AbstractAbstract 2083In the mammalian genome cytosine residues that are followed by guanine (5‘-CpG-3‘ dinucleotides) are frequently methylated, a modification that is associated with transcriptional silencing. Two genome-wide waves of demethylation, in primordial germ cells and in the early pre-implantation embryo, erase methylation marks and are each followed by de novo methylation, setting up a pattern subsequently inherited throughout development [1]. While no global methylation changes are thought to occur during further somatic development, methylation does alter at gene-specific loci, contributing to tissue-specific patterns of gene expression.We set out to study dynamic changes in DNA methylation during erythropoiesis. We used flow cytometry and the cell surface markers CD71 and Ter119 to subdivide freshly isolated fetal liver cells into a developmental sequence of six subsets, from the least mature Subset 0 (S0), to the most mature Subset 5 (S5) [2]. We measured DNA methylation in genomic DNA prepared from freshly sorted S0 to S5 cells. Surprisingly, we found that demethylation at the erythroid-specific β-globin locus control region (LCR) was coincident with progressive genome-wide methylation loss. Both global demethylation as well as demethylation at the β-globin LCR began with the upregulation of CD71 at the onset of erythroid terminal differentiation, and continued with erythroid maturation, with global hypomethylation persisting during enucleation.We employed several distinct methodologies to measure global DNA methylation level. Using Enzyme-Linked Immunosorbent Assay (ELISA), we found that genomic DNA isolated from increasingly mature erythroblasts had progressively reduced binding to a 5-methylcytosine-specific antibody. We also used the LUminometric Methylation Assay (LUMA) to compare the genome-wide cleavage of CCGG sites by each of the isoschizomers HpaII and MspI, which are methylation sensitive and insensitive, respectively. Both the ELISA and LUMA assays showed a global, progressive and significant loss of DNA methylation with erythroid differentiation: 70% of CpG dinucleotides genome-wide were methylated in S0, decreasing to 40–50% by S4/5 (p<0.01). Further, using pyrosequencing of bisulfite-converted DNA, we found a similar decrease in CpG methylation in the promoters of genes whose transcription is silenced with erythroid maturation, notably PU.1 and Fas.To characterize the global loss in methylation further, we examined the status of imprinted genes and of repetitive transposable elements, since both represent genetic loci that are usually stably and highly methylated in somatic cells. We found loss of methylation in imprinted loci, including PEG3 and the H19 Differentially Methylated Region (DMR). We also found a significant loss of methylation at the Long Interspersed Nuclear Element (LINE-1), a repetitive retrotransposon, whose methylation level decreased from over 90% in S0 cells, to 70% in S4/5.Mechanistically, global demethylation was associated with a rapid decline in the DNA methyltransferases DNMT3a and DNMT3b. However, exogenous re-expression of these enzymes in vitro was not sufficient to reverse the process. Both global and erythroid-specific demethylation required rapid DNA replication, triggered with the onset of erythroid terminal differentiation. We were able to slow down demethylation quantitatively by slowing down the rate of DNA replication with aphidicolin, an inhibitor of DNA polymerase α.Global loss of DNA methylation was not associated with a global increase in transcription, as determined by GeneChip analysis, nor was it associated with increased transcription of the LINE-1 retrotransposon. We propose that global demethylation is a consequence of global cellular mechanisms required for the rapid demethylation and induction of β-globin and other erythroid genes. Our findings suggest mechanisms of global demethylation in development and disease, and show that contrary to previously held dogma, DNA demethylation occurs globally during physiological somatic cell differentiation.