Crispr-Cas9 Saturating Mutagenesis Reveals an Achilles Heel in the BCL11A Erythroid Enhancer for Fetal Hemoglobin Induction (by Genome Editing)

Common genetic variation associated with fetal hemoglobin (HbF) level and β-hemoglobin disorder clinical severity marks an erythroid enhancer within the BCL11A gene. The 12 kb intronic enhancer contains three ~1 kb erythroid DNase I hypersensitive sites (DHSs), termed +55, +58, and +62. Here we utilized a human adult-stage erythroid cell line to show by CRISPR-Cas9 mediated targeted deletion that the composite enhancer is required both for BCL11A expression and HbF repression. Because deletion of the entire enhancer is currently too inefficient to consider for a gene editing approach to hemoglobin disorders, we sought to define the critical features of the enhancer in its natural genomic context. We designed and synthesized a tiling pooled guide RNA (gRNA) library to conduct saturating mutagenesis of the enhancer sequences in situ using the CRISPR-Cas9 gene editing platform. The gRNAs direct Cas9 cleavage and non-homologous end-joining repair at discrete sites throughout the enhancer. By comparing the representation of lentiviral gRNA integrants in high and low HbF pools of the adult erythroid cells, we generated a functional map approaching nucleotide resolution of sequences within the enhancer influencing BCL11A regulation. We observed several discrete enhancer regions required for maximal expression. The largest effect was observed by producing mutations within a narrow functional core of the +58 DHS. These sequences include a GATA1 motif conserved among vertebrates located within a primate-specific context. This region constitutes an Achilles Heel for functional inactivation of the enhancer. We also identified rare genetic variants within the +58 DHS core in individuals with sickle cell disease that are associated with HbF level, independent of all known associations of common genetic variants. In parallel, we performed a similar saturating CRISPR mutagenesis screen of the corresponding murine Bcl11a enhancer. To our surprise, despite low-resolution evidence of conservation by primary sequence homology, syntenic genomic position, and shared chromatin signature, the mouse enhancer sequence determinants of BCL11A expression showed substantial functional divergence. The +58 orthologous sequences were dispensable whereas the +62 orthologous sequences were critically required in murine adult erythroid cells. These results were validated by producing targeted deletions in mouse and human adult erythroid cell lines. Furthermore we subjected cells to individual gRNAs to correlate individual nucleotide disruptions with loss of BCL11A expression. To substantiate the tissue-restricted effect of the enhancer mutations, we generated transgenic mice with deletion of the Bcl11a enhancer and found these sequences were dispensable for expression in developing neurons and B-lymphocytes (unlike conventional Bcl11a knockout) but essential for appropriate hemoglobin switching in vivo. We showed that in primary CD34+ hematopoietic stem and progenitor derived human erythroid precursors that delivery of an individual gRNA and Cas9 is sufficient to produce robust reinduction of HbF. These results validate the BCL11A erythroid enhancer as a promising therapeutic target. Our findings define the most favorable regions for generation of indel mutations in the BCL11A erythroid enhancer as a therapeutic genome editing strategy for HbF reinduction for the β-hemoglobin disorders.