Primary DNA sequence determines sites of maintenance and de novo methylation by mammalian DNA methyltransferases.

Analysis of the enzymatic methylation of oligodeoxynucleotides containing multiple C-G groups showed that hemimethylated sites in duplex oligomers are not significantly methylated by human or murine DNA methyltransferase unless those sites are capable of being methylated de novo in the single- or double-stranded oligomers. Thus, the primary sequence of the target strand, rather than the methylation pattern of the complementary strand, determines maintenance methylation. This suggests that de novo and maintenance methylation are the same process catalyzed by the same enzyme. In addition, the study revealed that complementary strands of oligodeoxynucleotides are methylated at different rates and in different patterns. Both primary DNA sequence and the spacing between C-G groups seem important since in one case studied, maximal methylation required a specific spacing of 13 to 17 nucleotides between C-G pairs.     

Hypothesis of the initiation of DNA methylation de novo and allelic exclusion by small RNA

We have discovered that 5'-CG-3' dinucleotide and 5'-CNG-3' trinucleotide are found in published sequences of small interfering RNA and microRNA more often than they should be found in a random sequence. This circumstance is evidence of an important biological purpose of 5'-CG-3' dinucleotides and 5'-CNG-3' trinucleotides in small RNA sequences. We suppose that small RNAs containing mentioned di- and trinucleotides participate in creation of chromatin marks of epigenetic information through high-specific search of DNA sequences liable to repression and through initiation of the methylation de novo of 5'-CG-3' and 5'-CNG-3' sites in DNA fragments, which appeared to be bound complementary with small RNA. Several genes can be inactivated simultaneously when they contain the motif which is recognized by small RNA. Allelic exclusion appears, to our opinion, as a result of initiation by small RNA of de novo DNA methylation of all alleles but one that exist in the cell. The predecessor of this small RNA is transcribed from the antiparallel allele chain. Those alleles are inactivated which antiparallel chain is less actively read by RNA-polymerase, which, as we suppose, releases DNA from attached to it small RNA in the process of transcribing. But the quantity of small RNA which is transcribed from just one allele is insufficient to overcome the level when the repression process of this allele de novo starts.     

De novo and maintenance DNA methylation by a mouse plasmacytoma cell DNA methyltransferase

A DNA methyltransferase of Mr = 140,000 that is active on both unmethylated and hemimethylated DNA substrates has been purified from the murine plasma-cytoma cell line MPC 11. The maximal rate of methylation was obtained with maintenance methylation of hemimethylated Micrococcus luteus or M13 DNAs. At low enzyme concentrations, the highest rate of de novo methylation occurred with single-stranded DNA or relatively short duplex DNA containing single-stranded regions. Strong substrate inhibition was observed with hemimethylated but not unmethylated DNA substrates. Fully methylated single-stranded M13 phage DNA inhibited neither the de novo nor the maintenance reactions, but unmethylated single-stranded M13 DNA strongly inhibited the maintenance reaction. The kinetics observed with hemimethylated and single-stranded substrates could be explained if the enzyme were to bind irreversibly to a DNA molecule and to aggregate if present in molar excess. Such aggregates would be required for activity upon hemimethylated but not single-stranded DNA. For de novo methylation of duplex DNA, single-stranded regions or large amounts of methyltransferase appear to be required. The relative substrate preference for the enzyme is hemimethylated DNA greater than fully or partially single-stranded DNA greater than fully duplex DNA.     

Recognition of duplex DNA by RNA polynucleotides.

We are interested in creating artificial gene repressors based on duplex DNA recognition by nucleic acids. Homopyrimidine RNA oligonucleotides bind to duplex DNA at homopurine/homopyrimidine sequences under slightly acidic conditions. Recognition is sequence-specific, involving rU.dA.dT and rC+.dG.dC base triplets. Affinities were determined for folded polymeric RNAs (ca. 100-200 nt) containing 0, 1 or 3 copies of a 21 nt RNA sequence that binds duplex DNA by triple helix formation. When this recognition sequence was inserted into the larger folded RNAs, micromolar concentrations of the resulting RNA ligands bound a duplex DNA target at pH 5. However, these binding affinities were at least 20-fold lower than the affinity of an RNA oligonucleotide containing only the recognition sequence. Enzymatic probing of folded RNAs suggests that reduced affinity arises from unfavorable electrostatic, structural and topological considerations. The affinity of a polymeric RNA with three copies of the recognition sequence was greater than that of a polymeric RNA with a single copy of the sequence. This affinity difference ranged from 2.6- to 13-fold, depending on pH. Binding of duplex DNA by polymeric RNA might be improved by optimizing the RNA structure to efficiently present the recognition sequence.     

