Replacement of 5-methylcytosine by cytosine: a possible mechanism for transient DNA demethylation during differentiation

DNA demethylation in vitro: involvement of RNA

An in vitro system for studying DNA demethylation has been established using extracts from tissue culture cells. This reaction, which is unusually resistant to proteinase K, takes place through the removal of a 5-methylcytosine nucleotide unit from the DNA substrate and its conversion to an RNase-sensitive form. It is likely that this represents the in vivo mechanism, as well, since extracts from L8 myoblasts specifically demethylate an alpha-actin gene, while extracts from F9 teratocarcinoma cells specifically demodify the Aprt CpG island. After pretreatment with proteinase K, these extracts demethylate both genes equally, suggesting that gene specificity may be controlled by protein factors.

Regulation of DNA methyltransferase during differentiation of F9 mouse embryonal carcinoma cells

DNA becomes demethylated when F9 mouse embryonal carcinoma cells differentiate into parietal endoderm. DNA methyltransferase (DNA-MTase) activity decreased by 50% during 1 week of differentiation. The level of DNA-MTase mRNA was also diminished accordingly, but the transcription rate of the DNA-MTase gene measured by run-on transcription was essentially unchanged, indicating regulation of DNA-MTase expression at a posttranscriptional step. The decline of DNA-MTase mRNA paralleled that of histone H3 mRNA in accord with the notion that DNA-MTase is preferentially expressed in the S phase of the cell cycle. Since DNA-MTase expression decreases in parallel with DNA synthesis, DNA demethylation during differentiation of F9 cells appears not to be due to limited expression of DNA-MTase. However, the plasmid pAFP7000CAT, alpha-fetoprotein (AFP), which is strongly de novo methylated when transfected into F9 stem cells became only weakly methylated after transfection into the F9 parietal endoderm derivative P1, indicating that the activity of DNA-MTase within parietal endoderm cells is more strongly diminished than is apparent from measurements of mRNA amounts and of overall DNA-MTase activity in vitro. The discrepancy between DNA-MTase expression and its actual activity within the cell indicates the existence of a novel mechanism controlling the activity of DNA-MTase.

Human chorionic gonadotropin-beta subunit gene expression in cultured human fetal and cancer cells of different types and origins

BACKGROUND

The authors' previous investigations using living cultured human cancer cells and cells isolated from cancer tissues, analytical flow cytometry, and monoclonal antibodies directed to epitopes located in five different sites of the human chorionic gonadotropin (hCG) molecule, identified the presence of membrane-associated hCG, its subunits and fragments, by cells from all cancers, irrespective of type and origin, indicating that the expression of these sialoglycoproteins is a common phenotypic characteristic of cancer. Although benign neoplasms do not express these compounds, cultured human embryonic and fetal cells also express the same materials. To corroborate these findings, five fetal cell lines and 28 cancer cell lines were randomly selected from those previously studied, to determine the presence of translatable levels of hCG-beta (hCG beta) mRNA.

METHODS

All cell lines were grown under identical conditions. Determination of hCG beta mRNA was made by extracting the total RNA from the cells, followed by synthesis of cDNA with RNase H- reverse transcriptase and polymerase chain reaction amplification using specific hCG beta-luteinizing hormone-beta (hLH beta) primers. The presence of amplified hCG beta cDNA was corroborated by hybridization of the product with an hCG beta-specific oligonucleotide and Southern blot analyses of the hybridization products. Gestational choriocarcinoma cells and HeLa adenocarcinoma of cervical cells, known producers of biologically active hCG, were positive control subjects, and human pituitary cells were used as negative control subjects.

RESULTS

The results showed single and multiple hCG beta gene activation by the fetal cells and the different types of cancer, indicating that at any given time, there is the possibility of activation of as many as four genes of the six genes of the hCG beta-hLH beta gene cluster, even though alternative gene splicing cannot be ruled out.

CONCLUSIONS

In addition to the authors' previous findings, the results of these studies support the concept that cancer is a problem of development and differentiation, and, to the authors' knowledge, prove definitively for the first time that synthesis and expression of hCG, its subunits, and its fragments, is a common biochemical denominator of cancer, providing the scientific basis for studies of its prevention and/or control by active and/or passive immunization against these sialoglycoproteins.

Overexpression of DNA methyltransferase in myoblast cells accelerates myotube formation

We overexpressed mouse DNA methyltransferase in murine C2C12 myoblast cells and tested the isolated clones for their ability to differentiate. Significant numbers of the clones showed distinct myotubes 24 h after the isolated transformants had been induced to differentiate, whereas the parent C2C12 cells did not form myotubes at this time point. Transfection of the vacant vector or the plasmid containing the reverse-oriented DNA methyltransferase cDNA did not provide significant numbers of transformants with the accelerated differentiation phenotype, suggesting that the effect is caused by the expression of DNA methyltransferase. The expressions of skeletal muscle myosin and creatine kinase in clones that showed the accelerated differentiation-phenotype were also induced about 24 h earlier and at higher levels relative to the parent C2C12 or the control cells, indicating that the entire process of myogenesis had been accelerated. All the methyltransferase-transfected clones, regardless of their phenotypes, demonstrated about threefold higher DNA methyltransferase activity and higher methylation levels than those of the clones transfected with vector alone or the reverse-oriented plasmid. At the early stage of transfection of the sense-oriented plasmid, high de novo methylation activities were detected. We consider it likely that this high de novo methylation activity is the reason for the high methylation levels and the accelerated myotube formation of the clones transfected with the sense-oriented plasmid. In some transformants which showed the accelerated differentiation phenotype, MyoD1 was already fully expressed under the growth conditions while, in control cells, MyoD1 was expressed at low levels. This elevated level of MyoD1 transcription could account for the accelerated myotube formation observed in the transformants. The methylation state of the HpaII sites in exon 1 through exon 2 of the MyoD1 gene and the expression of the MyoD1 transcript are positively correlated.

