Abstract

During hematopoiesis, all-trans-retinoic acid (ATRA), a natural derivative of vitamin A, has been shown to induce both myelomonocytic progenitor/stem cell differentiation and self-renewal. Although these opposing effects are likely to be partly due to developmental differences, it has been shown that pro- and anti-differentiation effects of ATRA are mediated by distinct retinoic acid receptor isotypes (RARαa and RARγ, respectively). With the exception of acute promyelocytic leukemia (APL) ATRA treatment as a single agent has not been successful in other types of acute myeloid leukemia (AML). We have hypothesized that one of the underlying reasons for poor response of non-APL AML to ATRA (pan-RAR agonist) is aberrant expression and/or activities of RAR isotypes favoring RARγ and cell growth versus differentiation. Consistently, we have reported that expression of RARαa isoforms, particularly ATRA-inducible RARαa2, are down-regulated in AML (Blood 2008; 111:2374). Epigenetic analysis of patient samples revealed that relative to normal CD33+ cells, the loss of RARαa2 in AML is associated with a diminution in histone H3K4me2 and an increase H3K27me3 on the RARA2 promoter (modifications associated with transcriptional activation and silencing, respectively). Interestingly, H3K4 demethylase LSD1 (AOF2) and the polycomb represive complex 2 (PCR2)-associated H3K27 methyltransferase EZH2 are highly expressed in AML (www.proteinatlas.org). Small molecules that target these enzymes are in development and, given the above results, we predict that the use of such agents in combination with ATRA will enhance the effects of ATRA-mediated induction of gene expression and differentiation of AML cells. To test this hypothesis, we used ATRA-responsive HL-60 AML cells and the TEX cell line. TEX cells are derived from primitive human cord blood cells immortalized by expression of the TLS-ERG oncogene. These cells, the ATRA-responsiveness of which is not known, mimic features of primary human AML and leukemia initiating cells (Leukemia. 2005; 19:1794). LSD1 activity was inhibited using monoaminoxidase inhibitor (MAOI) trans-2-phenylcyclopropylamine (Parnate, 1μM) in combination with pharmacological (1μM) and sub-optimal (0.1μM) concentrations of ATRA. Co-treatment with Parnate potentiated the HL-60 response to sub-optimal ATRA concentration. While ATRA appeared to be a less potent inducer of TEX cell differentiation, Parnate nevertheless enhanced their maturation at pharmacological ATRA concentrations and sensitized these cells to differentiation induction under sub-optimal ATRA levels. Additionally, we investigated the biguanide polyamine analogue 1,15-bis[N5-[3,3-(diphenyl) propyl]-N1-biguanido]-4,12-diazapentadecane (2d), which is structurally unrelated to Parnate, obtaining similar results. Biguanide polyamine analogue inhibitors of LSD1 may have several benefits over MAOIs, including DNA targeting due their cationic nature. We also tested 3-deazaneplanocin A (DZNep), which diminishes levels of H3K27 trimethylation via depletion of the EZH2 catalytic subunit of the PCR2. Consistent with our hypothesis and the above data, co-treatment of HL-60 and TEX cells with DZNep (0.05μM) and ATRA (0.1μM and/or 1μM) led to more robust differentiation response than when ATRA was used as a single agent. The use of ATRA in combination with DZNep and LSD1 inhibitors at the same time led to a better differentiation response, as measured by CD11b/CD11c expression, morphology and superoxide production (NBT assay), than when either drug alone was used with ATRA. The effects of these drug combinations on AML cell maturation were paralleled by synergistic induction of endogenous ATRA target genes and expected changes in the levels of H3K4/K27 methylation. At the concentrations used with ATRA neither Parnate, 2d nor DZNep induced differentiation when used as single agents, however, when used at higher concentrations both singly and in combination with ATRA, these drugs exerted cytotoxic effects. Importantly, the above described combination treatments were specific for AML blasts as they had no cytotoxic effects on normal CD33+/CD34+ cell populations. These data demonstrate existence of therapeutically relevant crosstalks between the ATRA-induced differentiation pathway and histone H3K4 and K27 methylation and that targeting LSD1 and/or EZH2 in combination with ATRA may represent a promising treatment for AML.