Gates Package — page 821

of -globin expressing red blood cells (RBCs). Reactivation of -globin protein expression was also shown by HPLC where -globin chains reached levels of 30% relative to - and -globin chains (Fig.2D). Efficient -globin reactivation resulted in a significant reduction of reactive oxygen species levels (ROS), a hall-mark of oxidative stress seen in -thalassemia and SCD (36) (Figs.3A, S3B). -globin re-activation by HDAd-EF1.ABE8e also improved erythroid differentiation/expansion of Thal and SCD CD34+ cells. To total number of erythroid cells increased 2- and 7-fold, for Thal and SCD samples between days 7 and 18 in settings with in vitro selection (Fig.3B). Improved erythropoiesis is also supported by microscopic analyses of cells at the end of ED which shows more differentiated cells (maturing orthochromatic erythroblasts, reticulocytes/pyrenocytes in the Thal (Fig.4A and SCD (Fig.4B) samples. HDAd-EF1.ABE8e transduction and base editing did not cause cytotoxicity as demonstrated by i) no reduction in the percentage of primitive HSCs (CD34+/CD38-/CD90+) (Fig.S4A) and ii) no significant effect on progenitor colony formation (Fig.S4B). In summary, transduction of CD34+ cells from -thalassemia and SCD patients with HDAd-EF1.ABE8e mediated efficient editing of the HBG1/2 target site and reactivation of -globin expression resulting in phenotypic amelioration of erythroid cells after in vitro ED of HSCs. Ex vivo HSC genome editing without O6BG/BCNU selection. We used transgenic mice carrying the wildtype 248-kb -globin locus (-YAC) including the LCR, - and -globin genes (22). They were crossed with human CD46-transgenic mice to allow for HDAd5/35++ transduction via CD46. First, we tested an ex vivo HDAd-EF1.ABE8e editing approach. Bone marrow lineage-negative (Lin−) cells (a fraction enriched for HSCs) from -YAC/CD46 mice were transduced with HDAd-EF1.ABE8e and transplanted into lethally irradiated C57BL/6 mice (Fig.5A). Engraftment, measured based on human CD46 expression on PBMCs, was >95% after week 8, indicating that ex vivo HSC transduction and editing did not affect HSC functions (Fig.5B). After transplantation, A>G conversion rates measured in PBMCs by Sanger sequencing were >90% at the first time point analyzed (4 weeks post-transplantation) and remained stable over the 16 week ("week 16-P", see Fig.5A) observation period (Fig.5C). As seen in in vitro studies with human CD34+ cells (e.g. Fig.1D), substantial bystander A>G editing was observed, specifically at position A5. A>G conversion rates in splenocytes, bone marrow mononuclear cells (MNCs), and Lin− cells as well as colonyforming units were in agreement with the PBMC data (Fig.5D). The comparison of editing rates in the transplanted cells (~38% -113 A>G) and in week-16P mice (>95% -113 A>G) likely indicates preferential transduction of repopulating HSCs and expansion of edited cells after transplantation (Fig.5E). The vast majority of edits were A>G conversions (Fig.5E). The rate of A>T and A>C conversions was ~3 and ~4 orders