Dnmt2 is not required for de novo and maintenance methylation of viral DNA in embryonic stem cells.

We have shown previously that de novo methylation activities persist in mouse embryonic stem (ES) cells homozygous for a null mutation of Dnmt1 that encodes the major DNA cytosine methyltransferase. In this study, we have cloned a putative mammalian DNA methyltransferase gene, termed Dnmt2 , that is homologous to pmt1 of fission yeast. Different from pmt1 in which the catalytic Pro-Pro-Cys (PPC) motif is 'mutated' to Pro-Ser-Cys, Dnmt2 contains all the conserved methyltransferase motifs, thus likely encoding a functional cytosine methyltransferase. However, baculovirus-expressed Dnmt2 protein failed to methylate DNA in vitro . To investigate whether Dnmt2 functions as a DNA methyltransferase in vivo , we inactivated the Dnmt2 gene by targeted deletion of the putative catalytic PPC motif in ES cells. We showed that endogenous virus was fully methylated in Dnmt2 -deficient mutant ES cells. Furthermore, newly integrated retrovirus DNA was methylated de novo in infected mutant ES cells as efficiently as in wild-type cells. These results indicate that Dnmt2 is not essential for global de novo or maintenance methylation of DNA in ES cells.     

Lsh is involved in de novo methylation of DNA

Deletion of Lsh perturbs DNA methylation patterns in mice yet it is unknown whether Lsh plays a direct role in the methylation process. Two types of methylation pathways have been distinguished: maintenance methylation by Dnmt1 occurring at the replication fork, and de novo methylation established by the methyltransferases Dnmt3a and Dnmt3b. Using an episomal vector in Lsh-/- embryonic fibroblasts, we demonstrate that the acquisition of DNA methylation depends on the presence of Lsh. In contrast, maintenance of previously methylated episomes does not require Lsh, implying a functional role for Lsh in the establishment of novel methylation patterns. Lsh affects Dnmt3a as well as Dnmt3b directed methylation suggesting that Lsh can cooperate with both enzymatic activities. Furthermore, we demonstrate that embryonic stem cells with reduced Lsh protein levels show a decreased ability to silence retroviral vector or to methylate endogenous genes. Finally, we demonstrate that Lsh associates with Dnmt3a or Dnmt3b but not with Dnmt1 in embryonic cells. These results suggest that the epigenetic regulator, Lsh, is directly involved in the control of de novo methylation of DNA.     

Inhibition of DNA methylation by chemical carcinogens in vitro

A diverse range of ultimate chemical carcinogens inhibited the transfer of methyl groups from S-adenosylmethionine to hemimethylated DNA in a reaction catalyzed by mouse spleen methyltransferase. The formation of alkali-labile sites in DNA lessened its ability to accept methyl groups in vitro, but the methylation reaction was much less sensitive to thymine dimers or double-strand breaks. Carcinogens induced the formation of alkali-labile DNA lesions, but the degree of methyltransferase inhibition observed was greater than that expected for this damage alone. Certain carcinogens were also capable of direct modification and inactivation of the methyltransferase enzyme. Benzo(a)pyrene treatment of living BALB/3T3 A31 clone 1-13 but not C3H/10T1/2 clone 8 cells resulted in a 12% decrease in total 5-methylcytosine content of cellular DNA. Carcinogenic agents may therefore cause heritable changes in 5-methylcytosine patterns in certain cell types by a variety of mechanisms, including adduct formation, induction of apurinic sites and single-strand breaks and direct inactivation of DNA methyltransferase.     

