Lipid nanoparticle-mediated epitope editing in human HSPCs enables resistance to CD45-directed CAR-T cell therapy

Abstract CD45-directed CAR-T cells (CART45) are a promising immunotherapy for a broad range of hematologic malignancies, given the pan-leukocyte expression of CD45. However, their clinical use is limited by cytotoxicity against normal hematopoietic cells expressing CD45. Epitope editing of CD45 in human hematopoietic stem and progenitor cells (HSPCs) offers a strategy to protect the normal hematopoietic system from CART45-mediated pancytopenia. While electroporation (EP) is widely used for CRISPR-based gene editing in HSPCs, it involves physical manipulation of cells and can reduce viability and function, particularly in primitive hematopoietic stem cells (HSCs). Lipid nanoparticles (LNPs) offer a non-viral, RNA-based delivery alternative that operates through endocytic uptake and membrane fusion, providing a gentle, scalable, and GMP-compatible platform. Although LNPs have been successfully used in other cell types, their use for ex vivo gene editing of human HSPCs remains limited. This study directly compares the editing efficiency of a Cas9-derived adenine base editor (ABE) targeting the CD45 epitope in human HSPCs delivered via LNP versus EP. To compile a preclinical data package for potential future clinical translation, we evaluated how each delivery method affects HSPC viability and preservation of stem cell properties such as multi-lineage differentiation and engraftment in immunodeficient mice. Human CD34⁺ HSPCs were edited ex vivo with an ABE and a guide RNA targeting the relevant CD45 epitope to install a non-synonymous mutation, delivered via either EP or LNP. Editing efficiency, colony-forming units, and immunophenotypic HSCs showed comparable results between both methods ex vivo. While electroporation achieved slightly higher editing efficiency, LNP delivery preserved cell viability, reaching a 1.5-fold increase in cell recovery relative to EP at 48 hours post-editing. Functional human HSC frequency, as determined by limiting dilution analysis (LDA) in immunodeficient mice (n=45) with equal serially diluted cell doses, was comparable between the two delivery methods. The estimated frequency of functional HSCs among the cells injected was 1 in 81,089 for EP- and 1 in 80,915 for LNP-mediated delivery. These findings indicate that the delivery of the base editing machinery via LNP supports long-term engraftment of HSCs at a similar level to EP. Both groups of cells gave rise to multilineage hematopoiesis in peripheral blood and long-term engraftment in bone marrow, assessed until 28 weeks in primary recipients. Both delivery methods achieved efficient and stable epitope editing, with editing efficiencies of 95% for EP and 81% for LNP before transplantation, which was maintained at 87% for EP and 78% for LNP post-transplantation. Furthermore, donor-matched CART45 cells selectively eliminated unedited cells while preserving edited hematopoietic populations using either EP or LNP delivery, both in ex vivo and in vivo settings. The total number of cells in the bone marrow derived from edited HSPCs was approximately 300-fold higher than that of unedited cells. This demonstrates that epitope-edited hematopoietic cells are effectively shielded from CAR-T cell-mediated cell killing. These findings demonstrate that our developed process using LNP-mediated base editing of CD45 enables highly efficient gene editing in HSPCs, improving cell recovery, preserving long-term HSC function, and simplifying the manufacturing process compared to EP. In addition, CD45-edited HSPCs via LNP are protected from CAR-T cell-mediated cytotoxicity. This pre-clinical LNP delivery process offers a clinically scalable alternative to electroporation for genetic engineering of HSPCs with potential applicability across a broad spectrum of hematologic malignancies and diseases.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.