SEARCH

SEARCH BY CITATION

Keywords:

  • DNA methylation;
  • genes;
  • methylome;
  • open chromatin;
  • Populus trichocarpa (black cottonwood);
  • shoot apical meristem

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
  • DNA methylation is involved in the control of plant development and adaptation to the environment through modifications of chromatin compaction and gene expression. In poplar (Populus trichocarpa), a perennial plant, variations in DNA methylation have been reported between genotypes and tissues or in response to drought. Nevertheless, the relationships between gene-body DNA methylation, gene expression and chromatin compaction still need clarification.
  • Here, DNA methylation was mapped in the noncondensed chromatin fraction from P. trichocarpa shoot apical meristematic cells, the center of plant morphogenesis, where DNA methylation variations could influence the developmental trajectory. DNase I was used to isolate the noncondensed chromatin fraction. Methylated sequences were immunoprecipitated, sequenced using Illumina/Solexa technology and mapped on the v2.0 poplar genome. Bisulfite sequencing of candidate sequences was used to confirm mapping data and to assess cytosine contexts and methylation levels.
  • While the methylated DNase I hypersensitive site fraction covered 1.9% of the poplar genome, it contained sequences corresponding to 74% of poplar gene models, mostly exons. The level and cytosine context of gene-body DNA methylation varied with the structural characteristics of the genes.
  • Taken together, our data show that DNA methylation is widespread and variable among genes in open chromatin of meristematic cells, in agreement with a role in their developmental trajectory.

Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

In plants, epigenetic reprogramming occurs at key developmental stages and in response to environmental stimuli (Feng et al., 2010b). These epigenomic events play important roles in genome protection, control of gene expression and inheritance via chromatin structural reworking (Teixeira & Colot, 2010). Among epigenetic marks, DNA methylation shows high stability through mitosis and meiosis and has been thoroughly investigated. In plants, methylated cytosines are found mainly in CG dinucleotides and to a lesser extent in CHG and asymmetric CHH contexts (where H is A, T, or C). Non-CG methylation is specific to plants and fungi, with the exception of mammal embryonic stem cells (Suzuki & Bird, 2008; Lister et al., 2009). The enzymatic machinery of methylation (Goll & Bestor, 2005; Teixeira & Colot, 2010) and demethylation (Penterman et al., 2007), driven by both developmental and environmental stimuli, shapes a methylome that is the set of nucleic acid methylation modifications in an organism's genome or in a particular cell.

The first methylome characterization was performed in Arabidopsis (Zhang et al., 2006) and revealed that DNA methylation is concentrated in heterochromatin and repeats but is also present on 30% of genes. Several other approaches have confirmed this distribution across the genomes of other species, including animals, and raised the problem of repeated sequence data processing, as a consequence of both their repetitive nature and their high DNA methylation (Beck & Rakyan, 2008; Cokus et al., 2008). Indeed, repeats are affected by high CG and non-CG methylation, while the gene body is only methylated in CG (Cokus et al., 2008). Despite its high conservation in many organisms, the functional role of gene-body methylation needs to be clarified.

DNase I is a pancreatic enzyme that preferentially digests with no sequence-based bias (Crawford et al., 2006) nucleosome-depleted DNA, while tightly packaged chromatin is more resistant to cleavage (Crawford et al., 2004). In plants, genome-scale assays revealed tissue-specific DNase hypersensitive sites in promoters and potential regulatory elements, but also within genes, allowing the preferential isolation of genes and the elimination of repeated regions (Zhang et al., 2012a,b). In addition, while DNA methylation and nucleosomes co-localize on exons, some nucleosome depleted regions are also methylated (Chodavarapu et al., 2010; Pecinka et al., 2010).

Draft sequencing of the poplar (Populus trichocarpa; western poplar) genome has led to the construction of 19 scaffolds which contain 370 Mb of sequence (out of 403 Mb) and the identification of > 40 000 protein-coding transcripts (http://www.phytozome.net/poplar; Tuskan et al., 2006). As a consequence of its genome sequence becoming available, along with various molecular tools, and its substantial genetic and phenotypic variation, Populus spp. has become widely used as a model tree (Tuskan et al., 2006; Jansson & Douglas, 2007). As perennial plants with long life-spans and generation times, trees have to face and acclimate to changing environments and are therefore models of interest for epigenetic studies (Hamanishi & Campbell, 2011). Interestingly, genetic variability of global DNA methylation, ranging from 4% to 12%, has been observed in Populus × euramericana hybrid shoot apices (Gourcilleau et al., 2010) and has been positively correlated with biomass production. Furthermore, differences in global DNA methylation paralleled transcriptome level trends in leaves of poplar genotypes during drought, suggesting an epigenomic basis for the clone history-dependent transcriptome divergence (Raj et al., 2011). Recently, drafts of genome-wide P. trichocarpa methylome revealed several particularities of this model, such as a high CHG methylation (Feng et al., 2010a), as well as a negative correlation between promoter and, particularly, gene-body DNA methylation and gene expression level (Vining et al., 2012). However, the relationships between gene-body DNA methylation and tissue-specific gene expression remain to be clarified.

In this context, our aim was to characterize the methylome of noncondensed chromatin to investigate gene-body DNA methylation characteristics in an open chromatin state. Moreover, the elimination of heterochromatic repeated loci would reduce the complexity of the whole-genome analysis. For this purpose, the chromatin fraction hypersensitive to DNase I was isolated from P. trichocarpa shoot apical meristem (SAM) cells and was used in MethylDNA Immunoprecipitation followed by Illumina/Solexa (Fasteris, Plan-les-Ouates, Switzerland) sequencing (MeDIP-SEQ). The mapping on the P. trichocarpa genome was confirmed by bisulfite sequencing of candidate mapped loci to assess the cytosine contexts and the corresponding methylation levels. Taken together, our results showed that methylated DNase I hypersensitive sites covered 2% of the genome but were regularly distributed along poplar scaffolds and strongly enriched in genes, particularly in exons. Furthermore, gene-body DNA methylation in the noncondensed chromatin fraction was dependent on structural gene characteristics, redundancy in the genome, tissue-specific pattern of expression and among poplar genotypes.

Materials and Methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Plant material and growth conditions

Unrooted dormant stem cuttings of Populus trichocarpa (Torr. & Gray) clone 101-74 (used for the DNase-MeDIP-SEQ procedure and subsequent manipulations) and Populus hybrid (Populus deltoides × Populus nigra) clones Carpaccio and Soligo (only used for RT-PCR and bisulfite sequencing analysis of the PtMET1a (POPTR_0019s00240) and PtMET1b (POPTR_0004s14140) genes) were obtained from National Institute for Agronomic Research (INRA) nurseries (Orléans, France in 2008 and Nancy, France in 2006, respectively). Ten-year-old mother plants are cut off each spring. Cuttings were 22 cm long and 0.5–2 cm wide and were harvested in February from plants grown in INRA nurseries. Unrooted cuttings (20 for 101-74 and 12 for each hybrid clone) were then stored for 1 month in a cold (2°C) dark room. Before potting, cuttings were soaked in water in the vertical position to encourage rooting. For Carpaccio and Soligo cuttings, growth conditions have been previously described (Gourcilleau et al., 2010). Rooted cuttings were planted in 20-l pots containing a mixture of peat and sand (50/50, v/v) amended with magnesian chalk (60 g per 100 l) and fertilized with a slow-release fertilizer (Nutricote T100; 13/13/13/2, N/P/K/Mg + trace elements; FERTIL SA, Boulogne Billancourt, France) according to (Gourcilleau et al., 2010). All cuttings were placed in a glasshouse where the temperature was maintained in the range 15–27°C and where they were exposed to natural daylight (ranging from 350 to 900 μmol m−2 s−1). Plants were manually watered every second day for 5 wk. Active buds from 12 to 20 individuals of each genotype were collected, and leaves were removed to isolate the shoot apex (Fig. 1a). Shoot apices were then cut in half with a scalpel and cleared of all visible differentiated tissues under a binocular magnifying glass. Samples that contained a shoot apical meristem (Fig. 1a) were then frozen in liquid nitrogen and stored at −80°C.

image

Figure 1. The DNase I-MethylDNA Immunoprecipitation followed by Illumina/Solexa sequencing (DNase I-MeDIP-SEQ) approach. (a) Description of the DNase I-MeDIP-SEQ approach. The pictures show the shoot apical meristem isolation procedure, from left to right. Arrows with red dotted lines show sections of vegetal material; dotted line surrounds the meristem used for the DNase I-MeDIP-SEQ procedure. White flashes represent DNase I action on hypersensitive sites, that is, nucleosome-free DNA. (b, c) Distribution of sequences from the DNase I-MeDIP-SEQ fraction on Populus trichocarpa scaffolds. Number of (b) MeDIP contigs and (c) number of 36-mers reads are shown for each scaffold (v2 genome version) depending on its size in Mb. MeDIP contigs are contiguous loci where one or several 36-mer reads were mapped. A significant Pearson linear correlation (< 0.01) between the number of MeDIP contigs and scaffold size is indicated by **. 8, 9, 14 and 17 are scaffold numbers.

