Abstract
Cell and gene therapies for hematologic malignancies currently rely on mobilization, apheresis, ex vivo genetic manipulation, and reinfusion of engineered cells. This multistep process is associated with significant logistical and clinical challenges, including manufacturing delays, disease progression during production, and toxicities related to conditioning regimens and repeated cell manipulation. In vivo nucleic acid delivery represents a promising alternative and has been explored using viral vectors and lipid nanoparticles (LNPs). Although antibody-conjugated LNPs can enhance cellular specificity, they suffer from manufacturing complexity and instability following freeze-thaw cycles, limiting their translational feasibility despite preclinical efficacy.
To address these limitations, we previously developed a hematopoietic stem and progenitor cell (HSPC)–specific LNP platform without the need for targeting moieties, identified through in vivo high-throughput barcoded screening and tested in rhesus monkeys (Kim et al., Nature Biotechnology, 2024). In the current study, we evaluated whether this antibody-free, HSPC-selective LNP could achieve in vivo mRNA delivery to human HSPCs in the absence of conditioning or ex vivo manipulation, using NBSGW humanized hematopoietic mouse models. Mice were humanized under non-myeloablative conditions with a single dose of busulfan (12.5 mg/kg) followed by transplantation of one million mobilized human CD34+ cells. Individual models were generated from three healthy donors of varying ages. After 16 weeks, mice demonstrated robust engraftment with human chimerism exceeding 85% of bone marrow mononuclear cells (BMMCs) and approximately 5% hCD34+ cells among hCD45+ cells, along with evidence of multilineage maturation, including erythroid and megakaryocytic compartments. Mice were treated with 2 mg/kg of LNP-formulated aVHH mRNA, and gene expression was analyzed 16 hours post-injection. Flow cytometry revealed aVHH protein expression in over 30% of hCD45+ bone marrow cells, with less than 5% expression in mCD45+ cells, indicating species-specific uptake. Among HSPC subsets, aVHH expression was robust, with a median of 36.76 % (range 22.2~64.8%) in HSCs (CD34+CD38–CD90+CD45RA–), and comparable expression in multipotent progenitors (MPPs) and multi-lymphoid progenitors (MLPs). Committed progenitor populations also showed high levels of expression, including 70.45% (51.4~84.6%) in common myeloid progenitors (CMPs), 59.55% (37.5~84.3%) in granulocyte-monocyte progenitors (GMPs), and 52.15% (40.2~80.0%) in megakaryocyte-erythroid progenitors (MEPs). In contrast, more differentiated lineages showed significantly lower expression: 29.04% (14.4–52.4%) in CD11b+ myeloid cells, 11.4% (2.01–26.6%) in NK cells, 5.1% (2.23~9.29%) in B cells, and 3.7% (0.39~7.52%) in T cells. Erythroid cells (CD71+GlyA+, CD71–GlyA+) and megakaryocytes (CD41+) demonstrated <5% aVHH positivity. Gene transfer efficiency was lower in the older donor (age 74), though lineage-specific delivery patterns remained similar to those of younger donors. Immunofluorescence analysis of non-hematopoietic tissues, including liver, spleen, and lung, showed strong co-localization of hCD45 and aVHH with minimal signal in mCD45+ cells, consistent with flow cytometry results. Notably, no significant mRNA delivery was observed in mouse hepatocytes. Single-cell transcriptomic analysis of BMMCs further confirmed that aVHH transcripts were most highly expressed in human HSPCs compared to progenitors and differentiated lineages. Moreover, no significant global transcriptomic differences were observed between aVHH-positive and -negative cells within each population, suggesting that any acute molecular changes induced by LNP delivery are minimal or normalize within 16 hours of administration. Single-cell multi-omic studies are ongoing to investigate the molecular mechanisms underlying HSPC selectivity of the LNP.
In summary, we demonstrate that this novel, antibody-free LNP platform achieves selective in vivo gene delivery to human HSPCs in humanized mouse models without the need for conditioning or mobilization. This LNP system is scalable, stable, and clinically tractable, and holds strong potential for future application in in vivo cell and gene therapies. Further preclinical studies are warranted to evaluate long-term safety, durability of gene expression, and therapeutic payload optimization in preparation for clinical translation.
