Antibody-Targeted LNPs Enable In Vivo CAR-T Cell Generation

In a 2025 study published in Science, Hunter and colleagues reported an antibody-targeted lipid nanoparticle platform that delivered CAR-encoding mRNA preferentially to CD8+ T cells, enabling functional CAR-T cells to be generated directly in vivo. The platform induced rapid B-cell depletion and tumor control in preclinical models. In cynomolgus monkeys, deep B-cell depletion was followed by the recovery of predominantly naïve B cells, suggesting the possibility of an immune-system reset. The study provides important preclinical evidence for targeted mRNA-LNP delivery as a transient, nonviral approach to in vivo CAR-T engineering.

Read the original article in Science: In vivo CAR T cell generation to treat cancer and autoimmune disease

Experimental Objective, Design, and Key Findings

To develop a systemically administered in vivo CAR-T platform, the researchers jointly optimized three components: the LNP formulation and targeting strategy, the CAR mRNA payload, and the dose regimen required to balance biological activity with inflammatory safety.

1) L829 Development and CD8-Targeted Delivery

To reduce the liver-biased distribution of conventional LNPs, the researchers developed a new ionizable lipid, L829. Compared with an ALC-0315-based benchmark formulation, L829 showed lower functional delivery to the liver and faster hepatic clearance. When functionalized with an anti-CD5 antibody, L829 nanoparticles increased reporter expression in the spleen without increasing liver expression.

The researchers subsequently used an anti-CD8 antibody to direct mRNA delivery toward CD8+ T cells. This progression highlights an important principle: effective immune-cell targeting depends not only on the targeting ligand but also on an LNP formulation with an appropriate biodistribution profile.

Fig. 1. Development and biodistribution of the L829-based targeted LNP platform.

2）CAR mRNA Optimization

In parallel, the researchers screened anti-CD19 CAR mRNAs with different untranslated regions and coding-sequence designs. The optimized CAR2 construct produced higher CAR expression in T cells and stronger activity against CD19-positive target cells.

These results show that delivery efficiency alone is not sufficient. The mRNA payload must also support adequate protein expression and biological activity after reaching the target cell.

Fig. 2. Enhancements in anti-CD19 CAR mRNA improve expression and function in T cells.

3）Functional Validation and Inflammatory Safety

The optimized platform was evaluated in human immune-cell samples, humanized mice, and cynomolgus monkeys. In vitro, CD8-targeted L829 tLNPs generated functional CAR-T cells from samples obtained from both healthy donors and patients with autoimmune diseases. These cells showed antigen-specific activity and depleted B cells, including autologous patient-derived B cells.

In humanized mice, systemic administration generated CAR-expressing T cells and induced rapid B-cell depletion. In a Nalm6 leukemia model, treatment also produced marked tumor clearance, indicating that transient CAR expression was sufficient to control a B-cell malignancy in this model.

For evaluation in non-human primates, the researchers delivered an anti-CD20 CAR mRNA using the CD8-targeted L829 platform. Treatment depleted B cells in the blood, spleen, lymph nodes, and bone marrow. B-cell recovery began approximately three weeks later and was dominated by naïve B cells, while memory B-cell populations remained substantially reduced.

The primate study also identified a dose-related safety concern. One animal receiving three infusions at 1.5 mg/kg developed an HLH-like inflammatory syndrome. A subsequent two-infusion regimen achieved comparable B-cell depletion with lower inflammatory responses, including reduced IL-6 elevation. This finding emphasizes that clinical translation will require careful control of dose and schedule to balance biological activity with inflammatory safety.

Fig.3 CAR-Tgeneration, B-cell depletion, tumor control, and B-cell reconstitution across preclinical models.

Conclusion: Implications for In Vivo Therapeutic Development

The broader significance of this study lies in the modular combination of an ionizable lipid, a cell-targeting ligand, and an mRNA payload. This design could potentially be adapted to different immune-cell populations and therapeutic targets. Transient CAR expression may be particularly relevant to autoimmune disease, where temporary B-cell depletion could support immune reconstitution without permanent genetic modification.

However, the recovery of predominantly naïve B cells should be viewed as an early indication rather than proof of an immune reset. Human dosing, inflammatory toxicity, repeat administration, and clinical durability remain unresolved. The study therefore provides a strong preclinical foundation while also showing that delivery efficiency, payload activity, and treatment regimen must be optimized together.

Levostar supports in vivo CAR-T research through CAR mRNA development, targeted LNP formulation, and preclinical evaluation. 

Learn more about the Levostar In Vivo CAR-T Platform.