Experimental data have shown that two of the most frequent genetic alterations in AML, the AML1-ETO (A/E) fusion gene and the FLT3 length mutation (FLT3-LM) are both mostly insufficient on their own to induce leukemia. These findings support the model that collaboration of two classes of genetic alterations, altering proliferation or differentiation, is necessary for leukemogenesis. When we first analyzed 135 patients with A/E positive AML, additional mutations affecting signal transduction were found in 38 % of all cases (FLT3-LM 10.3 %, KIT 8.1 % and NRAS 9.6 %). In contrast, none of the patient with A/E positive leukemia had alterations associated with transcriptional regulation such as MLL-PTD. To test the hypothesis that A/E collaborates with FLT3-LM in inducing acute leukemia, we transplanted mice with bone marrow (BM) cells retrovirally expressing A/E, FLT3-LM or both alterations. Mice transplanted with BM cells expressing A/E or FLT3-LM alone did not develop any disease. In contrast, mice (n=11) transplanted with BM cells expressing both alterations succumbed to an aggressive acute leukemia. Intriguingly, developing leukemias differed with regard to their phenotype with 7 animals developing AML and 4 animals developing ALL. Furthermore, the majority of AML cases showed simultaneous expression of lymphoid antigens as described in patients with A/E positive AML. The collaboration of A/E with FLT3-LM was depending on DNA binding activity of the fusion gene as the L148D point mutation in the Runx1 domain of the construct abrogated collaboration of A/E with the FLT3-LM in the CFU-S assay. Furthermore, inactivation of the kinase activity of the FLT3-LM (FLT3-LM K672R mutant) resulted in the complete loss of collaboration with the A/E fusion. Treatment of cells co-infected with A/E and FLT3-LM with the kinase inhibitor PKC412 resulted in a 62 % reduction of the CFU-S frequency. To further explore a possible contribution of retroviral insertional mutagenesis to the transformation process in this model, 10 retroviral integration sites were subcloned and sequenced from 4 leukemic mice: all 10 sites were unique with no indication of a common integration site associated with the leukemic transformation. Moreover, 5 sites were intergenic or not linked to known genes. The remaining sites were in introns in a 5′ to 3′ orientation most likely to lead to gene knockdown rather than activation. These data provide direct functional evidence for the oncogenic collaboration between A/E with a class of activating mutations, recurrently found in patients with t(8;21), and add experimental data to the clinical observation which demonstrated a significant inferior treatment outcome in patients with AML1-ETO and additional mutations of receptor tyrosine kinases.

Author notes

Corresponding author