CD33 has long been pursued as immunotherapeutic target in acute myeloid leukemia (AML)[1, 2]. Improved survival with gemtuzumab ozogamicin (GO) validates this approach [3]. Partly stimulated by GO’s success, several investigational CD33-directed therapeutics are currently in clinical testing [4]. However, CD33 expression on normal hematopoietic cells leads to “on-target, off-leukemia” toxicity with significant morbidity/mortality from profound cytopenias, limiting the use of CD33-directed immunotherapies [4]. This toxicity should be minimal if normal blood cells did not express the epitope targeted by these antibodies. Supporting the feasibility of CD33-engineering the hematopoietic system are the findings that CD33-deficient mice have a very mild phenotype and show no difference in cellular response to pro-inflammatory stimuli compared to wild-type animals, indicating functional degeneracy between CD33 and other proteins [5]. Moreover, recent studies have shown that CRISPR/Cas9-mediated disruption of the CD33 coding region in CD34+ hematopoietic stem and progenitor cells (HSPCs) may not affect engraftment [6], suggesting that the generation of CD33-manipulated hematopoiesis is a clinically viable strategy to protect from

“off-leukemia” cell toxicity of CD33-directed immunotherapy. Here we have investigated an alternative, precise CD33 genome-editing approach that would only eliminate exon 2 and therefore the V-set immunoglobulin-like domain, which is the target of all current clinical CD33-directed approaches. Our editing strategy is expected to result in expression of a naturally occurring shorter isoform of CD33 (CD33 ΔE2) but not full-length CD33 (CD33FL), which may minimize potential adverse effects associated with disruption of the entire CD33 locus. We used CRISPR/Cas9 [7–10] to accomplish this goal and functionally assessed genome-edited human hematopoietic cells in vitro and in immunodeficient mice. Human myeloid ML-1 cells and human fetal liver CD34+ HSPCs were used for our studies. ML-1 cells were maintained as described [11]. Human fetal liver CD34+ cells were enriched by immunomagnetic separation from tissue obtained from Advance Bioscience Resources Inc.(ABR, Alameda, CA). Cells were cultured in StemSpan SFEMII media (StemCell Technologies, Cambridge, WA) supplemented with penicillin/streptomycin (Life Technologies, Carlsbad, CA), Stem cell factor, Thrombopoietin (both PeproTech, Rocky Hill, NJ), and FLT3-L (Miltenyi Biotec, Auburn, CA). CRISPR/Cas9-editing was carried out by electroporation of purified Cas9 protein (TrueCut Cas9 V2; ThermoFisher Scientific, Waltham, MA) complexed with synthetic guide RNAs (sgRNAs; Supplementary Table 1), which were modified at the 5′ and 3′ ends with 2′-O-methyl-3′-phosphorothiate (Synthego, Redwood City, CA) using the ECM 380 Square Wave Electroporation system (Harvard Apparatus, Cambridge, MA)[12]. For evaluation of colony-forming units (CFUs), 1500 CD34+ cells were seeded in 3.5 mL ColonyGEL 1402 (ReachBio, Seattle, WA) and scored after 12–14 days. CFU DNA was extracted in QuickExtract (Epicentre, Madison, WI). We quantified drug-induced cytotoxicity as described previously [11, 13]. Briefly, parental and CRISPR-engineered ML-1 cells were incubated in 96-well round