Download figure to PowerPoint

Chromatin extraction and DNase I hypersensitive sites

Shoot apical meristems from 10 individuals (c. 500 mg fresh weight per extraction) of the 101-74 P. trichocarpa genotype were ground to a fine powder in liquid N2 and used for chromatin extraction. Six chromatin extraction protocols (Galbraith et al., 1983; Steinmuller & Apel, 1986; Lechner et al., 2000; Causevic et al., 2006; Loureiro et al., 2007) were tested in parallel (Supporting Information Table S1). Chemicals were purchased from Sigma-Aldrich (Saint-Quentin Fallavier, France). After extraction, the chromatin pellet was solubilized in DNase I (Roche, Meylan, France) digestion buffer composed of 10 mM TrisHCl (pH 7.8), 250 mM saccharose, 5 mM MgCl2, 5 mM KCl and 0.1% (v/v) 2-mercaptoethanol (Crawford et al., 2006). Different concentrations of DNase I, ranging from 0.1 to 1 units ml−1, and durations of incubation, from 0 to 10 min, were tested to achieve limited chromatin digestion at hypersensitive sites. The reaction was stopped by adding 0.5 volume of a solution containing 2% (w/v) NaDodSO4 and 80 mM EDTA equilibrated at pH 8.0 and 0.5 volume of 4 M NaCl. Genomic DNA was isolated after overnight proteinase K incubation at 37°C, phenol/chloroform/isoamyl alcohol (25 : 24 : 1) (v/v/v) extraction and ethanol precipitation. The DNase I hypersensitive fraction, ranging from 0.4 to 2.0 kb, was recovered after electrophoresis on a low melting point 0.8% agarose gel and ethidium bromide staining using the QIAEX II Gel extraction kit (Qiagen, Courtaboeuf, France) according to the manufacturer's recommendations.

Methyl DNA ImmunoPrecipitation (MeDIP)

Purified DNA (0.4–2 kb) extracted from the DNase I hypersensitive chromatin fraction was randomly nebulized using a Nebulizer kit (Invitrogen, Cergy-Pontoise, France) at a pressure of 30 psi for 2 min to obtain fragments ranging from 0.2 to 1 kb. Rabbit polyclonal anti-5mC antibodies (250 μg; Maine Biotechnology Service, Portland, Maine, USA) were cross-linked to 300 μl of a batch of N-hydroxysuccinimide (NHS)-activated Sepharose matrix (NHS HP SpinTrap kit; GE HealthCare, Aulnay-sous-Bois, France) in accordance with the manufacturer's instructions (www.gelifesciences.com/trap). Matrix–antibody coupling was estimated using PD-10 desalting columns (GE HealthCare) according to the manufacturer's recommendations and was estimated at > 70%. The nebulized DNA (2.5 μg) was solubilized in a 10 mM Na-phosphate buffer (pH 7.0) containing 140 mM NaCl and 0.05% (v/v) Triton-X (IP buffer 1X) and incubated with the affinity matrix under regular slow mixing for 1 h at room temperature and for a further 3 h at 4°C. The batch was washed three times with the IP buffer 1X before resuspension in a proteinase K digestion solution (50 mM Tris at pH 8.0, 10 mM EDTA, 0.5% (w/v) SDS and 20 mg ml−1 proteinase K) at 50°C for 3 h. DNA was then isolated following the classical procedure of phenol/chloroform/isoamyl alcohol extraction and ethanol precipitation. This purification was followed by a second step using a MinElute PCR purification kit (Qiagen). DNA bound (methylated DNA) and unbound (1 : 60 ratio) to the affinity matrix was recovered.

High-throughput sequencing of the MeDIP fraction (MeDIP-SEQ) and controls

The purified poplar methyl-DNA immunoprecipitated fraction was physically fragmented by nebulization into 50–200-bp fragments, end-repaired, ligated with bar-coded adapters, PCR-amplified, purified on an agarose gel and used to generate a DNA colony template library according to the Fasteris procedure (Fasteris, Plan-les-Ouates, Switzerland). Illumina/Solexa sequencing (Bennett, 2004) of the fraction was performed on a Genome Analyzer GAII (Illumina, San Diego, CA, USA) using a Chrysalis 36 cycles v 3.0 sequencing kit and a GAPipeline 1.3.2 data analysis pipeline (Table S2). Illumina quality control was performed by direct capillary sequencing of 30 sequences from the (Illumina, USA) preliminary DNA clone library and comparison with the P. trichocarpa genome (Blast). The methylation status of two of these sequences was confirmed by bisulfite sequencing (see primers in Table S3). Semiquantitative PCR was performed on random sonicated DNA and the DNA used for the Illumina sequencing (the DNase I-MeDIP-SEQ fraction) to amplify two sequences: one covered by the DNase I-MeDIP Illumina sequencing reads (POPTR_0006s21000; read depth = 3.4) and one not covered at all (POPTR_0011s13770). Random Illumina sequencing of P. trichocarpa genome was performed with genomic DNA extracted from young leaves of the P. tricocarpha 101-74 genotype on the GAII platform (Centre National de Génotypage, Evry, France; P. Faivre-Rampant et al., unpublished). 32 312 796 reads were generated, achieving a mean coverage of 16.44, covering 93.5% of the reference genome by at least one read.

Bioinformatics analyses

Short reads mapping

Mapping of the 36-mer reads from Illumina/Solexa sequencing or mapping after de novo assembly with the Exact DE Novo Assembler (EDENA, v2.1.1; Hernandez et al., 2008) was performed on v2.0 of the 403-Mb P. trichocarpa unmasked genome (http://www.phytozome.net/poplar.php; Tuskan et al., 2006) using Mapping and Assembly with Qualities (maq, v0.7.1; http://maq.sourceforge.net/) software and confirmed using eland v2.0 (Efficient Local Alignment of Nucleotide Data, casava v1.6.0, Illumina, USA). Two mismatches per 36-mer read were allowed in the first 24 bases on the reference sequence for maq and in the first 32 bases for eland. The sequencing data are available in the NCBI Sequence Read Archive (SRA) database (SRA number: SRA050249) and on the Popgenie website (http://www.popgenie.org/).

Sequence annotation

Analysis of the content of repeats was performed with RepeatMasker (http://www.repeatmasker.org/) using P. trichocarpa and Arabidopsis thaliana (data not shown) as DNA sources. A few repeated sequences were validated as transposable elements using PlotRep (http://repeats.abc.hu/cgi-bin/plotrep.pl) and RepPop (http://csbl.bmb.uga.edu/~ffzhou/RepPop/) with default parameters. Identification of transposase was performed using Fgenesh software (http://linux1.softberry.com/berry.phtml?topic=fgenesh&group=programs&subgroup=gfind) and confirmed by TBlastX analysis (e-value < 10−30) on a flowering plant transcripts database. Identification of potential small RNAs was carried out with psRNATarget (http://plantgrn.noble.org/psRNATarget/) on mRNA sequence with a maximum expectation of 3. Gene annotation from Phytozome poplar genome v2.0 (http://www.phytozome.net/poplar.php) was used. Gene ontology class enrichment was produced with AgriGO software (http://bioinfo.cau.edu.cn/agriGO/analysis.php). Nucleosome occupancy was a computational prediction produced by Segal Lab software version 3.0 (http://genie.weizmann.ac.il/software/nucleo_prediction.html). ‘Low redundancy’ and ‘high redundancy’ coding sequences were determined according to their e-values after a blastN comparison of 500 nucleotides centered on each sequence with P. trichocarpa v2.0 genome. If there were more than two results with an e-value under 10−50, sequences were considered as highly redundant and otherwise they were classified as lowly redundant.

