Inhibition of EZH2 Depletes MLL Fusion Leukemia Stem Cells Through Restoration of p16 Expression

Abstract 3510

Leukemia stem cells (LSCs) are resistant to conventional chemotherapy and persistent LSCs after chemotherapy are supposed to be a major cause of disease relapse or refractoriness. However, information on genetic or epigenetic regulation of stem cell properties is still limited and LSC-targeted drugs have scarcely been identified or used in clinical settings so far. Epigenetic regulators are associated with many cellular processes such as cell cycle, proliferation, and apoptosis. Of note are polycomb group proteins, because they potentially control stemness including activity of cancer stem cells, and can be pharmacologically targeted by a selective inhibitor of H3K27, 3-deazaneplanocin A (DZNep).

We first administrated DZNep to MLL-related leukemia mouse model in order to test whether DZNep has potential to eradicate LSCs of the leukemic mice. Remarkably, the leukemic granulocyte-macrophage progenitors (LGMPs) in MLL/AF9 positive cells were significantly decreased in number by administration of DZNep while AraC did not affect the number of LGMPs, which implied that LSCs were targeted by DZNep. These data were reproduced by transplantation assays using short hairpin RNA (shRNA)-mediated knockdown of EZH2, a major component of polycomb repressive complex 2 (PRC2) which is responsible for H3K27 tri-methylation. Significantly, DZNep administration to wild-type mice led to only mild suppression of hematopoiesis, suggesting that this agent spares normal hematopoietic stem cells while eliminating LSCs, which is consistent with a previous report that genetic depletion of EZH2 did not compromise adult hematopoiesis in mice. Serial replating assay of MLL/AF9-induced leukemia cells showed that DZNep treatment in vivo diminished their colony forming capacity. Limiting dilution transplantation assays revealed that frequency of LSCs was markedly reduced by DZNep administration. DZNep treatment or EZH2 knockdown significantly prolonged survival of MLL/AF9 and MLL/ENL leukemic mice. To elucidate a molecular mechanism underlying the effects of DZNep on LSCs, we investigated transcriptional or epigenetic changes during DZNep treatment and EZH2 knockdown. Gene expression profiling revealed that p16 was significantly upregulated by EZH2 knockdown or DZNep administration. Knockdown of p16 completely canceled the survival advantage of the leukemia mice which received DZNep in vivo and restored the colony forming capacity of leukemia cells transduced with shRNA for EZH2 in vitro. These results supported the idea that p16 upregulation derived from EZH2 attenuation is central to the LSC reduction. Next, we investigated epigenetic status around p16 promoter and transcription start site (TSS) by chromatin immunoprecipitation (ChIP) assays. In MLL/ENL leukemia cells, both H3K4 and H3K27 methylation marks were highly enriched around the TSS of p16, together with EZH2 and Bmi1, a component of PRC1. Therefore removal of EZH2 is supposed to convert the promoter of p16 from a bivalent to an active state. The results of ChIP assays also indicated that MLL/ENL fusion protein binds to p16 coding region. In order to clarify whether dependency on EZH2 is specific for MLL fusion leukemia or can be applied for other types of leukemia, we evaluated the consequence of EZH2 inhibition in several types of leukemia. DZNep or shRNA for EZH2 strongly suppressed the proliferation of leukemia cell lines and immortalized cells harboring MLL fusion genes with high specificity. Administration of DZNep or transduction of shRNA targeting EZH2 significantly prolonged survival of MLL/AF9 and MLL/ENL-induced leukemia mice while TEL/PDGFRA-AML1/ETO-induced leukemia was not sensitive to DZNep, although bone marrow (BM) cells from either mice became globally hypo-methylated on H3K27 by exposure to this drug. Serial replating assay with DZNep or EZH2-shRNA demonstrated high sensitivity to EZH2 inhibition of MLL/AF9-transduced BM cells but not of AML1/ETO-transduced BM cells, E2A/HLF-transduced BM cells, or normal c-kit+ BM cells. Thus, the anti-leukemia effect of EZH2 inhibition is thought to be specific for MLL fusion leukemia.

Collectively, our findings indicate that EZH2 is a potential therapeutic target of LSCs of MLL fusion leukemia to overcome the poor prognosis, encouraging the development of inhibitors against EZH2 with high specificity.