Introduction

DNA methylation is an important epigenetic mark for regulation of gene expression and heterochromatin silencing. In plants, cytosines in all contexts, including CG, CHG, and CHH (H is A, T, or C), are methylated. In Arabidopsis thaliana, DNA methylations in CG contexts are maintained by methyltransferase 1 (MET1), and those in non-CG contexts are maintained by chromomethylase 3. A Dnmt3 class DNA methyltransferase, i.e., domains rearranged methylase 2, is de novo methyltransferase (Chan et al. 2005; Law and Jacobsen 2010). In plants, de novo methylation is triggered by a 24-nt small RNA mediated pathway, called RNA-directed DNA methylation (Matzke et al. 2009). Methylated cytosines are not randomly distributed but mainly found in condensed chromosomal regions, called heterochromatin. Recent epigenomic research has revealed that DNA methylation is rich in pericentromeric heterochromatin, which consists of repetitive sequences and transposable elements. These regions are abundant in loci corresponding to siRNAs (Cokus et al. 2008; Lister et al. 2008).

In addition to DNA methyltransferases, a chromatin remodeling factor, decrease in DNA methylation 1 (DDM1), also regulates DNA methylation (Vongs et al. 1993). DDM1 is essential for maintenance of DNA methylation, and in a ddm1 mutant of A. thaliana, overall reduction of DNA methylation and reactivation of transposable elements have been observed (Miura et al. 2001; Singer et al. 2001; Tsukahara et al. 2009). Although ddm1 mutants show slight morphological change in the first generation, notable morphological abnormalities have been observed after repeated self-pollination for several generations (Kakutani et al. 1996; Kakutani 1997). The abnormalities in the ddm1 mutants are not linked to the DDM1 gene (Kakutani et al. 1999), and due to activation of endogenous transposons or misregulation of expression of genes near transposons (Soppe et al. 2001; Saze and Kakutani 2007; Fujimoto et al. 2008a; Tsukahara et al. 2009).

Recent studies have revealed some epigenetic modifications which affect agriculturally important traits (Manning et al. 2006; Hauben et al. 2009; Martin et al. 2009), suggesting the importance of epigenetics in agricultural science. However, there is little information on epigenetic regulation of gene expression in Brassica rapa, which includes various vegetables. In B. rapa, three DNA methyltransferase genes, i.e., BrMET1a, BrMET1b, and BrCMT, and two genes for chromatin remodeling factors, i.e., BrDDM1a and BrDDM1b, have been isolated, and hypomethylated plants with an RNAi construct of BrDDM1 (ddm1-RNAi plants) have been obtained (Fujimoto et al. 2006a, 2008b). The ddm1-RNAi plants have been shown to have demethylation in centromeric repeats and some retrotransposons (Fujimoto et al. 2006b, 2008b). However, there is little information on the regions methylated by BrDDM1s at the genome-wide level, and it is not known whether BrDDM1s play a role in DNA methylation in protein-coding genes. In the present study, we performed methylation-sensitive amplification polymorphism (MSAP) analysis to investigate target regions of BrDDM1s. MSAP is a modified method of amplification fragment length polymorphism analysis using HpaII/MspI instead of MseI, in which the difference of DNA methylation at 5′-CCGG-3′ sites can be detected as polymorphism (Xiong et al. 1999). HpaII and MspI are isoschizomers which recognize CCGG sites, but their respective sensitivities to DNA methylation differ. HpaII cannot digest CCGG sites when either cytosine is fully methylated, while MspI can digest C5mCGG but not 5mCCGG. Differences in DNA methylation between a hybrid and its parental lines in rice (Xiong et al. 1999), among different developmental stages in A. thaliana (Ruiz-Garcia et al. 2005), and between an artificial amphiploid and their parent species (Beaulieu et al. 2009) have been revealed by MSAP. In the present study, searching for sequences whose DNA methylation is regulated by BrDDM1a and/or BrDDM1b, we found that the BrDDM1 genes regulate DNA methylation of repetitive sequences such as transposons. In contrast, gene body methylation was not changed in a ddm1-RNAi plant, suggesting that DDM1s play a role in DNA methylation in repetitive sequences preferentially.