Histone tails regulate DNA methylation by allosterically activating de novo methyltransferase

Cytosine methylation of genomic DNA controls gene expression and maintains genome stability. How a specific DNA sequence is targeted for methylation by a methyltransferase is largely unknown. Here, we show that histone H3 tails lacking lysine 4 (K4) methylation function as an allosteric activator for methyltransferase Dnmt3a by binding to its plant homeodomain (PHD). In vitro, histone H3 peptides stimulated the methylation activity of Dnmt3a up to 8-fold, in a manner reversely correlated with the level of K4 methylation. The biological significance of allosteric regulation was manifested by molecular modeling and identification of key residues in both the PHD and the catalytic domain of Dnmt3a whose mutations impaired the stimulation of methylation activity by H3 peptides but not the binding of H3 peptides. Significantly, these mutant Dnmt3a proteins were almost inactive in DNA methylation when expressed in mouse embryonic stem cells while their recruitment to genomic targets was unaltered. We therefore propose a two-step mechanism for de novo DNA methylation - first recruitment of the methyltransferase probably assisted by a chromatin- or DNA-binding factor, and then allosteric activation depending on the interaction between Dnmt3a and the histone tails - the latter might serve as a checkpoint for the methylation activity.     

Preferential methylation of unmethylated DNA by Mammalian de novo DNA methyltransferase Dnmt3a

DNA methylation is an epigenetic modification of DNA. There are currently three catalytically active mammalian DNA methyltransferases, DNMT1, -3a, and -3b. DNMT1 has been shown to have a preference for hemimethylated DNA and has therefore been termed the maintenance methyltransferase. Although previous studies on DNMT3a and -3b revealed that they act as functional enzymes during development, there is little biochemical evidence about how new methylation patterns are established and maintained. To study this mechanism we have cloned and expressed Dnmt3a using a baculovirus expression system. The substrate specificity of Dnmt3a and molecular mechanism of its methylation reaction were then analyzed using a novel and highly reproducible assay. We report here that Dnmt3a is a true de novo methyltransferase that prefers unmethylated DNA substrates more than 3-fold to hemimethylated DNA. Furthermore, Dnmt3a binds DNA nonspecifically, regardless of the presence of CpG dinucleotides in the DNA substrate. Kinetic analysis supports an Ordered Bi Bi mechanism for Dnmt3a, where DNA binds first, followed by S-adenosyl-l-methionine.     

Identification of genes required for de novo DNA methylation in Arabidopsis

De novo DNA methylation in Arabidopsis thaliana is catalyzed by the methyltransferase DRM2, a homolog of the mammalian de novo methyltransferase DNMT3. DRM2 is targeted to DNA by small interfering RNAs (siRNAs) in a process known as RNA-directed DNA Methylation (RdDM). While several components of the RdDM pathway are known, a functional understanding of the underlying mechanism is far from complete. We employed both forward and reverse genetic approaches to identify factors involved in de novo methylation. We utilized the FWA transgene, which is methylated and silenced when transformed into wild-type plants, but unmethylated and expressed when transformed into de novo methylation mutants. Expression of FWA is marked by a late-flowering phenotype, which is easily scored in mutant versus wild-type plants. By reverse genetics we discovered the requirement for known RdDM effectors AGO6 and NRPE5a for efficient de novo methylation. A forward genetic approach uncovered alleles of several components of the RdDM pathway, including alleles of clsy1, ktf1 and nrpd/e2, which have not been previously shown to be required for the initial establishment of DNA methylation. Mutations were mapped and genes cloned by both traditional and whole genome sequencing approaches. The methodologies and the mutant alleles discovered will be instrumental in further studies of de novo DNA methylation.Key words: DNA methylation, Arabidopsis, de novo, genetic screen, whole-genome sequencing     