Prevention of DNA 5-methylcytosine reutilization in human cells

DNA methylation is an important process contributing to transcriptional regulation in animal and plant cells. The well known reutilization of DNA nucleotide bases indicated that DNA degradation occurs in many cells and tissues. On the other hand, the reutilization of 5-methyl-2'-deoxycytidine monophosphate in the DNA synthesis would have deleterious effects on gene regulation. Recent molecular insights into the exclusion of exogenous 5-methylcytosine from DNA are the subject of this review.

Enzymic removal of 5-methylcytosine from poly(dG-5-methyl-dC) by HeLa cell nuclear extracts is not by a DNA glycosylase

A recent report in this journal [Vairapandi, M. and Duker, N.J. (1993) Nucleic Acids Res. 21, 5323-5327) presented evidence of an activity in HeLa cell nuclear extracts that released radiolabeled material from a poly(dG.dC) polymer that had been methylated and simultaneously labeled on cytosine residues by incubation with a CpG-specific DNA methylase and [methyl-3H]S-adenosylmethionine. Based on chromatographic evidence that the released products were thymine and 5-methylcytosine and on f1p4olabeling data suggesting a concomitant increase in abasic sites, the authors concluded that the releasing activity was a 5-methylcytosine-specific glycosylase and that the solubilized 5-methylcytosine was converted to thymine by a nuclear deaminase. We have confirmed that HeLa nuclear extracts promote release of ethanol-soluble radioactivity from a methyl-labeled poly(dG-5-methyl-dC)polymer, but the products released were neither 5-methylcytosine nor thymine. Furthermore, free 5-methylcytosine was not deaminated by incubation with the nuclear extract. The labeled compound released initially from the polymer appeared to be 5-methyl-deoxycytidine monophosphate, which was converted to 5-methyl-deoxycytidine, thymidine monophosphate, and/or thymidine by further incubation with the nuclear extract. The activity responsible for the release, therefore, was a nuclease. Release of 32P-labeled nucleotides from a 32P-labeled poly(dG-dC) polymer suggested, furthermore, that the activity was not specific for methylated DNA.

DNA methylation is not involved in the structural alterations of ornithine decarboxylase or total chromatin of c-Ha-rasVal 12 oncogene-transformed NIH-3T3 fibroblasts

The ornithine decarboxylase (odc) gene is an early response gene, whose increased expression and relaxed chromatin structure is closely coupled to neoplastic growth. In various tumour cells, the odc gene displays hypomethylation at the sequences CCGG. Hypomethylation of genes is believed to correlate with chromatin decondensation and gene expression. Since a given pattern of DNA methylation may not be preserved in neoplastic cells, we studied the methylation status of odc gene at the CCGG sequences in c-Ha-rasVal 12 oncogene-transformed NIH-3T3 fibroblasts during the growth cycle and relative to their normal counterparts. We found that the methylation state of the odc gene and its promoter and mid-coding and 3' regions remain unaltered during the cell cycle. We also found that in ras oncogene-transformed cells, which display a more decondensed nucleosomal organization of chromatin than the normal cells, the CCGG sequences in bulk DNA and at the odc gene were methylated to the same extent as in the nontransformed cells. These data suggest that DNA hypomethylation at the CCGG sequences is not a prerequisite for chromatin decondensation and cell transformation by the c-Ha-rasVal 12 oncogene.

Eukaryotic DNA methyltransferases--structure and function

Methylation of DNA plays an important role in the control of gene expression in higher eukaryotes. This is largely achieved by the packaging of methylated DNA into chromatin structures that are inaccessible to transcription factors and other proteins. Methylation involves the addition of a methyl group to the 5-position of the cytosine base in DNA, a reaction catalysed by a DNA (cytosine-5) methyltransferase. This reaction occurs in nuclear replication foci where the chromatin structure is loosened for replication, thereby allowing access to methyltransferases. Partly as a result of their recognising the presence of a methylcytosine on the parental strand following replication, these large enzymes are able to maintain the distribution of methyl groups along the DNA of somatic cells and, thereby, maintain tissue-specific patterns of gene expression.

DNA methylation changes during mouse spermatogenesis

Genomic imprinting in mammals is thought to be mediated by differences in the methylation level of cytosine residues in the genome. These differences in DNA methylation are thought to be generated during the development of the germ line. To characterize the profile of global methylation of the mouse genome during male gametogenesis, we have quantified the relative level of methylation in individual cells during meiosis and spermatogenesis. A decrease in the level of DNA methylation is observed from meiotic cells to elongated spermatids. The erasure of the somatic pattern of methylation during spermatogenesis suggests the existence of a subsequent mechanism generating the parental specific methylation patterns leading to genomic imprinting of specific alleles.

