Efficient and Specific In Vivo Genetic Engineering of Human Hematopoietic Stem Progenitor Cells without Selective Conditioning

Achieving in vivo genetic engineering of hematopoietic stem progenitor cells (HSPC) has the potential to transform treatment for hematological disorders. An ideal in vivo delivery platform should provide access to resting HSPC at low doses, in different in vivo compartments and with high specificity. We developed a multi-model strategy to assess efficiency and specificity of access to human HSPC, without reliance on conditioning to selectively enrich for gene-modified cells, and we achieved efficient genetic engineering of Lin-CD34+ cells and individual HSPC subtypes using highly potent lentiviral vectors (LV) and gene editing virus-like particles (VLP). We first conducted a high-resolution characterization of access to resting HSPC and individual FACS-sorted HSPC subtypes in vitro with different LV envelopes, including a detailed analysis of insertion site distribution in resting vs cytokine activated CD34+ cells. With a BaEVTR LV we achieved transduction of ~100% resting HSPC at MOI 66 (vs ~40% with a VSV-G LV at MOI 50,750) and ~20% of FACS-sorted hematopoietic stem cells (HSC) at MOI 32. This translated into 2-4% transduction of bone marrow (BM) huCD45+ cells when the same BaEVTR LV was dosed in NBSGW long-term humanized mice. However, such models carry a highly lymphoid-skewed human hematopoiesis and are not well suited for measuring access to all HSPC subtypes in all compartments. We therefore established 2 additional in vivo experimental scenarios. To measure LV access to HSPC in the peripheral blood (PB), we followed long-term NBSGW mice where BaEVTR LV was dosed intravenously (IV) immediately after HSPC infusion. In this model we achieved targeting of 23% early engrafting BM HSPC which was mirrored by a ~20% transduction in PB myeloid cells during the first 6 weeks, later stabilizing at 5-7% from week 8 up to week 16. To then measure access to HSPC in the BM niche, we characterized early human HSPC dynamics in the BM of NBSGW mice along 5 timepoints over the first 19 days after CD34+ cell infusion, with or without AMD3100/G-CSF mobilization. We then dosed BaEVTR LV at D7 post-humanization, a timepoint when we established that engrafted human cells have a more physiological lineage distribution, are few and remained confined in the BM even upon mobilization. In this model we demonstrated that a single BaEVTR LV dose injected IV can reach 5-6.4% of BM-resident HSPC. Based on these data, we developed a BaEVTR VLP carrying a CRISPR/Cas9 editor targeting the B2M locus. This VLP displayed 93% editing efficiency in vitro in resting HSPC and 38 to 61% in FACS-sorted HSC at MOI 16 and MOI 83 respectively. When dosed in vivo in the D7 post-humanization model, our VLP achieved B2M KO in 16% and 31% BM-resident Lin-CD34+ cells with 1 and 3 consecutive IV injections respectively, including 14% of phenotypic HSC. Importantly, a comparable level of in vivo editing of BM Lin-CD34+ cells (12%) was achieved upon equivalent pilot single VLP dosing in long-term humanized mice despite them carrying substantially higher numbers of human HSPC in the BM and a complete and stable PB human blood cell engraftment. Secondary transplantation experiments to confirm editing of long-term HSC are ongoing. We proved delivery of genetic payloads to resting HSPC in vitro and access to HSPC in different compartments in vivo with high efficiency. Adding a specific tropism for HSPC has the potential to enhance on-target bioavailability impacting effectiveness and safety of our delivery systems. To this aim, we are now testing LV and VLP carrying HSPC-targeted envelopes engineered with viral fusion proteins (retargetable fusogens). We will present comprehensive data showing that a LV pseudotyped with a CD133-targeted fusogen combines in vitro and in vivo potency on par with BaEVTR LV with high specificity for CD133+ cells, as seen upon titrations on 8 cell lines and 10 primary human cell types as well as in spike-in experiments aimed at measuring CD133+ HSPC transduction when diluted in a mixture of CD133- human blood cells in vitro and in NBSGW mice. Furthermore, we will show that a CD117-targeted LV can transduce, in vivo, a cell population exclusively located in the BM and comprising as little as 0.3% of total huCD45+ cells with a 175-fold increased specificity as compared to envelopes with broad specificity. These results set the basis for generating a potent candidate for in vivo delivery of genetic payloads to HSPC with high efficiency and specificity.