5-Methylcytosine DNA glycosylase participates in the genome-wide loss of DNA methylation occurring during mouse myoblast differentiation.

Changes in gene expression during mouse myoblast differentiation were monitored by DNA microarray hybridisation. Four days after the onset of differentiation 2.37% of the genes increased in activity from a value of zero, whereas during the same time 1.68% of total genes had decreased expression. During the first 24 h of differentiation an average of 700 000 CpG sites per haploid genome were demethylated. Maximal loss of DNA methylation is attained after 2 days of differentiation, followed by a gradual remethylation. The highest demethylation is observed in highly repeated DNA sequences, followed by single copy sequences. When DNA replication is inhibited by aphidicolin or L-mimosine this genome-wide demethylation is still observed. During the first 3 h of differentiation there is an increase in the number of hemimethylated CpG sites, which disappear rapidly during the course of genome-wide hypomethylation. Transfection of cells with an antisense morpholino oligonucleotide to 5-methylcytosine DNA glycosylase (G/T mismatch DNA glycosylase) decreases both the activity of the enzyme and genome-wide demethylation. It is concluded that the genome-wide loss of DNA methylation in differentiating mouse myoblasts occurs in part by formation of hemimethylated CpG sites, which can serve as the substrate for 5-methylcytosine-DNA glycosylase.

Reversal of methylation-mediated repression with short-chain fatty acids: evidence for an additional mechanism to histone deacetylation.

We have constructed a stable cell line, human embryonal kidney 293M+, containing a lacZ reporter gene controlled by an in vitro methylated hormone-responsive enhancer. Methylation of the enhancer-promoter abolishes lacZ expression controlled by ponasterone A (an analogue of ecdysone). Ponasterone A-induced expression is restored by the short-chain fatty acids valeric > butyric > propionic > acetic acid, but not by the histone deacetylase inhibitors trichostatin A and suberoylanilide hydroxamic acid (SAHA). lacZ expression is restored to levels approaching that from an unmethylated counterpart. Incubation with short-chain fatty acids alone does not promote demethylation of the lacZ promoter, however, some demethylation (30%) is observed when transcription is triggered by addition of ponasterone A. Similar levels of hyperacetylated histones H3 and H4 were observed in cells treated with short-chain fatty acids, trichostatin A or SAHA. In vivo DNase I footprinting indicates a more open chromatin structure at the promoter region for butyric acid-treated cells. A synergistic effect in reversing the methylation-mediated repression of the lacZ gene is obtained by combined treatments with the normally ineffective compounds trichostatin A and the short-chain fatty acid caproic acid. Our results suggest the existence of an alternative silencing mechanism to histone deacetylation in executing methylation-directed gene silencing.

Overexpression of 5-methylcytosine DNA glycosylase in human embryonic kidney cells EcR293 demethylates the promoter of a hormone-regulated reporter gene.

We have shown that the DNA demethylation complex isolated from chicken embryos has a G(.)T mismatch DNA glycosylase that also possesses 5-methylcytosine DNA glycosylase (5-MCDG) activity. Herein we show that human embryonic kidney cells stably transfected with 5-MCDG cDNA linked to a cytomegalovirus promoter overexpress 5-MCDG. A 15- to 20-fold overexpression of 5-MCDG results in the specific demethylation of a stably integrated ecdysone-retinoic acid responsive enhancer-promoter linked to a beta-galactosidase reporter gene. Demethylation occurs in the absence of the ligand ponasterone A (an analogue of ecdysone). The state of methylation of the transgene was investigated by Southern blot analysis and by the bisulfite genomic sequencing reaction. Demethylation occurs downstream of the hormone response elements. No genome-wide demethylation was observed. The expression of an inactive mutant of 5-MCDG or the empty vector does not elicit any demethylation of the promoter-enhancer of the reporter gene. An increase in 5-MCDG activity does not influence the activity of DNA methyltransferase(s) when tested in vitro with a hemimethylated substrate. There is no change in the transgene copy number during selection of the clones with antibiotics. Immunoprecipitation combined with Western blot analysis showed that an antibody directed against 5-MCDG precipitates a complex containing the retinoid X receptor alpha. The association between retinoid receptor and 5-MCDG is not ligand dependent. These results suggest that a complex of the hormone receptor with 5-MCDG may target demethylation of the transgene in this system.

