Base editing: changing single letters of DNA

Our DNA is written with four letters, A, C, G and T, and a single wrong letter can contribute to an inherited disease. Base editing is a gene-editing method that changes one letter into another at a chosen spot in the genome, without cutting through both strands of the DNA double helix. That matters because many earlier gene-editing tools cut both strands, and the cell's repair of the break can leave unwanted changes. Base editing aims for a more precise correction. It is still early, and most results come from cells, animals and a handful of people.

How it works

A base editor pairs two parts. A guide system finds the target spot in the genome, and an enzyme attached to it changes one letter. A review by Rees and Liu describes base editors as a catalytically disabled CRISPR nuclease fused to a nucleobase deaminase, which makes point mutations without double-strand breaks (Rees and Liu, 2018). The two main families, cytosine base editors and adenine base editors, work on C and A letters.

Prime editing, described by Anzalone and colleagues in 2019, works differently. Its guide RNA carries both the target address and a template for the new sequence, so the editor can write a wider range of changes (Anzalone et al., 2019).

What has been shown

Lab work from 2016. Komor and colleagues reported a way to change a single target base in genomic DNA without cutting both strands (Komor et al., 2016). In four human and mouse cell lines, their second- and third-generation editors permanently corrected roughly 15 to 75 per cent of total cellular DNA, with insertions and deletions typically at 1 per cent or below.

Blood stem cells in mice, 2021. Newby and colleagues edited blood stem and progenitor cells from people with sickle cell disease outside the body, converting 80 per cent of the sickle version of the gene to a harmless variant (Newby et al., 2021). Sixteen weeks after these edited human cells were transplanted into immunodeficient mice, 68 per cent still carried the edit. In a second experiment using cells from mice engineered to carry human sickle cell disease, mice given edited cells showed near-normal blood measures and less spleen damage than mice given unedited cells. These were animal experiments, not treatment of people.

Prime editing in human cells, 2019. Anzalone and colleagues reported more than 175 edits in human cells, including insertions, deletions and all 12 types of point mutation, without double-strand breaks or donor DNA (Anzalone et al., 2019). This is also a cell study.

One infant, 2025. This is the only human case among the sources cited here, and it covers one patient. Musunuru and colleagues reported in the New England Journal of Medicine an infant with severe CPS1 deficiency, a rare metabolic condition with an estimated 50 per cent death rate in early infancy (Musunuru et al., 2025). A base-editing treatment, delivered by lipid nanoparticles, was developed for him after diagnosis. The paper reports two infusions at about 7 and 8 months of age. In the 7 weeks after the first infusion he could take more dietary protein and his nitrogen-scavenger medicine was reduced to half the starting dose, with no serious adverse events. The authors state that longer follow-up is needed.

What the hospital reports

The Children's Hospital of Philadelphia, in a press release dated 15 May 2025, says the treatment targeted the child's specific CPS1 variant and was delivered to the liver (Children's Hospital of Philadelphia, 2025). The hospital reports that he first received treatment at six to seven months old, in February 2025, with further doses in March and April 2025, three in all. It reports that he could eat more protein, needed less nitrogen-scavenger medicine and had no serious side effects. These are the hospital's statements; the peer-reviewed paper describes two infusions and a shorter follow-up period.

Limits and open problems

• Off-target edits. A 2020 review says guide-dependent off-target editing occurs relatively infrequently for both cytosine and adenine base editors, but still matters for medical use. It also describes guide-independent, genome-wide deamination by the cytidine deaminases used in cytosine base editors (Jeong et al., 2020).
• Bystander edits. Other C or A letters near the target can change too. The same review says narrowing the editing window is one way to reduce this. The first base editors acted on a window of about five letters (Komor et al., 2016).
• Which changes are possible. Base editors built on SpCas9 prefer NGG PAM sequences, which limits the genes they can reach. The review notes that DNA base editing is confined to transition mutations and cannot make insertions or deletions, for which prime editing was developed as an alternative (Jeong et al., 2020).
• Delivery. The 2025 case used a liver-directed route. The sources cited here do not show delivery to muscle, brain, lung or blood-forming stem cells in the body. The 2020 review also notes that long-lasting expression from plasmid or viral delivery frequently increases off-target effects.
• Cost and scale. The 2025 treatment was designed for one child's variant. The sources cited here give no costs and do not show that this can be repeated at scale.
• Long-term safety. The hospital says the child will need careful monitoring for the rest of his life, and the follow-up reported so far covers months, not years.
• Evidence size. The only clinical evidence cited in this article is one patient. The larger body of evidence here comes from cells and mice.

Speculative

• Personalised base editing for many rare inherited diseases that have no approved treatment. This rests on a single reported case.
• Prime editing to correct mutations that base editors cannot reach. The 2019 work was in cells only.
• A single course of blood stem cell editing to treat several inherited blood disorders. No human data cited here support this yet.

What to watch next

• Longer follow-up for the 2025 infant, and whether the same team treats a second patient with a bespoke editor.
• Peer-reviewed clinical reports with more than one patient, stating the condition and follow-up period.
• Delivery methods that reach tissues other than the liver.
• Published cost figures for individually designed editors.

This article is general information and is not medical advice.