Human DNA (cytosine-5)methyltransferase selectively methylates duplex DNA containing mispairs

Maintaining the unmethylated state

BACKGROUND

A remarkable correspondence exists between the cytogenetic locations of the known fragile sites and frequently reported sites of hypermethylation. The best-known features of fragile sites are sequence motifs that are prone to the spontaneous formation of a non-B DNA structure. These facts, coupled with the known enzymological specificities of DNA methyltransferase 1 (DNMT1), the ATP-dependent and actin-dependent helicases, and the ten-eleven translocation (TET) dioxygenases, suggest that these enzymes are involved in an epigenetic cycle that maintains the unmethylated state at these sites by resolving non-B structure, preventing both the sequestration of DNA methyltransferases (DNMTs) and hypermethylation in normal cells.

PRESENTATION OF THE HYPOTHESIS

The innate tendency of DNA sequences present at fragile sites to form non-B DNA structures results in de novo methylation of DNA at these sites that is held in check in normal cells by the action of ATP-dependent and actin-dependent helicases coupled with the action of TET dioxygenases. This constitutes a previously unrecognized epigenetic repair cycle in which spontaneously forming non-B DNA structures formed at fragile sites are methylated by DNMTs as they are removed by the action of ATP-dependent and actin-dependent helicases, with the resulting nascent methylation rendered non-transmissible by TET dioxygenases.

TESTING THE HYPOTHESIS

A strong prediction of the hypothesis is that knockdown of ATP-dependent and actin-dependent helicases will result in enhanced bisulfite sensitivity and hypermethylation at non-B structures in multiple fragile sites coupled with global hypomethylation.

IMPLICATIONS OF THE HYPOTHESIS

A key implication of the hypothesis is that helicases, like the lymphoid-specific helicase and alpha thalassemia/mental retardation syndrome X-linked helicase, passively promote accurate maintenance of DNA methylation by preventing the sequestration of DNMTs at sites of unrepaired non-B DNA structure. When helicase action is blocked due to mutation or downregulation of the respective genes, DNMTs stall at unrepaired non-B structures in fragile sites after methylating them and are unable to methylate other sites in the genome, resulting in hypermethylation at non-B DNA-forming sites, along with hypomethylation elsewhere.

Statistical inference of in vivo properties of human DNA methyltransferases from double-stranded methylation patterns

DNA methyltransferases establish methylation patterns in cells and transmit these patterns over cell generations, thereby influencing each cell's epigenetic states. Three primary DNA methyltransferases have been identified in mammals: DNMT1, DNMT3A and DNMT3B. Extensive in vitro studies have investigated key properties of these enzymes, namely their substrate specificity and processivity. Here we study these properties in vivo, by applying novel statistical analysis methods to double-stranded DNA methylation patterns collected using hairpin-bisulfite PCR. Our analysis fits a novel Hidden Markov Model (HMM) to the observed data, allowing for potential bisulfite conversion errors, and yields statistical estimates of parameters that quantify enzyme processivity and substrate specificity. We apply this model to methylation patterns established in vivo at three loci in humans: two densely methylated inactive X (Xi)-linked loci (FMR1 and G6PD), and an autosomal locus (LEP), where methylation densities are tissue-specific but moderate. We find strong evidence for a high level of processivity of DNMT1 at FMR1 and G6PD, with the mean association tract length being a few hundred base pairs. Regardless of tissue types, methylation patterns at LEP are dominated by DNMT1 maintenance events, similar to the two Xi-linked loci, but are insufficiently informative regarding processivity to draw any conclusions about processivity at that locus. At all three loci we find that DNMT1 shows a strong preference for adding methyl groups to hemi-methylated CpG sites over unmethylated sites. The data at all three loci also suggest low (possibly 0) association of the de novo methyltransferases, the DNMT3s, and are consequently uninformative about processivity or preference of these enzymes. We also extend our HMM to reanalyze published data on mouse DNMT1 activities in vitro. The results suggest shorter association tracts (and hence weaker processivity), and much longer non-association tracts than human DNMT1 in vivo.

ROS1 5-methylcytosine DNA glycosylase is a slow-turnover catalyst that initiates DNA demethylation in a distributive fashion

Arabidopsis ROS1 belongs to a family of plant 5-methycytosine DNA glycosylases that initiate DNA demethylation through base excision. ROS1 displays the remarkable capacity to excise 5-meC, and to a lesser extent T, while retaining the ability to discriminate effectively against C and U. We found that replacement of the C5-methyl group by halogen substituents greatly decreased excision of the target base. Furthermore, 5-meC was excised more efficiently from mismatches, whereas excision of T only occurred when mispaired with G. These results suggest that ROS1 specificity arises by a combination of selective recognition at the active site and thermodynamic stability of the target base. We also found that ROS1 is a low-turnover catalyst because it binds tightly to the abasic site left after 5-meC removal. This binding leads to a highly distributive behaviour of the enzyme on DNA substrates containing multiple 5-meC residues, and may help to avoid generation of double-strand breaks during processing of bimethylated CG dinucleotides. We conclude that the biochemical properties of ROS1 are consistent with its proposed role in protecting the plant genome from excess methylation.