5-Methylcytosine DNA glycosylase activity is also present in the human MBD4 (G/T mismatch glycosylase) and in a related avian sequence.

A 1468 bp cDNA coding for the chicken homolog of the human MBD4 G/T mismatch DNA glycosylase was isolated and sequenced. The derived amino acid sequence (416 amino acids) shows 46% identity with the human MBD4 and the conserved catalytic region at the C-terminal end (170 amino acids) has 90% identity. The non-conserved region of the avian protein has no consensus sequence for the methylated DNA binding domain. The recombinant proteins from human and chicken have G/T mismatch as well as 5-methylcytosine (5-MeC) DNA glycosylase activities. When tested by gel shift assays, human recombinant protein with or without the methylated DNA binding domain binds equally well to symmetrically, hemimethylated DNA and non-methylated DNA. However, the enzyme has only 5-MeC DNA glycosylase activity with the hemimethylated DNA. Footprinting of human MBD4 and of an N-terminal deletion mutant with partially depurinated and depyrimidinated substrate reveal a selective binding of the proteins to the modified substrate around the CpG. As for 5-MeC DNA glycosylase purified from chicken embryos, MBD4 does not use oligonucleotides containing mCpA, mCpT or mCpC as substrates. An mCpG within an A+T-rich oligonucleotide is a much better substrate than an A+T-poor sequence. The K:(m) of human MBD4 for hemimethylated DNA is approximately 10(-7) M with a V:(max) of approximately 10(-11) mol/h/microgram protein. Deletion mutations show that G/T mismatch and 5-MeC DNA glycosylase are located in the C-terminal conserved region. In sharp contrast to the 5-MeC DNA glycosylase isolated from the chicken embryo DNA demethylation complex, the two enzymatic activities of MBD4 are strongly inhibited by RNA. In situ hybridization with antisense RNA indicate that MBD4 is only located in dividing cells of differentiating embryonic tissues.

A CpG-rich RNA and an RNA helicase tightly associated with the DNA demethylation complex are present mainly in dividing chick embryo cells.

In the developing chicken embryo, active DNA demethylation requires both RNA and proteins (Nucleic Acids Res. 25, 2375-2380, 1997; ibid. 25, 4545-4550, 1997, FEBS Lett. 449, 251-254, 1999a). In vitro assays indicate that in the 5- and 12-day-old embryos the highest specific activity of 5-methylcytosine DNA glycosylase is found in the brain, the eyes and the skin. In situ hybridization with antisense CpG-rich RNA tightly associated to the DNA demethylation complex shows a restricted expression pattern only in proliferating tissues such as the neuroepithelia of the brain in 5-day-old embryos. The RNA is absent in differentiated tissues like the skeletal and heart muscle, liver and the crystallin-producing cells in the lens. The CpG-rich RNA is transcribed in a developmental stage-specific rather than in a cell-specific manner. In contrast transcripts of DNA methyltransferase are found in dividing and quiescent cells. In situ hybridization with a probe of a RNA helicase which is also associated with the DNA demethylation complex shows a very similar localization in mitotically active tissues as the CpG-rich RNA. The content of 5-methylcytosine in individual cells was determined with a specific monoclonal antibody and cytometric analysis on tissue sections. The results indicate that proliferating cells have on the average 15% more methylated cytosines than non-dividing cells. This represents roughly 3x10(6) more methylation sites per haploid genome.

5-methylcytosine-DNA glycosylase activity is present in a cloned G/T mismatch DNA glycosylase associated with the chicken embryo DNA demethylation complex.

We previously have shown that DNA demethylation by chicken embryo 5-methylcytosine DNA glycosylase (5-MCDG) needs both RNA and proteins. One of these proteins is a RNA helicase. Further peptides were sequenced, and three of them are identical to the mammalian G/T mismatch DNA glycosylase. A 3,233-bp cDNA coding for the chicken homologue of human G/T mismatch DNA glycosylase was isolated and sequenced. The derived amino acid sequence (408 aa) shows 80% identity with the human G/T mismatch DNA glycosylase, and both the C and N-terminal parts have about 50% identity. As for the highly purified chicken embryo DNA demethylation complex the recombinant protein expressed in Escherichia coli has both G/T mismatch and 5-MCDG activities. The recombinant protein has the same substrate specificity as the chicken embryo 5-MCDG where hemimethylated DNA is a better substrate than symmetrically methylated CpGs. The activity ratio of G/T mismatch and 5-MCDG is about 30:1 for the recombinant protein expressed in E. coli and 3:1 for the purified enzyme from chicken embryos. The incubation of a recombinant CpG-rich RNA isolated from the purified DNA demethylation complex with the recombinant enzyme strongly inhibits G/T mismatch glycosylase while slightly stimulating the activity of 5-MCDG. Deletion mutations indicate that G/T mismatch and 5-MCDG activities share the same areas of the N- and C-terminal parts of the protein. In reconstitution experiments RNA helicase in the presence of recombinant RNA and ATP potentiates the activity of 5-MCDG.

