges from 10-35 nucleotides (16 as optimal starting point) and should encode desired mutations with strong homology to the target strand; PBS length should be 8-17 nucleotides (13 as optimal) with 40-60% GC content to balance hybridization efficiency and secondary structure formation.

Prime editing relies on specialized guide RNAs (pegRNAs) containing two essential elements: a primer binding site complementary to the nicked DNA strand and a reverse transcriptase template region. The primer binding site must complement the guide RNA spacer, creating a stem-loop structure requiring precise optimization. The site must be long enough for stable reverse transcription initiation but short enough to avoid impeding spacer function. This delicate balance enables productive molecular events while avoiding nonproductive structures, representing a key technical challenge in prime editing implementation.

The Prime Editing Designer database (primeedit.nygenome.org) provides computational infrastructure for designing PEC RNAs tailored to specific genetic variants. Researchers can select between therapeutic applications (repairing mutations) or modeling applications (creating disease models). The platform accepts gene symbols, clinical variant IDs, or genomic coordinates, then generates candidate PEC RNAs with predicted efficiency, binding capability, and off-target scores. This computational infrastructure accelerates experimental planning by pre-screening potential editing strategies before laboratory implementation. Prime Editing has demonstrated therapeutic potential in treating spinal muscular atrophy (SMA), where researchers designed PEC RNAs to convert SMN2 into SMN1-like sequences, increasing SMN protein expression from approximately 1.7% to 91% of normal levels. The workflow involves designing candidate PEC RNAs, cloning into expression vectors, transducing target cells, and monitoring editing outcomes over 14-21 days using flow cytometry and other analytical techniques.

This section details the systematic engineering improvements to Prime editing technology. Researchers identified three bottlenecks: pegRNA degradation creating truncated RNAs that block productive editing, MMR system interference that reverts heteroduplexes back to original sequences, and suboptimal protein architecture. Structured pseudoknot motifs appended to pegRNA 3' ends protected against degradation, increasing editing efficiencies several-fold across cell types. Using RepairSeek screening, researchers discovered that MMR components strongly influence Prime editing outcomes—knocking down MMR genes increased efficiencies while reducing indels. A dominant negative MLH1 protein (mlh1dn) transiently inhibited MMR during editing, enhancing efficiency in MMR-competent cells. Without perturbing MMR, researchers designed edits with multiple silent mutations to evade MMR recognition. Codon optimization, Cas9 mutations, linker engineering, and nuclear localization signal optimization produced PE Max architecture. Combining epegRNAs, mismatch evasion, and PE Max achieved over 10-fold improvement in editing efficiency and edit-to-indel ratios. Advanced applications include precise replacement of pathogenic trinucleotide repeats causing Huntington's disease and Friedreich's ataxia, and Twin PE enabling knock-in of large gene-sized DNA fragments (several thousand base pairs) at targeted sites.

Two pegRNAs can be used to install both DNA strands edited simultaneously, obviating the need for the cell to perform second strand synthesis. This method, called twin prime editing, is especially efficient for performing prime editing of larger DNA sequences such as insertion of roughly 50 base pairs at PAM or Bxb1 recombinase landing sites. Twin prime editing can then be used with Bxb1 recombinase either in a two-step process or in a single transfection process to knock in large multi-kilobase gene-sized DNA at targeted sites of choice in the human genome, including at safe harbor loci or at sites of pathogen DNA loss. This enables RNA-programmed site-specific insertion of gene-sized DNA in mammalian cells.

Prime editing's ability to precisely replace pathogenic trinucleotide repeat numbers with non-pathogenic numbers offers hope for directly correcting devastating repeat expansion disorders. Researchers demonstrated precise replacement of pathogenic CAG repeats causing Huntington's disease and Friedreich's ataxia in both mouse embryonic stem cells and patient-derived fibroblasts. Twin PE uses two pegRNAs simultaneously to install both edited DNA strands, eliminating the need for cellular second-strand synthesis. This approach enables efficient editing of larger sequences including attB landing sites for ΦC31 recombinase. Combining Twin PE with ΦC31 recombinase enables knock-in of large gene-sized DNA fragments (several thousand base pairs) at targeted sites in the human genome, representing the first RNA-programmable site-specific insertion of gene-sized DNA in mammalian cells. In vivo Prime editing has been achieved using split AAV systems, plasmid injections, and adenoviral delivery in animal models of tyrosinemia and phenylketonuria.

