KDM4 inhibitor SD49-7 attenuates leukemia stem cell via KDM4A/MDM2/p21CIP1 axis

Abstract

Rationale: Traditional treatments for leukemia fail to address stem cell drug resistance characterized by epigenetic mediators such as histone lysine-specific demethylase 4 (KDM4). The KDM4 family, which acts as epigenetic regulators inducing histone demethylation during the development and progression of leukemia, lacks specific molecular inhibitors. Methods: The KDM4 inhibitor, SD49-7, was synthesized and purified based on acyl hydrazone Schiff base. The interaction between SD49-7 and KDM4s was monitored in vitro by surface plasma resonance (SPR). In vitro and in vivo biological function experiments were performed to analyze apoptosis, colony-formation, proliferation, differentiation, and cell cycle in cell sub-lines and mice. Molecular mechanisms were demonstrated by RNA-seq, ChIP-seq, RT-qPCR and Western blotting. Results: We found significantly high KDM4A expression levels in several human leukemia subtypes. The knockdown of KDM4s inhibited leukemogenesis in the MLL-AF9 leukemia mouse model but did not affect the survival of normal human hematopoietic cells. We identified SD49-7 as a selective KDM4 inhibitor that impaired the progression of leukemia stem cells (LSCs) in vitro. SD49-7 suppressed leukemia development in the mouse model and patient-derived xenograft model of leukemia. Depletion of KDM4s activated the apoptosis signaling pathway by suppressing MDM2 expression via modulating H3K9me3 levels on the MDM2 promoter region. Conclusion: Our study demonstrates a unique KDM4 inhibitor for LSCs to overcome the resistance to traditional treatment and offers KDM4 inhibition as a promising strategy for resistant leukemia therapy.

Keywords: Leukemia stem cells; Lysine-specific demethylase 4; Small molecular compounds.

Figure 1

KDM4 is a key regulator in leukemia progression. (A) Relative mRNA expression levels of KDM4A, KDM4B, and KDM4C in PMN-BM, AML1-ETO, APL, AML (16)/t(16:16), and AML t(11q23)/MLL. Data were from HemaExplorer (n = 3 in PMN-BM, 39 in AML1-ETO, 37 in APL, 28 in AML (16)/t(16:16), and 38 in AML t(11q23)/MLL, respectively). Data were normalized to PMN-BM. (B) mRNA expression levels of KDM4A, KDM4B, and KDM4C in umbilical cord blood and AML patients' bone marrow mononuclear cells (n = 15). (C) Workflow of the MLL-AF9 mode- related experimental design. (D) MLL-AF9 leukemia cell proliferation in shLuc and shKDM4A groups by cell counting (n = 3). (E) Apoptosis of MLL-AF9 leukemia cells in shLuc and shKDM4A groups (n = 3). (F) Colony numbers (left) and representative microscopy images of colony formation (right) in shLuc and shKDM4A groups (n = 3). 500 MLL-AF9 leukemia cells per well were seeded into a 24-well plate. Scale bar = 100 µm. (G) Percentage (middle panel) and absolute (right panel) numbers of c-Kit⁺Gr-1^- cell population of shLuc and shKDM4A MLL-AF9 leukemia cells (n = 3). Representative flow cytometric results are shown in the left panel. (H) Percentage of GFP⁺ cells in peripheral blood (PB) on days 4, 8, 12, and 16 was detected by flow cytometry in the ex vivo translation model (right panel, n = 3). Representative flow cytometric results are shown in the left panel. (I) Percentage of GFP⁺ cells in bone marrow (BM) was detected by flow cytometry in the ex vivo translation model (right panel, n = 6). Representative flow cytometric results are shown in the left panel. (J) Apoptosis of CD34⁺ UCB cells in shLuc, shKDM4A, and shKDM4C groups (left panel, n = 3). Representative flow cytometric results are shown in the left panel. All experiments were repeated three times. ns. no significance, *p < 0.05, **p < 0.01, ***p < 0.001, by unpaired Student's t-test, error bars denote mean ± SD.

Figure 2

Specificity of a small molecule inhibitor SD49-7 of KDM4s. (A) Structure of SD49-7. (B) KD value of SD49-7 or SD70 interaction with KDM4A or KDM4C, respectively, as detected by SPR assay. (C) SPR results of the binding activity of KDM4A or KDM4C at various concentrations of SD49-7 or SD70. (D) Western blot analysis of the levels of tri- or di- methylation of H3K9 or H3K36 regulated by SD49-7 in THP-1 (left panel) and MLL-AF9 GFP⁺ cells (right panel, SD49-7 concentration: 1 µM). β-actin was used as a loading control, and the experiment was repeated three times. (E) Effect of SD79-7 (1 µM) and SD70 (1.75 µM) on demethylase kinetics was evaluated by JMJD2 demethylase activity in vitro. THP-1 cell nuclear extract without inhibitors was used as a control. (F) Apoptosis of THP-1 cells (left panel) and MLL-AF9 leukemia cells (right panel) triggered by SD49-7 treatment for 24 h when KDM4A was silenced by shRNAs (n = 3). Data were normalized to the control shLuc or shKDM4A group. ns. no significance, *p < 0.05, **p < 0.01, by unpaired Student's t-test, error bars denote mean ± SD.