Quantification of 5-methylcytosine in DNA by the chloroacetaldehyde reaction.

The study of changes in genome-wide levels of DNA methylation has become a key focus for understanding the epigenetic regulation of gene expression. Many procedures exist to study DNA methylation, falling into two categories: gene-specific and genome-wide. Genome-wide methylation analysis is best performed by DNA hydrolysis followed by HPLC; however, it requires access to an HPLC machine, which is not always available. Alternative procedures, such as the radioactive labeling of CpG sites using SssI DNA methyltransferase, have been developed to address this problem, but it can only monitor CpG methylation changes, and CpNpG methylation is not detected. Here, we present a method for the analysis of DNA methylation in any sequence context by fluorescent labeling. We present control analyses using synthetic oligonucleotides of known methylation levels and a comparison of genomic DNA from two transgenic tobacco lines known to differ in their methylation levels. The results indicate that hygromycin-induced hypermethylation acts equally on all classes of methylatable cytosine, perhaps indicating a common mechanism.

A chicken embryo protein related to the mammalian DEAD box protein p68 is tightly associated with the highly purified protein-RNA complex of 5-MeC-DNA glycosylase.

We have shown previously that DNA demethylation by chick embryo 5-methylcytosine (5-MeC)-DNA glycosylase needs both protein and RNA. Amino acid sequences of nine peptides derived from a highly purified 5-MeC-DNA glycosylase complex were identified by Nanoelectrospray ionisation mass spectrometry to be identical to the mammalian nuclear DEAD box protein p68 RNA helicase. Antibodies directed against human p68 helicase cross-reacted with the purified 5-MeC-DNA glycosylase complex and immunoprecipitated the glycosylase activity. A 2690 bp cDNA coding for the chicken homologue of mammalian p68 was isolated and sequenced. Its derived amino acid sequence is almost identical to the human p68 DEAD box protein up to amino acid position 473 (from a total of 595). This sequence contains all the essential conserved motifs from the DEAD box proteins which are the ATPase, RNA unwinding and RNA binding motifs. The rest of the 122 amino acids in the C-terminal region rather diverge from the human p68 RNA helicase sequence. The recombinant chicken DEAD box protein expressed in Escherichia coli cross-reacts with the same p68 antibodies as the purified chicken embryo 5-MeC-DNA glycosylase complex. The recombinant protein has an RNA-dependent ATPase and an ATP-dependent helicase activity. However, in the presence or absence of RNA the recombinant protein had no 5-MeC-DNA glycosylase activity. In situ hybridisation of 5 day-old chicken embryos with antisense probes of the chicken DEAD box protein shows a high abundance of its transcripts in differentiating embryonic tissues.

A re-investigation of the ribonuclease sensitivity of a DNA demethylation reaction in chicken embryo and G8 mouse myoblasts.