Methylation in fertilised and parthenogenetic preimplantation mouse embryos

DNA methylation is one of the proposed biochemical mechanisms involved in cell differentiation and in genomic imprinting, and DNA methyltransferase (DMT) is a key enzyme in the embryo since mutation of its gene is lethal early in development. In order to verify that non-viability of uniparental embryos was not due to a defect in the regulation of DMT activity, we compared the metabolism of methylation in parthenogenetic embryos (maternal genome) and in fertilised embryos (maternal and paternal genomes). As regards total methylation, estimated by a measure of S-adenosyl methionine (SAM) and S-adenosyl homocysteine (SAH) formation, no significant difference was found between the two kinds of embryos during preimplantation development. Mean values were 4.5 +/- 0.6 fmol (SAM+SAH)/h per 2-cell embryo and 0.40 +/- 0.05 fmol SAH/h per 2-cell embryo, i.e. a SAH/(SAM+SAH) ratio of 9%; there was no detectable SAH formation in blastocysts. The same observation can be made for DMT activity, with mean values of: 7.8 fmol/h per oocyte, 8.5 fmol/h per 2-cell embryo, 6.1 fmol/h per 4-cell embryo, 4.1 fmol/h per morula, and no detectable activity in blastocysts. Total methylation as well as DNA methylation is characterised by a progressive drop in activity during preimplantation development.

Enzymic removal of 5-methylcytosine from DNA by a human DNA-glycosylase

DNA 5-methylcytosine is a major factor in the silencing of mammalian genes; it is involved in gene expression, differentiation, embryogenesis and neoplastic transformation. A decrease in DNA 5-methylcytosine content is associated with activation of specific genes. There is much evidence indicating this to be an enzymic process, with replacement of 5-methylcytosine by cytosine. We demonstrate here enzymic release of 5-methylcytosines from DNA by a human 5-methylcytosine-DNA glycosylase activity, which affords a possible mechanism for such replacement. This activity generates promutagenic apyrimidinic sites, which can be related to the high frequency of mutations found at DNA 5-methylcytosine loci. The recovery of most released pyrimidines as thymines indicates subsequent deamination of free 5-methylcytosines by a 5-methylcytosine deaminase activity. This prevents possible recycling of 5-methylcytosine into replicative DNA synthesis via a possible 5-methyl-dCTP intermediate synthesized through the pyrimidine salvage pathway. Taken together, these findings indicate mechanisms for removal of 5-methylcytosines from DNA, hypermutability of DNA 5-methylcytosine sites, and exclusion of 5-methylcytosines from DNA during replication.

Dynamics of DNA methylation during development

DNA methylation plays a role in the repression of gene expression in animal cells. In the mouse preimplantation embryo, most genes are unmethylated but a wave of de novo methylation prior to gastrulation generates a bimodal pattern characterized by unmethylated CpG island-containing housekeeping genes and fully modified tissue-specific genes. Demethylation of individual genes then takes place during cell type specific differentiation, and this demodification may be a required step in the process of transcriptional activation. DNA modification is also involved in the maintenance of gene repression on the inactive X chromosome in female somatic cells and the marking of parental alleles at genomically imprinted gene loci.

Nuclear extracts of chicken embryos promote an active demethylation of DNA by excision repair of 5-methyldeoxycytidine

Here I show that nuclear extracts of chicken embryos can promote the active demethylation of DNA. The evidence shows that in hemimethylated DNA (i.e., methylated on one strand only) demethylation of 5mCpG occurs through nucleotide excision repair. The first step of demethylation is the formation of specific nicks 5' from 5-methyldeoxycytidine. Nicks are also observed in vitro on symmetrically methylated CpGs (i.e., methylated on both strands) but they result in breakage of the oligonucleotide with no repair. No specific nicks are observed on the nonmethylated CpG. Nicks are strictly 5mCpG specific and do not occur on 5mCpC, 5mCpT, 5mCpA, or 6mApT. The effect of nonspecific nuclease(s) has been ruled out. The nicking of mCpG takes place in the presence of 20 mM EDTA irrespective of the nature of the sequence surrounding the 5mCpG. No methylcytosine glycosylase activity could be detected. The repair is aphidicolin and N-ethylmaleimide resistant, suggesting a repair action by DNA polymerase beta. In extracts of chicken embryos, the excision repair of mCpG is highest between the 6th and the 12th day of development, whereas it is barely detectable in nuclear extracts from different organs of adults. The possible implications of 5mCpG endonuclease activity in active demethylation of DNA during differentiation is discussed.

Identification of two novel mouse nuclear proteins that bind selectively to a methylated c-Myc recognizing sequence

The c-Myc recognizes the sequence CACGTG (Blackwell, T. K., Kretzner, L., Blackwood, E.M., Eisenman, R. N., and Weintraub, H. (1990) Science 250, 1149-1151), and its binding is inhibited by methylation of the core CpG (Prendergast, G. C. and Ziff, E. B. (1991) Science 251, 186-189). We identified two novel nuclear proteins, MMBP-1 and MMBP-2, that bound specifically and under physiological salt condition to the c-Myc binding motif of which cytidine in the CpG sequence was methylated. MMBP-1 was about 42 kD and MMBP-2 was about 63 kD. MMBP-1 was found in specific cells, while MMBP-2 was found in all the cell lines tested, suggesting that MMBP-1 may modulate the role of MMBP-2 in tissue specific manner. We propose that the two proteins play a role in the regulation of c-Myc function through stabilizing or destabilizing the methylation state of the c-Myc binding motif.

Neural modulation of immunity: conditioning phenomena and the adaptability of lymphoid cells

The behavioral conditioning of alterations in the immune response is one pillar supporting the growing edifice of central nervous system (CNS) modulation of immunity. The mechanisms underlying such conditioning phenomena are not understood. In this communication, we attempt to develop a theoretical position based on the concept of phenotypic and functional adaptability of lymphoid cells. We propose that these cells can learn to associate responsiveness to antigens and to other "immunoactive" agents, with responsiveness to signals originating in the CNS delivered via neuroendocrine or autonomic nervous channels. Neural/endocrine signals act on the immune system in conjunction with immunological stimuli, in a way that leads to "storage" of the association (memory) of these two kinds of stimuli in the immune system rather than in the brain.

