Comparing NSphs with different cell fates enabled the identification of numerous T-DMRs in genes at different relative positions from TSSs. DNA methylation and demethylation occurred in a developmental stage-dependent manner, and changes in DNA methylation at these T-DMRs resulted in variable methylation in AdBr cells that shifted the DNA methylation profile as a whole. The hypomethylated status of most NSph-T-DMRs was reflected in the DNA methylation profile of the AdBr to different degrees in a locus-specific manner. The previous genome-wide methylation analyses of NPCs
[16–18] emphasized preexisting epigenetic marks, such as bivalent histone modifications on poised genes involved in early differentiation processes and demethylated promoters of astrocyte-specific genes in progenitor cells preceding expression in differentiated cells. DNA methylation status in NSphs and gene expression in the AdBr have led to the hypothesis that a considerable number of T-DMRs identified in this study are epigenetically marked prior to gene expression. The developmental-stage specific DNA methylation marks could be useful for identify and evaluation of NPCs established from not only fetus but also stem cells as pluripotent stem cells and those from adult tissues.
We observed developmental position changes such as 5′ distal hypomethylated T-DMRs in the NSphs and hypomethylated T-DMR marks 3′ proximal downstream of TSSs in the fully developed brain. These T-DMRs were often located around CGIs, which is in contrast to a previous genome-wide analysis of NPCs indicating biased DNA methylation changes to low-CpG promoters
[17, 18]. T-DMRs found in the Ntng1 locus could be classified into the previously described class of T-DMRs downstream of TSSs of CGI genes, in which hypomethylation was well correlated with gene expression
[7, 20]. T-DMRs have been identified at the edges of CGIs
, and DMRs around CGIs, named as CpG island shores, are identified in not only normal tissues but also cancer cells
. The biased distribution of E14Hypo-T-DMRs to the relatively 5′-distal positions from TSSs suggested that hypomethylation of these 5′-distal T-DMRs in the progenitor cells are epigenetic marks that lead to expression in differentiated cells, which exhibit hypomethylation of T-DMRs at 3′ downstream of TSS.
Systematic biases are inevitable for any genome-wide DNA methylation analysis: both Microarray-based Integrated Analysis of Methylation by Isoschizomers (MIAMI)
 and Reduced Representation Bisulfite Sequencing (RRBS)
, methods used in the previous epigenomic study in NPCs, are inevitably focusing on CGIs because of the biased distribution of MspI recognition sites that they uses for enrichment of fragments
; Methylated DNA immunoprecipitation (MeDIP)
, is known to have bias to high density CpG promoter
. Approximately 50% of promoters are associated with CGIs. D-REAM also has
[7, 25]. Only limited numbers of genes are coincided to be predicted to have DMRs in NSph: Gfap, which have been shown to have DMR hypomethylated in E14.5NSph
, was not included in our gene list because of lacking HpyCH4IV site in the proximal promoter region: DMRs on Ntng1, which was identified in this study, is an example of novel T-DMRs not described in the previous studies.
Several converging lines of evidence have indicated the significance of DNA methylation in normal brain function. Mutations in Dnmt1, Dnmt3b, and Mecp2 result in functional and/or morphological abnormalities in human and mouse brain
[26–28]. Mutations in the human orthologs of some genes carrying NSph-T-DMR, such as LIG4 and NFIA, are associated with neurological disorders
[29, 30]. Similar phenotypes are observed in mice harboring mutations in these genes
[4, 31]. Targeted mutation of 2 genes with E11Hypo-T-DMRs, Dclk1 and Nrcam, results in axonal defects in mice
[32, 33]. Disorganized DNA methylation profiles have been reported in cloned animals
, chemically treated cells
, and in certain diseases
[36, 37]. Epimutations in tumor suppressor genes are involved in carcinogenesis
. Therefore, the establishment of DNA methylation status at T-DMRs in these genes indicates the possibility that epimutations at T-DMRs could be involved in neurological disorders without genetic alterations.