Distribution of 5-methylcytosine in the DNA of somatic and germline cells from bovine tissues.

Genomic DNA of calf thymus contains 1.5 times as much 5-methylcytosine as similar sperm DNA, but the major EcoRI repeat fragment from satellite I of thymus contains ten times as much 5-methylcytosine as the corresponding fragment from sperm DNA. Restriction enzyme analyses of the total DNA and the satellite I fragment show that three HpaII sites in the fragment are completely unmethylated in sperm but fully methylated in thymus DNA. Under-methylation of many sites in the satellite DNAs can probably account for the lower level of methylation of sperm DNA rather than hemimethylation as previously suggested. These results are also discussed in relation to maintenance and de novo (initiation-type) methylases.     

Pattern of undermethylation of the major satellite DNA of mouse sperm.

Enzymatic hydrolysis and base analysis by high performance liquid chromatography showed that mouse satellite DNA had 30-50% less 5-methylcytosine in sperm than in somatic tissue (1.59 mols % vs 2.40-3.11 mols %). Maxam-Gilbert sequencing and analysis of the intensity of the cytosine bands indicated that the level of methylation of the eight CpGs of the consensus sequence in sperm satellite DNA ranged from 0 to about 50%, considerably lower than the levels reported in somatic tissues. The Mn1I site containing one of these CpGs was cut much more extensively in satellite DNA from sperm than from liver, confirming the undermethylation of this site in sperm DNA.     

Demethylation of satellite I DNA during senescence of bovine adrenocortical cells in culture.

Over the finite proliferative life span of cultured bovine adrenocortical cells, satellite I DNA shows a progressive and extensive loss of methylation at CCGG sites. This was shown by Southern blotting after digestion with the methylation-sensitive enzyme HpaII alone, which provides a sensitive indicator of methylation loss, or digestion with the combination of EcoRI and HpaII, which provides a quantitative indication of loss of methylation. Bovine tissues, including adrenal cortex, all showed a much higher level of satellite methylation than cultured adrenocortical cells. After adrenocortical cells are placed in culture, some demethylation of satellite I is seen as early as 10 population doublings. By 80 population doublings, loss of satellite DNA methylation is extensive. The loss does not appear to prevent continued cell division, since an extended life span clone of bovine adrenocortical cells transfected with SV40 T antigen showed a similar pattern of extensive demethylation. Satellite demethylation has been reported in aging in vivo and the present cell culture system may provide an in vitro model for this form of genetic instability.     

Alu repeated DNAs are differentially methylated in primate germ cells.

A significant fraction of Alu repeats in human sperm DNA, previously found to be unmethylated, is nearly completely methylated in DNA from many somatic tissues. A similar fraction of unmethylated Alus is observed here in sperm DNA from rhesus monkey. However, Alus are almost completely methylated at the restriction sites tested in monkey follicular oocyte DNA. The Alu methylation patterns in mature male and female monkey germ cells are consistent with Alu methylation in human germ cell tumors. Alu sequences are hypomethylated in seminoma DNAs and more methylated in a human ovarian dysgerminoma. These results contrast with methylation patterns reported for germ cell single-copy, CpG island, satellite, and L1 sequences. The function of Alu repeats is not known, but differential methylation of Alu repeats in the male and female germ lines suggests that they may serve as markers for genomic imprinting or in maintaining differences in male and female meiosis.     

Methylation patterns of repetitive DNA sequences in germ cells of Mus musculus.

The major and the minor satellite sequences of Mus musculus were undermethylated in both sperm and oocyte DNAs relative to the amount of undermethylation observed in adult somatic tissue DNA. This hypomethylation was specific for satellite sequences in sperm DNA. Dispersed repetitive and low copy sequences show a high degree of methylation in sperm DNA; however, a dispersed repetitive sequence was undermethylated in oocyte DNA. This finding suggests a difference in the amount of total genomic DNA methylation between sperm and oocyte DNA. The methylation levels of the minor satellite sequences did not change during spermiogenesis, and were not associated with the onset of meiosis or a specific stage in sperm development.     

Methylation of satellite sequences in mouse spermatogenic and somatic DNAs.