Biological aspects of cytosine methylation in eukaryotic cells

The existence in eukaryotes of a fifth base, 5-methylcytosine, and of tissue-specific methylation patterns have been known for many years, but except for a general association with inactive genes and chromatin the exact function of this DNA modification has remained elusive. The different hypotheses regarding the role of DNA methylation in regulation of gene expression, chromatin structure, development, and diseases, including cancer are summarized, and the experimental evidence for them is discussed. Structural and functional properties of the eukaryotic DNA cytosine methyltransferase are also reviewed.

Persistence or loss of preimposed methylation patterns and de novo methylation of foreign DNA integrated in transgenic mice

In cultured mammalian cells, foreign DNA can be integrated into the host genome. Foreign DNA is frequently de novo methylated in specific patterns with successive cell generations. The sequence-specific methylation of promoter sequences in integrated foreign DNA is associated with the long-term inactivation of eukaryotic genes. We have now extended these experiments to studies on transgenic mice. As in previous work, a construct (pAd2E2AL-CAT) has been used which consists of the late E2A promoter of adenovirus type 2 (Ad2) DNA fused to the prokaryotic gene for chloramphenicol acetyltransferase (CAT). This construct has been integrated in the non-methylated in the 5'-CCGG-3' premethylated form in the genomes of transgenic mice. DNA from various organs was analyzed by HpaII/MspI cleavage to assess the state of methylation in 5'-CCGG-3' sequences. The results demonstrate that the transgenic construct is in general stable. Non-methylated constructs have remained partly non-methylated for four generations or can become de novo methylated at all or most 5'-CCGG-3' sequences in the founder animal. Preimposed patterns of 5'-CCGG-3' methylation have been preserved for up to four generations beyond the founder animal. In the testes of two different founder animals and two F1 males, the transgenic DNA has become demethylated by an unknown mechanism. In all other organs, the transgenic DNA preserves the preimposed 5'-CCGG-3' methylation pattern. In the experiments performed so far we have not observed differences in the transmission of methylation patterns depending on whether the transgene has been maternally or paternally inherited. The 5'-CCGG-3' premethylated transgene does not catalyze CAT activity in several organs, except in one example of the testes of an animal in which the transgenic construct has become demethylated. In contrast, when the nonmethylated construct has been integrated and remained largely non-methylated, CAT activity has been detected in extracts from some of the organs.

Estradiol down regulates the binding activity of an avian vitellogenin gene repressor (MDBP-2) and triggers a gradual demethylation of the mCpG pair of its DNA binding site

A negative regulating protein (MDBP-2) from rooster liver nuclear extracts binds preferentially to a methylated promoter region 5'TTCACCTTmCGCTATGAGGGGGATCATACTGG3' of the avian vitellogenin II gene (Nucleic Acids Res. 19, 1029-1034, 1991). Treatment of adult and immature roosters with estradiol results in a 90% decrease in the binding activity of MDBP-2 within three days. This corresponds to the level found in egg laying hens. The decrease in the binding activity of MDBP-2 precedes the onset of vitellogenin gene transcription. At the same time, there is a two-fold increase in the binding activity of NHP-1 (tested with the same oligonucleotide as for MDBP-2), a protein thought to be involved in the active demethylation of DNA. The methylated oligonucleotide binds either MDBP-2 or NHP-1 and there is no complex formation between the two proteins and DNA. Estradiol treatment does not change the equilibrium binding constant of MDBP-2 which is about 10(-9)M for the methylated oligonucleotide. The early kinetics of demethylation of the mCpG pair in the binding site of MDBP-2 was studied by means of genomic sequencing. A low level of demethylation of mCpG starts gradually on both DNA strands already 4 hours after estradiol treatment during the lag phase of vitellogenin mRNA synthesis. It is concluded that the lowering of the binding activity of MDBP-2 may have a stronger effect on the derepression of the gene than the slow demethylation of MDBP-2 DNA binding site. The role of the methylated CpG is to assure a high binding affinity of the repressor to DNA.

DNA methylation and gene expression

A large body of evidence demonstrates that DNA methylation plays a role in gene regulation in animal cells. Not only is there a correlation between gene transcription and undermethylation, but also transfection experiments clearly show that the presence of methyl moieties inhibits gene expression in vivo. Furthermore, gene activation can be induced by treatment of cells with 5-azacytidine, a potent demethylating agent. Methylation appears to influence gene expression by affecting the interactions with DNA of both chromatin proteins and specific transcription factors. Although methylation patterns are very stable in somatic cells, the early embryo is characterized by large alterations in DNA modification. New methodologies are now becoming available for studying methylation at this stage and in the germ line. During development, tissue-specific genes undergo demethylation in their tissue of expression. In tissue culture cells this process is highly specific and appears to involve an active mechanism which takes place in the absence of DNA replication. The X chromosome undergoes inactivation during development; this is accompanied by de novo methylation, which appears necessary to stably maintain its silent state. As opposed to the programmed changes in DNA methylation which occur in vivo, immortalized tissue culture cells demonstrate alterations in DNA modification which take place over a long time scale and which appear to be the result of selective pressures present during the growth of these cells in culture.

