Chromosomal translocation (8;21) results in the formation of the AML1-ETO (AE) leukemia-associated fusion protein. Roughly 70% of AE+ patient samples also harbor an alternatively spliced variant of AE, termed AE9a. AE9a promotes the leukemic transformation of mouse hematopoietic stem/progenitor cells without the need for secondary external cooperating oncogenes. By comparison, leukemia development in response to expression of full length AE requires one or more additional events. We extended these findings into a human cell system to determine whether AE9a is able to promote the transformation of human CD34+ cord blood (CB) cells to acute myeloid leukemia (AML). Transduction of CD34+ CB cells with AE9a promotes long-term culture potential, enhanced replating capacity, and expansion of CD34+ progenitor cells similar to full length AE. While we have not yet been able to generate AE9a leukemia in xenograft assays (despite several attempts), subcutaneous injection of AE9a cells promotes formation of a robust granulocytic sarcoma which is not seen using cells expressing full length AE. Several of our analyses have highlighted the importance of increased AE9a fusion protein expression in its enhanced function relative to full length AE, a concept that was also observed using the MLL-AF9 leukemia oncogene (Chen et. al, Cancer Cell, 2008). Indeed, flow cytometric sorting of freshly transduced AE or AE9a cultures based on high and low fusion protein expression revealed the importance of expression level for both AE and AE9a. High AE expression completely blocked colony formation and decreased cell expansion. These effects correlated with a dramatic upregulation of the cell cycle inhibitor p21. Conversely, high levels of AE9a did not upregulate p21 to the same degree, and colony formation and cell expansion were not as dramatically affected. In contrast, flow sort for cells expressing low levels of AE or AE9a had significantly less remarkable effects on cell function and p21 expression. Unexpectedly, AE9a-low cells eventually came to resemble the AE9a-high in terms of AE9a protein levels. These effects were not seen for full length AE sorted cultures. To determine whether similar effects were occurring in the murine AML model, we examined the effects of low and high expression levels of AE9a in mouse fetal liver cells. Corroborating our findings with the human cell model, AE9a-low samples promoted significantly less AML as compared to AE9a-high cells. This was not due to lack of engraftment of these cells or to loss of AE9a expression. Similar effects were found in vitro, where AE9a-low cells were unable to replate in methylcellulose assays while AE9a-high cells as well as AE-high and AE-low cells showed enhanced replating ability. Strikingly, the AML that did develop in mice transplanted with AE9a-low cells reverted to a very high level of AE9a protein expression, implying that increased expression of AE9a is essential for function. To better understand the mechanistic link between increased AE9a expression and AML development, we performed global gene expression analysis of murine progenitor cells expressing AE or AE9a (after three weeks of in vivo expansion). Analysis revealed a subset of genes that were highly repressed in AE9a expressing cells as compared to full length AE. Given that AE9a lacks the carboxy-terminal NHR3 and NHR4 domains of AE that interact with the NCoR and SMRT corepressors, the finding of significantly more repressed genes in AE9a cells is quite surprising. Network analysis has uncovered potential signaling pathways that may be involved in AE9a mediated transformation. Uncovering the mechanism of AE9a function could lead to a better understanding of t(8;21)-associated function in human AML.
No relevant conflicts of interest to declare.
Asterisk with author names denotes non-ASH members.