The distribution of 5-methyl cytosine (5-MeC) residues in a highly repetitive sequence, mouse major satellite, was examined in germinal versus somatic DNAs by digestion with the methylation sensitive isoschizomers Msp I and Hpa II and Southern blot analysis, using a cloned satellite probe. DNA from liver, brain, and a mouse fibroblast cell line, C3H 10T1/2, yielded a multimeric hybridization pattern after digestion with Msp I (and control Eco RI) but were resistant to digestion with Hpa II, reflecting a high level of methylation of the satellite sequences. In contrast, DNA from mature sperm was undermethylated at these same sequences as indicated by the ability of Hpa II to generate a multimeric pattern. DNAs from purified populations of testis cells in different stages of spermatogenesis were examined to determine when during germ cell differentiation the undermethylation was established. As early as in primitive type A, type A, and type B spermatogonia, an undermethylation of satellite sequences was observed. This suggest that this highly specific undermethylation of germ cell satellite DNA occurs very early in the germ cell lineage, prior to entry into meiosis.     

CpG island methylation pattern in different human tissues and its correlation with gene expression

The study on DNA methylation pattern in different human tissues attracts increasing interest nowadays, but a systematic analysis of CpG island methylation pattern between both somatic tissues and gametocyte is still lacking. In this work, we analyzed the CpG island methylation data of sperm and other 11 somatic tissues from Human Epigenome Project, and found that the CpG island methylation profiles are highly correlated between somatic tissues, while the methylation profile in sperm is quite distinct. Furthermore, we observed that in the six tissues investigated, there is no obvious correlation between the methylation level of promoter CpG islands and corresponding gene expression across different tissues.     

Varied patterns of DNA methylation change between different satellite regions in bovine preimplantation development

Global reduction of DNA methylation, a part of genome reprogramming processes, occurs in a gradual manner until before implantation and is recognized as a conserved process in mammals. Here, we reported that in bovine, satellite regions exhibited varied patterns of methylation changes when one-cell egg advanced to the blastocyst; a maintenance methylation was observed in satellite I sequences, a decrease in alpha satellites, and an increase in satellite II regions. Cloned embryos exhibited similar changes for DNA methylation in the satellite I and alpha. We also observed that the satellite I and alpha sequences were methylated more in inner cell mass region of the blastocyst whereas the satellite II showed selective demethylation in this region. Together, these findings point that individual satellite sequences carry their own methylation patterns under the pressure of global demethylation, suggesting that local methylation control system acts on the satellite regions in early bovine embryos.     

Relative levels of methylation in human growth hormone and chorionic somatomammotropin genes in expressing and non-expressing tissues.

It has been shown that the extent of methylation of cytosine in vertebrate DNA is inversely correlated with gene expression. We studied cytosine methylation in and around the homologous human growth hormone (GH) and chorionic somatomammotropin (CS) genes to determine if these genes are undermethylated in DNA from tissues in which they are expressed (pituitary and placenta, respectively) compared to other tissues. Hpa II and Hha I (which cleave only unmethylated 5' CCGG 3' and 5' GCGC 3' respectively) and Msp I (which cleaves CCGG and CmeCGG) were used to digest DNA samples followed by gel electrophoresis, Southern transfer and hybridization with a GH cDNA probe. The extent of methylation of Hpa II and Hha I sites in the GH and CS genes was leukocyte much greater than pituitary greater than placenta = hydatidiform mole. Taken as a whole, our data support the hypothesis that undermethylation is a necessary but not sufficient condition for gene expression since placental and pituitary DNAs are less methylated than leukocyte DNA in this region. However, the correlation between gene expression and undermethylation is imperfect since (1) hydatiform mole DNA has a very similar methylation pattern compared to placental DNA even though moles make little or no CS and (2) the level of methylation of the GH gene compared to the CS gene does not vary in a tissue-specific manner.     

Satellite 2 methylation patterns in normal and ICF syndrome cells and association of hypomethylation with advanced replication

