通过RetroT全RNA基因组整合平台进行体内CAR T细胞工程改造以实现持久的抗肿瘤免疫

Adoptive T cell therapies have transformed the treatment of hematologic malignancies, yet broader application remains constrained by ex vivo manufacturing, viral gene delivery, and genome-editing approaches that introduce double-strand breaks (DSBs) and associated genotoxic risk. To address these limitations, here we present an in vivo CAR T cell engineering approach based on CD8-targeted lipid nanoparticles delivering linear CAR mRNA. This strategy enabled direct, transient CAR expression in cytotoxic T cells in vivo, resulting in controlled immune activation without viral vectors, genome modification, or lymphodepleting conditioning. In both mouse models and non-human primates, transient in vivo CAR programming produced potent immune activity and robust B cell depletion with favorable tolerability, demonstrating that RNA-based in vivo CAR delivery can achieve clinically meaningful efficacy while preserving reversibility and safety. While repeat dosing of transiently CARs robustly depletes target cells, oncology applications often require sustained effector function and durable tumor surveillance. To extend the in vivo CAR paradigm to cancer while preserving the safety, scalability, and manufacturing advantages of RNA-based delivery, we developed RetroT, a second-generation, all-RNA, non-viral genome-integration platform. RetroT co-opts the LINE-1 target-primed reverse transcription and combines CRISPR-guided nickase targeting with engineered RNA primer-binding sites to enable precise, DSB-free genomic insertion of therapeutic transgenes. Through systematic RNA engineering that includes optimization of primer-binding site architecture, suppression of dsRNA byproducts, and controlled ORF2p stoichiometry, we achieved an approximately 50-fold improvement in integration efficiency, enabling scalable generation of stably engineered T cells. In primary human T cells, RetroT mediated efficient, site-specific integration of a CD19 CAR cassette, achieving up to 12% CAR-positive cells. RetroT-engineered CAR T cells exhibited potent antigen-dependent cytotoxicity, sustained activation, and robust effector cytokine secretion upon engagement with CD19⁺ leukemia cells. Genome-wide analyses demonstrated precise junction fidelity, stable transgene copy number, and no detectable off-target integrations or partial insertions. In an NSG leukemia xenograft model, a single infusion of RetroT-engineered CAR T cells resulted in significant tumor burden reduction, demonstrating durable in vivo antitumor activity. Collectively, these data define a stepwise in vivo CAR engineering strategy in which transient mRNA-based programming enables rapid, controllable immune activity, while RetroT supports durable, stable CAR expression for oncology indications requiring long-term persistence.