Expression data

Shoot apical meristem microrray expression data were obtained from the Popgenie database, using the tissue comparison experiment data (Street et al., 2008), available in the Umeå Plant Science Centre database (http://www.upscbase.db.umu.se/) with the ref number UMA-0020, sample L1 (apical region). The data used were relative expression compared with other tissues. Digital northern heat maps representing the library distribution of expressed sequence tags (ESTs) representing gene models within PopulusDB were produced using the PopGenIE DigitalNorthern tool (http://popgenie.org/tool/digitalnorthern).

Bisulfite sequencing and methyl-sensitive PCR amplification

Genomic DNA was extracted from poplar shoot apical meristems using an RNase A digestion, phenol/chloroform extraction and ethanol precipitation procedure (Causevic et al., 2005). Controls for bisulfite conversion efficiency and amplification and the primer design, PCR and sequencing procedures have been recently described (Trap-Gentil et al., 2011). Quantification of cytosine methylation percentages for each cytosine position was obtained from three biological and two technical replicates for each gene and each genotype. Methyl-sensitive PCR (MS-PCR) using McrBC (New England BioLabs, Ipswich, UK) was performed to verify the bisulfite sequencing data according to the detailed procedure of Trap-Gentil et al. (2011). Primers used for bisulfite sequencing and MS-PCR are listed in Table S3. Three biological (SD = 3.72%) and three technical (SD = 0.99%) replicates were performed for each sequence.

RNA extraction and semiquantitative RT-PCR

Total RNAs were isolated from poplar shoot apical meristems using Nucleospin® RNA Plant (Macherey-Nagel, Hoerdt, France). Approximately 500 ng of total RNA was reverse-transcribed using the High-Capacity cDNA Reverse Transcription kit (Applied Biosystems, Courtaboeuf, France). Constitutively expressed genes encoding, respectively, a tubulin and an actin, POPTR_0001s25410 and POPTR_0019s02630, were used as internal standards to normalize the amount of mRNA in the PCR reaction according to Trap-Gentil et al. (2011). POPTR_0001s35160 amplification was also used to confirm that there was no genomic DNA contamination, as primers were designed based on a coding sequence containing a short intron (209 bp). Furthermore, a control without reverse transcriptase during the cDNA synthesis was performed to confirm the absence of genomic DNA in all the total RNA preparations. Primers for PCR were designed using eprimer3 software (http://mobyle.pasteur.fr/cgi-bin/MobylePortal/portal.py?form=eprimer3) and are listed in Table S3. The number of PCR cycles was adjusted to avoid reaching saturation. PCR products were visualized on a 8% polyacrylamide gel after ethidium bromide staining. Three biological and two technical replicates were performed for each gene and genotype.

Statistical analyses

Statistical analyses were performed using the spss statistical software package (SPSS version 11.0.1 PC; SPSS, Chicago, IL, USA). Means are expressed with their standard errors and compared by analysis of variance (one-way ANOVA; general linear model (GLM) procedure). Fraction effects on mapping distribution were evaluated using the χ2 homogeneity test. Relationships between pairs of continuous variables were analyzed by linear regression analysis (Pearson's correlation coefficients (rP)). Statistical tests were considered significant at: *, < 0.05; **, < 0.01; or ***, < 0.001.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

MeDIP-sequencing of the DNase I hypersensitive chromatin fraction extracted from shoot apical meristem cells

Shoot apical meristems from P. trichocarpa genotype 101-74 were isolated from all differentiated tissues (Fig. 1a). Chromatin extraction was optimized by testing six protocols to achieve efficient DNase I digestibility and high quality (purity, concentration and integrity) of the extracted DNA (Table S1 and the 'Materials and Methods' section). The most efficient and reliable protocol (Causevic et al., 2006; Table S1) was further used to isolate DNase I hypersensitive sites ranging from 0.4 to 2.0 kb (c. 30% of total DNA according to gel scan analysis), representing 2.5 μg of DNA, which was then gel-extracted, sonicated (0.2–0.8 kb) and immunoprecipitated with 5-methylcytosine antibodies (MeDIP approach; see the 'Materials and Methods' section for conditions). Purified methylated DNA (10 ng), representing 0.4% of the initial amount, was then sequenced using Illumina/Solexa technology and generated over 8 million reads of 36 bases (MeDIP reads). The sequencing data are available in the NCBI SRA database (SRA number: SRA050249) and on the Popgenie website (http://www.popgenie.org/; Fig. 2c).

image

Figure 2. Scaffold distribution of the DNase I-MethylDNA Immunoprecipitation followed by Illumina/Solexa sequencing (DNase I-MeDIP-SEQ) fraction. (a) Distribution of MeDIP reads, genes, repeats and sRNA densities along scaffold 12 (v2.0 genome version). Repeats, gene densities and sRNA mapping were obtained from the Popgenie database (http://www.popgenie.org/popgenie1//; Klevebring et al., 2009). (b) Locus on scaffold 12 (v2.0 genome) with a Long Terminal Repeat (LTR) and a gene model (POPTR_0012s00830). Exons and introns of the gene model are represented by light gray rectangles and black lines, respectively. The arrow indicates gene orientation. MeDIP contigs are represented by gray rectangles whose width is the covered region and height is read depth (read depth = number of reads/peak surface). Black rectangles represent nucleosome occupancy > 0.5. (c) Popgenie Gbrowser view of the locus from scaffold 12 shown in (b). The first track is v2.0 transcripts, the second is MeDIP contigs mapping with their number of MeDIP reads and length, and the last track is repeated sequences as detected by RepeatMasker.

Download figure to PowerPoint

Mapping of DNase I-MeDIP-SEQ reads on the P. trichocarpa genome assembly

After filtration, 180 650 MeDIP contigs (729 378 36-mers total; 8.7% of total reads), composed of 1–15 991 MeDIP reads, with a mean value of four reads, were mapped on the P. trichocarpa v2.0 genome assembly (Table S2). MeDIP contigs ranging from 36 to 5078 bp covered 1.9% of the P. trichocarpa v2.0 genome assembly (7 834 392 bp). Their read depth, defined as the number of MeDIP reads divided by the length of the covered genomic region, varied from 1 to 459, with a mean value of 3.35.

Validation of MeDIP-SEQ and mapping

Immunoprecipitation was assessed by semiquantitative PCR on two loci methylated or not according to prior bisulfite sequencing and showed amplification in the MeDIP fraction only for the methylated sequence (POPTR_0006s21000; Fig. S1a) but not for the nonmethylated sequence (PtTUB2, POPTR_0001s25410; Trap-Gentil et al., 2011; Fig. S2a). Bisulfite sequencing and MS-PCR (Fig. S2b, Table S3) of 34 regions (13 586 bp in total) randomly selected among genic, intergenic and repeated sequences confirmed that all sequences mapped by MeDIP reads were methylated. In addition, a positive linear correlation (Pearson coefficient, rP = 0.77; < 0.05) was established between the density of MeDIP reads and the methylcytosine percentage obtained by bisulfite sequencing. Finally, the mapping of MeDIP reads showed differential distributions on the P. trichocarpa genome compared with a random sequencing (Illumina) of the total genome, as evaluated by the χ2 homogeneity test, for example throughout the first 1 Mb of scaffold 1 (χ2 = 17.78*).

Mapping of DNase I-MeDIP-SEQ reads at scaffold space level

A linear positive correlation (rP = 0.99; < 0.01) was established between the number of mapped MeDIP contigs and the scaffold size of P. trichocarpa v2.0 (Fig. 1b). However, no correlation could be detected between the number of mapped MeDIP reads by scaffold and their sizes (Fig. 1b). Indeed, scaffolds 8, 9, 14 and 17 showed a high number of MeDIP reads with respect to their sizes (Fig 1c). These reads mapped on repeated sequences (Fig. S3a) and more precisely on rDNA sequences (18S/26S and 5S rDNA), which were particularly abundant in these four scaffolds (Fig. S3b).

Heterogeneity of the DNase-MeDIP read distribution was observed all along each scaffold and within them (Fig. 2). A linear correlation was established between the distribution of DNase-MeDIP reads and sRNA along scaffolds (rp = 0.501; < 0.001 on scaffold 12), while no relationships could be established with the densities of repeats or genes (Fig. 2a). The ratio between the densities of DNase I-MeDIP reads and the whole-genome MeDIP reads (Vining et al., 2012; http://http:poplar.cgrb.oregonstate.edu) was calculated and compared with the repeat density for scaffold 12 (Fig. S3c). A significant negative correlation was detected (rS = −0.191; < 0.05) in relation to the very low density of DNase I-MeDIP reads compared with whole-genome MeDIP reads in repeat-rich regions.