Mutation in the DNMT3B DNA methyltransferase gene is a common cause of ICF (immunodeficiency, centromeric heterochromatin, facial anomalies) immunodeficiency syndrome and leads to hypomethylation of satellites 2 and 3 in pericentric heterochromatin. This hypomethylation is associated with centromeric decondensation and chromosomal rearrangements, suggesting that these satellite repeats have an important structural role. In addition, the satellite regions may have functional roles in modifying gene expression. The extent of satellite hypomethylation in ICF cells is unknown because methylation status has only been determined with restriction enzymes that cut infrequently at these loci. We have therefore developed a bisulfite conversion-based method to determine the detailed cytosine methylation patterns at satellite 2 sequences in a quantitative manner for normal and ICF samples. From our sequence analysis of unmodified DNA, the internal repeat region analyzed for methylation contains an average of 17 CpG sites. The average level of methylation in normal lymphoblasts and fibroblasts is 69% compared with 20% in such cells from ICF patients with DNMT3B mutations and 29% in normal sperm. Although the mean satellite 2 methylation values for these groups do not overlap, there is considerable overlap at the level of individual DNA strands. Our analysis has also revealed a pattern of methylation specificity, suggesting that some CpGs in the repeat are more prone to methylation than other sites. Variation in satellite 2 methylation among lymphoblasts from different ICF patients has prompted us to determine the frequency of cytogenetic abnormalities in these cells. Although our data suggest that some degree of hypomethylation is necessary for pericentromeric decondensation, factors other than DNA methylation appear to play a major role in this phenomenon. Another such factor may be altered replication timing because we have discovered that the hypomethylation of satellite 2 in ICF cultures is associated with advanced replication.     

Different methylation pattern of melon satellite DNA sequences in hypocotyl and callus tissues

Satellite DNA sequences from Cucumis melo have been examined with respect to modification at CCGG sequences in hypocotyls and in callus tissues. For this purpose, restriction fragments given by HpaII and MspI were compared (both enzymes recognize CCGG sequences but have different sensitivity to methylation at this site). Whereas the methylation level of satellite DNA sequences is on average higher in hypocotyls than in callus tissues, the comparison of partially methylated repeat units of satellite DNA reveals that in callus tissues, all methylated restriction sites are doubly methylated.     

Hypermethylation of human DNA sequences in embryonal carcinoma cells and somatic tissues but not in sperm.

Certain human DNA sequences are much less methylated at CpG sites in sperm than in various adult somatic tissues. The DNA of term placenta displays intermediate levels of methylation at these sequences (Sp-0.3 sequences). We report here that pluripotent embryonal carcinoma (EC) cells derived from testicular germ cell tumors are hypermethylated at the three previously cloned Sp-0.3 sequences and seven newly isolated sequences that exhibit sperm-specific hypomethylation. In contrast to their hypermethylation in EC cells, the Sp-0.3 sequences are hypomethylated in a line of yolk sac carcinoma cells, which like placenta, represent an extraembryonic lineage. These DNA sequences, therefore, appear to be subject to coordinate changes in their methylation during differentiation, probably early in embryogenesis, despite their diversity in copy number (1 to 10(4] and primary structure. Two of these Sp-0.3 sequences are highly homologous to DNA sequences in human chromosomal regions that might be recombination hotspots, namely, a cryptic satellite DNA sequence at a fragile site and the downstream region of the beta-globin gene cluster.     

Modifications in major satellite methylation in the nucleus of a two-cell mouse embryo with respect to developmental conditions

A study was conducted of the degree of DNA methylation in the nucleus, in particular, of the major satellite in two-cell mouse embryos developing in the maternal organism, in standard culture medium M16, used for cultivating mouse embryos; and M2 media used for manipulating embryos in air. Two-cell embryo nuclei at 44–46 h after injections of chorionic hormone were investigated. The results are evidence for the dependence of the major satellite’s methylation level on the developmental conditions of embryos. The methylation level of the nuclear DNA was shown to increase with a deterioration of environmental conditions. It was reported that in the case of cultivation in M2 media unsuitable for long cultivation, the DNAís methylation level, the major satellite in particular, was higher compared to other embryo groups. Accordingly, not only a significant number of genes but also sequences of satellite DNA are involved in epigenetic regulation.     

Developmental differences in methylation of human Alu repeats.

Alu repeats are especially rich in CpG dinucleotides, the principal target sites for DNA methylation in eukaryotes. The methylation state of Alus in different human tissues is investigated by simple, direct genomic blot analysis exploiting recent theoretical and practical advances concerning Alu sequence evolution. Whereas Alus are almost completely methylated in somatic tissues such as spleen, they are hypomethylated in the male germ line and tissues which depend on the differential expression of the paternal genome complement for development. In particular, we have identified a subset enriched in young Alus whose CpGs appear to be almost completely unmethylated in sperm DNA. The existence of this subset potentially explains the conservation of CpG dinucleotides in active Alu source genes. These profound, sequence-specific developmental changes in the methylation state of Alu repeats suggest a function for Alu sequences at the DNA level, such as a role in genomic imprinting.     

