Assessing Stealth and Sensed Base Editing in Human Hematopoietic Stem/Progenitor Cells

Genome editing represents a promising tool to manipulate human hematopoietic stem and progenitor cells (HSPCs) opening the possibility to correct hematopoietic diseases avoiding the risk of insertional mutagenesis and uncontrolled expression of the transgene, issues that emerged with retroviral and lentiviral gene therapy. Engineered nucleases such as CRISPR/Cas9 have enable targeted genetic manipulation in human HSPCs for therapeutic purposes. Still, nuclease-induced DNA double-strand breaks (DSBs) trigger p53-dependent DNA damage response affecting HSPC properties and may lead to unintended chromosomal rearrangements. Base editing (BE) holds the promise for precise editing by the introduction of specific single-nucleotide variants (SNVs) while bypassing the requirement for DSBs. In particular, base editors are composed by: i) a deamination domain that directly modifies nucleotides comprised within a defined editing window in one of the two genomic strands, and ii) a nickase Cas9 that introduces a single-strand break (SSB) on the other strand to promote more efficient base editing. Depending on the type of modification introduced editors are classified in Cytosine (C-) BE (C-G transition to T-A) and Adenine (A-) BE (A-T transition to G-C). However, a comprehensive characterization of efficiency, tolerability and genotoxicity of CBE and ABE in human HSPCs is lacking and is required to instruct the rationale towards safe and effective clinical translation. Here, we developed an optimized mRNA-based protocol for BE in human HSPCs and compared CBE4max, ABE8.20-m and Cas9 nuclease by targeting the same locus (B2M) using the same sgRNA. Common outcome for all editors is disruption of targeted gene expression, which is measured by flow cytometry and Next Generation Sequencing. ABE8.20-m showed higher efficiency than CBE4max and Cas9 nuclease at the target locus (up to 90, 40 and 50%), which was consistent across HSPC subpopulations comprising the most primitive compartment endowed with long term repopulation potential and cell sources (such as cord blood- and mobilized peripheral blood-derived HSPCs). Importantly, Cas9, but not CBE4max and ABE8.20m, treated HSPCs showed lower in-vitro clonogenic capacity than mock electroporated cells. Transcriptional analyses uncovered that CBE4max, but not ABE8.20-m, triggered p53 pathway activation, albeit at lower extent as compared to Cas9 and presumably consequent to a fraction of single-strand nicks turning into DSB upon DNA replication. Additionally, BE, and particularly CBE4max, upregulated the expression of interferon-stimulated genes, which was not ascribed to mRNA delivery. Remarkably, despite edited HSPCs showed long-term multilineage capacity in xenotransplanted mice, CBE4max edited cells tended to decrease over time in the graft pointing to some detrimental response to the treatment of the long-term engrafting HSC subset. Overall, our results prompt further investigation on BE sensing in human HSPCs. On-going studies are aimed to investigate clonal dynamics and genome integrity of base-edited HSPCs with the final goal of building confidence for their perspective clinical translation.