The transcription factor RUNX1 is essential for definitive hematopoiesis and is perhaps the most common target for chromosomal translocations in acute leukemia, accounting for ~25% of the cases. Moreover, the RUNX1 point mutations have also been detected in AML with an overall frequency of 8.6% (Osato M, 2004). Therefore, the abnormalities of RUNX1 may contribute to over 30% of sporadic AML. The high frequency of disruption of RUNX1 suggests that it regulates key checkpoint genes that are responsible for leukemogenesis.

Neurofibromatosis type I (NF1) is a relatively common genetic disorder with an overall incidence of approximately one in 3000 worldwide. Affected individuals are prone to the development of benign tumors and malignant cancers. The disease is characterized by inactivating mutations of neurofibromatosis type 1 (NF1), thus NF1 is a tumor suppressor gene. Although NF1 mutations are associated with juvenile myelomonocytic leukemia, NF1 mutation is not observed in AML, implying that regulation of NF1 gene expression may exist in AML. Therefore, we asked if NF1 expression could be regulated by RUNX1. We identified 13 consensus RUNX1 binding sites in human NF1 promoter. Three of these sites were clustered near C/EBPα and ETS binding sites, similar to other genes that are co-regulated by these factors. This raised the possibility that NF1 is a transcriptional target of RUNX1, C/EBPα and/or ETS family members. In reporter assays, RUNX1 activated NF1 approximately 10- to 20-fold, but C/EBPα and ETS2 activated NF1 less than 7-fold. However, when C/EBPα was co-expressed with RUNX1, these factors dramatically cooperated to activate the promoter over 80-fold. While PU.1, an ETS family member, failed to synergize with RUNX1 to activate NF1, RUNX1 cooperated with ETS2 to activate the NF1 promoter over 90-fold. Through a series of deletion and point mutations, we defined the RUNX binding sites clustered between the AvrII and Tthlll I sites as required for the cooperative transactivation. Because RUNX1 is frequently disrupted in AML by the t(8;21), we further tested whether the RUNX1-MTG8 (R/M) fusion protein represses NF1. In reporter assays, R/M repressed NF1 up to 12-fold and this activity required DNA binding and the recruitment of co-repressors, as a point mutant that failed to bind DNA or a C-terminal deletion mutant that lacks several key co-repressor association domains fail to repress the NF1 promoter. In addition, we found that RUNX1 and R/M were associated with the endogenous NF1 promoter using chromatin immnoprecipitation analysis. Endogenous NF1 gene expression was also consistently repressed by R/M. Like mutational inactivation of NF1, R/M expression sensitized primary myeloid progenitor cells to GM-CSF, but not IL-3 in methylcellulose colony formation assay. Indeed, R/M expression in wild type and Nf1+/− myeloid progenitor cells isolated from d13.5 fetal livers of mouse embryos induced a similar number of colonies as complete deletion of Nf1. Taken together, these data indicate that NF1 is a direct transcriptional target of RUNX1 and the t(8;21) protein, linking reduction of NF1 to the molecular pathogenesis of AML.

Author notes

Corresponding author

Sign in via your Institution