Genome-Wide DNA Methylation Patterns Reveal Specific Signatures for HSC, CMP and Erythroblasts.

Abstract 391

DNA methylation is a reversible epigenetic modification that is required for proper mammalian development and is proposed to contribute to the pathogenesis of hematologic diseases including leukemia and bone marrow failure syndromes. Elucidating the pathways and genes regulated by DNA methylation during hematopoiesis may reveal new therapeutic targets for disease. Because the phenotype and activity of hematopoietic stem cells (HSC) and hematopoietic progenitor cells of many different lineages have been defined by both in vitro and in vivo assays, hematopoiesis is an excellent model for investigating epigenomic changes during differentiation. HSCs have the ability to self-renew and to generate blood cells of all lineages, which allows them to repopulate recipients after stem cell transplantation. The common myeloid progenitor (CMP) gives rise to all myeloid cell types including neutrophils, monocytes, platelets, and red blood cells, but cannot self renew or repopulate. In contrast to the multipotent HSC and CMP, erythroblasts (ERY) are terminally committed cells that become mature enucleated red blood cells. These three cell types represent unique stages of lineage commitment with distinct transcriptional programs, and potentially unique epigenomic signatures. In contrast to human HSC, which are defined by the absence of several cell surface markers, mouse HSC have the cell surface phenotype of lineage marker negative (Lin-) c-kit+ Sca-1+ and can be positively selected. For this reason we chose the mouse model for genome-wide methylation profiling. Murine HSC and CMP (Lin- c-kit+ Sca-1-) cells were enriched from adult mouse bone marrow with flow cytometry. Erythroblasts (CD71+/Ter119+) were positively selected from E13.5 mouse fetal livers. Genomic DNA isolated from each enriched cell population was sheared to 200-300 bp fragments. MBD2, one of five endogenous mammalian methyl CpG binding domain proteins, binds methylated DNA sequences with broad affinity. Methylated DNA fragments were enriched from the genomic DNA using a tagged, recombinant MBD2 pulldown kit (Active Motif). After pulldown, enrichment of known methylated sequences regulating the imprints of Snrpn and Rasgrf was validated by qPCR. Two biological replicates of HSC, CMP, and ERY methylated sequences and negative control supernatant fractions were submitted for high-throughput sequencing with the Illumina Genome Analyzer platform. Raw sequence data containing 32-46 × 10⁶ reads of 36-50 base pairs were obtained for each sample. The Eland program was used to map 41-59% of reads to unique sequences in the mouse genome. Model-based Analysis of ChIP-Seq (MACS) was used to estimate the mean and variance of the sequence tag distribution across the genome and define peaks below the significance threshold of p<10^-5. The number of methylation peaks decreased as cells differentiated, with 64,000 peaks identified in HSC (24,000 unique), 41,000 peaks in CMP (2000 unique), and 23,000 peaks in ERY (1000 unique). Approximately 20,000 peaks were common between all cell types with 57% of these peaks residing in RefSeq genes, 8% in regions adjacent to RefSeq genes (<10 kb), and 35% of methylation peaks in intergenic regions. Comparison of HSC expression data from Akashi et al (Blood 101: 383, 2003) to our HSC genic methylation peaks revealed that 2/3 of HSC genic peaks are within transcriptionally silent genes while 1/3 of HSC genic peaks are within expressed genes. Although DNA methylation is often associated with gene silencing, the important developmental gene Gata2 contains methylation peaks in HSC and CMP, cells that express Gata2, that are absent in ERY, where Gata2 is repressed. A Gata1-Fog1-Mbd2 complex has been described by Rodriguez et al (EMBO 24: 2354, 2005), therefore providing a link between DNA methylation and proteins known to bind at the Gata2 locus. Grass et al (Mol. Cell. Biol. 26:7056, 2006) determined that Gata2 is regulated by long-range interactions of GATA protein complexes, and consistent with this observation, distinct methylation patterns are observed up to 100 kb upstream of the Gata2 gene. Our genome-wide analysis supports an association of methylation with gene silencing but also suggests that DNA methylation is a dynamic epigenetic mark that influences hematopoietic differentiation. The changes in DNA methylation we observe around Gata2 may also contribute to long-range chromatin organization.