Hematopoietic stem cells (HSCs) should be the main target for accumulation of mutational events, which eventually leads to formation of leukemic stem cells. These leukemogenic mutations have been classified at least into class I (providing the proliferative and survival advantage) and class II (impairing the differentiation activity) gene abnormalities. It has been proposed that acquisition of both class I and class II mutation are essential for the development of leukemia. Although several experimental animal studies suggest this model, there is no direct evidence that class I and class II mutations collaborate to contribute to development of human leukemias. Here we demonstrate that the acquisition of 8;21 translocation, which encodes the AML1-ETO (a class II chimeric fusion gene), and of mutational c-Kit (a class I mutation) is sequentially occurred in human acute myelogenous leukemia (AML). It has been shown that in t(8;21) AML patients treated with chemotherapy, a small amount of AML1-ETO mRNA was never disappeared even in patients maintaining remission for more than 10 years. We have demonstrated that this AML1-ETO mRNA in “cured” patients is derived from t(8;21)+ HSCs that consisted only a few percent of HSCs in remission (Miyamoto et al., PNAS 2000; 97: 7521–7526). The t(8;21)+ HSCs possessed normal differentiation at least into myeloerythroid cells and B cells. These data strongly suggest that acquisition of the AML1-ETO fusion is not sufficient for development of t(8;21) AML, and that t(8;21)+ HSCs are preleukemic HSCs.
We hypothesized that acquisition of additional class I mutation might transform the AML1-ETO+ preleukemic HSCs into AML stem cells. We therefore searched for class I mutations in t(8;21) AML samples, and found that in 13 out of 33 t(8;21) AML patients, AML cells have c-Kit mutations (but not other class I such as FLT3-ITD and N-Ras mutations) at diagnosis. We then tested whether the AML1-ETO+ preleukemic HSCs in remission marrow have the c-Kit mutation. Six out of these 13 t(8;21) AML patients with c-Kit mutation maintaining long-term remission were enrolled in this study. To confirm the coexistence of AML1-ETO and c-Kit mutation in single leukemic stem cells, CD34+CD38− AML cells were purified from the bone marrow of patients at diagnosis, and tested for the presence of AML1-ETO and c-Kit mutation by single cell PCR. In all of 910 single CD34+CD38− AML cells, both AML1-ETO and c-Kit mutations were detected. Then, CD34+CD38− HSCs in remission were tested for the presence of AML1-ETO and c-Kit mutation. In 1728 single CD34+CD38− HSCs of remission marrow, 0.9% (16 cells) of these cells expressed AML1-ETO. Surprisingly, none of these AML1-ETO+ preleukemic HSCs possessed c-Kit mutation, indicating that AML1-ETO+ clones in long-term remission are independent from the original t(8;21) AML clones in terms of the presence of c-Kit mutation. We then performed colony-forming assays to evaluate the differentiation potential of these AML1-ETO+ preleukemic HSCs. HSCs of remission marrow-derived colonies were picked up, and tested for the presence of AML1-ETO and c-Kit mutation. In 7187 colonies formed in the culture of remission marrow, 1.2% (89 colonies) of these colonies were positive for AML1-ETO, and all of these colonies were negative for c-Kit mutation. These data collectively suggest that the acquisition of c-Kit mutation is the second step for formation of t(8;21) AML stem cells: Normal HSCs acquire t(8;21) and express resultant AML1-ETO (Class II) but it is not sufficient for full transformation into AML stem cells. These preleukemic HSCs possess normal differentiation activity, but additional c-Kit mutation (Class I) might be critical in transforming into AML stem cells. This is the first clear-cut evidence that HSCs transform into AML stem cells by stepwise acquisition of Class I and Class II mutations.
No relevant conflicts of interest to declare.
Asterisk with author names denotes non-ASH members.