Abstract
Treatment-related acute myeloid leukemia (t-AML) arises as a result of prior exposure to radiation, alkylator chemotherapy or topoisomerase II inhibitors. The overall prognosis is poor, with a median survival of only 6-10 months. Genetic background influences the risk of acquiring t-AML, yet specific susceptibility factors have not been wellcharacterized. We utilized a mouse model to identify novel risk factors for t-AML, which allowed us to control for environmental factors and test the effect of different genetic backgrounds. Cohorts of twenty inbred mouse strains (including several genetically diverse wild-derived strains) were treated with N-ethyl-N-nitrosourea (ENU), a potent alkylating agent in mice. As we previously reported, six of these mouse strains were susceptible to alkylator-induced myeloid leukemia. In the current study, we searched for candidate susceptibility factors that could explain this phenotype. Genome-wide association analysis using strain-specific leukemia incidence and a panel of single nucleotide polymorphism (SNP) markers revealed a peak spanning 1.07 Mb on mouse chromosome 3. This region contains six genes, including myeloid leukemia factor 1 (Mlf1). Mlf1 is a strong candidate gene since it is a translocation partner in the rare t(3;5) (q25.1;q34) associated with AML and MDS in which NPM1 is fused to nearly full-length MLF1. We first sequenced the Mlf1 locus in all 20 strains and found SNPs and in/dels at a frequency of 1/100 bp, but no nonsynonymous coding changes. Next, we asked whether Mlf1 is expressed in the cellular compartment relevant for initiation of t-AML. RNA was isolated from bone marrow cells from 18 of the 20 previously characterized strains (N=4 mice per strain) sorted into lineage (GR-1, CD19, B220, CD3, CD4, CD8, TER119, and IL-7Rα) negative/kit+, whole bone marrow, or lineage+ fractions. Quantitative RTPCR analysis demonstrated that Mlf1 is only expressed in the more immature lineage-/kit+ cells. In addition, Mlf1 is expressed in a strain-dependent fashion with a 10-fold difference of expression across strains that ranged from below the level of detection to 2.3% of the GAPDH signal. To further define the pattern of Mlf1 expression, the lineage-/ kit+ population was purified into HSC, CMP, GMP, and MEP subsets. The highest Mlf1 expression was found in the common myeloid progenitors (Lin-/c-Kit+/Sca-1-/CD34+/ FcγR-). There was not a direct correlation between RNA expression levels and t-AML susceptibility. Possible explanations for this discordance include heterogeneity of the cell populations analyzed, differences between cells at baseline vs. after exposure to a genotoxin, and potential differences between MLF1 mRNA and protein levels. Finally, to address how different levels of Mlf1 expression might affect t-AML susceptibility, we utilized an in vitro overexpression assay. Primary bone marrow cells were transduced with either an MLF1 ires YFP-MSCV retrovirus or control YFP-MSCV retrovirus with similar multiplicities of infection. YFP+ cells infected with the control virus expanded to nearly 4-fold greater levels than cells overexpressing MLF1 when analyzed at 72 hours. We show that the decreased accumulation of MLF1-expressing cells is due to, at least in part, a 3.4-fold increase in apoptosis (p<0.001, 44% in MLF1 ires YFP vs.13% in control). We propose a model in which the relative abundance of MLF1 in CMPs is a determinant that influences whether or not cells exposed to genotoxic stress undergo MLF1 induced apoptosis. The identification of the molecular basis of t-AML susceptibility may lead to strategies that reduce the incidence of this disease.
Cooperating cancer-gene identification through oncogenic-retrovirus–induced insertional mutagenesis