Secondary structure at a hot spot for DNA methylation in DNA from human breast cancers

The VNTR at c-Ha-ras resides in a hotspot for DNA methylation on chromosome 11 in human tumors, where it is flanked by two MspI restriction sites. We have investigated the nature of the MspI site polymorphism at the c-Ha-ras VNTR observed in variety of tumors including breast cancer.We find that the MspI site 5' to the VNTR is present in a Non-B DNA structure with single-strand character that renders it accessible to bisulfite modification under native conditions, while the MspI site 3' to the VNTR appears to reside in a normal B-form structure that is inaccessible to bisulfite. The non-B DNA structure accounts for the observed polymorphism since MspI cannot cleave single-stranded DNA and control experiments show that the MspI sites were neither mutated nor abnormally methylated. Southern blotting showed that structural polymorphism was present in tumor DNA and tumor adjacent normal tissue DNA but absent from lymphocyte DNA from the same patients. We conclude that the non-B DNA structural polymorphism detected in human tumors near the c-Ha-ras VNTR is a self-perpetuating epigenetic mark that manifests itself spontaneously during breast carcinogenesis in a methylation hot spot.

Anomalies épigénétiques et carcinogenèse hépatique

The molecular mechanisms involved in liver carcinogenesis are poorly understood. Over the past decade, epigenetic changes (DNA methylation) have received increasing attention for their potential involvement in the development of hepatocarcinoma. The DNA methylation level is influenced by environmental factors (folate and methionine diet), as well as by genetic factors (methylenetetrahydrofolate reductase/MTHFR polymorphisms). These findings provide new insight into the understanding of liver carcinogenesis. Interventional studies are now required to determine the role of folate supplementation in the development of liver tumors in targeted patients.

Transgene-induced CCWGG methylation does not alter CG methylation patterning in human kidney cells

Several reports suggest that C(m)CWGG methylation tends not to co-exist with (m)CG methylation in human cells. We have asked whether or not methylation at CCWGG sites can influence CG methylation. DNA from cells expressing an M.EcoRII-GFP fusion was actively methylated at CCWGG sites. CG methylation as measured by R.HpaII/R.MspI ratios was unchanged in cells expressing the transgene. Cloned representatives of C(m)CWGG methylated DNA often contained, or were adjacent to an ALU repeat, suggesting that M.EcoRII-GFP actively methylated gene-rich R-band DNA. The transgenic methyltransferase applied C(m)CWGG methylation to a representative human promoter that was heavily methylated at CG dinucleotides (the SERPINB5 promoter) and to a representative promoter that was essentially unmethylated at CG dinucleotides (the APC promoter). In each case, the CG methylation pattern remained in its original state, unchanged by the presence of neighboring C(m)CWGG sites. Q-PCR measurements showed that RNA expression from the APC gene was not significantly altered by the presence of C(m)CWGG in its promoter. Kinetic studies suggested that an adjacent C(m)CWGG methylation site influences neither the maintenance nor the de novo methylation activities of purified human Dnmt1. We conclude that C(m)CWGG methylation does not exert a significant effect on CG methylation in human kidney cells.

The use of prokaryotic DNA methyltransferases as experimental and analytical tools in modern biology

Prokaryotic DNA methyltransferases (MTases) are used as experimental and research tools in molecular biology and molecular genetics due to their ability to recognize and transfer methyl groups to target bases in specific DNA sequences. As a practical tool, prokaryotic DNA MTases can be used in recombinant DNA technology for in vitro alteration and enhancing of cleavage specificity of restriction endonucleases. The ability of prokaryotic DNA MTases to methylate cytosine residues in specific sequences, which are also methylated in eukaryotic DNA, makes it possible to use them as analytical reagent for determination of the site-specific level of methylation in eukaryotic DNA. In vivo DNA methylation by prokaryotic DNA MTases is used in different techniques for probing chromatin structure and protein-DNA interactions. Additional prospects are opened by development of the methods of DNA methylation targeted to predetermined DNA sequences by fusion of DNA MTases to DNA binding proteins. This review will discuss the application of prokaryotic DNA MTases of Type II in the methods and approaches mentioned above.

Mechanisms of DNA damage, DNA hypomethylation, and tumor progression in the folate/methyl-deficient rat model of hepatocarcinogenesis

Using the folate/methyl-deficient rat model of hepatocarcinogenesis, we obtained evidence that may provide new insights into a major unresolved paradox in DNA methylation and cancer research: the mechanistic basis for genome-wide hypomethylation despite an increase in DNA methyltransferase activity and gene-specific regional hypermethylation. Previous studies revealed that the methyltransferase binds with higher affinity to DNA strand breaks, gaps, abasic sites, and uracil than it does to its cognate hemimethylated CpG sites, consistent with its ancestral function as a DNA repair enzyme. These same DNA lesions are an early occurrence in models of folate and methyl deficiency and are often present in human preneoplastic cells. We hypothesized that the high-affinity binding of the maintenance DNA methyltransferase to unrepaired lesions in DNA could sequester available enzyme away from the replication fork and promote passive replication-dependent demethylation. In support of this possibility, we found that lesion-containing DNA is less efficiently methylated than lesion-free DNA from folate/methyl-deficient rats and that an increase in DNA strand breaks precedes DNA hypomethylation. Despite an adaptive increase in DNA methyltransferase activity, hemimethylated DNA from folate/methyl-deficient rats is progressively replaced by double-stranded unmethylated DNA that is resistant to remethylation with dietary methyl repletion. In promoter regions, the inappropriate binding of the DNA methyltransferase to unrepaired lesions or mispairs may promote local histone deacetylation, methylation, and regional hypermethylation associated with tumor suppressor gene silencing. These insights in an experimental model are consistent with the possibility that DNA lesions may be a necessary prerequisite for the disruption of normal DNA methylation patterns in preneoplastic and neoplastic cells.

