Structural and Functional Characterization of the NHR2 and Runt Domains of AML1/ETO.

AML1/ETO is the chimeric fusion protein resulting from the t(8;21) found in AML of the M2 subtype. It contains the N-terminal 177 amino acids of RUNX1 and virtually all (575aa) of ETO. The RUNX1 component includes the Runt domain, which mediates both DNA binding and heterodimerization with CBFβ, but lacks the more C-terminal sequences required for transactivation. AML1/ETO occupies RUNX target genes in vivo and is associated with a repressive chromatin structure characterized by reduced levels of acetylated histone H3. AML1/ETO is thought to repress transcription by recruiting a SMRT (N-CoR)/Sin3A/HDAC complex to chromatin via sequences in ETO. ETO is the human homologue of the Drosophila Nervy protein and shares 4 regions of homology with Nervy called Nervy Homology Regions (NHR) 1–4. Deletion studies have shown that three of the AML1/ETO domains essential for its repressive function are the Runt domain, NHR2, and NHR4. The NHR2 domain is a hydrophobic heptad repeat that mediates oligomerization of AML1/ETO, interaction with ETO family members, and also with mSin3A and HDACs. We recently solved an x-ray structure of the NHR2 domain and found it to be an alpha-helical tetramer. Based on this structure we have introduced amino acid substitutions into the NHR2 domain that disrupt tetramer formation but not AML1/ETO stability. These mutations impair the ability of AML1/ETO to inhibit the differentiation of GR-1+/Mac-1+ cells following retroviral transduction into primary mouse bone marrow cells, and also inhibit the serial replating ability of AML1/ETO expressing bone marrow cells in vitro. We additionally show that mutations reported by Amann et al. (Mol Cell Biol. 21, 6470, 2001) to disrupt mSin3A binding to NHR2 do not affect the biological activity of AML1/ETO in vitro. We also introduced mutations in the Runt domain of AML1/ETO that disrupt CBFβ binding by defined amounts (40-fold, 200-fold, 500-fold), and demonstrated that CBFβ binding by AML1/ETO is essential for its dominant negative activity. The latter results suggest that small molecules designed to selectively impair heterodimerization of AML1/ETO with CBFβ could potentially block AML1/ETO’s dominant negative activity.