DNA methylation controls the inducibility of the mouse metallothionein-I gene in lymphoid cells

The W7 mouse thymoma cell line does not express the metallothionein-I (MT-I) gene in the presence of either cadmium or glucocorticoids, unlike most other cell lines. This cell line was therefore used as a model system for studying the role of DNA methylation on MT-I gene expression. The extent of DNA methylation within the MT-I gene and its flanking regions was determined by comparing the cleavage patterns generated by the isoschizomeric restriction enzymes Hpa II and Msp I. In W7 cells, all of the Hpa II sites in the vicinity of the MT-I gene are methylated, whereas in cells that have an expressible MT-I gene (for example, Friend erythroleukemia cells) all of these Hpa II sites are unmethylated. When W7 cells are treated for a few hours with 5-azacytidine, the MT-I gene becomes inducible by both cadmium and glucocorticoids. Addition of hydroxyurea along with 5-azacytidine prevents MT-I gene induction, suggesting that incorporation of 5-azacytidine into DNA is required before this gene can be activated. To determine whether 5-azacytidine treatment changes the methylation pattern near the MT-I gene, we treated W7 cells with 5-azacytidine and selected inducible cells in 10 μM cadmium. All of the Hpa II sites within the MT-I gene are unmethylated in these cadmium-resistant W7 cells. In addition, flanking DNA sequences are also undermethylated in a pattern similar to that seen in Friend erythroleukemia cells that express the MT-I gene. The possible significance of methylation as a mechanism of gene commitment during cell differentiation is discussed.     

Analysis of the methylation pattern of c-Ha-ras oncogene in human prostatic cancer

To determine the methylation pattern of c-Ha-ras oncogene in prostatic tissue and to identify possible changes of methylation associated with cancer, high molecular weight DNA was extracted from 7 normal and 6 carcinomatous human prostates. Analysis of the samples was performed by cleaving DNA with the restriction endonucleases Msp I, Hpa II and Cfo I, and by Southern hybridizing the DNA digests with the 32P-labelled c-Ha-ras (pT24-C3) probe. Several discrete fragments were obtained with Hpa II and Cfo I digestion while the Msp I pattern showed fewer and smaller bands. Therefore, c-Ha-ras appears to be partially methylated. While a considerable polymorphism of the sequence 5'-CCGG-3' was observed at several Msp I sites in all cases, no significant differences could be evidenced in the methylation patterns of normal and neoplastic prostatic DNA samples extracted and purified from each patient.     

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.     

The somatic replication of DNA methylation

We have tested the hypothesis that DNA methylation patterns are replicated in the somatic cells of vertebrates. Using M-Hpa II, the modification enzyme from Haemophilus parainfluenzae which methylates the internal cytosine residues in the sequence 5'CCGG 3' GGCC, we methylated bacteriophage phi X174 RF DNA and the cloned chicken thymidine kinase (tk) gene in vitro and then introduced these DNAs and unmethylated controls into tk- cultured mouse cells by DNA-mediated transformation. Twenty-five cell generations later, the state of methylation of transferred DNA was examined by restriction endonuclease analysis and blot hybridization. We conclude that methylation at Hpa II sites is replicated by these cultured cells but not with 100% fidelity. We have also noted that methylation of the cloned chicken tk gene decreases its apparent transformation efficiency relative to unmethylated molecules.     

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.     

Mythylation of human HLA-D/DR genes: derangement in chronic lymphocytic leukemia

HLA-DR antigens are expressed as differentiation markers in certain human leukemias. To investigate whether DNA methylation plays a role in expression of DR genes in leukemia, we analyzed methylation patterns of the DR-alpha and D/DR-beta genes in the DR antigen-positive and -negative B-cell lines, in normal adults and in chronic lymphocytic leukemia (CLL) patients using Southern blot hybridization of DNA digested with Msp I and Hpa II. The DR-alpha and D/DR-beta genes of a DR antigen positive B-cell line, T5-1, were heavily methylated, while those of DR antigen-negative variant, 6.1.6, were hypomethylated. Blood cells collected from four normal adults contained different levels of DR-alpha and D/DR-beta mRNAs, but their relative amounts were about the same among the individuals. By contrast, the relative amounts of these mRNAs in CLL cells varied widely, indicating aberrant expression of one or both of these genes in CLL. The DR-alpha gene in four normal adults and six CLL patients produced only a 3 kb hybridizable band after Msp I digestion. Normal adult DR-alpha genes were resistant to Hpa II digestion, suggesting that all Hpa II sites are methylated. In contrast, digestion of CLL DNA with Hpa II yielded various bands of larger sizes which differed among the CLL patients, suggesting that Hpa II sites are differentially methylated in the CLL DNA. In the case of D/DR-beta genes, normal adult DNA gave Msp I bands which were slightly polymorphic among four individuals tested. In contrast, CLL DNA showed a high degree of restriction fragment length polymorphism (RFLP) on Msp I digestion. We speculate that the high RFLPs in the CLL DNA may result from differential methylation in CpG clusters in the D/DR-beta genes, and that this characteristic may be of use for diagnosis of CLL.     