Methods for the design and analysis of oligodeoxynucleotide-based DNA (cytosine-5) methyltransferase inhibitors

Several second-generation inhibitors of DNA (cytosine-5) methyltransferases based on studies of modified synthetic oligodeoxynucleoides have been described. As an aid to studies of these inhibitors, we present an electronic structure-based algorithm that can be used as a method for predicting the nature of the expected inhibition by any noncytosine nucleotide target. Targeting by the major human enzyme (hDnmt1) is governed by the presence of a three-nucleotide motif. In hemimethylated DNA, this motif consists of a 5-methylcytosine targeting signal that causes the enzyme to probe the opposite strand for a normally paired guanosine or inosine residue and attempt to methylate the residue 5' to that site. As a demonstration of the method, we apply these rules to the design and characterization of a novel oligodeoxynucleotide inhibitor of hDnmt1. This inhibitor takes advantage of the three-nucleotide recognition motif characteristic of hDnmt1 and shows that the enzyme is inhibited in vitro by non-CG methylation which targets the enzyme to normally basepaired but unproductive nucleotides such as dG, dA, and dT. Kinetic analysis at constant S-adenosyl-L-methionine concentration shows that representative inhibitory oligodeoxynucleotides are best viewed as weakly productive components of systems containing two DNA substrates. This model suggests that the most effective inhibitors are those with very low apparent Vmax and very low Km values. Oligodeoxynucleotides containing mispaired and unproductive targets such as dG, dA, dT, and dU are also inhibitory as secondary substrates for the human enzyme. Biologically, fail-safe mechanisms identified by the ab initio approach appear to be active in preventing potentially mutagenic deamination of dihydrocytosine and enzymatic methylation of dU.

Zebularine: a novel DNA methylation inhibitor that forms a covalent complex with DNA methyltransferases

Mechanism-based inhibitors of enzymes, which mimic reactive intermediates in the reaction pathway, have been deployed extensively in the analysis of metabolic pathways and as candidate drugs. The inhibition of cytosine-[C5]-specific DNA methyltransferases (C5 MTases) by oligodeoxynucleotides containing 5-azadeoxycytidine (AzadC) and 5-fluorodeoxycytidine (FdC) provides a well-documented example of mechanism-based inhibition of enzymes central to nucleic acid metabolism. Here, we describe the interaction between the C5 MTase from Haemophilus haemolyticus (M.HhaI) and an oligodeoxynucleotide duplex containing 2-H pyrimidinone, an analogue often referred to as zebularine and known to give rise to high-affinity complexes with MTases. X-ray crystallography has demonstrated the formation of a covalent bond between M.HhaI and the 2-H pyrimidinone-containing oligodeoxynucleotide. This observation enables a comparison between the mechanisms of action of 2-H pyrimidinone with other mechanism-based inhibitors such as FdC. This novel complex provides a molecular explanation for the mechanism of action of the anti-cancer drug zebularine.

Analysis of hypermethylation in the RPS element suggests a signal function for short inverted repeats in de novo methylation

A repetitive DNA sequence (RPS) from Petunia hybrida had previously been shown to enhance expression variegation in petunia and tobacco and to carry a hot spot for de novo DNA methylation. Here we show that a strong de novo hypermethylation site is located within a palindromic segment of the RPS and present indirect evidence, based on sequence homologies to other repeat units within the RPS, for the formation of secondary structures at the methylation site in vivo. We demonstrate that the palindromic RPS element, which is moderately to highly repetitive in petunia, does not predominantly localise to constitutive heterochromatin. To test whether the RPS is subject to de novo methylation due to its repetitive nature or to intrinsic signals within the RPS, we integrated the RPS into the genome of Arabidopsis thaliana, a plant lacking homology to the RPS. Our data indicate that the palindromic element also acts as a de novo hypermethylation site in the non-repetitive genomic background of Arabidopsis, strongly suggesting a signal function of the palindromic RPS element.

