Differences Between Normal and Leukemic Stem Cell-Specific Methylome Indicates Aberrantly Silenced Genes Involved in the Pathogenesis of Malignant Evolution

Due to technical limitations of methods used to study aberrant DNA hypermethylation, only a small numbers of promoters of empirically-selected genes have been investigated historically. Nevertheless, these studies demonstrated that aberrant methylation constitutes an important pathophysiologic mechanism of malignant evolution in MDS and AML. Novel methods utilizing genomic arrays allow for analysis of very large number of CpG islands and establishment of disease or tissues specific methylomes. In this study we hypothesized that a concordant methylation patterns exist that characterize hematopoietic progenitor and stem cells and genome-wide analysis of methylation patterns will allow for identification of a stem cell methylome that is distinct from that found in leukemia. To investigate this, chromatin immunoprecipitation promoter array-based profiling (ChIP-DSL 20,000 promoter array, Aviva Systems Biology) was used to identify genomewide methylated promoters in CD34 cells. We first sought to identify the normal stem cell methylome, using a pooled set of purified CD34 cells. We identified a relatively small set of 534 hypermethylated promoters (2.7% of genes analyzed with a cutoff of 2.6-fold methylation enrichment). Cytoskeleton remodeling, transcription regulation and nucleotide catabolism genes were significantly overrepresented (p=4.2E⁻⁵, p=2.2E⁻⁴, P=3.6E⁻³, Benjamini correction [BC] p=.008; p=.016; p=.069, respectively, using the DAVID database). When pooled CD34 cells were treated with decitabine; hypomethylation of 195 previously methylated genes (37%) was observed. The hypomethylated genes primarily functioned in basic metabolic processes (p=1.1E⁻⁴) and DNA repair/replication (p=3.0E⁻⁵). Subsequently, we analyzed triplicate samples of both immortalized AML cell lines (UT7, TF1, U937, HL60, KG1, K562 and Kazumi) and primary AML cells. Globally, the prevalence of hypermethylated sites in AML was infrequent: on average 2.8% or 552±181 of CpG islands were hypermethylated in cell lines compared to CD34 cells. Only 2431 (12%) promoters were hypermethylated in any AML cell line. Functional classification of hypermethylated genes showed that these genes were significantly enriched for cell development/differentiation (p=4.1E⁻⁵, BC p=.069), transcription regulation (p=4.8E⁻⁵, BC p=.027) and apoptosis (p=.01, BC p=.78) The methylation signature across several AML cell lines was variable: 115 genes were concordantly hypermethylated across the 7 cell lines and showed preponderance of transcription factor promoters (p=.0073) ); however, a functionally cancer-specific pattern was not obvious, suggesting that the establishment of phenotype in individual AML cell lines results from unique methylation events. Decitabine treatment of Kasumi cells resulted in hypomethylation of 75 genes involved in terminal differentiation (p=8.1E⁻⁴), transcription regulation (p=1.5E⁻²) and proliferation (p=4.3E⁻²). Analysis of the methylation pattern in AML patients produced similar results; 117 genes were concordantly hypermethylated in at least 60% of AML patients and were involved in mRNA transcription (p=.00073), leukocyte regulation (p=.007) and the Wnt signaling pathway (p=.01). Concordant hypermethylation of genes within Wnt pathways suggests involvement in leukemic evolution. In conclusion, the AML methylome is characterized by a highly variable hypermethylation of genes with consistent cellular functions including transcription, cell differentiation, apoptosis and leukocyte regulation.