Materials and methods

Plant materials

An F1 cultivar of B. rapa, ‘Osome’ (Takii Seed Co. Kyoto, Japan), was used as a wild-type (WT) plant, and a ddm1-RNAi plant, which has been transformed with an RNAi construct of the common sequence between BrDDM1a and BrDDM1b (Fujimoto et al. 2008b), and self-pollinated T1 plants were also used. Genomic DNAs for MSAP analysis were isolated from leaves by the cetyl trimethyl ammonium bromide method (Murray and Thompson 1980).

MSAP analysis

MSAP analysis was performed following Xiong et al. (1999). For MSAP analysis, genomic DNA isolated from the WT plant and a T0-2 plant of the ddm1-RNAi plant, in which DNA methylation status was lowest among T0 plants produced by Fujimoto et al. (2008b), were used. DNAs (100 ng) were digested with HpaII/EcoRI or MspI/EcoRI (TaKaRa Bio, Shiga, Japan). Oligonucleotides used in the MSAP analysis are shown in Electronic Supplemental Table 1. Polymerase chain reaction (PCR) products were electrophoresed in 5% polyacrylamide gel containing 8.5 M urea. After the electrophoresis, DNA fragments were stained by the silver staining method. MSAP analysis was repeated twice using the same primer pair, and reproducible polymorphic bands were counted.

Characterization of polymorphic bands

Bands showing polymorphism were cut from polyacrylamide gel and heated with 50 μl of PCR buffer at 65°C for 2 h. DNAs were reamplified by PCR using HM+0/Eco+0 primers listed in Electronic Supplemental Table 1. The PCR condition was 94°C for 1 min, followed by 45 cycles of 94°C for 30 s, 58°C for 30 s, and 72°C for 1 min, and final extension at 72°C for 3 min. Reamplified PCR products were cloned using pGEM-T Easy System I (Promega) and the nucleotide sequences were determined with a CEQ 2000XL DNA Analyzer (Beckman Coulter, USA). Sequence data were analyzed using Sequencher (Gene Codes Corporation, USA), and BLAST search of DDBJ was performed (http://www.blast.ddbj.nig.ac.jp/top-j.html).

To characterize the flanking sequences, suppression PCR was performed following Siebert et al. (1995). Amplified PCR products were cloned and sequenced.

Southern blot analysis

Genomic DNAs (2 μg) isolated from leaves of WT plants were digested with EcoRI (TaKaRa Bio) for analysis of copy number of DDM1 target sequences. For analysis of DNA methylation status, genomic DNAs isolated from the WT plants, the T0-2 plant of the ddm1-RNAi plants, and T1 progeny of the T0-2 plant were digested with HpaII or MspI. Used T1 progeny were T1-1, T1-2, and T1-3, which have the RNAi construct and are hypomethylated as T0-2 (Fujimoto et al. 2008b). Digested DNAs were electrophoresed on 1.0% agarose gel and transferred onto a nylon membrane (Nitran, Whatman, UK). Probes were labeled by PCR using the PCR dig-labeling mix (Roche, Switzerland), HM+0/Eco+0 primers listed in Electronic Supplemental Table 1, and cloned polymorphic sequences as templates. After pre-hybridization at 65°C for 6 h, the membrane was hybridized with a digoxigenin-labeled probe at 65°C for 12 h. After hybridization, the membrane was washed twice in a solution of 0.5 × SSC containing 0.1% SDS for 20 min at 65°C and rotated in blocking buffer containing anti-digoxigenin AP Fab fragment (Roche) at room temperature for 30 min. CSPD (Roche) was used as an alkaline phosphatase substrate and emitted light was exposed to Fuji Medical X-Ray Film (FUJIFILM, Tokyo, Japan).