Fragile X syndrome and Friedreich's ataxia: two different paradigms for repeat induced transcript insufficiency

DNA repeat expansion is the genetic basis for a growing number of neurological disorders. While the largest subset of these diseases results in an increase in the length of a polyglutamine tract in the protein encoded by the affected gene, the most common form of inherited mental retardation, fragile X syndrome, and the most common inherited ataxia, Friedreich's ataxia, are both caused by expansions that are transcribed but not translated. These expansions both decrease expression of the gene in which the expanded repeat is located, but they do so by quite different mechanisms. In fragile X syndrome, CGG. CCG expansion in the 5' untranslated region of the FMR1 gene leads to hypermethylation of the repeats and the adjacent CpG-rich promoter. Methylation prevents the binding of the transcription factor alpha-Pal/NRF-1, and may indirectly affect the binding of other factors via the formation of transcriptionally silent chromatin. In Friedreich's ataxia, GAA. TTC expansion in an intron of the FRDA gene reduces expression by interfering with transcription elongation. The model that best describes the available data is transcription-driven formation of a transient purine. purine. pyrimidine DNA triplex behind an advancing RNA polymerase. This structure lassoes the RNA polymerase that caused it, trapping the enzyme on the template.

An unexpectedly high excision capacity for mispaired 5-hydroxymethyluracil in human cell extracts

The oxidation of thymine in DNA can generate a base pair between 5-hydroxymethyluracil (HmU) and adenine, whereas the oxidation and deamination of 5-methylcytosine (5mC) in DNA can generate a base pair between HmU and guanine. Using synthetic oligonucleotides containing HmU at a defined site, HmU-DNA glycosylase activities in HeLa cell and human fibroblast cell extracts have been observed. An HmU-DNA glycosylase activity that removes HmU mispaired with guanine has been measured. Surprisingly, the HmU:G excision activity is 60 times greater than the corresponding HmU:A activity, even though the expected rate of formation of the HmU:A base pair exceeds that of the HmU:G base pair by a factor of 10(7). The HmU:G mispair would arise from the 5mC:G base pair, and, if unrepaired, would give rise to a transition mutation. The observation of an unexpectedly high HmU:G glycosylase activity suggests that human cells may encounter the HmU:G mispair much more frequently than expected. The conversion of 5mC to HmU must be considered as a potential pathway for the generation of 5mC to T transition mutations, which are often found in human tumors.

Examination of the DNA substrate selectivity of DNA cytosine methyltransferases using mass tagging

The biological significance of cytosine methylation is as yet incompletely understood, but substantial and growing evidence strongly suggests that perturbation of methylation patterns, resulting from the infidelity of DNA cytosine methyltransferase, is an important component of the development of human cancer. We have developed a novel in vitro assay that allows us to quantitatively determine the DNA substrate preferences of cytosine methylases. This approach, which we call mass tagging, involves the labeling of target cytosine residues in synthetic DNA duplexes with stable isotopes, such as (15)N. Methylation is then measured by the formation of 5-methylcytosine (5mC) by gas chromatography/mass spectrometry. The DNA substrate selectivity is determined from the mass spectrum of the product 5mC. With the non-symmetrical duplex DNA substrate examined in this study we find that the bacterial methyltransferase HPA:II (duplex DNA recognition sequence CCGG) methylates the one methylatable cytosine of each strand similarly. Introduction of an A-C mispair at the methylation site shifts methylation exclusively to the mispaired cytosine residue. In direct competition assays with HPA:II methylase we observe that the mispaired substrate is methylated more extensively than the fully complementary, normal substrate, although both have one HPA:II methylation site. Through the use of this approach we will be able to learn more about the mechanisms by which methylation patterns can become altered.

Mechanistic link between DNA methyltransferases and DNA repair enzymes by base flipping

Rotation of a DNA nucleotide out of the double helix and into a protein binding pocket ("base flipping") was first observed in the structure of a DNA methyltransferase. There is now evidence that a variety of proteins, particularly DNA repair enzymes, use base flipping in their interactions with DNA. Though the mechanisms for base movement into extrahelical positions are still unclear, the focus of this review is how base recognition is modulated by the stringency of binding to the extrahelical base(s) or sugar moiety.

Mechanism-based inhibition of C5-cytosine DNA methyltransferases by 2-H pyrimidinone

DNA duplexes in which the target cytosine base is replaced by 2-H pyrimidinone have previously been shown to bind with a significantly greater affinity to C5-cytosine DNA methyltransferases than unmodified DNA. Here, it is shown that 2-H pyrimidinone, when incorporated into DNA duplexes containing the recognition sites for M.HgaI-2 and M.MspI, elicits the formation of inhibitory covalent nucleoprotein complexes. We have found that although covalent complexes are formed between 2-H pyrimidinone-modified DNA and both M.HgaI-2 and M.MspI, the kinetics of complex formation are quite distinct in each case. Moreover, the formation of a covalent complex is still observed between 2-H pyrimidinone DNA and M.MspI in which the active-site cysteine residue is replaced by serine or threonine. Covalent complex formation between M.MspI and 2-H pyrimidinone occurs as a direct result of nucleophilic attack by the residue at the catalytic position, which is enhanced by the absence of the 4-amino function in the base. The substitution of the catalytic cysteine residue by tyrosine or chemical modification of the wild-type enzyme with N-ethylmaleimide, abolishes covalent interaction. Nevertheless the 2-H pyrimidinone-substituted duplex still binds to M.MspI with a greater affinity than a standard cognate duplex, since the 2-H pyrimidinone base is mis-paired with guanine.