Mapping of DNase I-MeDIP-SEQ reads at the coding sequence space level

Mapped MeDIP reads were distributed between intergenic regions (61%), genes (26%) and repeated sequences (13%; Fig. 3a). A depletion of c. 30% of repeated sequence was observed compare with the whole genome. Most of the repeats corresponded to retro-elements (58%) and DNA transposons (29%; Fig. 3b). It was noteworthy that 74% of v2.0 model genes (41 377 models) were mapped by MeDIP reads in the body region (74%; Fig. 3c) and/or in the ± 2 kb 5′ and 3′ flanking regions (59%; Fig. 3d). The gene-body regions were mostly mapped on exons (89%; Fig. 3e) close to the beginning or the end of predicted nucleosome sites (see one example in Fig. 2b).

image

Figure 3. Nature of sequences in the DNase I-MethylDNA Immunoprecipitation followed by Illumina/Solexa sequencing (DNase I-MeDIP-SEQ) fraction. (a) Pie chart representing proportion of regions covering v2.0 gene models, repeats or intergenic loci in the DNase I-MeDIP-SEQ fraction. (b) Pie chart representing proportion of different repeat classes in the DNase I-MeDIP-SEQ fraction. (c) Pie chart representing proportion of v2.0 gene models contained or not in the DNase I-MeDIP-SEQ fraction. Genes with at least one mapped read in one of their parts are considered as contained in the DNase I-MeDIP-SEQ fraction. Body, 2-kb upstream and 2-kb downstream regions are taken as parts of the gene. (d) Venn diagram with number of v2.0 genes whose body, 2-kb upstream and 2-kb downstream parts were mapped with at least one read. (e) Venn diagram with number of v2.0 genes whose exon(s) or intron(s) was mapped with at least one read.

Download figure to PowerPoint

The distribution of MeDIP reads along gene models showed a peak in the body part compared with flanking regions, with strong variations at the borders of the genes (Fig. 4). This distribution was affected by gene size (χ2 = 6.68, < 0.01; Fig. 4a). Thus, long genes (over 2 or 5 kb), which were found in higher proportions in the DNase I-MeDIP-SEQ fraction than in the whole genome (Fig. S4), exhibited fewer MeDIP reads in the central part of the gene body and more in the flanking regions than small genes (under 2 kb; Fig. 4a).

image

Figure 4. Sequencing read profile along gene models (from 2 kb upstream to 2 kb downstream). For each gene part, read intensity is shown as a percentage of total reads from 2 kb upstream to 2 kb downstream. (a) Genes are separated according to their size: [0 kb, 2 kb[, continuous line; [2 kb, 5 kb[, dashed line; [5 kb, +∞[, dotted line. The χ2 value was calculated on the read distribution inside the gene body and was significant at < 0.01. (b) Genes are separated according to their expression log value in the apical region (gene expression microarray experiments on tissue comparison; Street et al., 2008): ]−∞,−3], continuous line; ]−3, +3], dashed line; ]+3, +∞[, dotted line.

Download figure to PowerPoint

Concerning rDNA clusters, only some 5S clusters were mapped (data not shown), while all major 18S-5.8S-26S clusters (on scaffolds 14 and 17) and minor dispersed 18S/26S loci (on scaffolds 8 and 9) were covered by DNase I-MeDIP-SEQ reads (Fig. 5a). In addition, more MeDIP reads were mapped between 18S, 5.8S and 26S units than inside them (Fig. 5a). Bisulfite sequencing showed that this distribution is related to the density of methylation of these sequences (Fig. 5b). Variations in the levels of methylation of non-CG sites were also found all along a 18S-5.8S-26S unit and associated with a CHH methylation restricted to CAA sites.

image

Figure 5. DNA methylation of rDNA in the DNase I-MethylDNA Immunoprecipitation followed by Illumina/Solexa sequencing (DNase I-MeDIP-SEQ) fraction. (a) Locus on scaffold 17 containing rDNA clusters. MeDIP contigs are represented by white rectangles whose width is the covered region and height is the read depth (read depth = number of reads/peak surface). 18S-5.8S-26S ribosomal transcription units were localized by blast of 18S (blue traits), 5.8S (black traits) and 26S (red traits) poplar sequences (AF206999.1, AJ006440.1 and AF479118.1 GenBank accession numbers, respectively) in the v2.0 genome. The arrow shows the orientation of units. Gray regions separate each ribosomal transcription unit. (b) Bisulfite sequencing analysis of DNA methylation in Populus trichocarpa shoot apex of loci ‘a’, ‘b’ and ‘c’ corresponding, respectively, to the 5′ region of a ribosomal unit, the body of 18S rDNA and the 3′ region of a ribosomal unit. CG, CHG and CHH sites are represented by black diamonds, white squares and gray triangles, respectively. DNA methylation mean values are given for each sequence. Sequence of CHH contexts methylated above 35% is shown.

Download figure to PowerPoint

Expression and gene ontology analysis of genes mapped by DNase I-MeDIP-SEQ reads

The relative tissue-specific expression level in SAM compared with other poplar tissues (Street et al., 2008) was shown to affect the MeDIP read distribution along genes (Fig. 4b). Thus, SAM-specific overexpressed genes (log2 values > +3) exhibited two internal peaks at the beginning and at the end (the highest) of the body gene, respectively, and very few mapped MeDIP reads in the flanking regions (Fig. 4b). Furthermore, the enrichment ratio for genes in the DNase I-MeDIP-SEQ fraction compared with the whole genome was also affected by their relative tissue-specific expression level in SAM depending on the region of the gene considered (exon, intron, or 5′ or 3′ flanking region; Fig. 6). Indeed, for all regions, expressed genes with no tissue specificity (log2 values c. 0) were found in conserved ratios (c. 1) between the DNase I-MeDIP-SEQ fraction and the whole genome. In contrast, genes with SAM-specific expression (overexpressed with log2 > +4 or repressed with log2 < −4) had variable ratios ranging from 0 (not found in the DNase I-MeDIP-SEQ fraction) to 2 (enriched in the DNase I-MeDIP-SEQ fraction; Fig. 6). Among them, genes mapped on intronic regions were mostly under-represented in the DNase I-MeDIP-SEQ fraction. Digital northern analysis showed that genes mapped on intronic regions exhibited a variable level of expression in SAM (number of ESTs in each cDNA library; Fig. S5). To validate our approach, we confirmed the relative expression profiles of 24 genes by semiquantitative RT-PCR (Table S4) and by correlating the public data with personal transcriptomic microarray data on SAM (data not shown).

image

Figure 6. Relative expression of genes from the DNase I-MethylDNA Immunoprecipitation followed by Illumina/Solexa sequencing (DNase I-MeDIP-SEQ) fraction compared with the whole genome in the apical region of vegetative buds. Relative expression is given as normalized log2 values. The comparison between the DNase I-MeDIP-SEQ fraction and the whole genome is expressed as the ratio between relative frequencies for each expression class. The comparison was performed for genes whose exons, introns, 5′ regions or 3′ regions are found in the DNase I-MeDIP-SEQ fraction. The expression data are extracted from the Street et al. (2008) tissue comparison study.

Download figure to PowerPoint

A gene ontology study (singular enrichment analysis) for mapped genes showed that all biological categories were represented but with variations depending on the gene regions mapped. Interestingly, many expressed genes involved in the determination and the control of SAM activity, such as LEAFY, CLAVATA2 and WUSCHEL, were found in the DNase I-MeDIP-SEQ fraction (Table S4) and confirmed by bisulfite sequencing. Mapping in the 5′ and 3′ flanking regions gave a similar distribution between GO categories (rp = 0.77; < 0.01) but this distribution was different for genes mapped in their body (rp = −0.69; < 0.01). Genes mapped in flanking regions were depleted (< 50%) in transcription regulator activity, biological regulation and regulation of biological process, while genes mapped in their bodies were depleted in electron carrier activity and response to stimulus (Fig. S6). Detailed GO analysis (singular enrichment analysis) also identified over-represented classes of proteins: helicase, ATP-binding and ATP-ase activity for introns and exons and RNA-directed DNA polymerase activity for 5′ and 3′ regions (Figs S7–S10).

