psCas12f 초음파 유전자편집 — 단일 AAV, 치료비를 바꿀 3가지 조건psCas12f Ultrasound Gene Editing: Three Conditions for Lower Treatment Costs

The paper appeared online in Cell on September 28, 2026, followed by a Nature research highlight on October 2. The advance is evidence for controlling where editing occurs in animals. It is not evidence of a human cure or of measured healthcare savings. Source: Nature Research Highlight — 2026-10-02 Source: Addgene — 논문 및 공개 플라스미드

The Lin lab and a first author from KAIST

Lin’s lab has a long history of engineering proteins that switch on and off with light. Its research record includes a photoswitchable Cas9 project in 2017. The new work pushes the same question toward a smaller editor and deeper tissue. Source: Lin Lab — 이전 연구와 Song 이력

Song joined the Lin lab as a visiting postdoctoral scholar from KAIST in 2021. In his own account of the paper, he credits both Lin’s mentorship and the collaboration with Guosong Hong. That collaboration is what connects focused ultrasound and mechanoluminescence to the editing platform. Source: Song·Wu의 연구 설명 — 연구자 직접 공개 Source: Lin Lab — 이전 연구와 Song 이력

Coauthor Xiang Wu has also described the collaboration on ultrasound-triggered light activation. Stanford Bio-X names Marigold Malinao, Sa Cai, Michael Westberg and Hanlee Ji among the participants. That mix tells you the project joined protein engineering to materials and ultrasound work. The public accounts do not establish each author’s experimental assignments. Source: Stanford Bio-X — 연구진과 쥐 실험 Source: Song·Wu의 연구 설명 — 연구자 직접 공개

Why psCas12f needs ultrasound to switch on deep inside the body

Gene editing changes DNA at a selected location. Cas12f is a compact DNA-cutting protein, while a guide RNA directs it to its target. Lin and Song added light-dependent control to create psCas12f. The point is to decide when and where the editor works after it has already been delivered. Source: Song·Wu의 연구 설명 — 연구자 직접 공개 Source: Addgene — 논문 및 공개 플라스미드

Light is the obstacle. It scatters as soon as it meets tissue and does not travel far below the surface. The collaboration with Hong pairs focused ultrasound with particles that emit light under mechanical stimulation. Ultrasound sent from outside the body generates light at the target, and that local light switches on the photoswitchable editor. Ultrasound itself does not rewrite DNA. Source: Song·Wu의 연구 설명 — 연구자 직접 공개

1. 1
   AAV delivery

   Genetic instructions for editor and guide RNA

2. 2
   Focused ultrasound

   Stimulation at a selected region

3. 3
   Mechanoluminescence

   Particles emit light

4. 4
   psCas12f activation

   Editing induced in the target region

The value here is room to separate where a vector goes from where it acts. If editing can be switched on only in a chosen region among the cells a vector reaches, there are more ways to treat one tissue and spare another. But an ultrasound focus still contains a mix of cell types. Selecting a region is not the same thing as selecting only diseased cells.

What single-AAV delivery solves — and what the mouse result leaves open

AAV is a virus-based vehicle used to carry genetic instructions into cells. Its cargo space is fixed, so a bulky editing system complicates the design. What Song describes is fitting the compact photoswitchable editor and its guide RNA into one AAV architecture. Related constructs from the paper are available through Addgene. Source: Song·Wu의 연구 설명 — 연구자 직접 공개 Source: Addgene — 논문 및 공개 플라스미드

“Single” here does not mean one viral particle per patient, nor a guaranteed one-shot treatment. Against designs that split components across vectors, it may ease the requirement that several vectors arrive at the same cell. Dose and production cost do not follow from the word alone.

Lin and Song’s result confirms editing in targeted mouse skeletal muscle and brain regions. A therapeutic effect in humans sits at a different level entirely. Full-text and supplementary access was unavailable for this article, so the reporting was cross-checked against the institution, the authors’ own accounts and the deposited research tools. Editing percentages, animal counts and off-target rates could not be verified and are therefore omitted. Source: Stanford Bio-X — 연구진과 쥐 실험 Source: Addgene — 논문 및 공개 플라스미드

One more distinction matters. A switch that turns the editor off is not an undo button for DNA that has already been changed. Leakage when the switch is off, and effects that persist, deserve as much attention as clean activation in the intended region.

