Increased Potency and Uniformity of Fetal Hemoglobin Induction from Base Editing Compared to Cas9 Nuclease

Genetic manipulations to increase fetal hemoglobin (HbF, α2γ2) in red blood cells (RBCs) can alleviate β-thalassemia and sickle cell disease (SCD). We and others demonstrated that CRIPSR/Cas9 disruption of γ-globin promoter binding motifs for the repressors BCL11A or ZBTB7A raises erythroid HbF expression to potentially therapeutic levels. Moreover, disruption of an erythroid BCL11A gene enhancer to induce HbF produced clinical improvement in humans with SCD. However, Cas9-induced double-stranded breaks (DSBs) can cause potentially deleterious chromosomal abnormalities. Base editing represents a promising approach to install precise nucleotide alterations through DSB-independent mechanisms. In contrast to Cas9 generated indels, base editors can recreate naturally occurring base pair substitutions that cause hereditary persistence of fetal hemoglobin (HPFH), including those that create de novo binding motifs for erythroid transcriptional activators at the γ-globin promoter. Base editing and Cas9 nuclease-based approaches to induce HbF have not been compared directly and the best therapeutic strategy is not known. We compared two Cas9-based strategies and three base editing strategies for HbF induction by modifying CD34+ hematopoietic stem and progenitor cells (HSPCs) and examining HbF expression in erythroid progeny generated by in vitro differentiation. We used Cas9 nuclease to disrupt the BCL11A gene erythroid enhancer or the BCL11A binding motif in the γ-globin promoter. Alternatively, we used the adenine base editor ABE7.10 to create base pair substitutions at γ-globin promoter positions -198, -175, and -113, to generate de-novo binding motifs for transcriptional activators KLF1, TAL1, and GATA1, respectively. The -175 A>G edit was most potent, producing an approximately 1% rise in HbF per % editing. In contrast, Cas9 disruption of the BCL11A enhancer or the BCL11A binding motif in the γ-globin promoter caused 0.33% and 0.30% increases in HbF per % editing, respectively. Cas9 nuclease but not base editing induced a TP53-mediated DNA damage response, as reflected by increased expression of CDKN1(P21) mRNA. To compare further the different edits, we analyzed burst forming unit-erythroid (BFU-E) colonies derived from modified CD34+ cells. Erythroid colonies with -175 A>G edit in all four γ-globin genes expressed 81±7% HbF compared to 14±11% in unedited controls. In contrast, HbF in BFU-E colonies was 33±19% after Cas9 disruption of the BCL11A repressor binding motif in the promoters of all γ-globin alleles and 50±14% after Cas9 disruption of the erythroid enhancer in the BCL11A gene. Among Cas9 mutations disrupting the γ-globin BCL11A binding motif, significantly increased HbF occurred with a 13 bp deletion but not with -1, -2, -3 or +1 indels. This finding is unexpected because elimination of BCL11A binding is believed to be sufficient for γ-globin induction. All high frequency indels targeting the BCL11A enhancer induced HbF similarly, albeit with greater variability than in colonies with the -175 A>G base edit. In normal donor CD34+ cells, ABE7.10 installed γ-globin -175 A>G at efficiencies of 43±6% (n=6). The newer, more active ABE8e produced a 1.4-fold improvement in editing efficiency and a 1.3-fold increase in HbF induction relative to ABE7.10. In preliminary studies, we achieved on-target editing of 62% with ABE8e in SCD patient CD34+ HSPCs. At 16 weeks after xenotransplantation of the modified HSPCs into NBSGW mice, on-target editing of donor-derived cells was 53±11%. Base edited erythroid cells isolated from recipient mouse bone marrow expressed 53±5% HbF, versus 2±1% in unedited cells. After xenotransplantation of the same donor CD34+ cells that were treated with Cas9 nuclease to disrupt the γ-globin promoter BCL11A binding motif, editing retention was 92±4%, with 29±12% HbF in donor derived erythroblasts. Thus, the -175 A>G edit produces superior HbF induction. Our data identifies a base editing strategy for potent, uniform induction of HbF and provides insights into γ-globin gene regulation by showing that disruption of the promoter BCL11A binding motif is not sufficient for transcriptional activation. More generally, we demonstrate that diverse indels generated by Cas9 nuclease cause unexpected variation in biological outcomes that can be circumvented by base editing, with important implications for therapeutic gene editing efforts.