Tumor Suppressor FOXO1 Serves As a Critical Oncogenic Mediator in AML1-ETO Leukemia

AML1-ETO (AE) fusion protein, produced by the chromosome translocation (8;21), accounts for 10%-15% of acute myeloid leukemia (AML). We use CD34+ human umbilical cord blood (CB) cells retrovirally transduced with AE to model AE leukemia (AE cells). This model represents the pre-leukemic stage of AE leukemia since AE by itself does not lead to overt disease but largely enhances the self-renewal and impedes the differentiation of CB cells. However the molecular mechanism by which AE deregulates the hematopoietic program is still largely unknown.

The FOXO subfamily of forkhead transcription factors, which in mammals includes FOXO1, FOXO3, FOXO4, and FOXO6, are well known tumor suppressor proteins due to their ability to arrest cell cycle and induce apoptosis. Although FOXO1 is silenced in many types of solid tumors, unexpectedly, we found expression of FOXO1 (but not other family members) significantly upregulated in AE AML patient samples compared to other AML subtypes. Increased protein expression was confirmed by western blot in a cohort of patient samples and our AE cells. AE bound to a region within intron 1 of the FOXO1 gene as shown by chromatin immunoprecipitation, the sequence of which includes several AML1 binding motifs. The expression of FOXO1 was dependent on the presence of AE as shown by overexpression and knockdown studies. These findings suggest that FOXO1 could be a direct target of AE. Knockdown of FOXO1 by shRNA resulted in decreased proliferation of AE cells, and conditional deletion of Foxo1 in murine AE cells led to decreased CFU replating in methylcellulose compared to non-deleted control. In contrast, conditional deletion of Foxo3 using the same approach did not significantly influence the clonogenicity of AE cells. These data indicate FOXO1 is critical for AE cell growth.

Interestingly, overexpression of FOXO1 in CD34+ CB cells promoted their long-term proliferation in liquid culture, inhibited erythroid differentiation while promoting myeloid lineage development and enhanced their replating ability in colony forming assays. FOXO1-expressing cells also achieved a higher engraftment level when transplanted into immunodeficient mice. These effects partially phenocopy the outcomes observed upon AE expression in CB cells. These effects require the transcriptional activity of FOXO1, as shown with a FOXO1 mutant lacking DNA binding capacity, and are not seen upon FOXO3 expression. Although inhibition of reactive oxygen species (ROS) production is regarded as a general function of FOXO family proteins, ROS regulation is unlikely to be the major downstream mechanism, since ROS levels were only transiently and slightly decreased upon FOXO1 expression. In addition, treatment of CB cells with the ROS scavenger N-acetyl-cysteine could not recapitulate FOXO1’s phenotype.

To gain molecular insight into FOXO1’s function in CB cells, RNA-Seq experiments were performed in triplicate. CD34+ CB cells were transduced with empty vector (MIT), FOXO1 DNA-binding-deficient mutant (FOXO1 DB), wildtype FOXO1 (FOXO1 WT) or AE. The MIT and FOXO1 DB groups showed very similar gene expression profiles. 1349 genes were upregulated and 1113 genes downregulated in the AE group while FOXO WT showed 608 genes increased and 490 genes decreased (P<=0.01, fold change>=1.5). To our surprise, 282 upregulated and 231 downregulated genes were common targets of AE and FOXO WT, suggesting that about 20% of AE downstream targets are potentially regulated via FOXO1. More strikingly, pathway enrichment analysis showed that published AE gene signature datasets were enriched in the FOXO1-WT gene list. Among genes upregulated by FOXO1, several have been reported as AE transcriptional targets, such as POU4F1 and JUP. In addition, some FOXO1 target genes, SOX4 and HLF for example, have been implicated in leukemogenesis.

In summary, we find FOXO1 is required for AE cell growth, and overexpression of FOXO1 in CB cells can partially recapitulate the AE phenotype at both the morphological and molecular level. Taken together, these data suggest that instead of acting as a tumor suppressor, FOXO1 is a critical oncogenic mediator of the AE leukemia program. It is possible that FOXO1 serves as one signaling hub of the AE molecular network, similar to the role MEIS1-HOXA9 play in MLL-fusion AML. Fully dissecting the AE-FOXO1 pathway will enhance our understanding of AE leukemogenesis and provide potential therapeutic targets.