Identification of Hoxa9 Targets in Myeloid Progenitors Using Conditional Expression.

Deregulation of HOXA9 expression appears to be a common mechanism of transformation in both lymphoid and myeloid acute leukemias. Although HOXA9 overexpression is often associated with MLL rearrangements, deregulation is also seen in leukemias with other translocations or lacking known translocations, where overexpression is highly correlated with treatment failure. A wealth of evidence indicates that in isolation, HOXA9 is a relatively weak oncoprotein that at physiologic levels expands stem cells. However the combination of high level HOXA9 expression in concert with MEIS1 is potently leukemogenic. The molecular basis for this cooperativity and relevant downstream effectors are essentially unknown. To begin to resolve this important issue we immortalized murine hematopoietic precursors by retroviral transduction of bone marrow with a conditionally active form of Hoxa9. The C-terminus of Hoxa9 was fused to a modified estrogen receptor ligand binding domain (ER) so that protein activity is contigent upon 4-OHT. Proliferation and maintenance of myeloblast phenotype required continuous exposure to the 4-OHT precursor tamoxifen. Withdrawal of tamoxifen resulted in differentiation within 96 h, as evidenced by growth arrest, expression of myeloid/monocytic differentiation markers, and maturation to macrophage morphology. Similar experiments to establish conditional Meis1 regulation by ER fusion at either the N or C terminus were not successful as the fusions were non-transforming. Direct and indirect targets of Hoxa9 were identified by measuring gene expression differences after tamoxifen withdrawal. At 72 h, 346 genes showed >2-fold change (adjusted p<0.05). At later points, a cohort of genes were upregulated, corresponding with the onset of myeloid maturation. The Ets family member Ehf showed the largest drop (10-fold) with Hoxa9 withdrawal; EHF has been implicated in carcinomagenesis; its role in leukemia remains unstudied. Flt3, a known target of joint Hoxa9/Meis1 regulation with prognostic significance, shows an approximately 3-fold decrease with withdrawal of Hoxa9 in the presence or absence of Meis1. Cancer outlier profile analysis (www.oncomine.org) reveals that Flt3 and other targets identified in our analysis, including Rage, S100a10, and Bambi, are highly correlated with HOXA9 in AML/ALL microarray datasets, suggesting a downstream role in HOXA9-mediated leukemogenesis. Genes up-regulated with Hoxa9 withdrawal include JunB, a previously described target negatively regulated by Hoxa9 that antagonizes leukemic self-renewal (Steidl et al., Nat Genet (2006) 38 :1269–77). The targets Mylk and Laptm4b are notable for belonging to the 11-gene “immediate self-renewal signature” of MLL-AF9 leukemic stem cells, which also includes Hoxa9 and Meis1 (Krivtsov et al., Nature (2006) 442: 818–22), suggesting a key role for Hoxa9 in the expression hierarchy within this signature. The presence of a subset of targets, including Flt3 and Sox4, that are dependent on both Hoxa9 and Meis1 expression is consistent with a model in which their synergy involves co-regulation of similar targets, rather than parallel transcription programs. This model is currently being tested using global ChIP on chip and other approaches.