Traditional CRISPR–Cas9 transformed genetics by giving scientists a programmable way to cut DNA at selected locations. Once the DNA is cut, the cell’s repair machinery can disrupt a gene or, under certain conditions, incorporate a corrected sequence.
This approach has produced the first approved CRISPR therapy, but a double-strand DNA break is not always the ideal way to change a single genetic letter. Cellular repair can introduce unpredictable insertions, deletions or larger rearrangements around the cut site.
Base editing and prime editing were developed to make more targeted changes without creating the standard double-strand breaks used by conventional CRISPR–Cas9. Base editors chemically convert selected DNA bases, while prime editors copy new genetic information into a chosen location.
These technologies are not error-free, and neither has replaced traditional CRISPR. Instead, they expand the genome-editing toolkit, allowing researchers to match the editing method to the type of mutation being treated.
Why DNA breaks can be unpredictable
Traditional CRISPR–Cas9 generally uses two main components:
- A guide RNA that directs the system to a selected DNA sequence
- A Cas9 enzyme that cuts both strands of DNA
After the cut, the cell must repair the break. One common repair pathway, nonhomologous end joining, reconnects the DNA ends but frequently creates small insertions or deletions known as indels. These changes can disable a gene, which is useful when gene disruption is the goal.
Making a precise correction is more difficult. Researchers may provide a donor DNA template and attempt to direct the cell toward homology-directed repair. However, this pathway is inefficient in many cell types, particularly cells that are not actively dividing.
Double-strand breaks may also produce:
- Large deletions
- Chromosomal rearrangements
- Loss of genetic material
- Activation of cellular stress responses
- Different repair outcomes in different cells
These risks do not mean that traditional CRISPR is inherently unsafe. The first approved CRISPR therapy uses double-strand cutting successfully. However, when the goal is to correct one DNA letter, creating a full break may be more disruption than necessary.
What is base editing?
Base editing directly converts one DNA base into another at a targeted location. It combines a modified Cas protein with an enzyme that chemically changes a nucleotide.
Most base editors use a Cas9 nickase. Instead of cutting both DNA strands, the nickase cuts only one strand. A guide RNA positions the editor over the selected genomic sequence, and an attached deaminase enzyme performs the chemical conversion.
The first widely used cytosine base editor was reported in 2016. It converted a cytosine–guanine base pair into a thymine–adenine pair without requiring a conventional double-strand break or donor DNA template.
Two major categories followed:
Cytosine base editors
- Cytosine base editors, or CBEs, can generally produce the conversion:
-
-
- C•G → T•A
-
This can also be described as a G-to-A change when viewed from the opposite DNA strand.
Adenine base editors
- Adenine base editors, or ABEs, can generally produce:
- A•T → G•C
The first adenine base editor was reported in 2017 and expanded the number of disease-associated mutations that could potentially be corrected.
Additional experimental base editors are being developed to perform other conversions, including certain transversions. However, cytosine and adenine base editors are currently the most clinically advanced.
Base editing does not “cut and replace” DNA
It is tempting to describe base editing as correcting a typographical error, but the process is more chemically specific than using a genomic eraser.
A base editor usually acts within an editing window containing several nucleotides. If multiple compatible bases fall within that window, the editor may modify more than one. These unintended nearby conversions are called bystander edits.
The exact outcome depends on:
- The target sequence
- The position of the base within the editing window
- The guide RNA
- The type of Cas protein
- The deaminase enzyme
- The cell type
- The cell’s DNA-repair activity
A mutation may be chemically correctable but positioned poorly for the available editor. Alternatively, the desired base may be accompanied by another editable base that should remain unchanged.
Base editing is therefore highly programmable, but it is not universally precise at every genomic location.
The first clinical uses of base editing
Base editing entered human clinical research through ex-vivo cell therapy. In one early application, donor-derived T cells were edited at multiple locations and engineered to recognize cancer cells in patients with relapsed T-cell acute lymphoblastic leukemia.
An early clinical report demonstrated that base-edited CAR T cells could be manufactured and administered to patients, providing evidence that multiplex base editing was possible in a therapeutic cell product. The study was small and involved seriously ill patients, so larger trials are required to establish safety and effectiveness.
Base editing has since moved inside the body.
