Regulation of Myelopoiesis by Differentiation Stage-Specific Activities of HOXA9 and HOXA10

Abstract SCI-29

HOXA9 and HOXA10 are homeodomain (HD) transcription factors that are implicated in control of myelopoiesis and contribute to myeloid leukemogenesis. These proteins are expressed coordinately during hematopoiesis, with maximal expression in granulocyte/monocyte progenitor (GMP) cells. Engineered overexpression of Hoxa9 or Hoxa10 in primary bone marrow cells expands the GMP population in vitro, and results in myeloproliferation in murine bone marrow transplant experiments. Mice transplanted with Hoxa9- or Hoxa10-overexpressing bone marrow develop acute myeloid leukemia (AML) over time. Consistent with this, increased and sustained expression of a set of HD proteins, including HOXA9 and HOXA10, is found in a subset of human AML, including AML with MLL gene translocations (11q23-AML). Since the DNA-binding HDs of HOXA9 and HOXA10 are highly conserved, we hypothesize that they recognize a common set of target genes. However, since HOXA9 and HOXA10 diverge outside the HD, we considered the unexplored possibility that they perform different functions in regulating such genes. To identify molecular mechanisms for HOX-induced GMP expansion and leukemogenesis, we performed a chromatin immunoprecipitation-based screen for HOXA10 target genes. Gene ontology studies determined that the identified set is enriched for genes encoding growth factors and receptors, including fibroblast growth factor 2 (FGF2). We found that production of FGF2 by Hoxa10-overexpressing GMP stabilizes β-catenin and induces proliferation in an autocrine manner. We also found that HOXA9 and HOXA10 activate common FGF2 cis elements. The Hoxa10-target-gene set is also enriched for HD-transcription factors, including CDX4. We determined that Cdx4 transcription is activated by HOXA10 in GMP, but repressed by HOXA9 in differentiating myeloid cells. CDX4 activates transcription of both Hoxa9 and Hoxa10, identifying a HOX-CDX cross-regulatory mechanism. This mechanism may be influenced by Fgf2, since Hoxa10 and Cdx4 are β-catenin target genes, but β-catenin activity decreases Hoxa9 expression. Gene expression profiling studies indicate that HOXA9, HOXA10, CDX4, and FGF2 are increased in 11q23-AML, suggesting clinical relevance. Arih2 (encoding the E3 ligase Triad1) is another common HOXA9 and HOXA10 target gene that may influence Fgf2 activity. We found that Arih2 transcription is repressed by HOXA9 in myeloid progenitors, but activated by HOXA10 in differentiating phagocytes. FGF receptors are destabilized by ubiquitination, and we found increased FGF-R ubiquitination in Hoxa10-overexpressing cells. Therefore, Triad1-dependent regulation of FGF-R stability is another mechanism for control of FGF2 activity and myeloproliferation by HOXA9 and HOXA10. Therefore, HOXA9 and HOXA10 regulate a common set of target genes that control GMP expansion in a manner that is antagonistic for some genes and cooperative for others. Clinical correlative studies suggest that coordinate control of these genes by HOXA9 and HOXA10 is dysregulated in HOX-overexpressing leukemia. Understanding HOX-regulated gene networks may identify therapeutic targets for HOX-overexpressing leukemias. For example, blocking FGF-related signaling pathways may ameliorate cytokine hypersensitivity in such leukemias, and would be a topic of interest for additional studies.