Regulation of myeloid leukaemia by the cell-fate determinant Musashi

Abstract

Chronic myelogenous leukaemia (CML) can progress from a slow growing chronic phase to an aggressive blast crisis phase, but the molecular basis of this transition remains poorly understood. Here we have used mouse models of CML to show that disease progression is regulated by the Musashi-Numb signalling axis. Specifically, we find that the chronic phase is marked by high levels of Numb expression whereas the blast crisis phase has low levels of Numb expression, and that ectopic expression of Numb promotes differentiation and impairs advanced-phase disease in vivo. As a possible explanation for the decreased levels of Numb in the blast crisis phase, we show that NUP98-HOXA9, an oncogene associated with blast crisis CML, can trigger expression of the RNA-binding protein Musashi2 (Msi2), which in turn represses Numb. Notably, loss of Msi2 restores Numb expression and significantly impairs the development and propagation of blast crisis CML in vitro and in vivo. Finally we show that Msi2 expression is not only highly upregulated during human CML progression but is also an early indicator of poorer prognosis. These data show that the Musashi-Numb pathway can control the differentiation of CML cells, and raise the possibility that targeting this pathway may provide a new strategy for the therapy of aggressive leukaemias.

Figure 1. Expression of Numb impairs blast crisis CML development

a, CML cells were immunostained with anti-Numb antibody (red) and DAPI (green pseudocolor) and b, fluorescence intensity quantified *p<0.05. c, CML cells were analyzed by Western blot for Numb expression. d, Cells infected with BCR-ABL, NUP98-HOXA9 and either control vector or Numb were transplanted and survival monitored (control, n=18 and Numb, n=19). e, Representative and f, average frequency of Lin⁻ cells from control or Numb expressing leukemias **p<0.001. g, Lin⁻ cells from primary leukemias were serially transplanted and survival monitored (Vector, n=14 and Numb, n=15, **p<0.001). h, Hematoxylin and Eosin stained spleen sections from control vector or i, Numb expressing leukemias or j, surviving mice. Immature myeloid cells (red arrowheads) and lymphoid follicles (black arrowheads). Magnification: 10×. Error bars in all bar graphs are s.e.m. Data shown is representative of three to four independent experiments.

Figure 2. The RNA binding protein Musashi is highly expressed in immature normal and leukemic cells and is regulated by HoxA9

a, Musashi (Msi) expression in whole bone marrow (WBM), KLS cells, chronic and blast crisis CML, olfactory bulb (OB), –reverse transcriptase (-RT in OB) and water. Tbp, TATA binding protein. b–e, Realtime RT-PCR analysis of Msi2 expression in b, KLS cells (n=3), Lin⁺ cells (n=2) **p<0.001, c, blast crisis phase (n=9), chronic phase (n=6) **p<0.001, d, lin⁻ chronic and blast crisis phase cells relative to normal KLS and lin⁺ cells (lin⁺, n=2 and others, n=3), e, lin⁻ (n=5) or lin⁺ (n=5) blast crisis CML cells *p=0.039. Error bars represent s.e.m. f, Control vector- or Msi2-expressing CML cells were stained with anti-Numb antibody (red) and DAPI (green pseudocolor) and g, fluorescence intensity quantified **p<0.001. h, Msi2 expression in KLS cells transduced with either control vector or NUP98-HOXA9 retrovirus along with BCR-ABL *p=0.017. i–l, HoxA9 binds to the Msi2 promoter. Murine Msi2 gene structure: exons (numbered boxes), transcription start site (TSS; +1) and the direction of transcription (flag), putative HOX binding element 5.7kb upstream of TSS (oval) and +110kb site with no HoxA9 binding sequence (open rectangle). i, ChIP was performed either with IgG control or anti-HoxA9 antibody for j, Flt3, a known HoxA9 target gene, as a positive control and k, Msi2 −5.7kb region or l, Msi2 +110kb region. m, KLS cells from Msi2 genetrap reporter mice were transduced with BCR-ABL with either control vector or NUP98-HOXA9 and β-galactosidase reporter activity quantified (n=2 each, *p=0.011).

Figure 3. Loss of Musashi impairs the development and propagation of blast crisis CML

a, Frequency of KLS cells in mice of indicated genotypes (+/+, n=4, +/Gt, n=3 and Gt/Gt, n=4). b, Survival curve of mice transplanted with BCR-ABL and NUP98-HOXA9 infected +/+ or Gt/Gt KLS cells (+/+, n=15 and Gt/Gt, n=14, *p=0.0159). c, Colony-forming ability of blast crisis CML cells transduced with control shRNA (shLuc) or Msi2 shRNA (shMsi) **p<0.001. d, Survival curve of mice transplanted with established blast crisis CML cells infected with control shLuc or shMsi (n=13 each, *p=0.0267). e, Wright's stain of leukemic cells from mice transplanted with control shLuc or shMsi infected blast crisis CML. Immature myeloblasts (closed arrowheads), differentiating myelocytes and mature band cells (open arrowheads). Magnification: 100×. f, Survival curve of mice transplanted with Lin⁻ cells from primary shRNA expressing leukemias (Control, n=15 and shMsi, n=16, **p<0.001). Data shown is representative of two to three independent experiments.

Figure 4. Musashi expression is upregulated during human CML progression

a and b, PCR analysis of MSI2 expression in chronic and blast crisis CML patient samples from a, the Korean Leukemia Bank, Korea (n=9 per cohort, Mann-Whitney U test **p<0.001) and b, the Hammersmith MRD Lab Sample Archive, United Kingdom (n=6 per cohort, Mann-Whitney U test, **p<0.001). c–f, Microarray analysis of expression of c, MSI2 d, NUMB e, HOXA9 (all p<.001) and f, HES1 (p=.68) in bone marrow and peripheral blood samples from 42 chronic (red), 17 accelerated (green) and 31 blast crisis phase (blue) patients in the United States. g, Proposed model for the role of Musashi and Numb in CML progression.