A personalized treatment for one infant
In 2025, researchers reported a landmark personalized base-editing treatment for an infant with severe carbamoyl-phosphate synthetase 1, or CPS1, deficiency.
CPS1 deficiency disrupts the urea cycle, preventing the body from safely processing nitrogen. Ammonia can accumulate to toxic levels, causing brain injury, coma or death.
Researchers identified the infant’s specific mutation, designed an adenine base editor and packaged the editing components inside lipid nanoparticles that targeted the liver. From diagnosis to treatment, development took approximately six months.
Following treatment, the infant tolerated more dietary protein and required less ammonia-lowering medication. The child was also better able to tolerate common illnesses without severe ammonia accumulation.
This was an N-of-1 treatment developed for a single patient. It did not establish that the therapy is safe or effective for other patients with CPS1 deficiency. Long-term monitoring is required to determine durability and identify delayed effects.
Nevertheless, the case demonstrated that a genome-editing platform could potentially be adapted rapidly for an ultra-rare mutation.
Base editing for high cholesterol
Base editing is also being tested for more common conditions.
VERVE-102 is an investigational in vivo adenine base-editing therapy designed to disrupt a splice site in the liver’s PCSK9 gene. People with naturally occurring PCSK9 loss-of-function variants tend to have lower LDL cholesterol and a reduced lifetime risk of cardiovascular disease.
In a 2026 phase 1 study, 35 adults with heterozygous familial hypercholesterolemia or premature coronary artery disease received a single infusion. The highest dose produced average reductions of approximately 88% in PCSK9 and 62% in LDL cholesterol. Reductions appeared sustained during available follow-up, which had reached at least one year for some participants.
The results are promising but preliminary. Larger studies must determine whether the treatment prevents heart attacks and strokes and whether permanently editing PCSK9 offers an acceptable risk–benefit balance compared with effective, reversible cholesterol medications.
What is prime editing?
Prime editing was first reported in 2019 as a more versatile “search-and-replace” system. It can potentially create all 12 possible single-base substitutions, as well as selected small insertions and deletions.
A prime editor includes the following:
- A Cas9 nickase
- An engineered reverse-transcriptase enzyme
- A prime-editing guide RNA, or pegRNA
The pegRNA performs two jobs. It directs the editor to the target site and carries a template containing the desired genetic change.
After reaching the target, the Cas9 nickase makes a single-strand nick. The reverse transcriptase then uses the pegRNA template to write the proposed sequence into DNA. Cellular repair pathways help resolve the edited and unedited DNA strands, ideally retaining the new sequence.
Prime editing does not require a separately delivered donor DNA template and does not normally create the standard double-strand break produced by conventional Cas9.
Why prime editing is more versatile
Standard cytosine and adenine base editors are limited to particular chemical conversions. Prime editing can theoretically perform:
- All possible base-to-base substitutions
- Small targeted insertions
- Small targeted deletions
- Combinations of substitutions and insertions or deletions
This matters because disease-associated variants are not limited to transitions such as C-to-T or A-to-G. Some require a transversion—such as changing an A to a T—or the insertion or removal of several DNA letters.
In the original laboratory study, prime editing was used to model or correct variants associated with sickle cell disease, Tay–Sachs disease and other conditions in human cells.
However, theoretical versatility does not guarantee efficient editing in every cell or at every genomic site.
Prime editing’s technical challenges
Prime editing is more complex than conventional CRISPR or base editing. The editor is physically large, making delivery difficult, especially with viral vectors that have limited cargo capacity.
Efficiency can also vary according to:
- pegRNA stability
- Length and design of the reverse-transcription template
- Target sequence
- Chromatin accessibility
- Cell type
- DNA-repair pathways
- Delivery method
Some prime-editing strategies nick the second DNA strand to encourage the cell to preserve the edited version. This can increase efficiency but may also produce more indels or, in rare cases, create double-strand breaks when the two nicks occur close together.
Prime editing generally produces fewer byproducts than conventional double-strand-break-mediated correction in many experimental settings, but it does not eliminate unintended outcomes.
Prime editing reaches clinical research
The first reported clinical use of prime-edited cells involved PM359, an investigational ex-vivo therapy for chronic granulomatous disease.
This rare inherited immune disorder can result from mutations that impair the NADPH oxidase complex used by immune cells to kill certain bacteria and fungi. PM359 is designed to correct a common mutation in the NCF1 gene within a patient’s blood-forming stem cells.