Protein-mediated base flipping

Since the discovery that the DNA methyltransferase M.HhaI utilises a base flipping mechanism to expose its target cytosine during catalysis, this phenomenon has been observed in other nucleic acid modifying enzymes. The crystallographic analyses of such enzyme-DNA complexes have revealed the molecular features of extrahelical base stabilisation, but have been less informative about the flipping process itself.

Involvement of DNA methylation in human carcinogenesis

It is now generally accepted that the presence of 5-methylcytosine (5mC) in human DNA has both a genetic and an epigenetic effect on cellular development, differentiation and transformation. First, 5mC is more unstable than its unmethylated counterpart cytosine. Hydrolytic deamination of 5mC leads to a G/T mismatch and subsequently, if unrepaired, to a C-->T transition mutation. Sites of DNA methylation are mutational hotspots in many human tumors. Second, DNA methylation of promoter regions is often correlated with the down regulation of the corresponding gene. Both of these effects have fundamental consequences for basic functions of the cell like cellular differentiation, the development of cancer and possibly other diseases, and on the evolutionary process. Recent hypotheses also propose a role for methylation in the process of aging. In this review we will describe recent findings and hypotheses about the function of 5mC in DNA with the focus on its involvement in human carcinogenesis.

Hairpin induced slippage and hyper-methylation of the fragile X DNA triplets

The fragile X triplet repeats, (GCC)n x (GGC)n are located at the 5' untranslated region of the FMR-1 gene. Inordinate repeat expansion and hyper-methylation of the CpG islands inside the repeat lead to the suppression of the FMR-1 gene and the subsequent onset and progression of the disease. Previously, we have shown that the (GCC)n strand of the fragile X repeat readily forms hairpin structures under physiological conditions (Chen et al., Proc. Natl. Acad. Sci. USA, 92:5199-5203, 1995: Mariappan et al., Nucl. Acid Res. 24:784-792, 1996). Here, we show by an in vitro assay that formation of the (GCC)n hairpins leads to slippage during replication. The slippage structure is a three-way junction with two Watson-Crick, (GCC)n x (GGC)n, arms and a third (GCC)n hairpin arm. Formation of such slippage structures during replication may explain the observed length polymorphism of the fragile X repeat. We have also studied the substrate efficiency of these three-way junctions toward the human methyltransferase. the enzyme that methylates the CpG sites in DNA. These methylation studies show that the slippage structures induced by the (GCC)n hairpins are 10-15 times more efficient substrates than either the corresponding Watson-Crick duplexes or the (GCC)n hairpins. We demonstrate by appropriate designs that the exceptional substrate efficiency of the three-way junction slippage structures is due to two factors: (i) the presence of the (GCC)n hairpin in which CpG sites are more accessible for methylation than the CpG sites in the Watson-Crick duplex and (ii) the ability of the (GCC)n hairpin in these three-way junctions to move along the Watson-Crick arms that facilitates conversion of low-affinity Watson-Crick CpG sites into high-affinity hairpin CpG sites. Therefore, we suggest that the formation of the (GCC)n hairpins during replication can explain both length polymorphism and hyper-methylation of the fragile X repeats.

Stalling of human DNA (cytosine-5) methyltransferase at single-strand conformers from a site of dynamic mutation

Single-strand conformers (SSCs) from the C-rich strand of the triplet repeat at the FMR-1 locus are rapidly and selectively methylated by the human DNA (cytosine-5) methyltransferase. The apparent affinity of the enzyme for the FMR-1 SSC is about tenfold higher than it is for a control Watson-Crick paired duplex. The de novo methylation rate for the SSC is over 150-fold higher than the de novo rate for the control duplex. Methylation of what is generally called a hemi-methylated duplex occurs with a rate enhancement of over 100-fold, while methylation of what can be viewed as a hemi-methylated FMR-1 SSC is actually slower than the de novo rate. The pronounced inhibition of the methyltransferase by the methylated SSC suggests that the enzyme has a higher affinity for the methylated product of its reaction with the SSC than it has for the unmethylated SSC substrate. Gel retardation studies show that the methyltransferase binds selectively to SSCs from the C-rich strand of the FMR-1 triplet repeat. This suggests a two-step stalling process in which the human methyltransferase first selectively methlyates and subsequently stalls at the C-rich strand SSC. Stalling may reflect the inability of the enzyme to release a DNA product that is fixed in a conformation resembling its transition state by the unusual structure of the substrate. In particular, the data suggest that DNA methyltransferase may physically participate in biological processes that lead to dynamic mutation at FMR-1. In general, the data raise the possibility that a two-step stalling process occurs at secondary structures associated with chromosome instability, chromosome remodelling, viral replication or viral integration and may account for the local hypermethylation and global hypomethylation associated with viral and non-viral tumorigenesis.