In contrast to other Xenopus genes the estrogen-inducible vitellogenin genes are expressed when totally methylated

The methylation-sensitive restriction enzymes Hha I and Hpa II were used to analyze the methylation pattern of four Xenopus laevis genes in DNA of embryos, of erythrocytes, and of untreated and estrogen-treated hepatocytes. Within these four genes all sites tested are fully modified in embryonic DNA. However, the adult β1-globin gene is unmethylated in DNA of erythrocytes, where it is expressed, and the 68 kd albumin gene, active only in hepatocytes, is specifically hypomethylated in hepatic DNA. The vitellogenin genes A1 and A2, in hepatocytes simultaneously expressed upon estrogen treatment, are heavily methylated in all adult tissues, irrespective of expression. Our results reveal that specific genes can be actively transcribed even when they are fully methylated and that changes in the methylation pattern are not a general prerequisite for gene activation.     

Direct visualization of the sites of DNA methylation in human,and mosquito chromosomes

Human and mosquito fixed chromosomes were digested with restriction endonucleases that are inhibited by the presence of 5-methylcytosine in their restriction sites (Hha I, Hin PI, Hpa II), and with endonucleases for which cleavage is less dependent on the state of methylation (Taq I, Msp I). Methylation-dependent enzymes extracted low DNA amounts from human chromosomes, while methylation-independent enzymes extracted moderate to high amounts of DNA. After DNA demethylation with 5-azacytidine the isoschizomers Hpa II (methylation-dependent) and Msp I (methylation-independent) extracted 12-fold and 1.4-fold amounts of DNA from human chromosomes, respectively. These findings indicate that human DNA has a high concentration of Hpa II and Msp I restriction sites (CCGG), and that the internal C of this sequence is methylated in most cases, while the external cytosine is methylated less often. All the enzymes tested released moderate amounts of DNA from mosquito chromosomes whether or not the DNA was demethylated with 5-azacytidine. Hpa II induced banding in the centromere chromosome regions. After demethylation with 5-azacytidine this banding disappeared. Mosquito DNA has therefore, moderate to high frequencies of nonmethylated CpG duplets. The only exception is the centromeric DNA, in which the high levels of C methylation present produce cleavage by Hpa II and the appearance of banding. Centromere regions of human chromosomes 1 have a moderately low concentration of Hpa II-Msp I restriction sites.     

Demethylation of specific sites in the 5'-flanking region of the CYP17 genes when bovine adrenocortical cells are placed in culture.

DNA methylation of CYP17 (steroid 17 alpha-hydroxylase) was studied in bovine adrenocortical cells, which lose the capacity to express this tissue-specific gene in culture by phenotypic switching. Restriction enzyme digestions, and sequencing of a lambda clone of a second CYP17 gene (CYP17A2), showed that there are at least three CYP17 genes in the bovine genome. Southern blotting of DNA digested with Msp I or Hpa II together with Eco RI was used to investigate the methylation status of Hpa II sites at -1.0 kb (H1), -1.8 kb (H2), and -2.3 kb (H4) in CYP17A1 and CYP17A2 and at -0.7 kb (H0) in CYP17A3. In cells and tissues other than white blood cells, H0 was nonmethylated whereas H1 was always methylated; H2 and H4 showed variation in methylation status among different cells and tissues. In particular, whereas H4 was methylated in the bovine adrenal cortex in vivo, there was a rapid and complete demethylation at H4 when adrenocortical cells were placed in culture. Sites downstream from H4 did not change methylation over the first six passages in culture; additionally, the coding region of CYP17 remained fully methylated under all conditions. In contrast to adrenocortical cells, DNA from fibroblasts was nonmethylated at H2, whereas all downstream sites were fully methylated. Digestion with another methylation-sensitive enzyme, Bsa HI, which has a site between H2 and H4, showed that this region is methylated in intact adrenal cortex but nonmethylated both in cultured adrenocortical cells and in fibroblasts. The specific changes in methylation at this site and at H4 in adrenocortical cells indicate a reproducible, environmentally determined change in methylation in adrenocortical cells when they are placed in culture.     