Characterization of the level, target sites and inheritance of cytosine methylation in tomato nuclear DNA

The tomato nuclear genome was determined to have a G + C content of 37% which is among the lowest reported for any plant species. Non-coding regions have a G + C content even lower (32% average) whereas coding regions are considerably richer in G + C (46%). 5-methyl cytosine was the only modified base detected and on average 23% of the cytosine residues are methylated. Immature tissues and protoplasts have significantly lower levels of cytosine methylation (average 20%) than mature tissues (average 25%). Mature pollen has an intermediate level of methylation (22%). Seeds gave the highest value (27%), suggesting de novo methylation after pollination and during seed development. Based on isoschizomer studies we estimate 55% of the CpG target sites (detected by Msp I/Hpa II) and 85% of the CpNpG target sites (detected by Bst NI/Eco RI) are methylated. Unmethylated target sites (both CpG and CpNpG) are not randomly distributed throughout the genome, but frequently occur in clusters. These clusters resemble CpG islands recently reported in maize and tobacco. The low G + C content and high levels of cytosine methylation in tomato may be due to previous transitions of 5mC----T. This is supported by the fact that G + C levels are lowest in non-coding portions of the genome in which selection is relaxed and thus transitions are more likely to be tolerated. This hypothesis is also supported by the general deficiency of methylation target sites in the tomato genome, especially in non-coding regions. Using methylation isoschizomers and RFLP analysis we have also determined that polymorphism between plants, for cytosine methylation at allelic sites, is common in tomato. Comparing DNA from two tomato species, 20% of the polymorphisms detected by Bst NI/Eco RII could be attributed to differential methylation at the CpNpG target sites. With Msp I/Hpa II, 50% of the polymorphisms were attributable to methylation (CpG and CpNpG sites). Moreover, these polymorphisms were demonstrated to be inherited in a mendelian fashion and to co-segregate with the methylation target site and thus do not represent variation for transacting factors that might be involved in methylation of DNA. The potential role of heritable methylation polymorphism in evolution of gene regulation and in RFLP studies is discussed.

Programmed demethylation in CpG islands during human fetal development

The mechanism for establishing the DNA methylation patterns observed in adult mammalian tissues is not well understood. To determine when adult patterns are established for housekeeping genes, we examined the clustered CpGs in genes on the human active X chromosome (PGK, G6PD, P3, GdX, HPRT) and the autosomal gene, DHFR. We find unique methylation patterns present at the P3 locus in all tissues analyzed from 6- to 9-week fetal specimens, and at the HPRT locus in adrenal gland DNA at this stage of development. Adult patterns are established subsequently by demethylating specific CpGs. Our results show that demethylating events affecting CpG islands are programmed during mammalian fetal development. They suggest that the process of de novo methylation in the fetus methylates at least some sites in the 3' region of the CpG islands in active genes and that adult patterns are established at 6-14 weeks developmental age by sequence-specific demethylation.

Polymerase chain reaction-aided genomic sequencing of an X chromosome-linked CpG island: methylation patterns suggest clonal inheritance, CpG site autonomy, and an explanation of activity state stability

The 5' region of the gene encoding human X chromosome-linked phosphoglycerate kinase 1 (PGK1) is a promoter-containing CpG island known to be methylated at 119 of 121 CpG dinucleotides in a 450-base-pair region on the inactive human X chromosome in the hamster-human cell line X8-6T2. Here we report the use of polymerase chain reaction-aided genomic sequencing to determine the complete methylation pattern of this region in clones derived from X8-6T2 cells after treatment with the methylation inhibitor 5-azacytidine. We find (i) a clone showing full expression of human phosphoglycerate kinase is fully unmethylated in this region; (ii) clones not expressing human phosphoglycerate kinase remain methylated at approximately 50% of CpG sites, with a pattern of interspersed methylated (M) and unmethylated (U) sites different for each clone; (iii) singles, defined as M-U-M or U-M-U, are common; and (iv) a few CpG sites are partially methylated. The data are interpreted according to a model of multiple, autonomous CpG sites, and estimates are made for two key parameters, maintenance efficiency (Em approximately 99.9% per site per generation) and de novo methylation efficiency (Ed approximately 5%). These parameter values and the hypothesis that several independent sites must be unmethylated for transcription can explain the stable maintenance of X chromosome inactivation. We also consider how the active region is kept free of methylation and suggest that transcription inhibits methylation by decreasing Em so that methylation cannot be maintained. Thus, multiple CpG sites, independent with respect to a dynamic methylation system, can stabilize two alternative states of methylation and transcription.

DNA demethylation in erythroleukaemia cells

Despite a fall in the proportion of CGs methylated, evidence has not been obtained for significant demethylation of prelabelled DNA when mouse erythroleukaemia cells are induced to differentiate. There is, however, a delay in the methylation of the DNA that is synthesised in the early period of induction, leading to its undermethylation by 30-50% and this may be a contributory cause of the observed fall in CG methylation.

Eukaryotic DNA methylation: facts and problems

Patterns of DNA methylation in complex genomes like those of mammalian cells have been viewed as indicators of different levels of genetic activities. It is as yet unknown how these complicated patterns are generated and maintained during cell replication. There is evidence from many different biological systems that the sequence-specific methylation of promoters in higher eukaryotes is one of the important factors in controlling gene activity at a long-term level. In general, the fifth nucleotide 5-methyldeoxycytidine can be considered as a modulator of protein-DNA interactions. The degree and direction of this modulation has to be assessed experimentally in each individual instance. The establishment of de novo patterns of DNA methylation is characterized by the gradual non-random spreading of DNA methylation by an essentially unknown mechanism. In this review, some of the general concepts of DNA methylation in mammalian systems are presented, and research currently performed in the authors' laboratory has been summarized.

