Gene expression changes associated with progression of Irf8–/– CML-like disease into blast crisis after β-catenin activation

Progression and disease relapse of chronic myeloid leukemia (CML) depends on leukemia-initiating cells (LIC) that resist treatment. Using mouse genetics, we observed that compound constitutive activation of β-catenin and deletion of Irf8 results in progression of CML-like disease into fatal blast crisis, elevated leukemic potential of BCR-ABL-induced LICs, and accumulation of Imatinib-resistant LICs in GMP-like populations. We found that progression of the disease is tightly connected to the magnitude of gene expression and that activated β-catenin enhances a pre-existing Irf8-deficient gene signature that was defined as a “progression specific signature” (PSS). We identified β-catenin as an amplifier of disease progression and as a critical step in the shift of CML to blast crisis. Collectively, our data uncover Irf8 as a roadblock for β-catenin-driven leukemia and imply both factors as targets in combinatorial therapy. We used microarrays to identify the gene expression signature in GMPs underlying CML-like disease progression and identified distinct classes of up- and down-regulated genes during this process defined as a “progression specific signature (PSS)”. GMP populations were sorted from three to four independent pools of control MxCre–Ctnnb1(Ex3)fl/+, Ctnnb1(Ex3)∆/+, Irf8–/–Ctnnb1(Ex3)fl/+ and Irf8–/–Ctnnb1(Ex3)∆/+ mice for RNA extraction and hybridization on Affymetrix Mouse 430_2 chip arrays (MG-430 PM peg arrays; Affymetrix GeneChip). Cells were sorted using FACSAria (BD Biosciences Immunocytometry Systems, San Jose, CA) flow cytometers and GMPs defined as lineage depleted Lin– Sca-1–c-Kit+CD34+FcR II/IIIhi population. All mice were treated with 400µg polyinosinic-polycytidylic acid (poly(I:C)) (Amersham) on day 0, 3 and 5 to induce β-catenin activation (MxCre-dependent excision of Exon3 in the β-catenin gene). Harvesting of bone marrow cells were performed 10-14 days after the last poly(I:C) injection. We used heterozygous inducible MxCre+Ctnnb1(Ex3)fl/+ mice because of the dominant effect from a single activated Ctnnb1 allele. To validate excision efficiency, genomic DNA from harvested cells was subjected to PCR, as previously described (Huelsken et al., 2001; Scheller et al., 2006).