FLT3-ITD Has Developmental Context Specific Effects on Hematopoiesis and Myeloid Leukemogenesis

Much progress has been made toward identifying the mutations that cause human acute myeloid leukemia (AML), and these studies have shown that pediatric and adult AML are often caused by different mutations. Genetic differences between pediatric and adult AML may underlie differences in outcomes and necessitate different treatment strategies, yet we have few insights into why these differences occur. One possibility is that the mechanisms that regulate normal hematopoiesis change with age, and mutations therefore have age-specific phenotypes in pre-leukemic progenitors. To fully understand how AML evolves in children and adults, and how targeting individual pathways might impact cell physiology at different ages, it is important to understand how somatic mutations interface with the normal, temporally dynamic programs that regulate hematopoiesis.

The FLT3-Internal Tandem Duplication (FLT3-ITD) mutation is common in adult AML but rare in early childhood AML (30-40% of adult AML, 5-10% of AML in children <10 years old, <1% of infant AML). FLT3-ITD mutations occur late in the clonal evolution of AML cells, and they are thought to drive cell proliferation and survival. In mice, FLT3-ITD has been shown to deplete adult hematopoietic stem cells (HSCs) by promoting myeloid differentiation. This may explain why the mutation occurs late in clonal evolution - HSCs must first acquire mutations that enhance self-renewal - but it also raises the question of why infant HSCs, which have an inherently higher self-renewal capacity, do not give rise to FLT3-ITD positive AML more often than is observed.

We used FLT3-ITD knock-in mice to test whether FLT3-ITD has developmental context specific effects on hematopoiesis. In adult mice, FLT3-ITD depleted the HSC pool and expanded multipotent progenitor (MPP) and myeloid progenitor populations consistent with prior studies. In fetal mice, FLT3-ITD had no effect on HSC or MPP numbers, HSC function (as determined by limit dilution transplants) or myelopoiesis. FLT3-ITD did not affect hematopoiesis until shortly after birth. These temporal differences were evident even in the presence of cooperating Runx1 mutations. To understand why fetal and adult progenitors responded differently to FLT3-ITD, we characterized signal transduction and gene expression in fetal, neonatal and adult progenitors. We found that STAT5 was activated by FLT3-ITD at all stages of development, but MAPK was activated only in post-natal progenitors concordant with the onset of HSC and myeloid phenotypes. To our surprise, conditional Stat5a/b deletion exacerbated the HSC depletion and myeloid expansion phenotypes of adult FLT3-ITD mice rather than rescuing them. This suggests that STAT5 helps to maintain adult, FLT3-ITD mutant progenitors in an undifferentiated state even as other effectors promote myeloid differentiation. We next used microarrays to test whether FLT3-ITD has age-specific effects on gene expression in HSCs and MPPs, and to identify normal temporal changes in gene expression that may modulate the FLT3-ITD phenotypes. These studies made several key points: 1) In wild type HSCs, most fetal-specific genes were inactivated and most adult-specific genes were activated between birth and P14. This transition was earlier than prior studies have suggested, and it correlated with the age at which FLT3-ITD induced HSC depletion and myeloid expansion. 2) FLT3-ITD did not alter gene expression until after birth, coincident with onset of the HSC depletion and myeloid expansion phenotypes. 3) FLT3-ITD target genes were more differentially expressed in MPPs than in HSCs, consistent with recent data suggesting that MPPs are a cell of origin for FLT3-ITD driven AML. 4) Most, but not all, FLT3-ITD target genes were STAT5 dependent. Our analyses have identified novel, adult-specific candidate effectors of FLT3-ITD. Moreover, our findings raise the question of whether fetal genetic programs can suppress FLT3-ITD driven leukemogenesis, and we have begun to address this question with gain of function models. AML cells may exhibit "context addiction" (i.e.a sustained requirement for normal adult gene products and a toxic response to fetal gene products), that could be exploited therapeutically.