ETO2 Recruitment of NuRD to the γ-Globin Locus Involves Multivalent Interaction with GATAD2A

Increasing the expression of fetal hemoglobin ameliorates the symptoms of sickle cell disease and β-thalassemia. Therefore, blocking fetal hemoglobin silencing in adult erythrocytes represents a potent strategy for treating these diseases. Extensive research over the past five decades have uncovered critical transcription factors involved in silencing fetal hemoglobin, including Bcl11A and Zbtb7A (LRF), among others. The Nucleosome Remodeling and Deacetylase (NuRD) complex represents a central co-regulatory factor recruited by these proteins and DNA methylation to silence gene expression. Hence, NuRD represents a potential target for blocking silencing by multiple factors simultaneously, thereby leading to robust induction of fetal hemoglobin. We previously have shown that disrupting the MBD2-NuRD complex induces high levels of fetal hemoglobin in both the HUDEP-2 tissue culture cell line and primary patient-derived bone marrow stem cells [Yu et al. (2019) Haematologica, 104(12), 2361-2371].

Recent work by the Dean lab [Guo et al. (2020). Nucleic Acids Research, 48(18), 10226-10240] added to the list of factors that bind and recruit NuRD to the globin locus. They showed that ETO2, a well-established factor in erythroid differentiation, binds directly to NuRD, which required the tetramerization domain of ETO2. However, the specific contacts driving this interaction remain unknown. Based on previous work by the Bushweller [Liu et al. (2007) Cancer Cell, 11(6), 483-497] and Vermeulen [Spruijt et al. (2016) Cell Reports, 17(3), 783-798] groups, we hypothesized that ETO2 binds directly to the GATAD2A component of NuRD through the Nervy Homology Region-4 (NHR4) domain of ETO2 and multiple polyproline-leucine (PPPL) motifs in GATAD2A.

Here we use NMR analyses to show that the ETO2-NHR4 domain binds each of four polyproline-leucine motifs in GATAD2A. We measure binding by chemical shift analysis which reveals a weak affinity (K_D ~200-300 μM) for the individual peptides. We then show that the ETO2-NHR3 domain contributes to weak homo-oligomerization (dimer-tetramer formation) by gel filtration analysis. Importantly, multivalent interaction between the PPPLx4 region of GATAD2A and a construct including both the NHR3 and NHR4 domains of ETO2 leads to a dramatic increase in binding affinity (K_D ~ 4 μM) as measured by isothermal titration calorimetry. Furthermore, using an in-cell bioluminescent resonance energy transfer assay (NanoBRET), we show that including the NHR2 tetramerization domain augments association between these proteins. Mutating key residues in either GATAD2A-PPPLx4 or ETO2-NHR4 domains disrupts this association in cells. Finally, we show that enforced expression of a peptide from GATAD2A containing the four polyproline-leucine motifs consistently induces a low-level of fetal hemoglobin mRNA expression in K-562 cells.

Together, these findings establish that the ETO2-NHR4 domain plays a central role in recruitment of NuRD for gene silencing. Tetramerization of the NHR2 and NHR3 domains leads to multivalent association with GATAD2A, which dramatically increases binding affinity. This result has mechanistic implications for fetal hemoglobin silencing and erythroid differentiation, potentially leading to a novel therapeutic target for treating β-hemoglobinopathies. Furthermore, gene rearrangements involving the ETO family of proteins drive the development of acute myeloid leukemia. Our studies provide an explanation for how the stoichiometry of the fusion proteins may dictate which co-regulatory complexes are brought to a specific site and suggest a strategy for selectively targeting these complexes.

No relevant conflicts of interest to declare.