Control of methylation spreading in synthetic DNA sequences by the murine DNA methyltransferase

Methylation spreading, which involves a propensity for the mammalian DNA-(cytosine-5)-methyltransferase to de novo methylate cytosine-guanine dinucleotides (CpGs) near pre-existing 5-methylcytosine bases, has been implicated in the control of numerous biological processes. We have assessed methylation spreading by the murine DNA methyltransferase in vitro using synthetic copolymers and oligonucleotides which differ only in their methylation state. Double-stranded oligonucleotides were found to undergo higher levels of de novo methylation overall than otherwise identical single-stranded oligonucleotides. This difference reflects the greater number of de novo methylatable cytosine bases in double-stranded than single-stranded sequences. All tested oligonucleotides containing pre-existing 5-methyl-cytosine(s) were de novo methylated at several fold the rates of non-methylated controls. No mammalian proteins besides the DNA methyltransferase were required for this observed enhancement of de novo methylation. Studies using oligonucleotides differing in patterns of pre-methylation showed that methylation spreading can be initiated by hemimethylated or duplex methylated CpGs indicating that recognition of 5-methylcytosine by the enzyme is sufficient to stimulate methylation spreading. Double and single-stranded oligonucleotides with several bases between CpGs underwent considerably more de novo methylation per CpG than sequences containing sequential uninterrupted methylatable sites. Spacing preferences by the DNA methyltransferase were also observed in hemimethylated oligonucleotides, suggesting that this is a general property of the enzyme. Although methylation spreading outside of CpG dinucleotides was relatively rare, single-stranded DNA incurred higher levels of de novo methylation at sites other than CpG as compared to double-stranded DNA. This indicates less specificity of methylation spreading in single-stranded sequences. Finally, enhanced de novo methylation in the presence of fully methylated CpG sites in double-stranded oligonucleotides was not as high as the rates of methylation of hemimethylated CpGs in otherwise identical oligonucleotides. These studies provide further elucidation of the mechanisms and regulation of the methylation spreading process and its potential role in the biological processes it influences.

Stalling of human methyltransferase at single-strand conformers from the Huntington's locus

We describe evidence for a sequence of events in which the Human DNA(cytosine-5)methyl-transferase first methylates spontaneous single-stranded conformers (SSCs) and then stalls at the methylated site to produce a complex with the conformationally unusual DNA. This property of the enzyme is a result of its ability to respond to a general loss of symmetry at its CG recognition site. The data suggest that DNA methyltransferase, itself, may physically participate in biological processes that distinguish between DNA that is in the normal Watson-Crick paired conformation and DNA that is conformationally unusual (e.g. a hairpin loop or misassembled replication intermediate). The in vitro methylation of spontaneous SSCs from the Huntington's locus illustrates the phenomenon.

Mammalian DNA (cytosine-5-)-methyltransferase expressed in Escherichia coli, purified and characterized

Besides modulating specific DNA-protein interactions, methylated cytosine, frequently referred to as the fifth base of the genome, also influences DNA structure, recombination, transposition, repair, transcription, imprinting, and mutagenesis. DNA (cytosine-5-)-methyltransferase catalyzes cytosine methylation in eukaryotes. We have cloned and expressed this enzyme in Escherichia coli, purified it to apparent homogeneity, characterized its properties, and we have shown that it hemimethylates DNA. The cDNA for murine maintenance methyltransferase was reconstructed and cloned for direct expression in native form. Immunoblotting revealed a unique protein (M(r) = 190,000) not present in control cells. The mostly soluble overexpressed protein was purified by DEAE, Sephadex, and DNA cellulose chromatography. Peak methylating activity correlated with methyltransferase immunoblots. The purified enzyme preferentially transferred radioactive methyl moieties to hemimethylated DNA in assays and on autoradiograms. All of the examined properties of the purified recombinant DNA methyltransferase are consistent with the enzyme purified from mammalian cells. Further characterization revealed enhanced in vitro methylation of premethylated oligodeoxynucleotides. The cloning of hemimethyltransferase in E. coli should allow facilitated structure-function mutational analysis of this enzyme, studies of its biological effects in prokaryotes, and potential large scale methyltransferase production for crystallography, and it may have broad applications in maintaining the native methylated state of cloned DNA.

Hairpins are formed by the single DNA strands of the fragile X triplet repeats: structure and biological implications

Inordinate expansion and hypermethylation of the fragile X DNA triplet repeat, (GGC)n.(GCC)n, are correlated with the ability of the individual G- and C-rich single strands to form hairpin structures. Two-dimensional NMR and gel electrophoresis studies show that both the G- and C-rich single strands form hairpins under physiological conditions. This propensity of hairpin formation is more pronounced for the C-rich strand than for the G-rich strand. This observation suggests that the C-rich strand is more likely to form hairpin or "slippage" structure and show asymmetric strand expansion during replication. NMR data also show that the hairpins formed by the C-rich strands fold in such a way that the cytosine at the CpG step of the stem is C.C paired. The presence of a C.C mismatch at the CpG site generates local flexibility, thereby providing analogs of the transition to the methyltransferase. In other words, the hairpins of the C-rich strand act as better substrates for the human methyltransferase than the Watson-Crick duplex or the G-rich strand. Therefore, hairpin formation could account for the specific methylation of the CpG island in the fragile X repeat that occurs during inactivation of the FMR1 gene during the onset of the disease.