Methylation at the cytosine level: contexts and levels

The three cytosine methylation contexts displayed enrichment ratios ranging from 1.17 to 2.71 between the DNase I-MeDIP-SEQ fraction and the whole genome (Fig. 7a). For non-CG contexts, the highest ratios were found for contexts containing at least two cytosines (Fig. 7b,c).

image

Figure 7. Enrichment in cytosine methylation contexts in the DNase I-MethylDNA Immunoprecipitation followed by Illumina/Solexa sequencing (DNase I-MeDIP-SEQ) fraction. (a) Enrichment ratio of CG, CHG and CHH sites between the DNase I-MeDIP-SEQ fraction and v2.0 genome. This ratio is given for each (b) CHG and (c) CHH combination. The enrichment ratio is the ratio between each site frequency in the DNase I-MeDIP-SEQ fraction and the same frequency in the v2.0 genome.

Download figure to PowerPoint

Bisulfite sequencing of 23 randomly selected sequences found in the DNase I-MeDIP-SEQ fraction (10 312 bp total; Table S3), representing 373 CG, 445 CHG and 1133 CHH, confirmed that CG context was more frequently methylated (334 mCG; mean value: 80%) than CHG (239 mCHG; 40%) and CHH (166 mCHH; 20%) contexts (Fig. 8a). The levels and the context preferences of methylation were dependent on the type of sequences (gene coding protein, rDNA or transposable element (TE)) and the redundancy of genes in the genome (≤ 2 or > 2; Fig. 8b,c and Table S5). Indeed, coding sequences with one or two copies exhibited strong CG methylation (> 70%) and low non-CG methylation (< 10%) with preference for C(C/A)G sites but none for CHH contexts (Figs 8, 9a). Coding sequences with more than two copies as well as rDNA and TE exhibited high CG (> 80%) and CHG (> 60%) methylation with moderate methylation (> 20%) of CHH, except for rDNA, which showed a lower CHH methylation level (< 10%). All these sequences showed site-specific methylation on C(A/T)G and CAA sites (Figs 5, 8, S1). Computational analysis did not detect TE inside or within 500 bp around genes, whether they were highly redundant or not.

image

Figure 8. DNA methylation of CG and non-CG contexts in the DNase I-MethylDNA Immunoprecipitation followed by Illumina/Solexa sequencing (DNase I-MeDIP-SEQ) fraction. (a) Average DNA methylation values (bisulfite sequencing) of CG, CHG and CHH contexts in 23 sequences found in the DNase I-MeDIP-SEQ fraction and randomly chosen. Means are given with ± SE. Significant DNA methylation differences (< 0.05) among the three nucleotidic contexts, as determined by one-way ANOVA, are shown by letters ‘a’, ‘b’ and ‘c’. (b) Average DNA methylation values of CG, CHG and CHH in lowly (black) and highly (gray) redundant coding sequences, rDNA (white) and TE (hatched). Means are given with their standard errors. ‘Low redundancy’ coding sequences: = 8; ‘high redundancy’ coding sequences: = 6; rDNA:= 4; TE:= 3. For each cytosine context, significant DNA methylation differences (< 0.05) among the four sequence types, as determined by one-way ANOVA, are shown by letters ‘a’ and ‘b’. (c) Sequence logos showing preferentially methylated CHG and CHH sites in coding with ‘low redundancy’, coding with ‘high redundancy’, rDNA and TE sequences, from left to right. CHG and CHH sequence logos were obtained using GENIO/logo software (http://www.biogenio.com/logo/). Sequence logos were made with the same number of each CHG or CHH combination to eliminate site frequency-dependent bias. The number of analyzed sites in lowly redundant coding sequences: = 155 CHG and 463 CHH; in highly redundant coding sequences: = 176 CHG and 501 CHH; in rDNA:= 95 CHG and 221 CHH; in TE:= 55 CHG and 98 CHH.

Download figure to PowerPoint

image

Figure 9. DNA methylation and mRNA relative abundance of two poplar MET1 genes in different genetic backgrounds. (a) Genomic context of POPTR_0019s00240 (PtMET1a, left) and partial DNA methylation profile in the shoot apex of three genetic backgrounds (right). (b) Genomic context of POPTR_0004s14140 (PtMET1a, left) and partial DNA methylation profile in the shoot apex of three genetic backgrounds (right). The two genes are shown with their MeDIP contigs from the DNase I-MethylDNA Immunoprecipitation followed by Illumina/Solexa sequencing (DNase I-MeDIP-SEQ) fraction. The black bar above the gene represents the locus analyzed by bisulfite sequencing. The three genetic backgrounds are Populus trichocarpa (the 101-74 genotype) and P. × euramericana (Soligo and Carpaccio genotypes). CG, CHG and CHH sites are represented by black diamonds, white squares and gray triangles, respectively. CG sites mutated to TG in P. × euramericana compared with P. trichocarpa are shown by asterisks. Exon and intron regions are gray and white, respectively. DNA methylation mean values are given for each sequence. (c) mRNA relative abundance of POPTR_0019s00240 (PtMET1a) and POPTR_0004s14140 (PtMET1b) in P. trichocarpa (the 101-74 genotype) and P. × euramericana (Soligo and Carpaccio genotypes) shoot apex assessed by RT-PCR analyses. PtACTIN7 (POPTR_0019s02630) and PtTUBULIN-β2 (POPTR_0001s25410) cDNA amplification was used for normalization between samples. Amplification on genomic DNA was used as a positive control.

Download figure to PowerPoint

Genetic variations: DNA methylation and expression of DNA METHYLTRANSFERASE 1 (MET1) genes in P. trichocarpa and two P. × euramericana hybrid clones

The two P. trichocarpa MET1-like genes (83.9% nucleotidic similarity), PtMET1a (POPTR_0019s00240) and PtMET1b (POPTR_0004s14140), were mapped by DNase I-MeDIP-SEQ reads, particularly in exons, but not at the same positions (Fig. 9a,b). Bisulfite sequencing confirmed significant methylation only at CG sites between the two genes (Fig. 9a,b). The methylation profiles for these genes were assessed in two P. × euramericana hybrid clones, Soligo and Carpaccio, and showed both genetic and epigenetic variations associated with the level of expression. Thus, PtMET1a exhibited lower methylation levels in the hybrid clones in relation with mutations (CG to TG transitions) but higher levels of mRNA accumulation (Fig. 9c). By contrast, PtMET1b had no mutations among genotypes and showed no methylation or expression in 101-74 and Soligo, while Carpaccio was highly methylated in all cytosine contexts and exhibited expression (Fig. 9c).

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

DNase I-MeDIP-SEQ strategy for methylome assembly studies

Several methods, including DNA methylation microarray analysis, MeDIP-SEQ and direct bisulfite sequencing, have been used in recent years to assess the methylome in several organisms, such as rice (Oryza sativa), human and honeybee (Weng et al., 2009; Zemach et al., 2010). The choice of method is mainly dependent on the quality of genomic resources, the availability of molecular tools, and the objectives in terms of genome coverage and scale resolution, but costs and the need for further bioinformatics analyses must also be considered. In this context, MeDIP-SEQ was the most suitable method for genomic studies in poplar (Vining et al., 2012). Our objective was to establish a draft of the methylome that will highlight gene-body DNA methylation characteristics in poplar, a model tree. In this context, the strategy developed in the present work was to characterize by MeDIP-SEQ the methylome of the DNase I hypersensitive chromatin fraction, for two main reasons. First, this genomic fraction should be strongly enriched in genes and regulatory sequences, allowing focus on gene-body DNA methylation, and depleted in repeated loci, which cause difficulties in bioinformatics analyses. Indeed, the poplar genome is 30% heterochromatin and 33 Mb of unassembled sequences are not incorporated in the actual genome assembly (403 Mb). Secondly, use of this fraction will allow the mapping of epigenetic marks in a specific structural chromatin context: the open chromatin, permissive to transcription. For this purpose, experimental conditions for chromatin extraction, efficient DNase I digestibility and the extraction of high-quality DNA were optimized.

