The first generation of genetic medicines was built around a relatively straightforward idea. If a disease is caused by a defective gene, correct the gene itself. CRISPR made this idea increasingly practical, but permanently changing DNA is not always necessary to restore a biological function. In some cases, the same result may be achieved by correcting the RNA produced from the gene, leaving the genome untouched.

RNA editing is interesting because it changes the relationship between genetic precision and permanence. Rather than introducing a lasting alteration to DNA, a therapeutic molecule can direct a cell's existing machinery to modify a specific nucleotide in an RNA transcript. The edited RNA can then produce a different protein. Once the transcript is degraded, however, the correction is lost and newly transcribed RNA can carry the original mutation again. This creates a potential advantage in safety and control, but also raises a difficult question about whether a sufficiently durable therapeutic effect can be achieved.

The distinction between RNA editing and other RNA therapeutics is important. Antisense oligonucleotides can alter splicing or reduce the abundance of a transcript. Small interfering RNAs can trigger the degradation of specific messenger RNAs. Messenger RNA therapeutics introduce a transcript that cells translate into a protein. RNA editing takes a different approach by changing the information contained in an existing RNA molecule. Rather than replacing a protein or suppressing a gene, it attempts to alter the molecular message from which the protein is produced.

One of the most developed strategies uses adenosine deaminases acting on RNA, known as ADAR enzymes. These enzymes occur naturally in human cells and catalyse the conversion of adenosine to inosine in double stranded RNA. During translation, inosine is generally interpreted as guanosine. An A to I edit can therefore change the codon read by the ribosome and, in suitable circumstances, restore the intended amino acid.

The chemistry is straightforward to describe, but designing a therapeutic is considerably harder. ADAR enzymes do not automatically recognise disease causing mutations. They act on RNA structures, and their activity depends on the sequence surrounding the target, the structure of the RNA and the availability of the enzyme. Researchers have therefore developed guide oligonucleotides that bind to the target transcript and create a suitable substrate for endogenous ADAR.

A useful review by Pfeiffer and Stafforst in Nature Biotechnology examines the different approaches to programmable RNA base editing, including the use of endogenous enzymes and engineered editing systems. Their analysis highlights a central limitation of the field. It is not enough to identify a nucleotide that could theoretically be corrected. The therapeutic must edit that nucleotide efficiently and selectively in the relevant cells, while avoiding changes elsewhere in the transcriptome.

Read the review in Nature Biotechnology.

The transition from laboratory research to clinical testing has already produced an important milestone. In October 2024, Wave Life Sciences reported the first clinical proof of mechanism for an endogenous ADAR based RNA editing therapeutic. Its candidate, WVE-006, was developed for alpha-1 antitrypsin deficiency, an inherited disorder caused by pathogenic variants in SERPINA1.

The common Pi*ZZ form of the disease presents an interesting therapeutic problem because the mutation has consequences in two different organs. In the liver, the mutant protein misfolds and accumulates inside hepatocytes. In the lungs, insufficient functional alpha-1 antitrypsin reaches the circulation to protect tissue against proteases such as neutrophil elastase. A treatment that simply replaces the circulating protein may address the deficiency in the lungs but does not directly correct the underlying production of the misfolded protein in the liver.

WVE-006 is designed to edit the mutant SERPINA1 messenger RNA. The Z variant introduces an adenine at the position targeted by the treatment. ADAR mediated conversion of that adenine to inosine allows the editing machinery to restore the intended coding information. The DNA mutation remains present, but the edited transcript can produce functional M alpha-1 antitrypsin.

In the first two patients reported in 2024, a single dose was followed by a mean circulating concentration of approximately 6.9 micromolar of functional M alpha-1 antitrypsin at day 15. Total alpha-1 antitrypsin reached approximately 10.8 micromolar. The results suggested that an oligonucleotide could recruit endogenous editing machinery in humans and produce a measurable amount of functional protein.

Read the clinical proof of mechanism in Nature Biotechnology.

These results deserve attention, but they also need to be interpreted carefully. The initial report involved only two patients. It demonstrated that the therapeutic mechanism could operate in humans, not that the treatment could prevent liver disease, preserve lung function or improve long term survival. A change in a circulating biomarker is an important first step, but it is not equivalent to demonstrating clinical efficacy.