Chromosomal position and specific demethylation in enhancer sequences of germ line-transmitted retroviral genomes during mouse development.

The methylation pattern of the germ line-transmitted Moloney leukemia proviral genome was analyzed in DNA of sperm, of day-12 and day-17 embryos, and of adult mice from six different Mov substrains. At day 12 of gestation, all 50 testable CpG sites in the individual viral genomes as well as sites in flanking host sequences were highly methylated. Some sites were unmethylated in sperm, indicating de novo methylation of unique DNA sequences during normal mouse development. At subsequent stages of development, specific CpG sites which were localized exclusively in the 5' and 3' enhancer regions of the long terminal repeat became progressively demethylated in all six proviruses. The extent of enhancer demethylation, however, was tissue specific and strongly affected by the chromosomal position of the respective proviral genome. This position-dependent demethylation of enhancer sequences was not accompanied by a similar change within the flanking host sequences, which remained virtually unchanged. Our results indicate that viral enhancer sequences, but not other sequences in the M-MuLV genome, may have an intrinsic ability to interact with cellular proteins, which can perturb the interaction of the methylase with DNA. Demethylation of enhancer sequences is not sufficient for gene expression but may be a necessary event which enables the enhancer to respond to developmental signals which ultimately lead to gene activation.     

Hypomethylation of the interphotoreceptor retinoid-binding protein (IRBP) promotor and first exon is linked to expression of the gene.

The interphotoreceptor retinoid-binding protein (IRBP) is limited in expression to retinal photoreceptor cells and a subset of pinealocytes. We have obtained a genomic clone containing the entire coding region and 7 kb of 5' flanking sequence. As a first step in studying IRBP gene regulation we have examined the CpG methylation patterns of the entire IRBP gene in expressing and non-expressing human cells. This has been done by isolation of high molecular weight DNA from Y-79 cells grown in suspension or attached to poly-D-lysine, which synthesize IRBP at different levels, and from human lymphocytes, which were shown by northern analysis to lack IRBP message. The DNA was digested by either Hpa II, Msp I, or Hha I. Southern blots were prepared with these digests and hybridized with probes made from fragments covering the complete genomic clone. Probes from the first exon, the introns and the 3' end gave banding patterns which showed no differences between the expressing cells and the lymphocytes. A probe from the very 5' end did not give a clear banding pattern, probably due to the presence of repetitive elements in the probe. However, a Hind III probe covering the 5' flanking 3 kb and the beginning of the first exon hybridized with a 1.8 kb band in Hpa II digests of Y-79 cells which was not present in Hpa II digests of lymphocyte DNA. In addition, a 2.1-2.3 kb Hha I band was found only in the Y-79 DNA digests. Sequence analysis of the promoter region indicated that these bands were due to hypomethylation of sites within a CpG rich island from -1578 to -1108 in the promoter and hypomethylation of sites in the beginning of the first exon. A Hha I site between the CpG island and the first exon was not hypomethylated in the expressing Y-79 cells. We propose that hypomethylation of the CpG rich island of the IRBP promoter and the first exon is linked to the expression of this gene.     