Methylation of somatic vs germ cell DNAs analyzed by restriction endonuclease digestions.

Bacterial restriction endonucleases containing the dinucleotide CpG in their cleavage sequences were used to compare the methylation patterns of primarily repeated DNA sequences in (1) bovine somatic cell native DNAs vs bovine sperm cell native DNA and (2) native vs renatured bovine liver and sperm cell DNAs. The restriction patterns of sperm native DNA differ markedly from those of somatic cell native DNAs when using Hpa II, Hha I, and Ava I but not when using the enzymes Eco RI and Msp I. Digestion patterns of germ cell renatured DNA differed significantly from those of germ cell native DNA when using Hpa II but not when using Msp I or Eco RI. The results may not be due to artifacts of renaturation of the DNAs. The results are consistent with the concept that germ cell DNA may be strand asymmetrically hemimethylated. The data also suggest that methylation of the 5'-cytosine in the sequence CCGG renders this site insensitive to cleavage by Msp I.     

Methylated and unmethylated DNA compartments in the sea urchin genome.

Sea urchin (Echinus esculentus) DNA has been separated into high and low molecular weight fractions by digestion with the mCpG-sensitive restriction endonucleases Hpa II, Hha I and Ava I. The separation was due to differences in methylation at the recognition sequences for these enzymes because an mCpG-insensitive isoschizomer of Hpa II (Msp I) digested Hpa II-resistant DNA to low molecular weight, showing that many Hpa II sites were in fact present in this fraction; and because 3H-methyl methionine administered to embryos was incorporated into the high molecular weight Hpa II-, Hha I- and Ava I-resistant fraction, but not significantly into the low molecular weight fraction. The fraction resistant to Hpa II, Hha I and Ava I amounted to about 40% of the total DNA. It consisted of long sequence tracts between 15 and well over 50 kg in length, in which many sites for each of these enzymes were methylated consecutively. The remaining 60% of the genome, (m-), was not significantly methylated. Methylated and unmethylated fractions were considered to be subfractions of the genome because enriched unique sequences from one fraction cross-reassociated poorly with the other fraction and specific sequences were found in either (m+) or (m-) but not in both (see below). Similar (m+) and (m-) compartments were found in embryos, germ cells and adult somatic tissues. Furthermor, we found no evidence for changes in the sequence composition of (m+) or (m-) between sperm, embryo or intestine DNAs, although low levels of exchange would not have been detected. Using cloned Echinus histone DNA, heterologous 5S DNA and ribosomal DNA probes, we have found that each of these gene families belongs to the unmethylated DNA compartment in all the tissues examined. In particular, there was no detectable methylation of histone DNA either in early embryos, which are thought to be actively transcribing the bulk of histone genes, or in sperm and gastrulae, in which most histone genes are not being transcribed. In contrast to these gene families, sequences complementary to an internally repetitious Echinus DNA clone were found primarily in the methylated DNA compartment.     

Modeling genomic data with type attributes, balancing stability and maintainability

Background

The computational prediction of DNA methylation has become an important topic in the recent years due to its role in the epigenetic control of normal and cancer-related processes. While previous prediction approaches focused merely on differences between methylated and unmethylated DNA sequences, recent experimental results have shown the presence of much more complex patterns of methylation across tissues and time in the human genome. These patterns are only partially described by a binary model of DNA methylation. In this work we propose a novel approach, based on profile analysis of tissue-specific methylation that uncovers significant differences in the sequences of CpG islands (CGIs) that predispose them to a tissue- specific methylation pattern.

Results

We defined CGI methylation profiles that separate not only between constitutively methylated and unmethylated CGIs, but also identify CGIs showing a differential degree of methylation across tissues and cell-types or a lack of methylation exclusively in sperm. These profiles are clearly distinguished by a number of CGI attributes including their evolutionary conservation, their significance, as well as the evolutionary evidence of prior methylation. Additionally, we assess profile functionality with respect to the different compartments of protein coding genes and their possible use in the prediction of DNA methylation.

Conclusion

Our approach provides new insights into the biological features that determine if a CGI has a functional role in the epigenetic control of gene expression and the features associated with CGI methylation susceptibility. Moreover, we show that the ability to predict CGI methylation is based primarily on the quality of the biological information used and the relationships uncovered between different sources of knowledge. The strategy presented here is able to predict, besides the constitutively methylated and unmethylated classes, two more tissue specific methylation classes conserving the accuracy provided by leading binary methylation classification methods.