The distinction is particularly relevant for genetic medicines, where molecular results can appear convincing long before the relationship between those results and patient outcomes has been established. The next questions are whether editing remains consistent with repeated administration, whether the treatment is safe over longer periods and whether the resulting protein levels translate into meaningful clinical benefit.

Another difficulty is that RNA editing efficiency is not determined by the guide sequence alone. The local RNA structure influences how effectively ADAR can access the target. A guide that performs well in one transcript or cell type may be much less effective in another. Increasing the amount of therapeutic molecule may not solve the problem if the target structure is unfavourable or the relevant enzyme is limiting.

A study published in Nature Biotechnology in 2025 and included in the journal's 2026 volume investigated this problem through a strategy called MIRROR. Rather than relying only on conventional guide designs, the researchers used structural motifs derived from naturally occurring RNA substrates that are efficiently edited by ADAR. They reported improvements in editing efficiency of up to 5.7 fold in several experimental systems.

Read the MIRROR study in Nature Biotechnology.

The interesting aspect of this work is not simply the reported increase in efficiency. It suggests that RNA editing can be improved by learning from the structures that endogenous enzymes already recognise. This creates an opportunity for more rational therapeutic design, where guide molecules are developed according to experimentally established rules rather than selected entirely through trial and error.

However, these results should not be confused with clinical validation. Improved editing in cells and animal models does not establish that a candidate will be safe or effective in humans. The study addresses an important molecular engineering problem, but delivery, tissue specificity and the relationship between editing efficiency and therapeutic benefit remain separate challenges.

Delivery may ultimately determine which diseases can be treated. An oligonucleotide must reach the relevant tissue, enter the appropriate cells and become available in the compartment where its target RNA is found. Liver directed delivery has benefited from N-acetylgalactosamine conjugation, which allows oligonucleotides to be taken up by hepatocytes through a receptor mediated pathway. This helps explain why liver directed applications have progressed further than many programmes targeting other organs.

The same approach cannot simply be transferred to the brain, skeletal muscle or every other tissue. Each presents different biological barriers. A therapeutic designed for a liver disease may therefore be technically feasible while an equivalent approach for a neurological disorder remains difficult to deliver. The range of diseases that RNA editing can treat will depend as much on delivery technology as on the editing chemistry itself.

Specificity is another unresolved issue. ADAR enzymes participate in natural RNA editing throughout the body, and unintended editing could alter protein sequences or affect RNA processing. A guide molecule may also interact with transcripts other than its intended target. These risks require transcriptome wide analysis and careful evaluation in relevant biological systems. The fact that editing is temporary does not make an unintended edit harmless, particularly if it changes a protein involved in an important cellular process.

There is also a question of how RNA editing should compete with existing treatments. For some diseases, protein replacement or conventional RNA therapeutics may already provide an effective and relatively safe option. RNA editing needs to offer a meaningful advantage to justify the additional complexity of developing a sequence specific treatment. That advantage might be the restoration of a functional protein within the patient's own cells, the correction of a disease mechanism that existing medicines cannot address, or the ability to adjust treatment without permanently altering DNA.

Its temporary nature could be valuable where permanent genomic changes are undesirable. At the same time, repeated dosing may be necessary to maintain the effect. Whether this is an acceptable trade off will depend on the disease, the duration of the desired response and the safety of long term administration.

The technology may eventually extend beyond correcting inherited mutations. RNA editing could also be used to modify protein activity or regulate biological pathways where a temporary change is preferable to permanent genome modification. But this broader ambition needs to be approached cautiously. A technically editable transcript is not necessarily a therapeutically useful target. The target must be causally relevant to the disease, and changing it must produce a benefit that matters to the patient.

The most important question is therefore no longer whether RNA can be edited in living systems. Human proof of mechanism has established that the basic approach can work. The challenge is to determine where it offers a genuine clinical advantage, how reliably it can be delivered and whether its molecular effects translate into better outcomes.

RNA editing will not replace genome editing, antisense therapies or other forms of genetic medicine. Each acts at a different level and has a different set of advantages and limitations. Its value will depend on identifying the situations in which changing an RNA transcript is more useful than changing DNA, suppressing a gene or supplying a replacement protein.

The broader significance is that genetic medicine is becoming less dependent on a single idea of what it means to correct a mutation. Sometimes the most effective intervention may be permanent. In other cases, a temporary correction at the RNA level may provide a better balance between precision, reversibility and risk. The progress of RNA editing therapeutics will depend on how well researchers can establish that balance in real patients.