In prime editing, the gene is snipped off rather than cutting the DNA, and it is a gene editing method that can perform targeted small insertions, deletions, and base swapping in a precise way. Twin prime editing is a new CRISPR-based gene editing strategy that enables the manipulation of gene-sized chunks of DNA in human cells without cutting the DNA double helix.

Prime editing, introduced in late October 2019 by Dr. David Liu at MIT and Harvard, is a new gene editing technology that may correct 89% of the more than 75,000 disease-associated DNA variants listed in ClinVar. The technique uses methods similar to CRISPR but with higher accuracy. However, prime editing may not be able to make the very big DNA insertions or deletions that CRISPR-Cas9 is capable of, so it is unlikely to completely replace the established editing tool. Different genome editing platforms will still be needed for different types of edits.

Prime editing is a revolutionary CRISPR technology developed in 2019 that can produce all 12 types of base substitutions, precise insertions up to 44 base pairs, deletions up to 80 base pairs, and their combinations. It uses a fusion of nickase Cas9 with reverse transcriptase, complexed with a prime editing guide RNA containing a reverse transcriptase template and primer binding site. This technology enables precise genome modifications including transversions that were previously difficult to achieve with other methods.

Prime editing was translated to non-human primate models to assess clinical translatability. Using LNP delivery of prime editors targeting primate PCSK9, researchers achieved editing rates of up to 30% depending on dose, consistent with mouse study results. This resulted in significant reductions in serum PCSK9 and LDL cholesterol levels, demonstrating that prime editing can achieve therapeutic editing in primates. This successful translation supports advancing prime editing toward clinical trials for treating familial hypercholesterolemia.

Prime editing technology was invented only five years ago and is still primarily in preclinical development stages. Researchers are using multiple animal models of retinal degeneration to develop and test base editing and prime editing approaches. The progress with Luxturna and other viral delivery systems for RNA and DNA materials provides a foundation for accelerating genome editing translation. Once the first genome-edited treatment is approved, subsequent treatments for different mutations will follow a similar development pathway, potentially streamlining regulatory approval processes.

Prime editing combines nickase and reverse transcriptase to enable arbitrary edits (substitutions, insertions, deletions up to ~200 bp) through a flap-based mechanism. The pegRNA encodes both target specificity and edit template. PACE (phage-assisted continuous evolution) accelerated optimization, yielding compact variants (p6a/b) and specialized variants (p6c/d) for challenging templates. This enabled correction of CFTRΔF508 (85% of CF cases), restoring ion channel function comparable to existing drug cocktails. Prime editing achieves high fidelity through multiple hybridization checkpoints. FDA cleared the first Prime editing clinical trial for chronic granulomatous disease, demonstrating rapid clinical translation.

Prime editing was proposed to correct almost any type of DNA misspelling in living cells, envisioned as a DNA word processor unlike CRISPR-Cas9 (scissors) or base editors (pencils). The machine uses guide RNA to specify both where to edit and what the corrected sequence should be, encoded in a special extension. The system uses a protein that copies RNA sequences into DNA, attached to disabled CRISPR that nicks one DNA strand. In test tubes, this successfully copied corrected sequences onto target DNA. In yeast, the corrected flap replaced original DNA. In human cells, initial attempts failed until researchers discovered that a Goldilocks length runway was critical for breakthrough. Initial efficiency was terrible, with only one cell out of 100 showing the desired edit. Testing dozens of variants found a special combination of five changes achieving around 10 percent efficiency. Nicking the non-edited strand further improved editing to exceed 50 percent. Prime editing enables true search and replace genome editing, allowing replacement of any stretch of up to dozens of DNA letters with virtually any other stretch. This versatility corrects the types of misspellings accounting for the vast majority of known DNA errors causing genetic disorders. Prime editing has been used in animals to rescue phenylketonuria, tyrosinemia, and sickle cell disease. Four base editing clinical trials are underway in four countries treating diseases from cancer to sickle cell disease to high cholesterol. Beam Therapeutics received FDA clearance for the first base editing clinical trial in the U.S. using Nicole's base editor to treat sickle cell disease. The first clinical outcome was announced for Alyssa, a 13-year-old with T cell leukemia who achieved complete remission after treatment with base-edited T cells.

Translating gene editing to human clinical trials requires: (1) Large animal proof-of-concept studies (underway in 2017); (2) Safety studies in non-human primates; (3) Scaling up delivery systems while maintaining efficiency; (4) Confirming absolute safety of the technique. If all goes well, initial human trials could be envisioned within a few years. However, enormous responsibility accompanies this technology given the potential for permanent changes to the human genome.