Methylation of slipped duplexes, snapbacks and cruciforms by human DNA(cytosine-5)methyltransferase

When human DNA(cytosine-5)methyltransferase was used to methylate a series of snapback oligodeoxy-nucleotides of differing stem lengths, each containing a centrally located CG dinucleotide recognition site, the enzyme required a minimum of 22 base pairs in the stem for maximum activity. Extrahelical cytosines in slipped duplexes that were 30 base pairs in length acted as effective methyl acceptors and were more rapidly methylated than cytosines that were Watson-Crick paired. Duplexes containing hairpins of CCG repeats in cruciform structures in which the enzyme recognition sequence was disrupted by a C.C mispair were also more rapidly methylated than control Watson-Crick-paired duplexes. Since enzymes have higher affinities for their transition states than for their substrates, the results with extrahelical and mispaired cytosines suggest that these structures can be viewed as analogs of the transition state intermediates produced during catalysis by methyltransferases.

M.HhaI binds tightly to substrates containing mismatches at the target base

The (cytosine-5) DNA methyltransferase M.HhaI causes its target cytosine base to be flipped completely out of the DNA helix upon binding. We have investigated the effects of replacing the target cytosine by other, mismatched bases, including adenine, guanine, thymine and uracil. We find that M.HhaI binds more tightly to such mismatched substrates and can even transfer a methyl group to uracil if a G:U mismatch is present. Other mismatched substrates in which the orphan guanine is changed exhibit similar behavior. Overall, the affinity of DNA binding correlates inversely with the stability of the target base pair, while the nature of the target base appears irrelevant for complex formation. The presence of a cofactor analog. S-adenosyl-L-homocysteine, greatly enhances the selectivity of the methyltransferase for cytosine at the target site. We propose that the DNA methyltransferases have evolved from mismatch binding proteins and that base flipping was, and still is, a key element in many DNA-enzyme interactions.

Mechanism of human methyl-directed DNA methyltransferase and the fidelity of cytosine methylation

The properties of the methyl-directed DNA (cytosine-5-)-methyltransferase (EC 2.1.1.37) suggest that it is the enzyme that maintains patterns of methylation in the human genome. Proposals for the enzyme's mechanism of action suggest that 5-methyldeoxycytidine is produced from deoxycytidine via a dihydrocytosine intermediate. We have used an oligodeoxynucleotide containing 5-fluorodeoxycytidine as a suicide substrate to capture the enzyme and the dihydrocytosine intermediate. Gel retardation experiments demonstrate the formation of the expected covalent complex between duplex DNA containing 5-fluorodeoxycytidine and the human enzyme. Formation of the complex was dependent upon the presence of the methyl donor S-adenosylmethionine, suggesting that it comprises an enzyme-linked 5-substituted dihydrocytosine moiety in DNA. Dihydrocytosine derivatives are extremely labile toward hydrolytic deamination in aqueous solution. Because C-to-T transition mutations are especially prevalent at CG sites in human DNA, we have used high-performance liquid chromatography to search for thymidine that might be generated by hydrolysis during the methyl transfer reaction. Despite the potential for deamination inherent in the formation of the intermediate, the methyltransferase did not produce detectable amounts of thymidine. The data suggest that the ability of the human methyltransferase to preserve genetic information when copying a methylation pattern (i.e., its fidelity) is comparable to the ability of a mammalian DNA polymerase to preserve genetic information when copying a DNA sequence. Thus the high frequency of C-to-T transitions at CG sites in human DNA does not appear to be due to the normal enzymatic maintenance of methylation patterns.

Recognition of unusual DNA structures by human DNA (cytosine-5)methyltransferase

The symmetry of the responses of the human DNA (cytosine-5)methyltransferase to alternative placements of 5-methylcytosine in model oligodeoxynucleotide duplexes containing unusual structures has been examined. The results of these experiments more clearly define the DNA recognition specificity of the enzyme. A simple three-nucleotide recognition motif within the CG dinucleotide pair can be identified in each enzymatically methylated duplex. The data can be summarized by numbering the four nucleotides in the dinucleotide pair thus: 1 4/2 3. With reference to this numbering scheme, position 1 can be occupied by cytosine or 5-methylcytosine; position 2 can be occupied by guanosine or inosine; position 3, the site of enzymatic methylation, can be occupied only by cytosine; and position 4 can be occupied by guanosine, inosine, O6-methylguanosine, cytosine, adenosine, an abasic site, or the 3' hydroxyl group at the end of a gapped molecule. Replacing the guanosine normally found at position 4 with any of the moieties introduces unusual (non-Watson-Crick) pairing at position 3 and generally enhances methylation of the cytosine at that site. The exceptional facility of the enzyme in actively methylating unusual DNA structures suggests that the evolution of the DNA methyltransferase, and perhaps DNA methylation itself, may be linked to the biological occurrence of unusual DNA structures.

Interaction of oligonucleotides containing 6-O-methylguanine with human DNA (cytosine-5-)-methyltransferase [published erratumm appears in Biochemistry 1992 Aug 4;31(30):7008]

Thirty-base-pair synthetic oligonucleotide duplexes containing a single meG.C (meG = 6-O-methylguanine) or A.C base pair at the 16th position (i.e., 5'-CCCGTTTAAATATACXTATACCCGGGTACC-3', where X = A or meG) were used to study de novo methylation by the purified human DNA (cytosine-5)-methyltransferase isolated from CEM cells. Both duplexes containing meG.C and A.C base pairs show enhanced methyl group acceptor properties. Subsequent introduction of hemimethylated sites at the 15th position of the top strand (the C residue next to the abnormal base pair) and the 7th, 15th (which represents the C residue in the 6meG.C and A.C base pairs), and 27th positions of the bottom strand were used to study the maintenance methylation of the hemimethylated duplexes by the methylase. This revealed striking differences in the rate, amount, and sites of methylation, which are dependent on the position of the hemimethylated site in the duplex. The possible mechanism of action of the methylase is discussed. The data show that 6-O-methylguanine residues in DNA can have other genetic effects apart from their miscoding behavior and that meG.C and A.C base pairs exert different effects in terms of methylation.

