Epigenome-targeted CRISPR screen reveals stage-specific dependencies in human erythropoiesis

Erythropoiesis is a tightly regulated differentiation process that proceeds through discrete cellular states and culminates in terminal maturation and enucleation. While lineage-defining transcription factors are essential to erythroid gene regulation, including the master erythroid transcription factor GATA1, the mechanism of how epigenetic regulators shape cell state–specific transcriptional programs during erythroid maturation remains poorly understood. In this thesis, we systematically interrogated the epigenetic regulatory landscape of human erythropoiesis through CRISPR/Cas9-based functional screens. This work revealed two principal chromatin regulatory pathways that play central roles in erythroid differentiation: the histone acetyltransferase EP300 and the SMARCB1-containing cBAF (SWI/SNF) chromatin remodeling complex. Using synchronized human erythroid differentiation models, including BEL-A cells and primary CD34⁺ progenitors, we demonstrate that EP300 and SMARCB1 regulate erythropoiesis through distinct, cell state–dependent mechanisms. We show that EP300 is required at early stages of erythroid differentiation. Loss of EP300 disrupts early erythroid cell state transitions, leading to a broad failure in erythroid gene induction and impaired commitment to the erythroid lineage. In contrast, SMARCB1 is essential during terminal maturation. In primary CD34⁺ hematopoietic stem/progenitor cells, EP300 loss blocked progression at the basophilic–polychromatic transition, while SMARCB1 deficiency significantly reduced erythrocyte output and enucleation efficiency. Furthermore, using single-cell and bulk RNA sequencing, we further validated that EP300 governs differentiation-associated transcriptional programs. To understand the role of SMARCB1 in terminal erythropoiesis, we assessed chromatin accessibility across the genome and found that SMARCB1 loss does not alter global chromatin accessibility. We also found that GATA1 protein expression does not change with SMARCB1 perturbation. These findings indicated that SMARCB1 does not regulate erythroid maturation through changes in chromatin accessibility or transcription factor expression. To investigate the protein–DNA interactions underlying SMARCB1-dependent erythroid transcriptional programs, we performed Docking & Deamination-Seq (D&D-Seq), a deaminase-based footprinting approach that enables high-resolution mapping of transcription factor occupancy on chromatin. The results indicated that SMARCB1 knockout alters GATA1 occupancy at a subset of erythroid regulatory loci, including reduced binding at the SNCA locus and increased binding at TFR2. As SNCA is robustly induced during normal late-stage erythroid differentiation, these results demonstrate that SMARCB1 is required for proper GATA1-mediated activation of terminal maturation gene programs. Together, these findings establish EP300 as an early differentiation dependency and SMARCB1/cBAF as a late enucleation dependency, uncovering distinct chromatin-based mechanisms that coordinate sequential stages of human erythropoiesis.