Recently published results (Nucleic Acids Res. 26, 5573-5580, 1998) suggest that the ribonuclease sensitivity of the DNA demethylation reaction may be an experimental artifact due to the possible tight binding of the nucleases to the methylated DNA substrate. Using an improved protocol we show for two different systems that demethylation of hemimethylated DNA is indeed sensitive to micrococcal nuclease, requires RNA and is not an experimental artifact. The purified 5-MeC-DNA glycosylase from chicken embryos and G8 mouse myoblasts was first incubated for 5 min at 37 degrees C with micrococcal nuclease in the presence of Ca2+ in the absence of the DNA substrate. Upon blocking the nuclease activity by the addition of 25 mM EGTA, the DNA demethylation reaction was initiated by adding the labeled hemimethylated DNA substrate to the reaction mixture. Under these conditions the DNA demethylation reaction was abolished. In parallel controls, where the purified 5-MeC-DNA glycosylase was pre-incubated at 37 degrees C with the nuclease, Ca2+ and EGTA or with the nuclease and EGTA, RNA was not degraded and no inhibition of the demethylation reaction was obtained. As has already been shown for chicken embryos, the loss of 5-MeC-DNA glycosylase activity from G8 myoblasts following nuclease treatment can also be restored by the addition of synthetic RNA complementary to the methylated strand of the substrate DNA. No reactivation of 5-MeC-DNA glycosylase is obtained by complementation with a random RNA sequence, the RNA sequence complementary to the non-methylated strand or DNA, thus ruling out a non-specific competition of the RNA for the binding of the nuclease to the labeled DNA substrate.

Molecular analysis of animal and plant cells is facilitated by their attachment to collodion (cellulose nitrate) films.

A procedure for the efficient transfer of cell monolayers, cultured on glass coverslips, to microscopy slides has been developed. This technique involves the coating of the upper surface of an ethanol-fixed cultured cell layer with a film of collodion dissolved in n-amyl acetate. The dry collodion-coated cell layer can then be detached by rehydrating it for 1 h under water or phosphate-buffered saline and then carefully peeling it away from the coverslip using a pair of tweezers. Once a cell layer has been so mounted, it can be subjected to rough treatment such as proteolytic degradation (which greatly improves the signal-to-noise ratio in procedures like in situ hybridization [ISH] or primed in situ synthesis [PRINS]) without running the risk of cell detachment because of the partial or total degradation of the extracellular matrix. As an example of its application, we show a PRINS of telomeres from mouse fibroblasts. The high mechanical strength of collodion ensures that the structural integrity and morphology of the cell layer is maintained under experimental conditions where the collodion itself is insoluble. In addition to its use on cell layers, collodion can be used for the production of support films for (i) attaching suspension cell cultures, (ii) immobilizing cells normally cultured in suspension (such as tobacco BY2 cells or germinating tobacco pollen grains) or (iii) planting cryostat sections to microscopy slides. The value of this technique lies in its ease of use and the large number of different applications, in both the plant and animal fields of research, to which it may be applied.

A novel 52 kDa protein induces apoptosis and concurrently activates c-Jun N-terminal kinase 1 (JNK1) in mouse C3H10T1/2 fibroblasts.

A 52 kDa protein (p52) was purified from chicken embryos and its corresponding cDNA was cloned. The p52 cDNA is 1768 bp long and has an open reading frame of 465 amino acids. The sequence of the p52 cDNA shows significant homology with mouse and human cDNAs from the EST database, so do the deduced amino acid sequences, indicating the existence of human and mouse homologues of p52. Northern blot hybridization showed that the p52 mRNA was expressed in a wide range of embryonic and adult tissues. There was more p52 mRNA in embryonic heart and liver than in the brain or muscle. The adult testis had the highest level of p52 mRNA, whereas adult liver had the lowest. Expression of p52 in mouse C3H10T1/2 fibroblasts caused apoptotic cell death, upregulation of transcription factor c-Jun and activation of c-Jun N-terminal kinase 1 (JNK1). In addition, expression of Bcl-2, but not of the dominant negative mutant JNK1, can block the p52-mediated apoptosis. These results indicate that p52 may represent a new cell-death protein inducing apoptosis and activating JNK1 through different pathways.

Multiple domains are involved in the targeting of the mouse DNA methyltransferase to the DNA replication foci.

It has been shown that, during the S-phase of the cell cycle, the mouse DNA methyltransferase (DNA MTase) is targeted to sites of DNA replication by an amino acid sequence (aa 207-455) lying in the N-terminal domain of the enzyme [Leonhardt, H., Page, A. W., Weier, H. U. and Bestor, T. H. (1992) Cell , 71, 865-873]. In this paper it is shown, by using enhanced green fluorescent protein (EGFP) fusions, that other peptide sequences of DNA MTase are also involved in this targeting. The work focuses on a sequence, downstream of the reported targeting sequence (TS), which is homologous to the Polybromo-1 protein. This motif (designated as PBHD) is separated from the reported targeting sequence by a zinc-binding motif [Bestor , T. H. (1992) EMBO J , 11, 2611-2617]. Primed in situ extension using centromeric-specific primers was used to show that both the host DNA MTase and EGFP fusion proteins containing the targeting sequences were localized to centromeric, but not telomeric, regions during late S-phase and mitosis. Also found was that, in approximately 10% of the S-phase cells, the EGFP fusions did not co-localize with the centromeric regions. Mutants containing either, or both, of these targeting sequences could act as dominant negative mutants against the host DNA MTase. EGFP fusion proteins, containing the reported TS (aa 207-455), were targeted to centromeric regions throughout the mitotic stage which lead to the discovery of a similar behavior of the endogenous DNA MTase although the host MTase showed much less intense staining than in S-phase cells. The biological role of the centromeric localization of DNA MTase during mitosis is currently unknown.

