Ezh2 Loss Promotes the Transformation of Early T Cell Precursor Leukemia Via Suppressing Critical Genes for T-Cell Differentiation

Early T cell precursor (ETP) acute lymphoblastic leukemia (ALL) has been identified as a new pathologic entity with poor outcome in patients with T-ALL.  In contrast to cortical T-ALL, ETP-ALL has been characterized by the activating mutations in genes regulating cytokine signaling and the inactivating mutations in the polycomb repressor complex 2 (PRC2) component genes including EZH2.  EZH2 catalyzes trimethylation of histone H3 at lysine 27 (H3K27me3) that mediates silencing of critical genes for cell proliferation and differentiation.  Given that loss-of-function mutations in EZH2 are often found in ETP-ALL, EZH2 may play a tumor suppressor role in the pathogenesis of ETP-ALL.  To determine how EZH2 dysfunction promotes the development of ETP-ALL in vivo, we generated a novel mouse model of ETP-ALL by utilizing Ezh2 and p53 conditional knockout mice.  We harvested bone marrow cells from Cre-ERT (WT), Cre-ERT;Ezh2^flox/flox (Ezh2 KO), Cre-ERT;Trp53 ^flox/flox (p53 KO) and Cre-ERT;Trp53 ^flox/flox;Ezh2^flox/flox (DKO) mice, and transplanted them into lethally-irradiated Ly5.1⁺ recipient mice.  We deleted p53 and Ezh2 via administration of tamoxifen 4 weeks post-transplantation.  We observed that p53 KO mice died dominantly due to CD3⁺CD8⁺ cortical T-ALL (median survival; 137 days), as previously reported.  While 2 out of 8 DKO mice developed CD8⁺ cortical T-ALL, we found that 6 out of 8 DKO mice developed CD4^-CD8^- (DN) T-ALL with a longer latency (median survival; 189 days).  At the time of sacrifice, DKO mice showed significantly enlarged thymus due to the expansion of c-Kit⁺CD44⁺CD25^- DN1 and c-Kit⁺CD44⁺CD25⁺ DN2 cells, which were markedly compromised in differentiation into DN3 and further down to CD4⁺CD8⁺ (DP) cells.  Histological analysis demonstrated the expansion of cytoplasmic CD3⁺ (cyCD3) tumor cells in the thymus of DKO mice with DN T-ALL.  Thus, DKO mice faithfully recapitulated the phenotypic features of human ETP-ALL, indicating that the loss of Ezh2 is prerequisite to the development of c-Kit⁺CD44⁺CD25⁺cyCD3⁺DN ALL in mice lacking p53.

To understand the underlying molecular mechanism in the pathogenesis of Ezh2 loss-induced ETP-ALL, we performed gene expression analysis in DN1, DN2 and DN3 cells isolated from WT and DKO mice at pre-disease and ETP-ALL stages.  Since NOTCH1 activating mutations are not frequently found in patients with ETP-ALL, we observed no significant difference in activation of Notch1-target genes between WT and DKO ETP-ALL cells.  In contrast, Ras-target genes were significantly activated in DKO ETP-ALL DN2 cells relative to WT DN2 cells.  Thus, our ETP-ALL mouse model shows active Ras signaling but lacks Notch1 activation, consistent with the molecular features of human ETP-ALL.  Furthermore, we found that the genes critical for T-cell commitment, including Tcf7 or Bcl11b, were transcriptionally repressed in DKO ETP-ALL DN2 cells compared to WT DN2 cells, while the signature genes of both HSCs and myeloid cells were retained in DKO ETP-ALL DN2 cells.  Taken together, Ezh2 and p53 deletions cooperate to activate the function of HSCs and impede the transcriptional program of T-cell differentiation at the DN2 stage with sustaining myeloid potential.  To determine how H3K27me3 modification contributed to induce ETP-ALL in the absence of Ezh2, we performed H3K27me3-chromatin immunoprecipitation (ChIP) sequencing in WT and DKO ETP-ALL DN1/2 cells.  We found that H3K27me3 marks were lost or kept at low levels at the promoter regions of T-cell differentiation regulators in ETP-ALL cells.  Thus, we sought to determine whether altered DNA hypermethylation contributed to silencing the expression of T-cell differentiation regulators.  The transduction of either Tcf7 or Bcl11b expression alone was not sufficient to induce differentiation of DKO DN1/2 cells in vitro, however, we found that treatment of decitabine, a demethylating agent, clearly induced the differentiation of DKO DN1/2 cells beyond the DN3 stage in vitro, implying that Ezh2 loss and p53 loss cooperatively induced aberrant DNA hypermethylation, thereby impeding the differentiation of DN1/2 cells.  In conclusion, we demonstrated that combined deletion of Ezh2 and p53 altered the epigenetic regulation to an extent not seen in either deletion alone, and induced highly penetrant ETP-ALL characterized by the molecular profile similar to that in patients with ETP-ALL harboring mutations in the PRC2 components.