Figure 3

Pharmacological inhibition of KDM4s suppresses leukemia development. (A) Apoptosis of MLL-AF9 leukemia cells triggered by SD49-7 treatment for 24 h (n = 3). Data were normalized to the control. (B) Colony numbers of MLL-AF9 leukemia cells following 24 h pre-treatment with SD49-7 (1 µM) or SD70 (1.75 µM). 625 cells were seeded per well in a 24-wellplate after the treatment (n = 3). (C) Percentage (middle panel) and absolute (right panel) numbers of c-Kit⁺Gr-1^- cell population in MLL-AF9 leukemia cells after treatment for 24 h with SD49-7 (1 µM) or SD70 (1.75 µM), n = 3. Representative flow cytometric results are shown in the left panel. (D) MLL-AF9 leukemia cell proliferation measured by cell counting after SD49-7 (1 µM) or SD70 (1.75 µM) treatment for 24 h (n = 10). (E) Survival plot representing the percentage of surviving mice injected with MLL-AF9 leukemia cells treated with 1 µM SD49-7 or 1.75 µM SD70 (n = 10). (F) Limiting dilution transplantation of MLL-AF9 leukemia cells following 24-h treatment with SD49-7 (1 µM) or SD70 (1.75 µM). Cell dose: 50,000, 5,000, 500, 50, n = 8. ns. no significance *p < 0.05 by chi-square test. (G) Workflow of the PDX model-related experimental design. (H) Percentage of hCD45⁺ cells (right panel) in PB of the PDX model after week 4, 6, and 8 detected by flow cytometry (n = 6). Representative flow cytometric results at week 8 are shown in the left panel. (I) Percentage of hCD45⁺ cell (right panel) in BM were detected by flow cytometry in the PDX model (right panel, n = 6). Representative flow cytometric results are shown in the left panel. (J) Apoptosis of CD34⁺CD38^- UCB cells treated with SD49-7 (0, 0.5, 1, or 2 µM) or SD70 (0, 0.8755, 1.75, or 3.5 µM) for 24 h (n = 3). Vehicle was used as a negative control. ns. no significance, *p < 0.05, **p < 0.01, by unpaired Student's t-test, error bars denote mean ± SD.

Figure 4

KDM4A triggers MDM2 and p21 in leukemia cells. (A) Volcano plot showing the genes up-regulated or down-regulated by SD49-7 in THP-1 cells by cDNA microarray. Cut off: fold-change > 2, p ≤ 0.05. (B) Enrichment of KEGG analysis of differentially expressed genes. (C) GSEA enrichment of the hallmark p53 pathway. (D) Fold-change of mRNA levels of p53 and its targets (p21, PUMA, SESN2, and DR5) mediated by 24 h SD49-7 treatment in THP-1 cells (n = 3). Data were normalized to the control. (E) Enrichment of the MDM2 promoter in THP-1 cells (left panel) and MLL-AF9 leukemia cells (right panel) after a 24 h treatment of 1 µM SD49-7 (n = 3). IgG was used as a negative control. (F) Fold-change of MDM2 mRNA levels in THP-1 cells and MLL-AF9 leukemia cells following 24-h treatment with 1 µM SD49-7 (n = 3). Data were normalized to the control. (G) Western blot analysis of MDM2, p53, and p21 protein levels in THP-1 cells and MLL-AF9 leukemia cells following 24-h treatment with SD49-7 (1 µM in MLL-AF9 GFP⁺ cells). β-actin was used as a loading control. Experiments were repeated three times. (H) Enrichment of the MDM2 promoter in THP-1 cells (left panel) and MLL-AF9 leukemia cells (right panel) when KDM4A was abolished with shRNAs (n = 3). IgG was used as a negative control. (I) Fold-change of MDM2 mRNA levels in THP-1 cells and MLL-AF9 leukemia cells when KDM4A was abolished by shRNAs (n = 3). Data were normalized to the control. (J) Western blot analysis of MDM2 and p21 protein levels in THP-1 cells and MLL-AF9 leukemia cells when KDM4A was abolished by shRNAs. β-actin was used as a loading control. Experiments were repeated three times. Vehicle was used as a negative control. ns. no significance, *p < 0.05, **p < 0.01, ***p < 0.001, by unpaired Student's t-test, error bars denote mean ± SD.