Developmental changes in DNA methylation of the two tobacco pollen nuclei during maturation.

Changes in DNA methylation during tobacco pollen development have been studied by confocal fluorescence microscopy using a monoclonal anti-5-methylcytosine (anti-m5C) antibody and a polyclonal anti-histone H1 (anti-histone) antibody as an internal standard. The specificity of the anti-m5C antibody was demonstrated by a titration series against both single-stranded DNA and double-stranded DNA substrates in either the methylated or unmethylated forms. The antibody was found to show similar kinetics against both double- and single-stranded DNA, and the fluorescence was proportional to the amount of DNA used. No signal was observed with unmethylated substrates. The extent of methylation of the two pollen nuclei remained approximately constant after the mitotic division that gave rise to the vegetative and generative nuclei. However, during the subsequent development of the pollen, the staining of the generative nucleus decreased until it reached a normalized value of (1)/(5) of that of the vegetative nucleus. The use of a confocal microscope makes these data independent of possible focusing artefacts. The anti-histone antibody was used as a control to show that, while the antibody staining directed against 5-methylcytosine changed dramatically during pollen maturation, the histone signal did not. We observed the existence of structural dimorphism amongst tobacco pollen grains, the majority having three pollen apertures and the rest with four. However, the methylation changes observed occurred to the same extent in both subclasses.

Demethylation of DNA by purified chick embryo 5-methylcytosine-DNA glycosylase requires both protein and RNA.

We have previously purified and characterized a 5-methylcytosine (5-MeC)-DNA glycosylase from 12 day old chick embryos [Jost,J.P. et al. (1995) J. Biol. Chem. 270, 9734-9739]. The activity of the purified enzyme is abolished upon treatment with proteinase K and ribonuclease A. RNA copurifies with 5-MeC-DNA glycosylase activity throughout all chromatographic steps and preparative gel electrophoresis. RNA with a length of approximately 300-500 nucleotides was isolated from the gel purified enzyme. Upon extensive treatment with proteinase K, the gel eluted and labeled RNA did not show any significant change in molecular mass. The purified RNA incubated alone or in the presence of Mg2+and deoxyribonucleotide phosphates had no 5-MeC-DNA glycosylase or demethylating activities. However, activity of 5-MeC-DNA glycosylase could be restored when the purified RNA was incubated with the inactive protein, free of RNA.

Antibiotics induce genome-wide hypermethylation in cultured Nicotiana tabacum plants.

Plant genomic DNA methylation was analyzed by an improved SssI methyltransferase assay and by genomic sequencing with sodium bisulfite. Kanamycin, hygromycin, and cefotaxime (also called Claforan) are commonly used as selective agents for the production of transgenic plants. These antibiotics caused DNA hypermethylation in tobacco plants grown in vitro, which was both time- and dose-dependent. An exposure of the plantlets to 500 mg/liter cefotaxime for 1 month caused the de novo methylation of 3 x 10(7) CpG sites/haploid genome of 3.5 x 10(9) base pairs. It occurred in high, moderate, and low repetitive DNA and was not reversible upon the removal of the antibiotics. Reversion was only observed in progeny grown in the absence of drugs. Analysis of the promoter regions of two single-copy genes, an auxin-binding protein gene and the class I chitinase gene, showed the hypermethylation to be heterogeneous but biased toward CpGs. The hypermethylation of the class I chitinase and the auxin-binding protein promoters was not a consequence of a drug-induced gene amplification.

The methylated DNA binding protein-2-H1 (MDBP-2-H1) consists of histone H1 subtypes which are truncated at the C-terminus.