Bisulfite sequencing

Genomic bisulfite sequencing analysis was performed as described by Paulin et al. (1998). Genomic DNAs (1 μg) extracted from leaves of the WT plant and the T0-2 plant of the ddm1-RNAi plants were digested with EcoRI and SalI or with EcoRI and XhoI in 200 μl of the reaction mixture. After ethanol precipitation, DNAs were dissolved in 20 μl of water. After being heated at 94°C for 10 min and then cooled on ice, DNAs were denatured by the addition of 2.2 μl of 3 N NaOH and incubated at 37°C for 30 min. The denatured DNAs were dissolved in 208 μl of urea/bisulfite solution (7.5 g of urea (Wako, Osaka, Japan) and 7.6 g of sodium metabisulfite (MERCK, Germany) dissolved in 20 ml of water, adjusted to pH 5.0) and 12 μl of 10 mM hydroquinone (SIGMA, USA) and overlaid with mineral oil. The samples were subjected to 30 cycles of 95°C for 30 s and 55°C for 15 min, followed by 55°C for 15 h in a PCR instrument. After the reaction, DNAs were purified using the Gene Clean Kit (Q-BIOgene, USA) and eluted with 20 μl of water. For the desulfonation, 3 N NaOH was added and the DNAs were incubated at 37°C for 15 min. After ethanol precipitation, DNAs were eluted with 20 μl of water. PCR was performed in 50 μl of reaction mixture containing 5 μl of DNA as a template. The PCR condition was 94°C for 2 min followed by 40 cycles of 94°C for 20 s, 50°C for 30 s, and 72°C for 1 min, and a final extension at 72°C for 10 min. The primers employed are listed in Electronic Supplemental Table 1. The PCR products were gel-purified and cloned. Ten to twelve independent clones were sequenced. We confirmed perfect conversion of unmethylated cytosines to thymines in all eight clones of MSAPg-13 sequence.

Results

DNA methylation targets of BrDDM1s in B. rapa

To identify the DNA methylation targets of BrDDM1s, MSAP analysis was performed using genomic DNAs of a WT plant and a ddm1-RNAi hypomethylated plant. A total of 1,913 fragments were amplified by 72 primer pairs, and 94 of their fragments (4.9%) showed different patterns between the WT plant and the ddm1-RNAi plant. Eighty-six fragments (91.5%) and eight fragments (8.5%) newly appeared and disappeared, respectively, in the ddm1-RNAi plant (Fig. 1, Table 1). Thirty-four fragments in the 94 polymorphic fragments were sequenced (Table 2). BLAST search of DDBJ revealed that 14 sequences were similar to the transposon sequences, two were parts of the rDNA region, and one was a putative centromeric repeat. The remaining 17 sequences had no homology with known sequences. To confirm the differences of DNA methylation between the WT plant and the ddm1-RNAi plant, Southern blot analyses using HpaII and MspI were performed. All the 12 examined probes revealed higher levels of cleavage by the methylation-sensitive restriction enzyme, i.e., HpaII, in ddm1-RNAi plants than in the WT plants (Fig. 2, Table 2). Hypomethylation status of genomic regions was confirmed in both sequences newly appearing in the ddm1-RNAi plant and those disappearing in the ddm1-RNAi plant.

Fig. 1
figure 1

Polymorphic bands between a wild-type (WT) plant and a ddm1-RNAi plant in MSAP analysis. Arrowheads indicate polymorphic bands between the WT plant and the ddm1-RNAi plant. WT wild-type plant, ddm1 ddm1-RNAi plant, H HpaII/EcoRI digestion, M MspI/EcoRI digestion

Full size image
Table 1 The numbers of polymorphic bands between WT and ddm1-RNAi plant in MSAP
Full size table
Table 2 Annotation of sequences showing differences of DNA methylation in the ddm1-RNAi plant
Full size table
Fig. 2
figure 2

Detection of DNA methylation status of the target sequences of BrDDM1s in the ddm1-RNAi plants by Southern blot analysis. Genomic DNA was digested by HpaII (H) or MspI (M) and hybridized with polymorphic bands of MSAP analysis as probes