The DNase I-MeDIP-SEQ mapping yielded a proportion of mapped reads, a coverage, and a number and size of contigs in agreement with those reported previously in similar studies (Vining et al., 2012) and provides a useful public database (Sjödin et al., 2009) for other investigations. The procedure was also validated using PCR amplifications of methylated and nonmethylated sequences in the immunoprecipitated fraction, comparison with the mapping of a random sequencing of the P. trichocarpa genome and bisulfite sequencing of randomly selected regions. A linear distribution was observed for DNase I-MeDIP-SEQ contigs along scaffolds correlated to their size as well as a depletion of DNase I-MeDIP-SEQ contigs in putative highly methylated pericentromeric regions, which were strongly targeted in previous total methylome approaches, such as in poplar (Vining et al., 2012). The low genome coverage (1.9%) is consistent with the low global DNA methylation in poplar apices (c. 10%; Gourcilleau et al., 2010), mostly located in heterochromatic regions and nucleosome-bound DNA (Chodavarapu et al., 2010), but also with the use of 30% of extracted DNA after limited DNase I digestion performed according to the procedure of Crawford et al. (2006). With 1.9% of the P. trichocarpa genome covered, sequences of 30 679 genes (including 2 kb of upstream and downstream regions) were mapped, representing 74% of poplar v2.0 gene models (26% promoter-methylated genes and 55% body-methylated genes, including genes that were methylated at both features) and confirming the ability of our strategy to focus on gene-body DNA methylation vs repeated sequences. These percentages contrast with percentages of 5% obtained for promoter-methylated genes and 33% for body-methylated genes in A. thaliana (Zhang et al., 2006) and 17% for promoter-methylated genes and 15% for body-methylated genes in poplar using the whole-genome approach (Vining et al., 2012). However, quantitative differences in comparisons with these data should be interpreted with caution because procedures, calculations and statistical methods were different. Moreover, previous data on poplar relate to a tissue comparison and were obtained using a whole-genome MeDIP-SEQ approach (Vining et al., 2012). One other explanation could be that our mapping was concentrated on a very small fraction of the genome (1.9%) rich in genes, increasing the sensibility of the detection. It is also possible that these data are related to the undifferentiated or highly dividing characteristics of meristematic cells, already observed at the transcriptome and proteome level in maize (Zea mays) (Dembinsky et al., 2007). Further analyses will be needed to clarify this point. This kind of approach allows a low-cost and simplified methylome bioinformatic analysis that could be useful for organisms with a large content of repeated sequences (Morse et al., 2009; Schnable et al., 2009) or whose assembly is incomplete.

DNA methylation and chromatin structure

Methylated DNase I hypersensitive sites were interspersed across the genome, with a depletion in putative pericentromeric regions, a strong enrichment in genes and a large proportion of intergenic regions that might be distal and proximal to regulatory sequences (Crawford et al., 2006). Some of these interspersed hypersensitive sites were mapped on repeated sequences methylated in the three cytosine contexts, suggesting that, as in A. thaliana, the P. trichocarpa epigenome is a mosaic of domains with different chromatin states (Roudier et al., 2011). However, in some scaffolds, long adjacent loci corresponding to rDNA were mapped with strong density, raising questions about ribosomal chromatin organization. It must be admitted that only a few rDNA copies among thousands are in an open chromatin state and transcriptionally active (Tucker et al., 2010). This is contradictory with the finding of two 18S-5.8S-26S clusters (on scaffolds 14 and 17) and two additional dispersed 18S-5.8S-26S loci (on scaffolds 8 and 9) that were almost entirely covered by the DNase I-MeDIP-SEQ mapping. This could be explained by the incomplete genome assembly in rDNA loci with < 100 copies of 18S-5.8S-26S units found by blast localization, whereas this number has been reckoned at several thousand by restriction analyses (Faivre-Rampant et al., 1992). This inaccurate assembly might also explain why we found four rDNA-containing scaffolds while only three chromosomes were found using cytological approaches (Tuskan et al., 2006; Islam-Faridi et al., 2009). In addition, rDNA active copies are much more numerous in meristematic cells than in differentiated cells (Qian et al., 2006). In this context, our most interesting finding was that rDNA loci were more frequently mapped between 18S, 5.8S and 26S units than inside them by DNase I-MeDIP-SEQ reads, which is consistent with previous localization of DNase hypersensitive sites in barley (Hordeum vulgare) rDNA supposed to co-localize with PolI binding sites (Dimitrovna et al., 2009). In addition, CHH contexts were poorly methylated, which is consistent with 5S rDNA low CHH methylation as a consequence of demethylation targeting by DEMETER and DEMETER-LIKE enzymes in A. thaliana (Woo et al., 2008). Finally, these data show that high CG and CHG methylation can co-localize with low nucleosome occupancy (Mathieu et al., 2002).

Taken together, these data confirmed the relevance of methylome studies focused on a specific chromatin state to propose new functional hypothesis and improve our knowledge of epigenetic control.

Insights regarding gene-body DNA methylation

The DNase I-MeDIP-SEQ mapping on the gene body for half of the poplar model genes, preferentially in exons, showed that gene-body DNA methylation is more extensive than previously reported (Vining et al., 2012) and is associated with DNase I hypersensitive sites and the open chromatin state. In A. thaliana and rice, DNase I hypersensitive sites were detected mainly in the promoter but also in the gene body (Zhang et al., 2012a,b), potentially corresponding to origins of replication (ORIs) which are contained in the gene body (Costas et al., 2011). In active ORI sites, DNA is methylated and wrapped in nucleosomes with H2A.Z, a histone variant that compromises nucleosome stability and becomes hypersensitive to DNase (Jin et al., 2009). This phenomenon could be particularly common in poplar meristematic dividing cells, and further investigations are needed to determine the role of gene-body DNA methylation in replication.

In A. thaliana, gene-body methylation was detected in highly expressed and constitutively active genes, whereas promoter-methylated genes show a greater degree of tissue-specific expression (Zhang et al., 2006). By contrast, gene-body methylation in poplar was correlated to tissue-specific expression and had a more negative effect on transcription than promoter-gene methylation (Zilberman et al., 2007; Vining et al., 2012). The proportion of genes with reported expression in microarray analysis was slightly higher (38%) for genes mapped by DNase I-MeDIP-SEQ reads than for unmapped genes (30%), but no relationship with gene expression level was found. This could be explained by the different environment and Populus species used to generate microarray expression data (Street et al., 2008). To address this issue, a methylome/transcriptome cross study has been performed on the same biological material (C. Lafon-Placette, unpublished). However, variations in relation to their size or tissue-specific expression were observed. As long genes were previously shown to be more methylated in A. thaliana (Zhang et al., 2008), the depletion of reads in the central part of long poplar genes might reflect low chromatin accessibility for DNase I preventing binding of transcription factors to a cryptic initiation site (Zilberman et al., 2007). Methylation of SAM-specific overexpressed genes was low in the proximal promoter (< 1 kb), consistent with previous findings (Zhang et al., 2006), and high at the end of the gene body, in agreement with protection against cryptic initiation or a role in termination. Interestingly, genes with SAM-specific expression (under- or over-expressed) exhibited variations (enrichment or depletion) of methylation and/or DNase accessibility in their gene body and 5′/3′ flanking regions compared with ubiquitous genes. This analysis confirmed the possible role of DNA methylation in directing or maintaining tissue-specific gene expression, as proposed by Vining et al. (2012). However, genes mapped in intronic regions displaying SAM-specific expression patterns were strongly under-represented in our methylome. The co-localization of DNA methylation and the nucleosome on exons (Chodavarapu et al., 2010) could not alone explain the depletion in introns only for SAM-specific expressed genes. Recently, the role of intragenic DNA methylation in identification of coding regions was assessed in the rat, and suggested an important role for DNA methylation in the regulation of splicing events and the final constitution of the protein sequence (Sati et al., 2012). Our data suggest that this process will be affected by the gene tissue-specific expression profile.

Single- and two-copy genes were mostly methylated in CG contexts, as previously shown in many eukaryotes (Feng et al., 2010a), possibly through protection from non-CG methylation by the jmjC domain-containing protein INCREASE IN BONSAI METHYLATION 1, as described in A. thaliana (Miura et al., 2009). By contrast, redundant genes were methylated in the three contexts, probably as a result of small RNA targeting (Teixeira & Colot, 2010). Methylated contexts in redundant genes were different from those in lowly redundant genes and mostly similar to those in TEs and rDNA, suggesting a specific recognition of repeated genes by the DNA methylation machinery. Considering the recent genome duplication event that occurred in poplar (Tuskan et al., 2006), these data are consistent with theories involving epigenetics in rapid extinction of new duplicated genes to avoid expression redundancy, which could lead to phenotypic abnormalities (Abrouk et al., 2010). However, as only one copy has been studied for each gene, the investigation of DNA methylation in all copies will be necessary to reveal whether this pattern is specific to one copy or affects all copies.

