Mechanisms of Polycomb Group Protein-Mediated Fetal Hemoglobin Silencing

Abstract

Reactivation of fetal hemoglobin (HbF) is beneficial for patients with β-globin gene disorders, such as sickle cell disease and β-thalassemia. Polycomb repressive complex 1 (PRC1) and 2 (PRC2) are multi-subunit protein complexes that function as transcriptional repressors to maintain lineage-specific gene expression programs. Prior works showed that genetic or pharmacologic manipulation of select PRC2 subunits can induce HbF expression in adult erythroid cells, and an inhibitor targeting the EED subunit of PRC2 has entered a clinical trial. Yet the molecular basis underlying HbF regulation by PRC1/2 complexes is unknown. Using a CRISPR-Cas9-based genetic screen, we identified BMI1, a subunit of the PRC1 complex as novel HbF repressor. Disrupting BMI1 in the adult-type erythroid cell line HUDEP2 or primary human erythroblast cells robustly increased HbF production. We failed to detect BMI1 binding at the β-globin gene locus by CUT&RUN, indicating that BMI1 functions via an indirect mechanism. RNA-seq analysis of BMI1-deficient HUDEP2 cells uncovered upregulation of two known HbF regulators: IGF2BP1 and LIN28B, both of which induce HbF when forcibly expressed in adult erythroid cells. Depletion of either IGF2BP1 or LIN28B partially restored HbF silencing in BMI1-deficient HUDEP2 cells, whereas simultaneous depletion of IGF2BP1 and LIN28B lowered HbF production virtually to basal levels in these cells. In primary erythroid cells, IGF2BP3, which is functionally related to IGF2BP1, also contributes to the effects of BMI1 depletion. Together, these data indicate that IGF2BP1, IGF2BP3, and LIN28B are the critical BMI1 targets in human erythroid cells. Since BMI1 functions in two distinct PRC1 complexes (cPRC1 and ncPRC1), we aimed to define which of these is critical for HbF regulation. cPRC1 is defined by the presence of CBX proteins 2, 4, 6, 7, and 8, while ncPRC1 contains RYBP or its homolog YAF2. ncPRC1 and cPRC1 deposit the histone mark H2AK119ub1, and facilitate chromatin compaction, respectively. Disruption of CBX2/4/7/8 proteins but not RYAP/YAF2 in HUDEP2 cells activated HbF to levels similar to those observed upon BMI1 depletion, implicating cPRC1 in HbF control. Accordingly, CUT&RUN analysis in adult erythroid cells revealed a strong genome-wide co-localization of BMI1 and CBX2/4/7/8, including at the IGF2BP1, IGF2BP3, and LIN28B genes. Finally, BMI1 depletion does not lead to significant changes in H2AK119ub1 at these genes, in line with the fact that the repressive activity of the cPRC1 complex does not rely on its H2A ubiquitination activity. Together these results indicate that BMI1 represses HbF via the cPRC1 complex in adult erythroid cells. The BMI1-containing cPRC1 complex is known to repress transcription in conjunction with the PRC2 complex which deposits the histone mark H3K27me3. CUT&RUN profiling studies revealed focal loss of H3K27me3 in BMI1-deficient HUDEP2 cells, including at the CG-rich regions near the promoters of the IGF2BP1 and LIN28B genes. Therefore, we tested whether cPRC1 and PRC2 complexes converge on HbF regulation by regulating IGF2BP1 and LIN28B. Disruption in HUDEP2 cells of EZH2, the catalytic component of PRC2, led to global depletion of H3K27me3 along with upregulation of IGF2BP1 and LIN28B. Co-depletion of IGF2BP1 and LIN28B largely restored the EZH2 depletion-induced HbF upregulation. In summary, our studies further define the nature of PRC complexes involved in HbF regulation and identify the IGF2BP1, IGF2BP3, and LIN28B genes as critical links between PcG activity and HbF silencing. Our results may guide the development of improved PRC inhibitors for the treatment of β-globin disorders.