Full size image

Southern blot analyses using the annotated sequences newly appearing or disappearing in the ddm1-RNAi plant as probes showed multiple bands except for the reverse transcriptase regions, which showed 1–2 bands (Fig. 3, Table 2). Band patterns of multiple bands were categorized as multi I or multi II, multi I having many bands with a smear background and multi II having more than three bands with no smear background (Fig. 3). Seventeen probes of unknown sequences revealed ten sequences to have more than six copies, termed multi I, and seven sequences to have 1–3 copies (Table 2). Taken together, these findings indicate that BrDDM1s regulate DNA methylation of repetitive sequences in B. rapa.

Fig. 3
figure 3

Detection of target sequences of BrDDM1s by Southern blot analysis. Genomic DNA was digested by EcoRI and hybridized with polymorphic bands detected by MSAP analysis. The sequences were categorized into three classes by their band patterns. Multi I has many dense bands with a smear background, multi II has more than three dense bands with no smear background, and others have a few dense signals

Full size image

DNA methylation in protein-coding regions

DNA methylation in genic regions was analyzed. Recent studies have revealed that actively transcribed genes are methylated at their CG sites, which is called “body methylation” (Zhang et al. 2006; Zilberman et al. 2007). To obtain body methylated sequences, we characterized some more MSAP bands and got four candidates, i.e., MSAPg-4 (4), -12, -19, and -54 (1). These genes were expressed in leaves of WT plants, except for MSAPg-4 (4).

To confirm their DNA methylation status, bisulfite-sequencing analysis was carried out. A 197 bp sequence of MSAPg-4 (4), which corresponds to the 10th–11th intron of the putative serine carboxypeptidase 1 gene of A. thaliana, contained 42 cytosines, i.e., three in a CpG context and 39 in a non-CpG context, and two cytosines in the CpG context were methylated. In a 303 bp sequence of MSAPg-12, which corresponds to the 10th–12th exon of a putative protein gene (homologous to At3g45045 of A. thaliana), containing 58 cytosines, i.e., 8 in a CpG context and 50 in a non-CpG context, two cytosines in the CpG context and one in a non-CpG context were methylated (Fig. 4b). A 287 bp flanking sequence of MSAPg-19, which corresponds to the third exon to the third intron of a hypothetical protein gene (homologous to At4g14850 of A. thaliana), contained 51 cytosines, 13 of which were of a CpG context and 38 of which were of a non-CpG context. There were nine methylated sites, all of which were of a CpG context. A 270 bp flanking sequence of MSAPg-54 (1), which corresponds to putative trehalose-6-phosphate synthase, contained 51 cytosines, i.e., 14 in a CpG context and 37 in a non-CpG context. All cytosines in the CpG context were heavily methylated, and 16 cytosines in the non-CpG context were also methylated, their level of methylation being lower than that in the CpG context. DNA methylation in these four regions was confirmed not only in a CCGG context, which can be detected by MSAP analysis, but also at other CG sites. In these four sequences of the ddm1-RNAi plant, body methylation was not changed (Fig. 4). Regulation of DNA methylation in genic regions was contrary to that in repetitive sequences, i.e., the retrotransposon Tto1 (Fujimoto et al. 2008c) was hypomethylated in the ddm1-RNAi plants in all contexts (Fig. 4). These results indicate that BrDDM1s are not involved in DNA methylation of these regions.

Fig. 4
figure 4

Differences of DNA methylation levels between a wild-type (WT) plant and a ddm1-RNAi plant analyzed by bisulfite sequencing. Bar graphs show percentages of methylated cytosines in CG, CHG, and CHH context detected by bisulfite-sequencing analysis. Black and gray bars indicate the WT plant and the ddm1-RNAi plant, respectively. a MSAPg-4 (4), b MSAPg-12, c MSAPg-19, d MSAPg-54 (1), and e Tto1