CpG methylation inhibits proenkephalin gene expression and binding of the transcription factor AP-2

DNA methylation at HpaII (CmCGG) sites inhibits expression of a human proenkephalin-CAT fusion gene when it is transiently expressed in CV-1 cells or stably expressed in C6-glioma cells. The inhibitory effects of HpaII methylation have been mapped to a site within the human proenkephalin promoter located at position -72 relative to the start site of transcription. This region spans a cAMP and phorbol ester inducible enhancer and methylation at this position inhibits both basal transcription and transcription induced by either cAMP or TPA. The HpaII site is located within an element which binds the transcription factor AP-2. In vitro methylation at this HpaII site inhibits the binding of AP-2. These results suggest that CpG methylation inhibits proenkephalin gene expression by directly interfering with the binding of a positively acting transcription factor previously shown to be essential for maximal basal, cAMP, and TPA inducible transcription.

Reversible variations in the methylation pattern of carrot DNA during somatic embryogenesis

Twenty five clones from a carrot genomic library were used as probes to detect variations in the 5-methyl-cytosine pattern during somatic embryogenesis. The majority of them evidenced an invariant pattern. With two clones however, differences were found between the adulttype DNA pattern and the embryonic one. The level of transcription of the relevant DNA fragments during embryogenesis was investigated by Northern blotting. One of the two probes did not show signals in any of the developmental stages whereas the other showed developmentally regulated expression. This fact suggests a novel strategy for cloning developmentally expressed functions.

Hemimethylation of DNA prevents chromatin expression

The activity of hemimethylated herpes simplex virus thymidine kinase DNA and chromatin was analyzed by microinjection and thymidine incorporation into the DNA of thymidine kinase-negative Rat2 cells. Hemimethylated DNA was obtained by in vitro replication of single-stranded M13 DNA constructs and of chromatin produced by in vitro reconstitution of the DNA with purified chicken histone octamers. We found that methylation of either the coding or the noncoding DNA strand was sufficient to block expression of the hemimethylated chromatin. In contrast, the hemimethylated DNA was as active as the unmethylated control DNA after microinjection until chromatin formation occurred in the recipient cells. Microinjection of chromatin hemimethylated by bacterial Hae III methyltransferase excluded the possibility that inactivation was caused by symmetrical methylation of the injected molecules.

The significance of DNA methylation patterns: promoter inhibition by sequence-specific methylation is one functional consequence

This paper presents a review of previously published results from my laboratory on the inactivating or inhibiting function of sequence-specific methylation on promoter activity. In this study, viral promoters, mostly those from adenovirus type 2 (Ad2) or type 12 (Ad12), have been used. It has also been shown that the transcriptional block of these methylated viral promoters can, at least partly, be reversed by a transactivating protein or by the presence of a strong enhancer. We have also adduced evidence that the methylation of the late E2A promoter of Ad2 DNA at positions +6 and +24 from the cap site of this promoter interferes with the binding of one or several proteins at these particular sites, at least when 50 or 73 base-pair long fragments of this promoter have been used for studies on protein binding. With a 377 base-pair fragment, binding differences between the unmethylated and the 5'-CCGG-3' methylated late E2A promoter are not obvious. By applying the genomic sequencing technique developed by Church & Gilbert (1984), the patterns of methylation in all 5'-CG-3' dinucleotides in the late E2A promoter in the active or inactive state in different Ad2-transformed cell lines have been determined. It has been found that 5-methyldeoxycytidine residues introduced into foreign DNA, which is then integrated into the mammalian cell genome, can lead to the spreading of methylation starting from the point of initial methylation. We have begun to investigate whether certain patterns of methylation in mammalian DNA can also influence biological processes other than promoter activity. We have developed a cell-free system using nuclear extracts from hamster cells to study recombination between Ad12 DNA and hamster pre-insertion sites into which Ad12 DNA had previously integrated. The DNAs used in recombination experiments are in the unmethylated or the methylated form. Some speculative aspects have also been discussed in this review. Could existing patterns of methylation in mammalian (human) DNA represent composites of several interdigitating patterns each one of which might have a different signal value? Can a 5-methyldeoxycytidine group in DNA modulate DNA-protein interactions in a positive or negative way, for proteins which could have positively or negatively regulating functions? Patterns of methylation appear to be relatively stable over many years for cell lines propagated in culture. Are patterns of methylation stable also in different parts of the human chromosome? To what extent are these patterns inheritable?

Demethylation of genes in animal cells

Tissue-specific animal cell genes are usually fully methylated in the germ line and become demethylated in those cell types in which they are expressed. To investigate this process, we inserted a methylated IgG kappa gene into fibroblasts and lymphocytes at various stages of development. The results show that this gene undergoes demethylation only in the mature lymphocytes and therefore suggest that the ability to demethylate a gene is developmentally regulated. These studies were supported by similar experiments using the rat Insulin I gene, and in this case it appears that the cis-acting elements that control demethylation may be different from those responsible for gene activation. The ability to demethylate the housekeeping gene APRT is also under developmental control, because this occurs only in embryonic cells, both in tissue culture and in transgenic mice.