The methylated DNA binding protein-2-H1 (MDBP-2-H1), present in rooster liver, is a member of the histone H1 family which inhibits transcription by binding selectively to methylated promoters. Here we have determined the primary structure of MDBP-2-H1. A comparison between histone H1 and MDBP-2-H1 was achieved by analyzing reversed phase HPLC-purified and V8-digested proteins by mass spectrometry and/or microsequencing. In rooster liver the most abundant histone H1 subtypes are H1 01 and H1 11L. Similarly, MDBP-2-H1 contains the same subtypes of histone H1. The histone H1 subtype H1 01 in MDBP-2-H1 has 150 amino acids, whereas the full-size histone H1 01 is 218 amino acids. The difference in mass between the two proteins is explained by C-terminal truncation of histone H1 01.

The RNA moiety of chick embryo 5-methylcytosine- DNA glycosylase targets DNA demethylation.

We have previously shown that DNA demethylation by chick embryo 5-methylcytosine (5-MeC)-DNA glycosylase needs both protein and RNA. RNA from enzyme purified by SDS-PAGE was isolated and cloned. The clones have an insert ranging from 240 to 670 bp and contained on average one CpG per 14 bases. All six clones tested had different sequences and did not have any sequence homology with any other known RNA. RNase-inactivated 5-MeC-DNA glycosylase regained enzyme activity when incubated with recombinant RNA. However, when recombinant RNA was incubated with the DNA substrate alone there was no demethylation activity. Short sequences complementary to the labeled DNA substrate are present in the recombinant RNA. Small synthetic oligoribonucleotides (11 bases long) complementary to the region of methylated CpGs of the hemimethylated double-stranded DNA substrate restore the activity of the RNase-inactivated 5-MeC-DNA glycosylase. The corresponding oligodeoxyribonucleotide or the oligoribonucleotide complementary to the non-methylated strand of the same DNA substrate are inactive when incubated in the complementation test. A minimum of 4 bases complementary to the CpG target sequence are necessary for reactivation of RNase-treated 5-MeC-DNA glycosylase. Complementation with double-stranded oligoribonucleotides does not restore 5-MeC-DNA glycosylase activity. An excess of targeting oligoribonucleotides cannot change the preferential substrate specificity of the enzyme for hemimethylated double-stranded DNA.

In differentiating mouse myoblasts DNA methyltransferase is posttranscriptionally and posttranslationally regulated.

Upon the onset of mouse myoblast differentiation there is a rapid drop in DNA methyltransferase activity followed by a genome wide demethylation [Jost and Jost (1994) J. Biol. Chem. 269, 10040-10043]. Here we show by using specific antibodies directed against DNA methyltransferase that upon differentiation there was a rapid drop in nuclear DNA methyltransferase whilst the internal control histone H1 remained constant. The loss of nuclear methyltransferase was not due to a translocation of the enzyme from the nucleus to the cytoplasm where there was an increase in creatine phosphokinase protein. In vitro run on experiments carried out with growing and differentiating myoblast nuclei showed no difference in the rate of DNA methyltransferase mRNA synthesis. As measured by Northern blot hybridization the relative half life of DNA methyltransferase mRNA in growing and differentiating cells in the presence of Actinomycin D was 5 h and 1 h 30 min respectively, whereas in the same cells the half life of histone H4 mRNA was in both cases 80 min. As measured by a combination of pulse chase experiments with labeled leucine and immunoprecipitation, the relative half-life of DNA methyltransferase in growing and differentiating cells was approximately 18 h and 4 h 30 min respectively.

Non-symmetrical cytosine methylation in tobacco pollen DNA.

We have detected sequence-specific non-symmetrical cytosine methylation within a 140 bp region of the promoter for the tobacco auxin-binding protein gene T85 in pollen DNA. Direct sequencing of the population of bisulphite reaction products showed that, in this region. 10 out of a possible 49 cytosine residues were methylated at a high frequency in pollen whereas the corresponding region from somatic cells (leaf DNA) did not show a detectable level of methylation. The context of these sites was 1 x m5CpTpC, 1 x m5CpGpT, 1 x m5CpCpT, 2 x m5CpTpT, 2 x m5CpGpG, and 3 x m5CpApT of which only m5CpGpG and m5CpGpT fitted the consensus sequence for symmetrical methylation in plants.