Full size image

Discussion

DNA methylation is an important modification for epigenetic regulations of gene expression. It is involved in various phenomena, and some agriculturally important traits are also regulated by DNA methylation (Manning et al. 2006; Hauben et al. 2009; Martin et al. 2009). In the present study, MSAP analysis was applied to investigate the sequences whose methylation is controlled by BrDDM1s at the genome-wide level in B. rapa. Since MSAP analysis does not require information of genome sequences, it is suitable for analyzing the difference of DNA methylation in Brassica, whose genome sequencing has not been completed. The sequences with low methylation levels identified in the ddm1-RNAi plant included transposons, repetitive sequences in centromeres, and spacer sequences of rDNA. In addition to known sequences, Southern blot analysis revealed that most unknown sequences with low methylation in the ddm1-RNAi plants had high copy numbers in the B. rapa genome, suggesting a preference of BrDDM1s for DNA methylation in repetitive sequences. However, there is a possibility that genic regions were not detected because of their relatively small ratio in methylated sequences. To determine whether BrDDM1s are involved in DNA methylation in genic regions, we analyzed DNA methylation status in WT and ddm1-RNAi plant within body methylated regions. We found DNA methylation in protein-coding regions in B. rapa, and bisulfite sequencing results showed that most of the methylated contexts were CG sites. These results were similar to the results for other plants and other organisms (Tran et al. 2005; Zemach et al. 2010; Feng et al. 2010). The gene body methylation level in the ddm1-RNAi plant was similar to that in the WT plant, indicating that BrDDM1s play a role in DNA methylation in repetitive sequences preferentially.

In ddm1 mutants of A. thaliana, transposable elements are transcriptionally or transpositionally activated (Miura et al. 2001; Singer et al. 2001; Lippman et al. 2004; Tsukahara et al. 2009), and such transcriptional activation of transposable elements has also been observed in ddm1-RNAi plants of B. rapa (Fujimoto et al. 2008b). Changes of DNA methylation in transposable elements have potential to affect the transcription of protein-coding genes. Expressions of several genes were affected in the ddm1 mutants of A. thaliana, and such effects on gene expression were due to the change of DNA methylation in transposable elements that were found in their flanking regions (Saze and Kakutani 2007; Soppe et al. 2001; Kinoshita et al. 2007). Disruption of gene function by insertion of transposons in ddm1 mutants of A. thaliana has also been reported (Miura et al. 2001; Tsukahara et al. 2009). As transcription of transposable elements affects the gene expression of their flanking protein-coding genes (Kashkush et al. 2003), expression of the protein-coding genes might be changed by transcriptional activation of transposable elements in B. rapa. Although no change in gene body methylation in the ddm1-RNAi plant was detected, more detailed analysis might reveal other types of change in DNA methylation, such as de novo induction of DNA methylation. Further studies on whole genome transcriptome and methylome in B. rapa are needed to determine whether the defect of DDM1 function affects the expression of protein-coding genes directly or indirectly through hypomethylation of transposable elements.

Many single nucleotide polymorphisms (SNPs) in genes responsible for genetic traits important in plant breeding have been identified. Simple and efficient techniques for SNP analysis are applicable to conventional cross-breeding (Shirasawa et al. 2006). In addition to SNPs, epigenetic modification of DNA and/or histone also contributes to variation of some traits. In hypomethylated plants, some important genes, which are silenced epigenetically in normal conditions, may be activated. In contrast to mammals, epigenetic changes are inherited in plants, and therefore they could be treated as genetic changes as well as nucleotide-based genetic changes. Moreover, information in DNA methylation could be applied to make epi-markers (Takata et al. 2005). Although the main targets of BrDDM1 genes are repeat sequences, a recent study in A. thaliana has suggested that DDM1 is also involved in regulation of genes. Saze et al. (2008) have revealed that ddm1 ibm1 double mutants showed severe developmental abnormalities compared to ibm1 mutants, in which IBM1 is putative histone H3K9me2 demethylase and DNA methylation levels in genes are increased by the loss of the function of IBM1 (Miura et al. 2009). Studies on Brassica epigenome may provide important information for development of plant breeding.