Effects of the DNA Methylase Inhibitor Zebularine on Global Gene Expression in AML1-ETO-Positive Acute Myeloid Leukemia.

In acute myeloid leukemias (AML) containing the t(8;21)(q22;q22) translocation, the RUNX1 gene, encoding the transcription factor AML1, is fused to RUNX1T1 (ETO), resulting in a protein that retains the DNA binding domain of AML1 and, via ETO-mediated recruitment of corepressor complexes, represses AML1-dependent gene transcription. Because the recruited corepressor complex contains an active histone deacetylase (HDAC), and HDAC-mediated chromatin modifications are dynamically linked to DNA methylation, we were interested in defining the scope of epigenetic gene silencing in AML1-ETO-positive leukemias. To assess this, we examined the effects of zebularine, a newer DNA methylase inhibitor, on global gene expression in the AML1-ETO-positive AML cell line, SKNO-1. In a first step, we demonstrated the efficacy of zebularine treatment by confirming upregulation of CDKN2B, a gene known to be repressed by promoter methylation in AML1-ETO-positive AML. Likewise, the degree of methylation of the AKAP12 promoter, as measured by bisulfite sequencing, was reduced following treatment. We also examined the toxicity of zebularine in SKNO-1 and observed dose-dependent growth inhibition of cultures, owing to an increase in apoptosis (annexin V positivity) and decrease in viability (propidium iodide staining). Zebularine did not induce differentiation (CD11b expression) or alter the cell cycle (DNA analysis by flow cytometry). Using Affymetrix U133A arrays, we next generated triplicate genome-wide expression profiles of untreated or zebularine-treated SKNO-1 blasts. In parallel, we also obtained profiles of SKNO-1 cells treated with 5-aza-deoxycytidine (5-aza-dC), as well as expression profiles of a second AML1-ETO-positive leukemic cell line, Kasumi-1^SJ, treated with zebularine. Surprisingly, zebularine-induced changes in gene expression were distinct from the expression changes induced by 5-aza-dC, indicating significant differences in the effects of these putative demethylating agents on leukemia cells.To identify a signature of genes that might be epigenetically repressed in AML1-ETO-positive leukemias, we compared the untreated to treated replicate cultures within each cell line/drug combination using statistical approaches. Again, the lists of selected transcripts showed little overlap between zebularine and 5-aza-dC. The genes induced following drug treatment included a cell cycle regulator (CCND3), developmental transcription regulators (ID2, HOXA5), a membrane protein possibly involved in myeloma progression (LAPTM5), and genes functionally related to apoptosis (LGALS1, TNFSF10, SIPA1), protein /nucleic acid catabolism (PSMC3, RNASE2) or cellular defense mechanisms (EPHX1, TYROBP, TRGV9). We assessed the methylation status of the identified 5′ CpG islands of CCND3, ID2, HOXA5 and an intronic regulatory region of LAPTM5 in SKNO-1 cells, primary blasts from pediatric AML patients, or normal peripheral blood leukocytes. Remarkably, no evidence of aberrant methylation was found. From these data we conclude:

1. The cellular effects of zebularine and 5-aza-dC are more divergent than previously appreciated.

2. The observed changes in expression of CCND3, ID2, HOXA5 and LAPTM5 are likely due to methylation processes in yet undefined regulatory regions, or are the result of other transcriptional regulatory processes unrelated to changes in the methylation status of the transcriptional regulatory sequences of these genes.

3. In the two cell lines examined, a common AML1-ETO-dependent methylation “signature” does not exist.