Changes in DNA methylation during mouse embryonic development in relation to X-chromosome activity and imprinting

Changing DNA methylation patterns during embryonic development are discussed in relation to differential gene expression, changes in X-chromosome activity and genomic imprinting. Sperm DNA is more methylated than oocyte DNA, both overall and for specific sequences. The methylation difference between the gametes could be one of the mechanisms (along with chromatin structure) regulating initial differences in expression of parental alleles in early development. There is a loss of methylation during development from the morula to the blastocyst and a marked decrease in methylase activity. De novo methylation becomes apparent around the time of implantation and occurs to a lesser extent in extra-embryonic tissue DNA. In embryonic DNA, de novo methylation begins at the time of random X-chromosome inactivation but it continues to occur after X-chromosome inactivation and may be a mechanism that irreversibly fixes specific patterns of gene expression and X-chromosome inactivity in the female. The germ line is probably delineated before extensive de novo methylation and hence escapes this process. The marked undermethylation of the germ line DNA may be a prerequisite for X-chromosome reactivation. The process underlying reactivation and removal of parent-specific patterns of gene expression may be changes in chromatin configuration associated with meiosis and a general reprogramming of the germ line to developmental totipotency.

Tissue specific expression of avian vitellogenin gene is correlated with DNA hypomethylation and in vivo specific protein-DNA interactions

The avian vitellogenin gene is expressed only in the liver of egg-laying hens. It can, however, be activated in immature chicks or roosters by oestradiol. Parallel to the onset of transcription, there is a demethylation of specific mCpGs in the promoter region and in the oestrogen response element (ERE). The methylation pattern in the promoter region is hormone and expression specific, whereas in the ERE it is only hormone and not organ specific. The demethylation occurring in the promoter region is correlated with the appearance of DNase I hypersensitivity sites and changes in the specific protein-DNA interactions. In vivo genomic footprinting of the ERE with varying concentrations of dimethylsulphate revealed, upon gene activation, only minor changes in the protein-DNA interaction. We present evidence that there is another protein that binds with high affinity to the ERE, besides the oestrogen receptor.

DNA methylation. The effect of minor bases on DNA-protein interactions

DNA methylation is found almost ubiquitously in nature and the methyltransferases show evidence of a common evolutionary origin. It will be a fascinating study in protein evolution to follow the ways in which the structures of the various enzymes have developed. Although methylation may have a direct effect on DNA structure the evidence for the importance of this in vivo is accumulating only slowly. In contrast, there is now abundant evidence that methylation of DNA affects DNA-protein interactions and so may have a function in all processes in which such interactions occur. The binding of nucleases is affected in the processes of mismatch repair, DNA restriction and possibly demethylation during differentiation in vertebrates. The binding of transcription factors is affected by DNA methylation and the association of DNA with packaging and segregation proteins may play a part in the control of transcription and replication. The interplay of these effects makes DNA methylation a complex but rewarding area for study. Perhaps we should no longer refer to methylcytosine and methyladenine as minor bases, but rather as key bases which help regulate the functions of DNA.

Cytosine methylation and the fate of CpG dinucleotides in vertebrate genomes

The dinucleotide CpG is a "hotspot" for mutation in the human genome as a result of (1) the modification of the 5' cytosine by cellular DNA methyltransferases and (2) the consequent high frequency of spontaneous deamination of 5-methyl cytosine (5mC) to thymidine. DNA methylation thus contributes significantly, albeit indirectly, to the incidence of human genetic disease. We have attempted to estimate for the first time the in vivo rate of deamination of 5mC from the measured rate of 5mC deamination in vitro and the known error frequency of the cellular G/T mismatch-repair system. The accuracy and utility of this estimate (md) was then assessed by comparison with clinical data, and an improved estimate of md (1.66 X 10(-16) s-1) was derived. Comparison of the CpG mutation rates exhibited by globin gene and pseudogene sequences from human, chimpanzee and macaque provided further estimates of md, all of which were consistent with the first. Use of this value in a mathematical model then permitted the estimation of the length of time required to produce the level of "CpG suppression" currently found in the "bulk DNA" of vertebrate genomes. This time span, approximately 450 million years, corresponds closely to the estimated time since the emergence and adaptive radiation of the vertebrates and thus coincides with the probable advent of heavily methylated genomes. An accurate estimate of the 5mC deamination rate is important not only for clinical medicine but also for studies of gene evolution. Our data suggest both that patterns of vertebrate gene methylation may be comparatively stable over relatively long periods of evolutionary time, and that the rate of CpG deamination can, under certain limited conditions, serve as a "molecular clock".

Promoter inactivation or inhibition by sequence-specific methylation and mechanisms of reactivation

In studies on adenovirus promoters, predominantly on the late E2A promoter of adenovirus type 2 (Ad2), we have demonstrated by a number of experimental approaches that the sequence-specific methylation of three 5'-CCGG-3' sequences inactivates this promoter. Recently, we have developed a cell-free transcription system in which the methylation-inactivation of eukaryotic promoters can be studied in detail. It has also been shown that methylation-caused promoter inactivation can be reversed by the 289 amino acid E1A protein of Ad2 or of adenovirus type 5. In the presence of this protein with a transactivating effect, transcription is initiated at the authentic cap site of the methylated late E2A promoter. A similar reactivation of the methylated late E2A promoter can also be effected by a cis-acting genetic element, i.e., the strong enhancer of human cytomegalovirus. Further studies will be directed toward the biochemical mechanisms of promoter silencing by sequence-specific methylations.