Modulation of Escherichia coli DNA methyltransferase activity by biologically derived GATC-flanking sequences

Escherichia coli DNA adenine methyltransferase (EcoDam) methylates the N-6 position of the adenine in the sequence 5'-GATC-3' and plays vital roles in gene regulation, mismatch repair, and DNA replication. It remains unclear how the small number of critical GATC sites involved in the regulation of replication and gene expression are differentially methylated, whereas the approximately 20,000 GATCs important for mismatch repair and dispersed throughout the genome are extensively methylated. Our prior work, limited to the pap regulon, showed that methylation efficiency is controlled by sequences immediately flanking the GATC sites. We extend these studies to include GATC sites involved in diverse gene regulatory and DNA replication pathways as well as sites previously shown to undergo differential in vivo methylation but whose function remains to be assigned. EcoDam shows no change in affinity with variations in flanking sequences derived from these sources, but methylation kinetics varied 12-fold. A-tracts immediately adjacent to the GATC site contribute significantly to these differences in methylation kinetics. Interestingly, only when the poly(A) is located 5' of the GATC are the changes in methylation kinetics revealed. Preferential methylation is obscured when two GATC sites are positioned on the same DNA molecule, unless both sites are surrounded by large amounts of nonspecific DNA. Thus, facilitated diffusion and sequences immediately flanking target sites contribute to higher order specificity for EcoDam; we suggest that the diverse biological roles of the enzyme are in part regulated by these two factors, which may be important for other enzymes that sequence-specifically modify DNA.     

The changes in proviral chromatin that accompany morphological variation in avian sarcoma virus-infected rat cells

The clone All of avian sarcoma virus B77-infected Rat-1 cells comprises both morphologically normal and morphologically transformed derivatives. Transformed subclones, in which virus-specific RNA is readily detectable, contain a provirus that is very sensitive to DNase 1 digestion of chromatin, and show DNase 1 hypersensitive sites at the 5' end of the provirus and in 5' flanking cell DNA. Normal subclones with no detectable virus-specific RNA, whether infected cells that have never been transformed or revertants derived from transformed cells, contain a provirus that is far more resistant to DNase 1 digestion. Moreover this provirus lacks hypersensitive sites at its 5' end, although DNase 1 hypersensitive sites were detected at the 3' end of the provirus in either normal or transformed clones. The pattern of cytosine methylation in the proviral restriction sites of the isoschizomers Msp I and Hpa II differed between transformed and revertant clones; the revertants show additional methylation at some CpG doublets.     

Methylation Pattern of Radish (Raphanus sativus) Nuclear Ribosomal RNA Genes

The methylation pattern of radish Raphanus sativus nuclear rDNA has been investigated using the Hpa II, Msp I, and Hha I restriction enzymes. The presence of numerous target sites for these enzymes has been shown using cloned rDNA fragments. A large fraction of the numerous rDNA units are heavily methylated, being completely resistant to Hpa II and Hpa I. However, specific sites are constantly available in another fraction of the units and are therefore unmethylated. The use of different probes allowed us to demonstrate that hypomethylated sites are present in different regions. Major hypomethylated Hha I sites have been mapped in the 5′ portion of 25S rRNA coding sequence. Among the hypomethylated fraction, different methylation patterns coexist. It has been possible to demonstrate that methylation patterns are specific for particular units. The Hha I pattern of rDNA in tissues of different developmental stages was analyzed. Evidence for possible tissue specific differences in the methylation pattern is reported.     

Albumin and alpha-fetoprotein gene transcription in rat hepatoma cell lines is correlated with specific DNA hypomethylation and altered chromatin structure in the 5'' region.

We examined DNA methylation and DNase I hypersensitivity of the alpha-fetoprotein (AFP) and albumin gene region in hepatoma cell lines which showed drastic differences in the level of expression of these genes. We assayed for methylation of the CCGG sequences by using the restriction enzyme isoschizomers HpaII and MspI. We found two methylation sites located in the 5' region of the AFP gene and one in exon 1 of the albumin gene for which hypomethylation is correlated with gene expression. Another such site, located about 4,000 base pairs upstream from the AFP gene, seems to be correlated with the tissue specificity of the cells. DNase I-hypersensitive sites were mapped by using the indirect end-labeling technique with cloned genomic DNA probes. Three tissue-specific DNase I-hypersensitive sites were mapped in the 5' flanking region of the AFP gene when this gene was transcribed. Similarly, three tissue-specific DNase I-hypersensitive sites were detected upstream from the albumin gene in producing cell lines. In both cases, the most distal sites were maintained after cessation of gene activity and appear to be correlated with the potential expression of the gene. Interestingly, specific methylation sites are localized in the same DNA region as DNase I hypersensitive sites. This suggests that specific alterations of chromatin structure and changes in methylation pattern occur in specific critical regulatory regions upstream from the albumin and AFP genes in rat hepatoma cell lines.