Gates Package — page 819

binding to their respective promoter cis-regulatory elements (19). An approach based on RNA interference to silence bcl11a expression (19) or reactivation of -globin expression by CRISPR-Cas9 editing of the erythroid bcl11a enhancer (20) have delivered promising clinical data. Furthermore, preclinical studies to reactivate -globin by CRISPR/Cas9 editing of the BCL11A and ZBTB7A binding motifs within the HBG1/2 promoters have shown efficient HbF reactivation (21). Along this line, we used an in vivo HSC transduction approach to express a CRISPR/Cas9 specific to the BCL11A binding site within the HBG1/2 promoters and showed efficient −globin reactivation in transgenic mice carrying the wild-type 248-kb b-globin locus yeast artificial chromosome (-YAC mice) (22). However, we also detected an undesired deletion of the HBG1 gene due to CRISPR/Cas9 mediated double strand DNA breaks (DSBs) in both the HBG1 and the HBG2 promoter. There is also accumulating evidence that DSBs catalyzed by CRISPR/Cas9 can result in large genomic deletions, chromosomal rearrangements, as well as p53induced cell cycle arrest and apoptosis (23-25). To address these problems, we focused our work on base editors, enzymes that are capable of introducing precise cytidine or adenine substitutions with minimal occurrence of DSBs and indels at the target site (26-29). Here we employed an advanced adenine base editor version (ABE8e). ABE8e contains additional mutations that increase on-target activity and greatly decrease off-target activity (30). We targeted ABE8e to the BCL11A binding site in the HBG1/2 promoters (Fig.1A) to mediate a A>G conversion at the -113 position. The corresponding sgRNA was selected in an earlier study from a series of sgRNAs targeting the BCL11A binding motif as the most efficient in -globin reactivation (12). It is thought that the -113 A>G mutation does not disrupt BCL11A binding, but rather creates a de novo binding site for the transcriptional activator GATA1 ((T/A)GATA(A/G)) that outcompetes binding of the BCL11A (31). However, as a result of the relatively wide editing window of ABE8e, in addition to the -113A>G conversion, bystander editing at neighboring adenines is expected. (Fig.1A, right panel). Specifically, the - 116A>G conversion would destroy the BCL11A binding motif. Here, we evaluated a new HDAd-EF1.ABE8e vector in vitro in CD34+ cells from -thalassemia and SCD patients as well as in ex vivo and in vivo HSC gene therapy setting in the -YAC mouse model. In this context, we tested a new regimen of base editing that does not require the integration of the mgmtP140K cassette for in vivo selection. Results