Structure guided targeting of the GATAD2A-CHD4 interaction in the MBD2-NuRD complex results in high levels of HbF

Abstract It is well established that high levels of fetal hemoglobin (HbF) can overcome the pathophysiologic defects in sickle cell disease (SCD) and beta-thalassemia. Despite the recent success of gene therapy in increasing HbF in SCD and the usefulness of hydroxyurea in improving patient outcomes, there remains a pressing need for more effective small molecules to treat the vast majority of SCD patients who will not benefit from gene therapy in the foreseeable future. In addition to the strong negative regulatory transcription factors BCL11A (Sankaran et al, 2008) and ZBTB7A(LRF) (Masuda et al, 2016), the MBD2 NuRD complex, which recognizes and binds at sites of CpG methylation, is essential for HBG silencing in adult erythroid cells (Rupon et al, 2006; Yu et al, 2019; Shang et al, 2023). It has been shown previously that depletion of CHD4, a critical component of MBD2-NuRD which binds to the core complex through GATAD2A (Gnanapragasam et al, 2011), by as little as 50% induces high levels of HbF without blocking differentiation in primary adult human erythroid cells (Amaya et al, 2013; Lan et al, 2021; Vinjumar et al, 2021). To identify potential targets for small-molecule disruption of MBD2-NuRD, we employed the AlphaFold3 ML program to predict the structure of the protein-protein interaction between GATAD2A and CHD4 that tethers the latter to the MBD2-NuRD complex. This analysis indicated that a small helical region in the CR-2 domain of GATAD2A binds to the interface between the C-terminal domains 1 and 2 (CTD1/2) of CHD4. Intracellular NanoBRET assays confirmed strong binding between the entire CR-2 region or the small helix from this region and CHD4(CTD12) domains. Isothermal calorimetry likewise demonstrated tight binding between these domains in vitro. Furthermore, mutating amino acids in the CR-2 helix, which are predicted to be critical for the interaction, disrupted high-affinity binding to CHD4 (CTD1/2). Enforced expression of wild-type GATAD2A, but not GATAD2A mutated at these critical amino acids, resulted in silencing of HBG expression in HUDEP-2 cells in which GATAD2A was knocked out. Prime editing (PE) of the same critical amino acids in the endogenous GATAD2A(CR-2) region resulted in over 40% HBG/(HBG+HBB) RNA levels and likewise over 40 % HbF levels. Immunoprecipitation assays showed dissociation of CHD4 from GATAD2A in the edited cells. NOME SEQ assays demonstrated the same loss of nucleosomes over the HBG promoter as seen in MBD2 knockout cells, and chromatin immunoprecipitation assays showed markedly decreased occupancy of CHD4. RNA seq analysis was carried out to identify other genes affected by the PE editing. HBG1, HBG2, and BGLT3 were upregulated, and HBB was mildly downregulated, while neither CD235 nor CD71, markers of erythroid differentiation, were affected. Lentiviral vector-mediated expression of the small wild-type CR-2 peptide characterized in binding assays resulted in over 15% HBG/(HBG+HBB) RNA levels and dissociation of GATAD2A from CHD4, while in contrast expression of the corresponding peptide with mutations of critical amino acids at the interface with CHD4 (CTD1/2) did not affect either. Together, these experiments demonstrate an on-target effect of disrupting the GATAD2A interaction with CHD4 to relieve MBD2-NuRD mediated silencing of HBG expression. In summary, structural and biophysical studies guided by AlphaFold3 predictions and followed by functional genetic, biochemical, and cellular assays showed that a small peptide can disrupt the interaction between GATAD2A and CHD4 in the MBD2-NuRD complex, thereby inducing high levels of Hb F in adult human erythroid cells. Given the viability and mild phenotype of MBD2 knockout mice and the clinical success of multiple peptidomimetic drugs, these results identify a specific target for the development of small molecules to therapeutically induce high levels of Hb F in patients with SCD and beta-thalassemia.