In conclusion, although most genes in poplar SAM cells are partly in a DNA-methylated accessible chromatin state, this conformation can be modified depending on gene tissue-specific expression, gene size or redundancy and can be involved in the control of transcription, splicing or even replication (Zilberman et al., 2007).

Poplar, a model tree for epigenetic studies

Our data are consistent with recent data obtained in poplar, showing a high level of CHG methylation, an interaction between gene-body and promoter methylation, and tissue-specific expression (Feng et al., 2010a; Vining et al., 2012). We also describe new data: functional mapping of methylation on accessible chromatin, higher proportion of methylated genes, variations of gene-body DNA methylation depending on gene size, poplar genotypes and specific cytosine methylation context.

In particular, the investigation of two genes (PtMET1a and PtMET1b) showed a difference in DNA methylation patterns among poplar genotypes as a result of both genetic (CG[RIGHTWARDS ARROW]TG transitions) and epigenetic variations (C/mC) on exons at CG sites, paralleling expression differences. In poplar, epigenetic variations have previously been reported among hybrids (Gourcilleau et al., 2010) as well as among clone histories (Raj et al., 2011). The present study was performed on three genotypes (obtained from distinct nurseries) and two genes; genome-wide studies are now needed in several genotypes with characterized clone histories to confirm that epigenetic variations between genotypes or clone histories are extensive and affect related gene expression in poplar.

The investigation of variations of DNA methylation in response to environmental changes and between hybrids is a future challenge, and poplar is an appropriate model tree for such epigenetic studies. In addition, studies of SAM cells may answer questions about the impact of epigenetics on development, plasticity, seasonality, memory and inheritance, providing insights that may be of value for tree breeding in the context of global climate change.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

