Acute myeloid leukemia (AML) with MLL translocation is recognized as a group of aggressive hematopoietic malignancies with poor prognosis. An atypical homeodomain protein family (TALE family, three-amino-acid loop extension) has been reported to be involved in leukemogenesis and leukemia progression. For example, MEIS1 and PBX3 were abnormally expressed in examined AML patient samples when compared with their normal counterparts. While MEIS1 has been studied extensively, the precise role of PBX3 in leukemia development is still largely unknown.

In this study, we explored the epigenetic regulatory network of PBX3 and the specific role of PBX3 in leukemia progression. By analyzing the clinical database, we found that the high expression of PBX3 iscorrelated with poor prognosis of AML patients. In MLL-fusion induced AML mouse models that were established by retrovirus-mediated ectopic expression of MLL-AF9 or MLL-NRIP3 genes in murine BM progenitor cells (c-Kit+). PBX3 also showed high expression level in leukemia stem cells (LSC) compared with normal HSPCs. Unlike other homeobox genes such as HOXA9 and MEIS1 that were both highly expressed in LSKs and LSCs, PBX3 only showed extremely high expression in LSCs in comparison with HSPCs or other normal hematopoietic cells. To explore the role of PBX3 in leukemia, we analyzed the epigenetic modification patterns along PBX3 gene. By analyzing the public ChIP-seq database, we found that increased activation mark H3K79me2 and decreased repressive mark H3K27me3 exhibited on PBX3 in LSCs. Additionally, transcription activating modifications H3K79 hypermethylation of PBX3 was only observed in LSCs but not LSKs or GMPs, consistent with increased PBX3 expression level in LSCs compared with HSPCs. To further confirm this, we assessed H3K79me2 level of PBX3 in purified c-Kit+ cells which are enriched with stem cells from leukemic or control mice by ChIP-qPCR. Consistently, ChIP-qPCR results revealed the similar trend with dataset analysis. Moreover, we also detected the transcription repressive mark H3K9me3 on PBX3 in c-Kit+ cells and found a dramatic decrease of H3K9me3 in LSCs in comparison with HSPCs. Thus, our results suggested that high expression level of PBX3 in LSCs might be caused by elevated activation mark H3K79me2 and depressed repressive mark H3K27me3 and H3K9me3.

Considering that high expression level of PBX3 in LSCs may contribute to leukemic progression, we used the CRISPR/Cas9 system to delete PBX3 to examine its potential role in MLL-AF9 mediated AML development. We found that PBX3 inactivation significantly prolonged the survival of leukemic mice (median survival was 25 days of control, 31 days of SgPBX3-a, 30 days of SgPBX3-b, 10 mice per group). Morphological analysis revealed more leukemic blast cells with segment nuclei exhibited in PBX3 inactivation group, suggesting that PBX3 deficiency is associated with increased leukemic cell differentiation. Since LSC was considered to be sufficient to drive leukemogenesis and maintain leukemia progression, we investigated the impact of PBX3 inactivation on LSC capability. In vitro colony forming assay showed that PBX3 inactivation group formed significantly fewer colonies of total numbers and type A clones. Flow cytometric analysis showed decreased LSCs frequency in murine leukemia cells of PBX3 inactivation group. Moreover, limiting dilution transplantation assays revealed 4-7 fold decreased of functional LSCs after PBX3 inactivation with the 1/102 of LSCs in control leukemic bone marrow in comparison with 1/442 in SgPBX3-a and 1/741 in SgPBX3-b respectively. Further analysis suggested that depletion of functional LSCs in PBX3 inactivation group mainly through accelerating apoptosis of LSCs, evidenced by both increased Annexin V positive cell frequency of whole BM blasts and LSCs in PBX3 inactivation group.

In conclusion, we found PBX3 is epigenetically dis-regulated in LSCs of MLL-r AMLs and is essential for leukemia development. Inactivation of PBX3 could suppress leukemia progression by reducing LSCs frequency and impairing its function. More importantly, the differential expression of PBX3 in normal and malignant hematopoietic cells provides itself potentially as a prognostic marker and therapeutic target for MLL-rearranged leukemia.

HD.G and YJ.C contributed equally to this work


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.