The cells are collected, prime-edited outside the body and infused back after conditioning treatment. Preliminary company-reported results from the first treated participant in 2025 indicated engraftment and restoration of NADPH oxidase activity.
These findings represent very early clinical evidence. The results were initially reported by the developer rather than in a complete peer-reviewed clinical publication, and conclusions cannot be drawn from one patient. Larger cohorts and longer follow-up are essential.
Comparing the three approaches
|
Feature |
Traditional CRISPR–Cas9 |
Base editing |
Prime editing |
|
Primary DNA action |
Cuts both DNA strands |
Chemically changes a base, usually with a single-strand nick |
Nicks one strand and copies an encoded edit |
|
Donor DNA required |
Often required for precise correction |
No |
No separate donor template |
|
Main strength |
Efficient gene disruption |
Efficient selected base conversions |
Broad range of substitutions and small insertions or deletions |
|
Major limitation |
Repair may produce indels or rearrangements |
Limited conversions and bystander edits |
Complex design, delivery and variable efficiency |
|
Double-strand breaks |
Intentional |
Not standard, but rare breaks may occur |
Not standard, but some configurations may produce breaks |
|
Clinical maturity |
Approved therapy available |
Early clinical trials and personalized use |
Earliest clinical testing |
The choice is not simply between an older and a newer technology. Conventional CRISPR may remain the most appropriate tool when the goal is to disable a gene. Base editing may be preferable for a compatible single-letter conversion. Prime editing may be more appropriate for a change that base editors cannot create.
Does avoiding double-strand breaks eliminate risk?
No. Avoiding a conventional double-strand break may reduce certain risks, but base and prime editing introduce their own safety considerations.
Off-target editing
- The guide RNA may direct the editor to similar sequences elsewhere in the genome. The Cas component and attached editing enzyme can then modify unintended sites.
Bystander editing
- Base editors may convert additional compatible bases within the editing window. Some bystander changes may be harmless, while others could alter a protein or gene-regulatory sequence.
Guide-independent activity
- Some deaminase enzymes may act on DNA or RNA without being directed by the guide RNA. Newer editors are being engineered to reduce this activity.
Insertions and deletions
- Base and prime editors can still produce indels, particularly when DNA nicks are repaired unexpectedly.
Immune responses
- The body may recognize bacterial Cas proteins, delivery vehicles or edited cells as foreign.
Unexpected cellular effects
Even a perfectly targeted edit can have unforeseen biological consequences if the gene has additional functions in different tissues or stages of life.
Safety must therefore be assessed at three levels: whether the correct genetic location was targeted, whether the intended edit was produced cleanly, and whether changing that gene has the predicted biological effect.
Delivery remains a major barrier
Base and prime editors must reach the cells involved in the disease.
Ex-vivo editing provides easier access to blood stem cells and immune cells because they can be collected and manipulated in a laboratory. In-vivo editing requires a delivery system capable of reaching the correct organ.
Lipid nanoparticles currently perform particularly well in the liver, explaining why several early clinical programs target liver genes. Reaching the brain, muscle, lung, heart and bone marrow remains more difficult.
Prime editors are especially challenging because their molecular components are large. Researchers are exploring:
- Smaller Cas enzymes
- Split delivery systems
- Engineered lipid nanoparticles
- Virus-like particles
- Improved pegRNAs
- Transient RNA-based delivery
- Tissue-specific targeting molecules
The best editor is clinically useful only if it can be delivered safely to enough of the correct cells.
Durability and reversibility
Base and prime editing are intended to produce lasting DNA changes. If a long-lived stem cell is edited, the change may be transmitted to its descendant cells for years.
This durability could reduce or eliminate the need for repeated treatment. It also means that unexpected effects may be difficult to reverse.
A conventional medication can usually be stopped. A permanent genetic edit may remain after the editing machinery has disappeared. Long-term follow-up is therefore necessary even when early results appear favorable.
Researchers are also developing RNA editing and epigenetic editing, which may produce temporary or potentially reversible changes without altering the DNA sequence. These methods could be useful when permanent modification is unnecessary or presents too much risk.
Personalized editing for rare diseases
The personalized CPS1 treatment raises the possibility of developing gene editors for mutations affecting only one person or a few families.
