In Vivo Created CAR T Cells Eliminate Tumors in Mice

In a new study, an ingenious CRISPR-based tool was used to create CAR T cells in vivo instead of the usual in vitro approach. It showed higher efficacy across three cancer types, including a solid tumor [1].

CAR T therapies: promising but imperfect

Ideally, T cells, the killer cells of our adaptive immune systems, should be able to eliminate cancerous cells. In most cases, however, T cells either fail to recognize tumor cells as abnormal or are suppressed by the tumor microenvironment [2].

CAR T cell therapy solves this by genetically engineering T cells to express an artificial receptor – specifically, a chimeric antigen receptor (CAR) – that “instructs” them to attack cancer cells expressing a particular target protein (for example, CD19 on leukemia cells). Seven such therapies are now FDA-approved, and they can induce durable remissions in patients with blood cancers who had no other options [3].

The standard process works like this: the patient-derived T cells are taken to a specialized facility, a gene for the CAR is delivered into them using a retroviral or lentiviral vector, then the cells are expanded in culture and infused back into the patient. This process takes 3-5 weeks, costs hundreds of thousands of dollars per treatment, and produces results of variable quality; often, the cells just don’t expand well. Affordability issues aside, patients can simply die waiting.

Who ordered CRISPR delivery?

In a new study published in Nature, a group of researchers from the University of California San Francisco reports forgoing the standard process entirely by creating CAR T cells in vivo. A 2017 study demonstrated a CRISPR-based precision DNA-cutting system, combined with a DNA repair template, that can be used to insert the CAR sequence at a specific address in the T cell genome: the T cell receptor alpha constant (TRAC) locus [4]. Inserting the CAR gene here has several important advantages, such as tighter regulation of expression, which slows T cell exhaustion. However, until now, the system has not been deployed in vivo.

The authors achieved this feat by using two delivery systems. The first part used enveloped delivery vehicles (EDVs), virus-like particles engineered from viral structural proteins. EDVs delivered the CRISPR-based targeting and cutting complex. Another system, based on a different, adeno-associated, virus (AAV), delivered the CAR gene flanked by sequences matching the regions on either side of the CRISPR cut. When the cell’s own DNA repair machinery repairs the CRISPR-induced break, it uses the AAV-delivered template, and voilà: the CAR gene is inserted precisely at the TRAC locus.

After testing their concept in vitro, the researchers engrafted immunodeficient mice with a mix of human immune cells, including T cells, B cells, and monocytes (peripheral blood mononuclear cells, PBMCs). Then the mice received an intravenous injection of the EDV + AAV combination carrying the CD19-CAR.

Two weeks later, spleens were collected. The team found that TRAC-CAR T cells were present in the spleen, and those mice showed depletion of CD19-expressing B cells, demonstrating that the CAR T cells were indeed killing their targets. After several ingenious steps to tweak their design, the team achieved high levels of transfection with no evidence of systemic inflammation.

The phenotypic data suggested that the cells were not just present but functional, proliferating, and maintaining a memory-like profile (i.e., ready to engage the same antigen on rechallenge). The paper claims that this is the first targeted integration of a large DNA payload in primary human T cells in a living animal.

Knocking out real cancers

Next, the team challenged mice with aggressive leukemia. Three days later, human PBMCs were injected, and the experimental therapy was injected a day after that. This was repeated across four independent PBMC donors to assess reproducibility. 18 out of 20 mice achieved complete response (total tumor elimination) across all four donors.

The team then pitted their design against competition: a lentivirus-based design for in vivo CAR T generation, which is currently in Phase I clinical trials [5]. In vitro variants of both designs were tested as well. The TRAC-CAR T in vivo therapy vastly outperformed the rest of the field, with 6 out of 6 mice achieving complete response. In vivo TRAC-CART T expanded many times faster than in vivo lentiviral CAR T cells and showed higher and more uniform CAR expression. The authors suggest this is a direct consequence of random integration (the LLV approach) compared to site-specific integration at a single regulated locus (the TRAC-CAR T approach).

“What was especially remarkable was that the cells we’re generating in vivo actually look better than what we make in the lab,” said Justin Eyquem, PhD, an associate professor of medicine at UCSF and the senior author of the new paper. “We think that when cells are taken out of the body and grown in the lab, they lose some of their ‘stemness’ and proliferative capacity and that doesn’t happen here.”

The team then threw their invention at multiple myeloma: a different cancer type with a different CAR antigen. Here, too, the treatment led to complete responses in all eight mice.

The final test was against sarcoma. Solid tumor treatment has been much harder for CAR T therapy due to poor T cell infiltration, immunosuppressive tumor microenvironment, and antigen heterogeneity. Despite that, with one of the two donors, five out of six mice achieved complete responses. With the second donor, however, only three out of eight did, proving that donor variability remains a challenge.

“If we can translate this to humans, we could dramatically reduce costs, eliminate waiting times, and potentially allow community hospitals – not just major cancer centers – to offer these life-saving therapies,” said Eyquem. “That would truly democratize access to CAR-T cell therapy.”

Literature

[1] Nyberg, W. A., Bernard, P. L., Ngo, W., Wang, C. H., Ark, J., Rothrock, A., … & Eyquem, J. (2026). In vivo site-specific engineering to reprogram T cells. Nature, 1-10.

[2] Chen, D. S., & Mellman, I. (2013). Oncology meets immunology: the cancer-immunity cycle. immunity, 39(1), 1-10.

[3] Maude, S. L., Laetsch, T. W., Buechner, J., Rives, S., Boyer, M., Bittencourt, H., … & Grupp, S. A. (2018). Tisagenlecleucel in children and young adults with B-cell lymphoblastic leukemia. New England Journal of Medicine, 378(5), 439-448.

[4] Eyquem, J., Mansilla-Soto, J., Giavridis, T., Van Der Stegen, S. J., Hamieh, M., Cunanan, K. M., … & Sadelain, M. (2017). Targeting a CAR to the TRAC locus with CRISPR/Cas9 enhances tumour rejection. Nature, 543(7643), 113-117.

[5] Xu, J., Liu, L., Parone, P., Xie, W., Sun, C., Chen, Z., … & Mei, H. (2025). In-vivo B-cell maturation antigen CAR T-cell therapy for relapsed or refractory multiple myeloma. The Lancet, 406(10500), 228-231.