Three conditions for lower treatment costs

An economic comparison needs a defined alternative. Casgevy, approved by the FDA in December 2023, edits a patient’s blood stem cells outside the body and returns them after high-dose chemotherapy. It is a complex pathway, but its target diseases and treatment goals differ from this muscle-and-brain experiment. Reading psCas12f as a Casgevy replacement lines up the wrong comparison. Source: FDA — Casgevy 승인 당시 치료 절차

The first condition is reaching the needed clinical effect in humans with less invasive intervention. Avoiding an implanted light-delivery device would cut operating-room use, recovery and complication management. Long imaging and ultrasound sessions, or repeat visits, shave the savings back down. These are items for future comparison, not costs the paper measured.

The second condition is that simpler manufacturing outweighs the components it adds. One AAV architecture can trim the separate production, mixing and quality control that split-vector systems require. In exchange, this approach needs light-emitting particles and ultrasound equipment. That is why counting vectors cannot establish that costs have halved.

The third condition is a safety advantage proven well enough to lighten the testing and follow-up burden. FDA genome-editing guidance covers design, manufacturing and testing, nonclinical safety and trial design. Even with local control, our reading is that particle persistence and clearance, unwanted editing, and ultrasound effects on tissue all have to be assessed together. In early development that list of checks may get longer, not shorter. Source: FDA — 유전체편집 제품 가이던스 (2024)

The number that matters is not the drug price alone. It is the full cost of reaching a comparable outcome: vector and particle production, the procedure, safety testing and follow-up, and any further treatment. If travel and recovery time fall for patients and caregivers, there are benefits no invoice captures. There is still no basis for quoting a savings percentage.

Repeat activation is not repeat dosing

Song’s design raises an obvious follow-up: can an editor already delivered to the body be switched on again later? Reapplying ultrasound and administering another AAV dose are entirely separate problems. How much editor remains, whether more particles are needed, and whether further editing is clinically warranted each have to be established.

The FDA explains that immune responses and neutralizing antibodies against AAV can interfere with gene delivery. External activation does not remove that delivery barrier, and lower costs for repeat treatment do not follow automatically. Source: FDA — AAV 면역원성 연구

For Korean companies, the opening is in validation, not capacity

The adjacent industrial story is contract development and manufacturing for gene therapies — CDMO production and analytical services. SK pharmteco announced its SKyvec platform, including AAV services, on June 23, 2026. The company identifies US and European locations for it. Nothing there supports reading it as new domestic Korean capacity or as a contract tied to this paper. Source: SK pharmteco — SKyvec 출시 (2026-06-23)

To serve an approach like psCas12f, a manufacturer would have to show that the vectors it produces behave consistently. Assays for two things become valuable: how quiet the system stays without stimulation, and how much it edits once stimulated. Particle makers and ultrasound-device suppliers could be drawn into the same biological validation work. All of this assumes commercialization actually advances.

The nearer-term use is research tools rather than therapeutic revenue. Constructs deposited by the team are available for other laboratories to test. If perturbing a gene in one region gets easier, preparation and replication both get more productive. Open research materials still do not confer clinical-grade quality or commercial rights. Source: Addgene — 논문 및 공개 플라스미드

How much of the existing industry this displaces is not yet knowable. The order of questions is clear: benefit against standard care in a specific disease, regional control in human-scale tissue, and a genuinely simpler treatment pathway. What Lin and Song have supplied is a new tool for asking them. The answer on treatment cost will come from which burdens that tool actually removes for patients.

This article summarizes published research and its industrial context. It is not a recommendation to buy or sell any security.

The value of spatial control must be tested against the treatment burden it removes for patients.

Sources and further reading

• Cell 논문 원문 — Song et al. (2026)
• Stanford Bio-X — 연구진과 쥐 실험
• Nature Research Highlight — 2026-10-02
• Song·Wu의 연구 설명 — 연구자 직접 공개
• Addgene — 논문 및 공개 플라스미드
• Lin Lab — 이전 연구와 Song 이력
• FDA — Casgevy 승인 당시 치료 절차
• FDA — 유전체편집 제품 가이던스 (2024)
• FDA — AAV 면역원성 연구
• SK pharmteco — SKyvec 출시 (2026-06-23)