Abstract
Abstract 58
Histone demethylases are candidate therapeutic targets in cancer. We analysed a previously published microarray expression dataset of murine acute myeloid leukemias (AML) experimentally initiated with human MLL fusion oncogenes to identify histone demethylases whose expression correlated with the frequency of leukemia stem cells (LSCs). Only expression of Kdm1a correlated positively and significantly (p<0.001) with LSC frequency, a finding confirmed by quantitative PCR of selected leukemias. To investigate whether Kdm1a has a critical role in MLL LSCs, we targeted Kdm1a transcripts for knockdown in murine MLL-AF9 AML cells using lentivirally expressed shRNAs. Knockdown AML cells formed significantly fewer colonies in semi-solid culture compared with AML cells expressing a non-targeting control shRNA (12±7% of control; n=4; p=0.01), and the extent of Kdm1a knockdown correlated significantly (six shRNAs tested) with loss of AML cell colony forming (CFC) potential (p<0.005). Residual Kdm1a knockdown colonies predominantly exhibited “Type 2” morphology (i.e. containing macrophages), a finding confirmed by examination of cytospin preparations. The phenotype was rescued by forced expression of human KDM1A. Kdm1a knockdown MLL-AF9 AML cells were unable to initiate leukemia in secondary transplant assays, in contrast to control cells which readily induced short latency AML. These data demonstrate that Kdm1a is required to prevent differentiation of murine MLL-AF9 AML LSCs.
Tranylcypromine (TCP) is a licensed monoamine oxidase inhibitor that inhibits KDM1A, with an IC50 of 5–10uM. Treatment of murine MLL-AF9 AML cells, or human THP1 MLL-AF9 cells, with TCP phenocopied Kdm1a knockdown at a dose close to the KDM1A IC50; both murine and human MLL-AF9 cells exhibited reduced proliferation, fewer colonies and increased macrophage differentiation. Next, we tested the effect of KDM1A knockdown or TCP treatment (25uM) on primary human MLL leukemia cells. KDM1A knockdown reduced proliferation, induced morphological differentiation and induced apoptosis in the two samples of MLL-AF6 leukemia tested (5–20 fold fewer cells remained after a ten day culture versus control cells). Likewise, TCP treatment reduced proliferation 4-fold over five days (MLL-AF6 n=2; MLL-AF10 n=1), blocked CFC formation, reduced the proportion of blasts and increased the proportion of cells with features of monocyte/macrophage differentiation. Of note, addition of 25uM TCP to normal murine KIT+ BM stem and progenitor cells cultured in methylcellulose did not significantly reduce CFC frequency or size or induce macrophage differentiation, suggesting that MLL LSCs are more sensitive than normal BM cells to a given level of KDM1A inhibition.
To investigate the mechanism by which KDM1A inhibits a differentiation programme in MLL-AF9 LSCs, we performed exon array analysis of Kdm1a knockdown versus control LSCs 48 hours following lentiviral infection, prior to appearance of morphologically differentiated cells. Down regulated transcriptional regulators included Id2, Hmgb3 and Egr1, but not Hoxa, Meis1 or Myb. Up regulated transcriptional regulators included Irf8. Knockdown of Id2 in murine MLL-AF9 AML cells reduced CFC frequency. In contrast, forced expression of Id2 enhanced CFC frequency both of control MLL-AF9 AML cells and of Kdm1a knockdown LSCs, the latter representing a partial phenotypic rescue of Kdm1a knockdown.
These data demonstrate that KDM1A prevents differentiation of MLL LSCs by regulating the relative expression of pro- versus anti-differentiation transcriptional regulators and highlight it as a novel therapeutic target in MLL leukemias.
No relevant conflicts of interest to declare.
Author notes
Asterisk with author names denotes non-ASH members.