Synthesis of signals for de novo DNA methylation in Neurospora crassa

Most 5-methylcytosine in Neurospora crassa occurs in A:T-rich sequences high in TpA dinucleotides, hallmarks of repeat-induced point mutation. To investigate how such sequences induce methylation, we developed a sensitive in vivo system. Tests of various 25- to 100-bp synthetic DNA sequences revealed that both T and A residues were required on a given strand to induce appreciable methylation. Segments composed of (TAAA)(n) or (TTAA)(n) were the most potent signals; 25-mers induced robust methylation at the special test site, and a 75-mer induced methylation elsewhere. G:C base pairs inhibited methylation, and cytosines 5' of ApT dinucleotides were particularly inhibitory. Weak signals could be strengthened by extending their lengths. A:T tracts as short as two were found to cooperate to induce methylation. Distamycin, which, like the AT-hook DNA binding motif found in proteins such as mammalian HMG-I, binds to the minor groove of A:T-rich sequences, suppressed DNA methylation and gene silencing. We also found a correlation between the strength of methylation signals and their binding to an AT-hook protein (HMG-I) and to activities in a Neurospora extract. We propose that de novo DNA methylation in Neurospora cells is triggered by cooperative recognition of the minor groove of multiple short A:T tracts. Similarities between sequences subjected to repeat-induced point mutation in Neurospora crassa and A:T-rich repeated sequences in heterochromatin in other organisms suggest that related mechanisms control silent chromatin in fungi, plants, and animals.     

Identification of genes required for de novo DNA methylation in Arabidopsis

De novo DNA methylation in Arabidopsis thaliana is catalyzed by the methyltransferase DRM2, a homolog of the mammalian de novo methyltransferase DNMT3. DRM2 is targeted to DNA by small interfering RNAs (siRNAs) in a process known as RNA-directed DNA Methylation (RdDM). While several components of the RdDM pathway are known, a functional understanding of the underlying mechanism is far from complete. We employed both forward and reverse genetic approaches to identify factors involved in de novo methylation. We utilized the FWA transgene, which is methylated and silenced when transformed into wild-type plants, but unmethylated and expressed when transformed into de novo methylation mutants. Expression of FWA is marked by a late flowering phenotype, which is easily scored in mutant versus wild-type plants. By reverse genetics we discovered the requirement for known RdDM effectors AGO6 and NRPE5a for efficient de novo methylation. A forward genetic approach uncovered alleles of several components of the RdDM pathway, including alleles of clsy1, ktf1, and nrpd/e2, which have not been previously shown to be required for the initial establishment of DNA methylation. Mutations were mapped and genes cloned by both traditional and whole genome sequencing approaches. The methodologies and the mutant alleles discovered will be instrumental in further studies of de novo DNA methylation.     

Hypothesis of initiation of DNA methylation de novo and allelic exclusion by small RNAs

We have established that 5′-CG-3′ dinucleotide and 5′-CNG-3′ trinucleotide are found in published sequences of small interfering RNA and microRNA more often than they should be in random DNA sequences. This circumstance indicates the important biological role played by 5′-CG-3′ dinucleotides and 5′-CNG-3′ trinucleotides in small RNA sequences. We suggest that small RNAs containing these di- and trinucleotides participate in the creation of chromatin marks of epigenetic information through a highly specific search for repressible DNA sequences and through the initiation of the methylation de novo of 5′-CG-3′ and 5′-CNG-3′ sites in DNA fragments appearing to be bound complementary to small RNAs. Several genes can be inactivated simultaneously if they contain the motif recognized by small RNA. Allelic exclusion appears, in our opinion, as a result of initiation by small RNAs of DNA methylation de novo of all but one of the alleles that exist in the cell. The predecessor of this small RNA is transcribed from the antiparallel allele chain. Alleles whose antiparallel chains are less actively read by RNA polymerase, which, as we suggest, in the process of transcribing, releases DNA from small RNA bound to it, are inactivated. However, the quantity of small RNA transcribed from only one allele is insufficient to overcome the level above which the repression process of this allele is initiated de novo.     

Differential maintenance and de novo methylating activity by three DNA methyltransferases in aging and immortalized fibroblasts.