The DNA methylation system in proliferating and differentiated cells

The human melanoma cell line M21 can be induced to differentiate into oligodendrocyte-like cells with concommitant cessation of cell division. Cytosine-arabinoside, 5-aza-2'-deoxycytidine, hydroxyurea, aphidicolin, and phorbol-12-myristate-13-acetate were found to be potent differentiation inducers. We have analyzed the changes of methylation of DNA cytosines that occur after treatment of M21 cells with these compounds. Although DNA methylation levels remain unchanged in the presence of aphidicolin and phorbol ester, 5-aza-2'-deoxycytidine-induced differentiation of these cells results in a 40% DNA demethylation. On the other hand, hydroxyurea and cytosine-arabinoside treatment causes DNA hypermethylation, which, in the case of the cytidine analogue is of only transient nature. These results show that the differentiation of human melanoma cells can be accompanied by variable changes of DNA methylation levels. In another set of experiments, the DNA methylation levels have been analyzed during cytosine-arabinoside-induced differentiation of human K562 erythroleukemia cells. In this system, a transient DNA demethylation precedes the establishment of the differentiated phenotype. Since DNA replication is inhibited, this demethylation cannot be explained by inhibition of the maintenance activity of DNA methyltransferase, but is more likely caused by an active excision of 5-methylcytosine from DNA.

DNA methylation and cell memory

In this paper we address the question: How do replicating mammalian cells remember with high fidelity their proper state of differentiation? Several possible mechanisms for cell memory are discussed, and it is concluded that only mechanisms involving DNA methylation are supported by strong experimental evidence. This evidence is reviewed. The establishment and modulation of methylation patterns are discussed and a hemimethylation model for stem cells is presented. The overall conclusion is that, although little is yet known about the details, there should be little doubt about the existence of a methylation system functioning at least to aid cell memory.

In vitro DNA cytosine methylation of cis-regulatory elements modulates c-Ha-ras promoter activity in vivo

The effect of DNA cytosine methylation on promoter activity was assessed using a transient expression system employing pHrasCAT. This 551 bp Ha-ras-1 gene promoter region is enriched with 84 CpG dinucleotides, six functional GC boxes, and is prototypic of many genes possessing CpG islands in their promoter regions. Bacterial modification enzymes HhaI methyl transferase (MTase) and HpaII MTase, alone or in combination with a human placental DNA methyltransferase (HP MTase) that methylates CpG sites in a generalized manner, including asymmetric elements such as GC box CpG's, were used to methylate at different types of sites in the promoter. Methylation of HhaI and HpaII sites reduced CAT expression by approximately 70%-80%, whereas methylation at generalized CpG sites with HP MTase inactivated the promoter by greater than 95%. The inhibition of H-ras promoter activity was not attributable to methylation-induced differences in DNA uptake or stability in the cell, topological form of the plasmid, or methylation effects in non-promoter regions.

Genomic sequencing reveals a 5-methylcytosine-free domain in active promoters and the spreading of preimposed methylation patterns

Previous work demonstrated an inverse correlation between methylation at the three 5'-CCGG-3' sequences in positions +24, +6, and -215 relative to the cap site of the late E2A promoter of adenovirus type 2 (Ad2) DNA and its activity. In the study presented here, we used the genomic sequencing method to detect 5-methyl-2'-deoxycytidine (m5dC) residues in 5'-CG-3' sequences other than the 5'-CCGG-3' (Hpa II) sites. The patterns of methylation in all 5'-CG-3' sequences over a region of about 180 base pairs required for gene activity in the late E2A promoter of integrated Ad2 DNA were determined in cell lines that carry this promoter in an active or inactive state. In cell lines HE1 and uc2, the late E2A promoter is active and all thirteen 5'-CG-3' sequences between positions +24 and -160 are unmethylated. In cell line HE2, the same promoter is permanently shut off and all 5'-CG-3' sequences are methylated in both strands. Thus, the inverse correlation is perfect in these cell lines over a region of about 180 base pairs in the late E2A promoter. The same promoter segment was analyzed in cell lines mc23 and mc40, in which a late E2A promoter-chloramphenicol acetyltransferase (CAT) gene construct had been genomically fixed after in vitro 5'-CCGG-3' methylation and subsequent transfection. In cell line mc23, the preimposed methylation pattern was stable and the CAT gene was inactive. Genomic sequencing confirmed the presence of m5dC in the 5'-CCGG-3' sequences and revealed the spreading of methylation to neighboring 5'-CG-3' sequences along the entire promoter. Some of these sites were hemimethylated. In cell line mc40, several of the 120 integrated copies became demethylated in positions +24 and +6, but the promoter was methylated in some of the copies upstream of position -50. Cell line mc40 expressed the CAT gene.

Developmentally regulated 75-kilodalton protein expressed in LLC-PK1 cultures is a component of the renal Na+/glucose cotransport system

Na+/D-glucose symport is a secondary active glucose transport mechanism expressed only in kidney proximal tubule and in small intestine. A monoclonal antibody that recognized the Na+/glucose symporter of pig renal brush border membranes also recognized a 75-kD protein in apical membranes isolated from highly differentiated LLC-PK1 cultures, an epithelial cell line of pig renal proximal tubule origin. The 75-kD antigen was enriched from solubilized LLC-PK1 apical membranes by means of high-pressure liquid chromatography. The symporter antigen became apparent on the apical membrane surface after the development of a confluent monolayer in correlation with the expression of transport activity. Long-term treatment of cultures with the differentiation inducer hexamethylene bisacetamide was accompanied by a dramatically increased expression of the symporter antigen as detected quantitatively by Western blot analysis and qualitatively by immunofluorescence staining. The number of symporter-positive cells was dramatically increased after inducer treatment as predicted for differentiation-regulated expression. These results identify a 75-kD protein as a component of a developmentally regulated renal Na+/glucose symporter expressed in cell culture.