Abstract 211

The hematopoietic system is ideal for the study of epigenetic changes in primary cells because hematopoietic cells representing distinct stages of hematopoiesis can be enriched and isolated by differences in surface marker expression. DNA methylation is an essential epigenetic mark that is required for normal development. Conditional knockout of the DNA methyltransferase enzymes in the mouse hematopoietic compartment have revealed that methylation is critical for long-term renewal and lineage differentiation of hematopoietic stem cells (Broske et al 2009, Trowbridge el al 2009). To better understand the role of DNA methylation in self-renewal and differentiation of hematopoietic cells, we characterized genome-wide DNA methylation in primary cells representing three distinct stages of hematopoiesis. We isolated mouse hematopoietic stem cells (HSC; Lin- Sca-1+ c-kit+), common myeloid progenitor cells (CMP; Lin- Sca-1- c-kit+), and erythroblasts (ERY; CD71+ Ter119+). Methyl Binding Domain Protein 2 (MBD2) is an endogenous reader of DNA methylation that recognizes DNA with a high concentration of methylated CpG residues. Recombinant MBD2 enrichment of DNA followed by massively-parallel sequencing was used to map and compare genome-wide DNA methylation patterns in HSC, CMP and ERY. Two biological replicates were sequenced for each cell type with total read counts ranging from 32,309,435–46,763,977. Model-based analysis of ChIP Seq (MACS) with a significance cutoff of p<10−5 was used to determine statistically significant peaks of methylation in each replicate. Globally, the number of methylation peaks was highest in HSC (85,797peaks), lower in CMP (50,638 peaks), and lowest in ERY (27,839 peaks). Comparison of the peaks in HSC, CMP and ERY revealed that only 2% of the peaks in CMP or ERY are absent in HSC indicating that the vast majority of methylation in HSC is lost during differentiation. Comparison of methylation with genomic features revealed that CpG islands associated with promoters are hypomethylated, while many non-promoter CpG islands are methylated. Furthermore, methylation of non-promoter associated CpG islands occurs infrequently in cell-type specific peaks but is more abundant in common methylation peaks. When the DNA methylation patterns were compared to mRNA expression, we found that as expected, proximal promoter sequences of expressed genes were hypomethylated in all three cell types, while methylation in the gene body positively correlated with gene expression in HSC and CMP. Utilizing de novo motif discovery we found a subset of transcription factor consensus binding motifs that were overrepresented in methylated sequences. Motifs for several ETS transcription factors, including GABPalpha and ELF1 were found to be overrepresented in cell-type specific as well as common methylated regions. Other transcription factor consensus sites, such as the NFAT factors involved in T-cell activation, were specifically overrepresented in the methylated promoter regions of CMP and ERY. Comparison of our methylation data with the occupancy of hematopoietic transcription factors in the HPC7 cell line, which is similar to CMP (Wilson et al 2010), revealed a significant anti-correlation between DNA methylation and the binding of Fli1, Lmo2, Lyl1, Runx1, and Scl.

Our genome-wide survey provides new insights into the role of DNA methylation in hematopoiesis. Firstly, the methylation of CpG islands is associated with the most primitive hematopoietic cells and is unlikely to drive hematopoietic differentiation. We feel that the elevated genome-wide DNA methylation in HSC compared to CMP and ERY, combined with the positive association between gene body methylation and gene expression demonstrates that DNA methylation is a mark of cellular plasticity in HSC. Finally, the finding that transcription factor binding sites are over represented in the methylated sequences of the genome leads us to conclude that DNA methylation modulates key hematopoietic transcription factor programs that regulate hematopoiesis.

Disclosures:

No relevant conflicts of interest to declare.

Author notes

*

Asterisk with author names denotes non-ASH members.

Sign in via your Institution