A platform-based approach could reuse many components while changing the guide RNA or editing template for each mutation. This may reduce development time compared with building every therapy from the beginning.
However, personalized editing creates major regulatory and economic questions:
- How much evidence is required before treating one patient?
- Which preclinical tests can be standardized?
- How should rare off-target risks be evaluated?
- Who will pay for individualized development?
- How can treatment be offered equitably?
- How should lifelong follow-up be organized?
The scientific ability to design a treatment does not guarantee that healthcare systems can manufacture, evaluate and provide it.
The Bottom Line
Base editing and prime editing expand what CRISPR technology can do. Base editing can efficiently produce selected DNA-letter conversions without a standard double-strand break. Prime editing is more versatile, potentially supporting all base substitutions as well as small insertions and deletions. Early clinical experiences, including base-edited immune cells, in-vivo liver editing, a personalized treatment for CPS1 deficiency, and initial prime-edited stem-cell therapy—show that these tools are moving beyond laboratory research. However, greater precision does not mean perfect precision. Bystander edits, off-target activity, unexpected repair products, delivery barriers and permanent biological effects remain important concerns. The future of gene editing will not depend on finding one tool that replaces all others. It will depend on selecting the safest and most effective editor for each mutation, cell type and clinical situation.
This article is intended for scientific education and does not replace individualized medical advice. Most base-editing and prime-editing treatments discussed remain investigational and are not approved for routine clinical use.
References
Anzalone, A. V., Randolph, P. B., Davis, J. R., Sousa, A. A., Koblan, L. W., Levy, J. M., Chen, P. J., Wilson, C., Newby, G. A., Raguram, A., & Liu, D. R. (2019). Search-and-replace genome editing without double-strand breaks or donor DNA. Nature, 576, 149–157. https://doi.org/10.1038/s41586-019-1711-4
Chiesa, R., Georgiadis, C., Syed, F., Zhan, H., Etuk, A., Gkazi, S. A., Preece, R., Ottaviano, G., Braybrook, T., Chu, J., Kubat, A., Adams, S., Thomas, R., Gilmour, K., O’Connor, D., Vora, A., & Qasim, W. (2023). Base-edited CAR7 T cells for relapsed T-cell acute lymphoblastic leukemia. The New England Journal of Medicine, 389(10), 899–910. https://doi.org/10.1056/NEJMoa2300709
Gaudelli, N. M., Komor, A. C., Rees, H. A., Packer, M. S., Badran, A. H., Bryson, D. I., & Liu, D. R. (2017). Programmable base editing of A•T to G•C in genomic DNA without DNA cleavage. Nature, 551, 464–471. https://doi.org/10.1038/nature24644
Komor, A. C., Kim, Y. B., Packer, M. S., Zuris, J. A., & Liu, D. R. (2016). Programmable editing of a target base in genomic DNA without double-stranded DNA cleavage. Nature, 533, 420–424. https://doi.org/10.1038/nature17946
National Institutes of Health. (2025, June 3). Infant with rare disease receives customized gene therapy. NIH Research Matters. https://www.nih.gov/news-events/nih-research-matters/infant-rare-disease-receives-customized-gene-therapy
Prime Medicine, Inc. (2025, May 19). Prime Medicine announces breakthrough clinical data showing rapid restoration of DHR positivity after single infusion of PM359, an investigational prime editor for chronic granulomatous disease [Press release]. GlobeNewswire. https://www.globenewswire.com/news-release/2025/5/19/3083999/0/en/Prime-Medicine-Announces-Breakthrough-Clinical-Data-Showing-Rapid-Restoration-of-DHR-Positivity-After-Single-Infusion-of-PM359-an-Investigational-Prime-Editor-for-Chronic-Granuloma.htm
Vafai, S. B., Täubel, J., Ashdown, T., Patel, R. S., Diamondali, S., Cegla, J., Soran, H., Bashir, B., Abitbol, A., Gaudet, D., Lauzière, A., Brunham, L. R., Newby, D. E., Nicholls, S. J., Scott, R. S., Kerr, J., Tardif, J. C., Lunken, C., Humphries, S. E., Karsten, V., … Kathiresan, S. (2026). In Vivo Base Editing of PCSK9 with VERVE-102 for Hypercholesterolemia. The New England journal of medicine, 395(7), 648–659. https://doi.org/10.1056/NEJMoa2601283