A G4-DNA/B-DNA junction at codon 12 of c-Ha-ras is actively and asymmetrically methylated by DNA(cytosine-5)methyltransferase

Oligodeoxynucleotides spanning codon 12 of the human c-Ha-ras gene were found to be exceptionally good substrates for de novo methylation by human DNA(cytosine-5)methyltransferase. In the complex formed by two complementary 30mers, only the C-rich strand was methylated by the enzyme. Guanines at the 3' end of the G-rich strand of the complex could not be completely modified by dimethyl sulfate [corrected] suggesting tetrameric bonding at these G-residues. An eight-stranded structure, composed of four duplex DNAs at one end, joined to a G4-DNA segment at the other with the junction between the two DNA forms at codon 12, can account for our results.

Quantitative evaluation of Escherichia coli host strains for tolerance to cytosine methylation in plasmid and phage recombinants

Many strains of E. coli K12 restrict DNA containing cytosine methylation such as that present in plant and animal genomes. Such restriction can severely inhibit the efficiency of cloning genomic DNAs. We have quantitatively evaluated a total of 39 E. coli strains for their tolerance to cytosine methylation in phage and plasmid cloning systems. Quantitative estimations of relative tolerance to methylation for these strains are presented, together with the evaluation of the most promising strains in practical recombinant cloning situations. Host strains are recommended for different recombinant cloning requirements. These data also provide a rational basis for future construction of 'ideal' hosts combining optimal methylation tolerance with additional advantageous mutations.

Structure of mouse DNA (cytosine-5-)-methyltransferase

DNA (cytosine-5-)-methyltransferase was purified as a single polypeptide (190 kDa by SDS-PAGE) from mouse P815 mastocytoma cells. This enzyme transfers methyl groups to unmethylated as well as to hemimethylated DNA sites with a strong preference for the hemimethylated substrate. A structural analysis of the isolated enzyme by electron microscopical techniques was undertaken. On the basis of the results obtained, we propose a model for the enzyme structure. This model describes the enzyme as a hemi-elliptical globular structure with dimensions of 5.4-6.7 nm for the height h and 10.3-10.8 nm for the diameter d, respectively; this globular structure bears a small appendix at the flat side. A molecular mass of 235-250 kDa is calculated from the measured dimensions. Limited trypsin digestion of the enzyme led to a 160-kDa fragment which preserved the gross morphology of the original material. The possible structure function relationships are discussed.

RglB facilitated cloning of highly methylated eukaryotic DNA: the human L1 transposon, plant DNA, and DNA methylated in vitro with human DNA methyltransferase

In vitro methylation of Bluescribe plasmid DNA (pBS) with human placental DNA methyltransferase to 6% 5-methylcytosine (mC) reduced transformation efficiencies in rglB+ host strains C600 and DS410 by almost 2 orders of magnitude. By contrast, the rglB- derivative of DS410 showed no reduction in transformation efficiency with methylation while the rglB- derivative of C600 was partially tolerant to methylation. Further, we show that the 1.8 kilobase (kb) and 1.2 kb KpnI fragments derived from the human L1 repeat have respectively 18.3% and 2.3% mC in vivo. Using these hyper- and hypo-methylated genomic segments ligated into the pBS plasmid, transformants with the highly methylated 1.8 kb L1 insert were recovered at 17 to 40 fold higher frequency with the rglB- host strains than with the rglB+ hosts. In addition, recombinant phage (lambda 2001) containing inserts of plant genomic DNA with 26.7% mC (from Petunia hybrida) when plated on rglB- hosts gave titres up to 222 times higher than on the rglB+ strains.

Polyamines, DNA methylation and cell differentiation

The cellular concentration of AdoMet is normally very much higher than that of dcAdoMet, the aminopropyl group donor in polyamine synthesis. However, when cells are depleted of their putrescine and spermidine, i.e. the aminopropyl group acceptors, the dcAdoMet concentration increases dramatically, to a level that may greatly exceed that of AdoMet. Using a highly purified DNA methyltransferase and its preferred substrates, a defined hemimethylated duplex oligodeoxynucleotide or poly(dI-dC)-poly(dI-dC), we demonstrate that dcAdoMet is a poor methyl group donor, and that it starts to inhibit DNA methylation when its concentration exceeds that of AdoMet. At a dcAdoMet/AdoMet ratio of 5:1 there is very little methyl transfer. This study suggests that the antiproliferative and differentiative effects brought about by inhibitors of polyamine synthesis may be partly attributable to dcAdoMet-mediated inhibition of DNA methylation.