Targeting EZH1/2 to Restore Chemosensitivity in Dormant Myeloid Leukemia Stem Cells

Introduction: Treatment of acute myeloid leukemia (AML) in elderly patients has been challenging largely because of their intolerance to intensive therapy and development of therapy resistance. Epigenetic alterations, including aberrant DNA methylation and alterations in histone modifications, are frequently present in leukemia cells and associated with therapy resistance. Polycomb repressive complex 2 (PRC2) subunits, enhancer of zeste homolog (EZH) 1 and 2 (EZH1/2), play a critical role in gene silencing by methylation of histone H3 on Lys27 (H3K27me3). Knowing that deregulated H3K27me3 is linked to malignant hematopoiesis and PRC2 is required for AML cell survival (Basheer et al., 2019; Neff et al., 2012), we postulated that targeting both EZH1 and 2 concomitantly will be efficacious in disrupting oncogenic signaling. A recent study using a murine AML model has demonstrated that genetic deletion of EZH1/2 depleted quiescent leukemic stem cells (LSC) and prolonged survival of leukemia bearing mice (Fujita at al., 2018).

Hypothesis: Under the assumption that dormant LSC in the bone marrow are responsible for residual disease and subsequent relapse, we hypothesized that EZH1/2 inhibition promotes entry of LSC into the cell cycle restoring their chemosensitivity. To test this hypothesis, we utilized pharmacological inhibition of EZH1/2 by a potent and selective EZH1/2 dual inhibitor, valemetostat (DS-3201; Daiichi Sankyo).

Results: Previously, we demonstrated a remarkable increase in mobilized proliferating LSC from patients treated with valemetostat in a phase-I clinical trial (Chang et al., ASH 2020). In order to delineate the effects of EZH1/2 inhibition on AML LSC proliferation, we examined Ki67 positivity, DNA content, and p21 expression levels in primary LSC (CD45+CD34+CD38-CD123+) obtained from AML patients by flow cytometry. In vitro exposure to valemetostat increased Ki67 positivity (average increase: 62.2 % from 18.1 % to 29.5%, p = 0.03) and reduced p21 expression (average MFI reduction from 9203 ± 4402 to 6196 ±3915, p = 0.2) compared to vehicle treated controls, confirming that EZH1/2 inhibition can indeed recruit LSC effectively into cell cycle. At the molecular level, inhibition of EZH1/2 by valemetostat increased cMYC and Cyclin D expression along with decreasing the CDK inhibitory proteins, p27 and p18, indicating cell cycle progression and proliferation of primary LSC. Next, we tested if augmented cell cycle entry and LSC proliferation by EZH1/2 inhibition subsequently lead to improved drug sensitivity in AML. AML cell lines as well as primary AML samples were primed with valemetostat (24 hours and 72 hours, respectively) followed by 72-hour exposure of increasing concentrations of cytarabine (AraC), venetoclax (Ven), and azacytidine (5Aza), then subjected to flow cytometry analysis of 7AAD+/annexin V. Pharmacological inhibition of EZH1/2 significantly reduced IC50 values of all 3 drugs in AML cell lines (p=0.001, drug alone vs. drug+ valemetostat combination). Importantly, combinatorial treatment of EZH1/2 restored chemosensitivity of LSC not only to the anti-proliferative agent, AraC, but also to Ven (p<0.05) in primary AML samples. EZH1/2 mediated transcriptional changes in LSC analyzed by RNA-seq and single cell multi-omics will be reported at the ASH conference.

Conclusion: We have established that EZH1/2 is an important determinant of LSC proliferation and cellular sensitivity to anti-leukemia therapies. In previous research, we developed the FLAG-Ida protocol (Andreeff et al, 1990), based on granulocyte colony stimulating factor (G-CSF)-mediated recruitment of quiescent cells into the cell cycle and disruption of the SDF1-CXCR4 axis resulting in mobilization from the protective BM niche. FLAG-Ida is still considered one of the most efficacious protocols in AML therapy. Here we expect to expand the concept regarding proliferation of LSC through an alternative mechanism by epigenetic modification. We hypothesize that the combination of G-CSF with a pharmacological EZH1/2 inhibitor will result in further chemosensitization. We have ongoing in vivo experiments utilizing a PDX-AML murine model. Latest results will be presented at the ASH conference. If successful, this concept will further enhance the efficacy of FLAG-Ida chemotherapy and/or of BH-3 mimetic, apoptosis targeting therapies.