A gene editor that fixes the target base without scrambling its neighbors

What happens when a gene editor hits the right letter in your DNA but also rewrites the ones next door?

That question has shadowed base editing since the technology's earliest days. Base editors - molecular tools that chemically convert one DNA letter to another without cutting the double helix - have moved from laboratory curiosity to clinical trials in under a decade. They can treat sickle cell disease, lower cholesterol, fight leukemia. But they carry a liability: bystander edits, unwanted changes to neighboring bases that can be silent or, in the worst case, lethal to cells.

Now a team at UC San Diego, led by base editing pioneer Alexis Komor, has found a way to suppress those bystander edits without sacrificing the editing power clinicians need. The work, published in Nature Biotechnology, introduces a tool called ME-ABE - a minimally evolved adenine base editor.

The efficiency-precision tradeoff that plagued ABEs

Adenine base editors convert adenine (A) to guanine (G) at a target site. The problem arises when multiple adenines sit near the target. The editor does not always discriminate - it can convert neighboring A's along with the intended one. These bystander edits are an inherent risk because the enzyme's active zone, called the editing window, spans several bases.

One obvious fix is narrowing the editing window. Earlier versions of adenine base editors, particularly ABE7.10, had naturally narrower windows. But they also edited their targets less efficiently - roughly speaking, they missed their intended target too often to be clinically useful. The newer variants, ABE8.20 and ABE8e, solved the efficiency problem and entered clinical trials. But they widened the window, dramatically increasing bystander editing.

The field was stuck in a tradeoff: you could have precision or power, not both.

Fourteen mutations, five reversals, one new tool

Komor's postdoctoral researcher Mallory Evanoff took a counterintuitive approach. Instead of engineering forward - adding new modifications to improve the editor - she looked backward at ABE7.10's architecture.

ABE7.10 carries 14 engineered mutations in its deaminase enzyme, all identified through a process called directed evolution performed in E. coli bacteria. That technique won the 2018 Nobel Prize in Chemistry. But because the mutations were selected collectively in bacterial cells, no one had systematically determined what each individual mutation contributed - especially in human cells, where the editor actually needs to work.

Evanoff performed reversion analysis: she changed each of the 14 mutations back to its original wild-type sequence, one at a time, and tested the resulting editors in both human cells and E. coli. The results were revealing. Some mutations behaved similarly in both systems. Others had divergent effects - mutations that helped in bacteria did little or even hurt performance in human cells.

She identified five mutations that, when individually reverted, either had no impact or actually increased editing activity in human cells. Combining all five reversions produced ME-ABE: an editor with fewer engineered changes than ABE7.10, a narrow editing window, and efficiency on par with the ABE8 variants used in current clinical trials.

Decoupling two properties everyone assumed were linked

The significance is in what ME-ABE breaks apart. Every previous attempt to increase editing efficiency had come with a wider window and more bystander damage. ME-ABE achieves high on-target activity while maintaining the tight window of the older, less efficient tool. Komor called it one of the first times the field has decoupled these two properties.

For therapeutic applications, this matters enormously. An editor that efficiently installs the desired correction but simultaneously introduces unintended mutations nearby is a safety concern. Clinical regulators scrutinize off-target and bystander effects closely. ME-ABE offers a path to editors that satisfy both requirements.

Modeling disease, not just treating it

ME-ABE's precision also makes it valuable as a research instrument. When scientists want to understand which specific mutations drive a genetic disease, they need tools that can install one mutation at a time without altering surrounding sequence. If the editor introduces bystander changes, the experiment is contaminated - you cannot tell which mutation caused the observed effect.

Evanoff described ME-ABE as a tool that works in both directions: it helps researchers model potential disease-causing mutations in cell lines or animal models, and it helps them test corrections for those same mutations. The ability to do both with a single precise tool accelerates the pipeline from understanding a genetic disease to developing a therapy for it.

Millions of genetic variations exist between any two people. Most are harmless. Identifying which mutations or combinations actually drive disease - and proving it by installing them one at a time in model systems - requires exactly the kind of surgical precision ME-ABE provides.

What ME-ABE does not solve

ME-ABE addresses bystander editing in adenine base editors specifically. It does not tackle cytosine base editors, which have their own bystander issues. It also does not eliminate all off-target editing - changes at unintended genomic sites, as opposed to neighboring bases at the intended site. Those are separate challenges requiring different solutions.

The tool was characterized in cell culture. Moving from cell lines to animal models to clinical application introduces additional variables: delivery efficiency, immune response, tissue-specific editing rates. ME-ABE's narrow window and high efficiency are promising properties, but they do not automatically translate to a finished therapeutic.

Komor's lab is now working on evolving base editors directly in mammalian cells rather than bacteria - an approach that could yield tools even better matched to human cell biology from the start. They hope ME-ABE serves as a foundation for other groups developing precision gene-editing tools.

Plasmids for ME-ABE and other tools from the Komor lab are publicly available through AddGene.