The H19 gene is imprinted with preferential expression from the maternal allele. The putative imprinting control region for this locus is hypermethylated on the repressed paternal allele. Although maternal-specific expression of H19 is observed in mouse blastocysts that develop in vivo, biallelic expression has been documented in embryos and embryonic stem cells experimentally manipulated by in vitro culture conditions. In this study the effect of culture on imprinted H19 expression and methylation was determined. After culture of 2-cell embryos to the blastocyst stage in Whitten's medium, the normally silent paternal H19 allele was aberrantly expressed, whereas little paternal expression was observed following culture in KSOM containing amino acids (KSOM+AA). Analysis of the methylation status of a CpG dinucleotide located in the upstream imprinting control region revealed a loss in methylation in embryos cultured in Whitten's medium but not in embryos cultured in KSOM+AA. Thus, H19 expression and methylation were adversely affected by culture in Whitten's medium, while the response of H19 to culture in KSOM+AA approximated more closely the in vivo situation. It is unlikely that biallelic expression of H19 following culture in Whitten's medium is a generalized effect of lower methylation levels, since the amount of DNA methyltransferase activity and the spatial distribution of Dnmt1 protein were similar in in vivo-derived and cultured embryos. Moreover, imprinted expression of Snrpn was maintained following culture in either medium, indicating that not all imprinted genes are under the same stringent imprinting controls. The finding that culture conditions can dramatically, but selectively, affect the expression of imprinted genes provides a model system for further study of the linkage between DNA methylation and gene expression.

Imprinting, the differential expression of the two alleles of a gene based on their parental origin, requires that the alleles be distinguished or marked. A candidate for the differentiating mark is DNA methylation. The maternally expressed H19 gene is hypermethylated on the inactive paternal allele in somatic tissues and sperm, but to serve as the mark that designates the imprint, differential methylation must also be present in the gametes and the pre-implantation embryo. We now show that the pattern of differential methylation in the 5' portion of H19 is established in the gametes and a subset is maintained in the pre-implantation embryo. That subset is sufficient to confer monoallelic expression to the gene in blastocysts. We propose that paternal-specific methylation of the far 5' region is the mark that distinguishes the two alleles of H19.

The imprinted mouse H19 gene is hypermethylated on the inactive paternal allele in somatic tissues and sperm. Previous observations from a limited analysis have suggested that methylation of a few CpG dinucleotides in the region upstream from the start of transcription may be the mark that confers parental identity to the H19 alleles. Here we exploit bisulfite mutagenesis coupled with genomic sequencing to derive the methylation status of 68 CpGs that reside in a 4-kb region 5 to the start of transcription. This method reveals a 2-kb region positioned between 2 and 4 kb upstream from the start of transcription that is strikingly differentially methylated in midgestation embryos. At least 12 of the cytosine residues in this region are exclusively methylated on the paternal allele in blastocysts. In contrast, a 350-bp promoter-proximal region is less differentially methylated in midgestation embryos and, like most of the genome, is largely devoid of methylation on both alleles in blastocysts. We also demonstrate exclusive expression of the maternal H19 allele in the embryos that exhibit paternal methylation of the upstream 2-kb region. These data suggest that the 2-kb differentially methylated region acts as a key regulatory domain for imprinted H19 expression.