Genomic methylation, which influences many cellular processes such as gene expression and chromatin organization, generally declines with cellular senescence although some genes undergo paradoxical hypermethylation during cellular aging and immortalization. To explore potential mechanisms for this process, we analyzed the methylating activity of three DNA methyltransferases (Dnmts) in aging and immortalized WI-38 fibroblasts. Overall maintenance methylating activity by the Dnmts greatly decreased during cellular senescence. In immortalized WI-38 cells, maintenance methylating activity was similar to that of normal young cells. Combined de novo methylation activity of the Dnmts initially decreased but later increased as WI-38 cells aged and was strikingly elevated in immortalized cells. To further elucidate the mechanisms for changes in DNA methylation in aging and immortalized cells, the individual Dnmts were separated and individually assessed for maintenance and de novo methylating activity. We resolved three Dnmt fractions, one of which was the major maintenance methyltransferase, Dnmt1, which declined steadily in activity with cellular senescence and immortalization. However, a more basic Dnmt, which has significant de novo methylating activity, increased markedly in activity in aging and immortalized cells. We have identified this methyltransferase as Dnmt3b which has an important role in neoplastic transformation but its role in cellular senescence and immortalization has not previously been reported. An acidic Dnmt we isolated also had increased de novo methylating activity in senescent and immortalized WI-38 cells. These studies indicate that reduced genome-wide methylation in aging cells may be attributed to attenuated Dnmt1 activity but that regional or gene-localized hypermethylation in aging and immortalized cells may be linked to increased de novo methylation by Dnmts other than the maintenance methyltransferase.     

Recombinant human DNA (cytosine-5) methyltransferase. I. Expression, purification, and comparison of de novo and maintenance methylation.

A method is described to express and purify human DNA (cytosine-5) methyltransferase (human DNMT1) using a protein splicing (intein) fusion partner in a baculovirus expression vector. The system produces approximately 1 mg of intact recombinant enzyme >95% pure per 1.5 x 10(9) insect cells. The protein lacks any affinity tag and is identical to the native enzyme except for the two C-terminal amino acids, proline and glycine, that were substituted for lysine and aspartic acid for optimal cleavage from the intein affinity tag. Human DNMT1 was used for steady-state kinetic analysis with poly(dI-dC).poly(dI-dC) and unmethylated and hemimethylated 36- and 75-mer oligonucleotides. The turnover number (k(cat)) was 131-237 h(-1) on poly(dI-dC).poly(dI-dC), 1.2-2.3 h(-1) on unmethylated DNA, and 8.3-49 h(-1) on hemimethylated DNA. The Michaelis constants for DNA (K(m)(CG)) and S-adenosyl-L-methionine (AdoMet) (K(m)(AdoMet)) ranged from 0.33-1.32 and 2.6-7.2 microM, respectively, whereas the ratio of k(cat)/K(m)(CG) ranged from 3.9 to 44 (237-336 for poly(dI-dC).poly(dI-dC)) x 10(6) M(-1) h(-1). The preference of the enzyme for hemimethylated, over unmethylated, DNA was 7-21-fold. The values of k(cat) on hemimethylated DNAs showed a 2-3-fold difference, depending upon which strand was pre-methylated. Furthermore, human DNMT1 formed covalent complexes with substrates containing 5-fluoro-CNG, indicating that substrate specificity extended beyond the canonical CG dinucleotide. These results show that, in addition to maintenance methylation, human DNMT1 may also carry out de novo and non-CG methyltransferase activities in vivo.     

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.     

Murine DNA cytosine C(5)-methyltransferase: in vitro studies of de novo methylation spreading

The preference of murine DNA (cytosine-5)-methyltransferase (Dnmt1) for single stranded DNA substrates is increased up to 50-fold by the presence of a proximal 5-methyl cytosine (5(me)C). This modulation is distance-dependent and is due to an enhanced binding affinity and minor changes in catalytic efficiency. No modulation was observed with double stranded DNA. Modulation requires that the 5(me)C moiety be attached to the DNA strand containing the CpG methylation target. Our results support a model in which 5(me)C binding by the enzyme occurs to at least one site outside the region involved in CpG recognition. No modulation in response to 5(me)C is observed with the bacterial enzyme M.SssI, which lacks the large N-terminal regulatory domain found in Dnmt1. We suggest that this allosteric modulation involves the N-terminal domain of Dnmt1.