Cell and Gene Therapy · global
Forming CAR-T Cells Inside the Body: Dual-Vector Gene Editing Shows Anticancer Effects in Mice
Removing immune cells from the body for modification is a time-consuming and expensive part of CAR-T therapy. A UCSF team has tried delivering anticancer instructions directly inside the body, achieving results in humanized mice. Published in March 2026, the study still faces safety and manufacturing hurdles before human use.
For cancer patients who need CAR-T treatment, waiting is another hurdle: immune cells must first be collected, sent to specialized facilities for modification and expansion, and then infused back into the body. A research team at the University of California, San Francisco (UCSF) is exploring another route—delivering the modification tools directly into the body so that T cells acquire the ability to recognize cancer cells in place, potentially shortening this complex process.
These findings are not a new breakthrough that emerged only in October. The original study was published in *Nature* on March 18, 2026, and the University of California released an overview the following day. ScienceDaily covered the findings again in October. Led by Justin Eyquem, with William Nyberg and Pierre-Louis Bernard as co-first authors, the team has so far demonstrated experimental results in humanized mice, not treatment efficacy in patients.
At the heart of the study is a dual-vector system with complementary roles. An enveloped delivery vector transports CRISPR–Cas9 protein and guide RNA to cut DNA at a designated site; a separate, modified adeno-associated virus (AAV) delivers a DNA template encoding a chimeric antigen receptor (CAR). A CAR acts like a recognition device on the surface of a T cell, triggering an attack when it encounters the corresponding cancer cell marker.
The team chose to insert the CAR gene into the TRAC locus of the T-cell receptor alpha chain and have it use the existing promoter at that site, rather than carry its own promoter. This design combines control over which type of cell receives the payload with control over where expression is activated, aiming to achieve stable, T-cell-specific CAR expression. The vectors were also modified to improve delivery efficiency to T cells and resistance to preexisting neutralizing antibodies; these designs do not mean that all risks of unintended editing have been eliminated.
According to the University of California report, after a single injection of the two vectors, nearly all treated mice with aggressive leukemia cleared detectable cancer within two weeks. The paper also reported that all 8 treated mice in the multiple myeloma model achieved a complete response. In the sarcoma model, 5 of 6 mice and 3 of 8 mice achieved complete responses, depending on the donor of the human immune cells. These results support the platform's use with different CARs, while also showing that responses in solid tumors still vary. They do not allow a direct inference that all types of cancer could benefit.
Background Context
CAR-T research faces challenges in both recognition and manufacturing. Using AI to design binding proteins is one approach to improving how cells recognize tumors and launch an attack; this study focuses instead on delivering genetic instructions into cells and controlling their expression. The two approaches may complement each other, but the results each has achieved in mice cannot yet substitute for validation in humans.
Moving manufacturing inside the body would also change how quality control works: researchers would be unable to inspect cells batch by batch before reinfusion as they can with manufacturing outside the body. Human dosing, editing accuracy, long-term safety, and anticancer efficacy all require further study; the paper also notes that anti-AAV antibodies produced after injection may limit repeat dosing. The team has established Azalea Therapeutics to advance clinical development, but whether this approach can actually shorten waiting times and reduce costs still depends on the results of human trials and scalable manufacturing.