Epigenetic therapies are emerging as a promising therapeutic strategy for acute myeloid leukemia (AML), exemplified by advances in the development of inhibitors targeting DNMT3A, DOT1L and LSD1. We identified an essential role for the H3K9me3 histone demethylase, KDM4A, in maintaining AML cell survival with genetic depletion of KDM4A having no effect on normal hematopoiesis. Therefore, we hypothesise KDM4A inhibition may represent a novel and effective strategy to treat AML.
To address this, we developed a series of novel KDM4A inhibitors (KDM4Ai), based on the structure of pan inhibitor IOX1, and fully characterised their functional potential in AML cells representing major molecular subtypes, and primary patient blasts in comparison with healthy donor cells, as single agents or in combination with other anti-cancer drugs. To evaluate these compounds in physiological conditions, we utilised a stromal co-culture system mimicking the bone marrow microenvironment. Furthermore, we carried out global transcriptomic profiling by RNA-seq to elucidate the molecular consequences responsible for KDM4Ai induced leukemic killing.
As a mono-therapy, KDM4Ai induced leukemic cell differentiation and apoptosis in a broad spectrum of human AML cells, with an IC50 of 3.2µM ± 0.2 in MLL-AF9 driven THP1 cells after 48hr treatment (n=3), similar efficacy was observed in other human AML cell lines (n= ≤3) including Kasumi1 (2.69µM ± 0.1) , OCI-AML3 (4.9 µM ± 1.0) and MOLM13 (1.7 µM ± 0.7) and in primary patient blasts (3.8µM). A complete removal of colony forming potential was observed upon treatment (n = 2). The global expression of KDM4A's established substrate, H3K9me3 was upregulated by immunofluorescence and transcriptional changes in a 9-gene signature identified previously as direct KDM4A downstream targets, is indicative of an on-target effect by KDM4Ai. Importantly, KDM4Ai specifically reduced CD34+leukemic stem cell enriched population ( reduced by 6%). In contrast, a non significant reduction was observed on donor CD34+hematopoietic stem and progenitor cells proliferation and apoptosis suggesting a therapeutic window. Cytoprotection provided by stromal co-culture in both AML cell lines and primary samples resulted in a 50% decrease in apoptotic cells with maintenance of the CD34+compartment . Taking these results into account we identify importance of the microenvironment in drug mechanism and resistance.
To better understand the mechanism driving this selective anti-leukemic effect, we performed transcriptomic analysis on KDM4Ai treated THP1 cells (n=3). Corroborating the differentiation phenotype, pathway analysis showed an enrichment of IL4 & IL13 signalling (Enrichment score (ES) = - 0.53, q value = 0.025), and neutrophil degranulation (Enrichment score (ES) = - 0.51, q value = 0.025), this was accompanied by significant up-regulation of DNA damage response pathways (ES = 0.66, q value = 0.025) . These results were confirmed in AML cell lines, displaying accumulation > 20% of γH2AX by intracellular flow cytometry and PARP cleavage by western blot following treatment.
Based on these results we hypothesised that KDM4Ai may sensitise leukemic cells to DNA damage pathway inhibitors, such as PARP inhibitors (PARPi) (n = ≤3). While standard chemotherapies, such as cytarabine and azacitidine, in combination with KDM4Ai showed a largely additive effect, a dual inhibition of KDM4Ai at 3mM with 5mM olaparib (PARPi), exhibited a Combination Index of ~0.69 with a decrease in proliferation (>15% reduction) and increased apoptosis (>20%) in MLLr-AML cell lines compared with KDM4Ai alone (p=0.005). This effect was corroborated ex vivo using cells isolated from a patient derived xenograft model of MLL-AF10 (n=3). Taken together these results suggest a synergistic leukemic cell killing and have subsequently been subjected to global RNA-seq to confirm the detailed molecular mechanism underlying the synergistic effect.
Pharmacological inhibition of KDM4A using novel compounds effectively eliminated leukemic cells sparing normal hematopoietic cells with a synthetic lethality observed through combination with PARPi offering a promising therapeutic strategy in AML. Our data further support the essential role for KDM4A in myeloid oncogenesis, promoting future clinical evaluation of KDM4Ai its associated downstream targets as potential tractable therapeutic vulnerabilities in AML.
No relevant conflicts of interest to declare.
Author notes
Asterisk with author names denotes non-ASH members.