When sound becomes light: a new way to control gene editing
Light deep in the brain can be switched on from the outside — with the help of ultrasound and special nanoparticles. Researchers at Stanford University used this approach to trigger DNA editing in selected regions of the body in mice, including the brain. Activating the system required neither surgery nor implanted light sources.
The work was published on 28 September 2026 in the journal Cell. It was carried out by the teams of Michael Lin and Gusun Hong; the first author was postdoc Daesung Son.
In genetic engineering, precision begins with choosing a section of DNA and continues with choosing the site of intervention. The same gene can perform different tasks in different tissues. That is why it matters to researchers to control three parameters at once: which gene to modify, where to start the process, and when. The new development makes it possible to activate the editor remotely in a chosen area.
The method is based on a light-sensitive Cas12f editor from the CRISPR family and mechanoluminescent nanoparticles capable of emitting light under the action of mechanical vibrations. Ultrasound links these two components: it reaches the desired area, makes the particles glow, and the light switches on the molecular tool.
To create a controllable editor, the scientists used compact proteins of the Cas12f family. One of them, CasMINI, contains 529 amino acids — more than twice as few as the widely used SpCas9. A light-sensitive switch based on the Dronpa protein was added to the editor. Thanks to this, light became the signal that permits work on DNA.
This scheme helps overcome a limitation of light itself: tissues scatter and absorb it, making it difficult to reach deep structures. Ultrasound penetrates deeper and makes it possible to light up a source directly in the chosen area. The nanoparticles work like tiny internal flashlights.
Before this, the genetic instructions for the editor must be delivered into cells. The authors used a vector based on an adeno-associated virus — AAV. The compactness of the system made it possible to fit into a single viral particle the instructions both for the editor and for the guide RNA, which helps it recognize the required section of DNA. Delivery of the tool and control of its operation become two successive stages.
Spatial selectivity was tested using a fluorescent label. In the mice, a special section of DNA blocked the formation of a glowing protein. Successful editing removed this block, and the cells began producing a protein visible under the appropriate illumination. After ultrasound exposure to one side of the brain, a fluorescent signal appeared on the treated side, while the opposite side remained dark.
In this experiment, light performed two different tasks. The glow of the nanoparticles triggered the editor, while the fluorescent label showed where the DNA change had occurred. The authors also demonstrated the approach working in skeletal muscle and liver.
The technology's most immediate value is the ability to study gene functions more precisely. By triggering editing in a specific area of the brain, scientists will be able to study the contribution of local genetic changes to the workings of the nervous system. For medical use, the safety of the nanoparticles, of the editor delivery and of the intervention itself will need to be assessed separately, and therapeutic efficacy will also need to be tested.
At the same time, spatial selectivity and the precision of DNA modification are different characteristics. The ability to limit the area of activation does not in itself prove the absence of unwanted edits in other parts of the genome. Such effects require independent verification.
According to Hong, combining the two technologies produced a result exceeding the simple sum of their capabilities. Lin describes the prospect as "four-dimensional control": three spatial dimensions set the site of exposure, and time sets the moment of triggering. The choice of genetic target complements this control system.
Temporal control has an important boundary: DNA changes may persist after the exposure ends. The ability to switch the editor on at the right moment should be distinguished from the ability to undo the result of its work. That is why the team is also developing an approach to temporarily enhance or suppress gene activity while preserving the DNA sequence.
Thus researchers gain a more subtle way to ask questions of a living system: to modify a chosen gene in a specific place and observe the consequences. For future medicine, this is the prospect of local intervention, a path that runs through further studies of safety and efficacy.
