DNA methylation stability in mouse ES cells with ongoing passages (Px; Ndn probe a). (Left) DNA from IB10 ES cells (129/ola) were cleaved with BglII and with the methylation-sensitive restriction endonuclease EagI to study Ndn methylation. The ES cells were passaged in standard medium for 35 passages (∼90 days) without significant changes in methylation levels (∼15%). Methylation levels were assessed by relative hybridization intensities, as determined by scanning densitometry. (Right) Methylation analysis of RW4 ES cells (129/sv), which were cultured for 30 passages (∼80 days). The methylation patterns were stably transmitted and approximately the same as in the IB10 cell line, with ∼15% CpG methylation on the maternal allele (band on top of lanes).

The DNAs from differentiated ES cells show no significant changes in methylation levels of imprinted neuronal genes. For Southern blot analyses, 10 µg of genomic DNA from STS-induced ES cells (+) with neuronal morphology was cleaved with PvuII (for the Snurf/Snrpn DMR1 region) or with BglII (for the Ndn region) and then either with HpaII (H lanes), MspI (M lanes) or EagI (E lanes). (a) The Snurf/Snrpn DMR1 remained hypermethylated during STS-induced neurogenesis. (b) The hypomethylated EagI site within the Ndn gene (Ndn probe b) exhibited no significant methylation variation during in vitro neurogenesis. The HpaII restriction site remained completely unmethylated.

Epigenetic instability of the neuronal Ndn gene in mouse tumors. For Southern blot analyses, 5 µg of genomic DNA from tumor tissues was cleaved with XbaI for tumor A and BglII for tumor B and then with HpaII, MspI or EagI, respectively. (Left) The HpaII site of Ndn in DNA from control brain is partly methylated on the maternal allele and fully demethylated on the paternal allele. Southern blot analyses of tumor DNA, derived from a 129 mouse (tumor A), revealed de novo hypermethylation of the paternal allele (∼90% methylation). (Right) Tumor B, derived from the cerebral area of a DBA/2 mouse, exhibited fully de novo methylated alleles (100% methylation). The EagI site of Ndn in control DNA of BxDF₁ mice was allele specifically methylated, with a hypermethylated maternal allele and fully demethylated paternal allele. It is noteworthy that not all CpG dinucleotides within the Ndn gene are equally methylated in adult brain. Whereas the EagI and and SmaI (data not shown) sites show a fully developed imprinting pattern, the HpaII site displays only an incomplete methylation of the maternal allele, indicating that the methylation pattern for this CpG dinucleotide does not change much during development. H, HpaII; M, MspI; E, EagI.

(a) Structure of the bicistronic Snurf/Snrpn gene within the imprinted neuronal gene cluster on mouse chromosome 7C. Deletion, uniparental disomy (UPD) or inappropriate imprinting of the homologous chromosomal region in humans results in the neurogenetic PWS. Deletion of the DMR1 in mice, which spans the promoter and exon1 of Snurf/Snrpn, results in the disability to maintain the correct paternal methylation imprint during embryonic development (reviewed in 7). The bicistronic structure of this gene leads to two different protein products, SNURF and SmN. Vertical lines represent CpG dinucleotides. (b) Genomic structure of the intronless Ndn gene about 1 Mb telomeric to the IC. Ndn has DNA binding capacity and may play a role as a growth suppressor that interacts with the transcription factors E2F1 and p53 in virtually all post-mitotic neurons in the brain. H, HpaII/MspI; E, EagI; S, SacII; B, BglII; B_S, polymorphic BglII site to identify the parental origin of an allele in mice with mixed genetic background.

Stability of DNA methylation in imprinted neuronal genes in single cell-derived clonal ES cell lines. (a) Methylation-sensitive Southern blot analyses with restriction enzymes BglII and EagI revealed only small variations in methylation patterns of the Ndn gene in 10 Bruce 4 (C57BL/6) ES cell clones (Ndn probe a). (Bottom) The relative methylation levels in all ES cell subclones (K1–K10) varied around 15% (±5%) and were nearly identical to the methylation level observed in the original cell line (ES-P17). (b) Phosphoimager scan of a Southern blot with genomic DNA of 10 Bruce 4 subclones (5 µg/lane), cleaved with XbaI and the methylation-sensitive HpaII. The DNA on the membrane was hybridized with a mouse Snurf/Snrpn DMR1 probe. The first lane represents a control with the HpaII-isoschizomer MspI, which cleaves DNA also at methylated 5′-CCGG-3′ sites. The Snurf/Snrpn DMR1 was hypermethylated (65 ± 5%) in the Bruce 4 cell line ES-P17 and in all 10 subclones derived from these ES cells.

Southern blot analyses of the DNAs from ES clones transgenic for the Ad2-Luc and SV40-Luc constructs. (a) Map of the transgenes used in this study. Ad2-Luc contains the Ad2 E2A late promoter upstream of the luciferase reporter gene with the SV40 poly(A) signal downstream (23). SV40-Luc has the SV40 promoter upstream of the reporter gene and the SV40 enhancer 3′ of the poly(A) signal. (b) For Southern blot analyses, 10 µg of genomic DNA from non-transgenic controls and from transgenic ES cells was cleaved with BglII and then with EagI. The DNA on the membrane was hybridized with mouse Ndn probe b (see Fig. 1). Lane 1, testis DNA of control mice of the same genetic 129/ola background revealed a hypomethylated Ndn gene, indicating a maternal origin for the methylated allele in ES cells. Analysis with control brain DNA from a mixed genetic background [BxD(C57BL/6 × DBA/2)F1; lane 6] confirmed the maternally derived methylation imprint. Lane 2, the Ndn gene was imprinted in brain from control animals. Lanes 3 and 4, both transgenic constructs exhibited an allele-specific methylation pattern for the EagI site. However, the surrounding HpaII and SacII sites (see Fig. 1b) were not fully methylated on the maternal allele (data not shown). The methylation patterns were not significantly different compared with the non-transgenic ES cells (lane 5). The adenovirus construct in ES-A08 integrated via homologous recombination in intron 3 of the BLK gene on mouse chromosome 14, whereas the SV40 transgenic construct integrated randomly into the genome of ES cell line ES-S07. B6, C57BL/6 allele; D2, DBA/2 allele.

The Snrpn IC and the H19/Igf2 imprinting region may be differentially accessible to modifying enzymes. It seems possible that the overall chromatin structure in these regions may contain several dynamic levels of regulation. For example, a DNA sequence could be either in a methylated repressed state (heterochromatin) or in an active unmethylated conformation (euchromatin). Independent of these conformations, modifying enzymes could have different access to these chromosomal regions. Since the Snrpn IC and H19/Igf2 region are active in ES cells, they could both have an open chromatin structure on their active alleles, but may expose their CpG dinucleotides differently due to a different nucleosomal structure. The H19 region was reported to be accessible to methyltransferases in ES cells (50), which could lead to an increased susceptibility to epigenetic changes during in vitro culture. In contrast, the Snrpn IC could be already in an epigenetic maintenance mode, independent of its transcriptional activity. Both genomic regions have complex patterns of known epigenetic modifications (49,51,75–77). Histone H3 on the paternal allele has lysine 4 methylation and is acetylated, accompanied with a hypomethylation of the CpG dinucleotides. On the other hand, on the maternally inherited allele, chromatin is marked by hypermethylation on lysine 9 of H3 and hypermethylation of the DNA (75). Dark lollipops, CpG methylated; white lollipops, CpGs are unmethylated; Ac, histone-acetylation; Me, histone methylation; Lys, lysine residues.