The authors are grateful to D. Auguin and C. Hébrard (University Orléans, France) for helpful discussions. We thank A. Guichard (University Orléans), Patrick Poursat and M. C. Lesage-Descauses (INRA, Orléans), Fasteris (Plan-les-Ouates, Switzerland), V. Vidal and B. Jesson from Imaxio (Clermont-Ferrand) and K. Vining (Oregon University) for technical assistance.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
  • Abrouk M, Murat F, Pont C, Messing J, Jackson S, Faraut T, Tannier E, Plomion C, Cooke R, Feuillet C. 2010. Palaeogenomics of plants: synteny-based modelling of extinct ancestors. Trends in Plant Science 15: 479487.
  • Beck S, Rakyan VK. 2008. The methylome: approaches for global DNA methylation profiling. Trends in Genetics 24: 231237.
  • Bennett S. 2004. Solexa Ltd. Pharmacogenomics 5: 433438.
  • Causevic A, Delaunay A, Ounnar S, Righezza M, Delmotte F, Brignolas F, Hagège D, Maury S. 2005. DNA methylating and demethylating treatments modify phenotype and cell wall differentiation state in sugarbeet cell lines. Plant Physiology and Biochemistry: PPB/Société Française De Physiologie Végétale 43: 681691.
  • Causevic A, Gentil M-V, Delaunay A, El-Soud WA, Garcia Z, Pannetier C, Brignolas F, Hagège D, Maury S. 2006. Relationship between DNA methylation and histone acetylation levels, cell redox and cell differentiation states in sugarbeet lines. Planta 224: 812827.
  • Chodavarapu RK, Feng S, Bernatavichute YV, Chen P-Y, Stroud H, Yu Y, Hetzel JA, Kuo F, Kim J, Cokus SJ et al. 2010. Relationship between nucleosome positioning and DNA methylation. Nature 466: 388392.
  • Cokus SJ, Feng S, Zhang X, Chen Z, Merriman B, Haudenschild CD, Pradhan S, Nelson SF, Pellegrini M, Jacobsen SE. 2008. Shotgun bisulphite sequencing of the Arabidopsis genome reveals DNA methylation patterning. Nature 452: 215219.
  • Costas C, de la Paz Sanchez M, Stroud H, Yu Y, Oliveros JC, Feng S, Benguria A, López-Vidriero I, Zhang X, Solano R et al. 2011. Genome-wide mapping of Arabidopsis thaliana origins of DNA replication and their associated epigenetic marks. Nature Structural & Molecular Biology 18: 395400.
  • Crawford GE, Davis S, Scacheri PC, Renaud G, Halawi MJ, Erdos MR, Green R, Meltzer PS, Wolfsberg TG, Collins FS. 2006. DNase-chip: a high-resolution method to identify DNase I hypersensitive sites using tiled microarrays. Nature Methods 3: 503509.
  • Crawford GE, Holt IE, Mullikin JC, Tai D, Blakesley R, Bouffard G, Young A, Masiello C, Green ED, Wolfsberg TG et al. 2004. Identifying gene regulatory elements by genome-wide recovery of DNase hypersensitive sites. Proceedings of the National Academy of Sciences, USA 101: 992997.
  • Dembinsky D, Woll K, Saleem M, Liu Y, Fu Y, Borsuk LA, Lamkemeyer T, Fladerer C, Madlung J, Barbazuk B et al. 2007. Transcriptomic and proteomic analyses of pericycle cells of the maize primary root. Plant physiology 145: 575588.
  • Dimitrovna AD, Ananiev ED, Gecheff KI. 2009. DNase I hypersensitive sites within the intergenic spacer of ribosomal RNA genes in reconstructed barley karyotypes. Biotechnology & Biotechnological Equipment 23: 10391043.
  • Faivre-Rampant P, Jeandroz S, Lefevre F, Lemoine M, Villar M, Berville A. 1992. Ribosomal DNA studies in poplars: Populus deltoides, P. nigra, P. trichocarpa, P. maximowiczii, and P. alba. Genome 35: 733740.
  • Feng S, Cokus SJ, Zhang X, Chen P-Y, Bostick M, Goll MG, Hetzel J, Jain J, Strauss SH, Halpern ME et al. 2010a. Conservation and divergence of methylation patterning in plants and animals. Proceedings of the National Academy of Sciences, USA 107: 86898694.
  • Feng S, Jacobsen SE, Reik W. 2010b. Epigenetic reprogramming in plant and animal development. Science 330: 622627.
  • Galbraith DW, Harkins KR, Maddox JM, Ayres NM, Sharma DP, Firoozabady E. 1983. Rapid flow cytometric analysis of the cell cycle in intact plant tissues. Science (New York, N.Y.) 220: 10491051.
  • Goll MG, Bestor TH. 2005. Eukaryotic cytosine methyltransferases. Annual Review of Biochemistry 74: 481514.
  • Gourcilleau D, Bogeat-Triboulot M-B, Le Thiec D, Lafon-Placette C, Delaunay A, El-Soud WA, Brignolas F, Maury S. 2010. DNA methylation and histone acetylation: genotypic variations in hybrid poplars, impact of water deficit and relationships with productivity. Annals of Forest Science 67: 208.
  • Hamanishi ET, Campbell MM. 2011. Genome-wide responses to drought in forest trees. Forestry 84: 273283.
  • Hernandez D, François P, Farinelli L, Osterås M, Schrenzel J. 2008. De novo bacterial genome sequencing: millions of very short reads assembled on a desktop computer. Genome research 18: 802809.
  • Islam-Faridi MN, Nelson CD, DiFazio SP, Gunter LE, Tuskan GA. 2009. Cytogenetic analysis of Populus trichocarpa – ribosomal DNA, telomere repeat sequence, and marker-selected BACs. Cytogenetic and Genome Research 125: 7480.
  • Jansson S, Douglas CJ. 2007. Populus: a model system for plant biology. Annual Review of Plant Biology 58: 435458.
  • Jin C, Zang C, Wei G, Cui K, Peng W, Zhao K, Felsenfeld G. 2009. H3.3/H2A.Z double variant-containing nucleosomes mark ‘nucleosome-free regions’ of active promoters and other regulatory regions. Nature Genetics 41: 941945.
  • Klevebring D, Street NR, Fahlgren N, Kasschau KD, Carrington JC, Lundeberg J, Jansson S. 2009. Genome-wide profiling of Populus small RNAs. BMC Genomics 10: 620.
  • Lechner T, Lusser A, Pipal A, Brosch G, Loidl A, Goralik-Schramel M, Sendra R, Wegener S, Walton JD, Loidl P. 2000. RPD3-type histone deacetylases in maize embryos. Biochemistry 39: 16831692.
  • Lister R, Pelizzola M, Dowen RH, Hawkins RD, Hon G, Tonti-Filippini J, Nery JR, Lee L, Ye Z, Ngo Q-M et al. 2009. Human DNA methylomes at base resolution show widespread epigenomic differences. Nature 462: 315322.
  • Loureiro J, Rodriguez E, Dolezel J, Santos C. 2007. Two new nuclear isolation buffers for plant DNA flow cytometry: a test with 37 species. Annals of Botany 100: 875888.
  • Mathieu O, Yukawa Y, Sugiura M, Picard G, Tourmente S. 2002. 5S rRNA genes expression is not inhibited by DNA methylation in Arabidopsis. Plant Journal: For Cell and Molecular Biology 29: 313323.
  • Miura A, Nakamura M, Inagaki S, Kobayashi A, Saze H, Kakutani T. 2009. An Arabidopsis jmjC domain protein protects transcribed genes from DNA methylation at CHG sites. The EMBO Journal 28: 10781086.
  • Morse AM, Peterson DG, Islam-Faridi MN, Smith KE, Magbanua Z, Garcia SA, Kubisiak TL, Amerson HV, Carlson JE, Nelson CD et al. 2009. Evolution of genome size and complexity in Pinus. PLoS ONE 4: e4332.
  • Pecinka A, Dinh HQ, Baubec T, Rosa M, Lettner N, Mittelsten Scheid O. 2010. Epigenetic regulation of repetitive elements is attenuated by prolonged heat stress in Arabidopsis. The Plant Cell 22: 31183129.
  • Penterman J, Zilberman D, Huh JH, Ballinger T, Henikoff S, Fischer RL. 2007. DNA demethylation in the Arabidopsis genome. Proceedings of the National Academy of Sciences, USA 104: 67526757.
  • Qian J, Lavker RM, Tseng H. 2006. Mapping ribosomal RNA transcription activity in the mouse eye. Developmental Dynamics: An Official Publication of the American Association of Anatomists 235: 19841993.
  • Raj S, Brautigam K, Hamanishi ET, Wilkins O, Thomas BR, Schroeder W, Mansfield SD, Plant AL, Campbell MM. 2011. Clone history shapes Populus drought responses. Proceedings of the National Academy of Sciences, USA 108: 1252112526.
  • Roudier F, Ahmed I, Bérard C, Sarazin A, Mary-Huard T, Cortijo S, Bouyer D, Caillieux E, Duvernois-Berthet E, Al-Shikhley L et al. 2011. Integrative epigenomic mapping defines four main chromatin states in Arabidopsis. The EMBO Journal 30: 19281938.
  • Sati S, Tanwar VS, Kumar KA, Patowary A, Jain V, Ghosh S, Ahmad S, Singh M, Reddy SU, Chandak GR et al. 2012. High resolution methylome map of rat indicates role of intragenic DNA methylation in identification of coding region. PLoS ONE 7: e31621.
  • Schnable PS, Ware D, Fulton RS, Stein JC, Wei F, Pasternak S, Liang C, Zhang J, Fulton L, Graves TA et al. 2009. The B73 Maize genome: complexity, diversity, and dynamics. Science 326: 11121115.
  • Sjödin A, Street NR, Sandberg G, Gustafsson P, Jansson S. 2009. The Populus genome integrative explorer (PopGenIE): a new resource for exploring the Populus genome. New Phytologist 182: 10131025.
  • Steinmuller K, Apel K. 1986. A simple and efficient procedure for isolating plant chromatin which is suitable for studies of DNase I-sensitive domains and hypersensitive sites. Plant Molecular Biology 7: 8794.
  • Street N, Sjödin A, Bylesjö M, Gustafsson P, Trygg J, Jansson S. 2008. A cross-species transcriptomics approach to identify genes involved in leaf development. BMC Genomics 9: 589.
  • Suzuki MM, Bird A. 2008. DNA methylation landscapes: provocative insights from epigenomics. Nature Reviews Genetics 9: 465476.
  • Teixeira FK, Colot V. 2010. Repeat elements and the Arabidopsis DNA methylation landscape. Heredity 105: 1423.
  • Trap-Gentil M-V, Hébrard C, Lafon-Placette C, Delaunay A, Hagège D, Joseph C, Brignolas F, Lefebvre M, Barnes S, Maury S. 2011. Time course and amplitude of DNA methylation in the shoot apical meristem are critical points for bolting induction in sugar beet and bolting tolerance between genotypes. Journal of Experimental Botany 62: 25852597.
  • Tucker S, Vitins A, Pikaard CS. 2010. Nucleolar dominance and ribosomal RNA gene silencing. Current Opinion in Cell Biology 22: 351356.
  • Tuskan GA, DiFazio S, Jansson S, Bohlmann J, Grigoriev I, Hellsten U, Putnam N, Ralph S, Rombauts S, Salamov A et al. 2006. The genome of black cottonwood, Populus trichocarpa (Torr. & Gray). Science 313: 15961604.
  • Vining KJ, Pomraning KR, Wilhelm LJ, Priest HD, Pellegrini M, Mockler TC, Freitag M, Strauss S. 2012. Dynamic DNA cytosine methylation in the Populus trichocarpa genome: tissue-level variation and relationship to gene expression. BMC Genomics 13: 27.
  • Weng Y-I, Huang TH-M, Yan PS. 2009. Methylated DNA immunoprecipitation and microarray-based analysis: detection of DNA methylation in breast cancer cell lines. Methods in Molecular Biology (Clifton, N.J.) 590: 165176.
  • Woo HR, Dittmer TA, Richards EJ. 2008. Three SRA-domain methylcytosine-binding proteins cooperate to maintain global CpG methylation and epigenetic silencing in Arabidopsis (T Kakutani, Ed.). PLoS Genetics 4: e1000156.
  • Zemach A, McDaniel IE, Silva P, Zilberman D. 2010. Genome-wide evolutionary analysis of eukaryotic DNA methylation. Science 328: 916919.
  • Zhang X, Shiu S, Cal A, Borevitz JO. 2008. Global analysis of genetic, epigenetic and transcriptional polymorphisms in Arabidopsis thaliana using whole genome tiling arrays (GP Copenhaver, Ed.). PLoS Genetics 4: e1000032.
  • Zhang W, Wu Y, Schnable JC, Zeng Z, Freeling M, Crawford GE, Jiang J. 2012a. High-resolution mapping of open chromatin in the rice genome. Genome research 22: 151162.
  • Zhang W, Zhang T, Wu Y, Jiang J. 2012b. Genome-wide identification of regulatory DNA elements and protein-binding footprints using signatures of open chromatin in Arabidopsis. The Plant Cell 24: 27192731.
  • Zhang X, Yazaki J, Sundaresan A, Cokus S, Chan SW-L, Chen H, Henderson IR, Shinn P, Pellegrini M, Jacobsen SE et al. 2006. Genome-wide high-resolution mapping and functional analysis of DNA methylation in Arabidopsis. Cell 126: 11891201.
  • Zilberman D, Gehring M, Tran RK, Ballinger T, Henikoff S. 2007. Genome-wide analysis of Arabidopsis thaliana DNA methylation uncovers an interdependence between methylation and transcription. Nature Genetics 39: 6169.

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.

FilenameFormatSizeDescription
nph12026-sup-0001-FiguresS1-S10-TableS1-S4-S5.pdfapplication/PDF1365K

Fig. S1 DNA methylation of a transposable element and a highly redundant gene in the DNase I-MeDIP-SEQ fraction.

Fig. S2 Control PCR for sequence enrichment of the DNase I-MeDIP-SEQ fraction.

Fig. S3 Repeat content of the DNase I-MeDIP-SEQ fraction.

Fig. S4 Enrichment of long genes in the DNase I-MeDIP-SEQ fraction.

Fig. S5 Digital northern heat map representation of genes whose introns are in the DNase I-MeDIP-SEQ fraction.

Fig. S6 General view of enriched GO molecular functions in genes whose (a) upstream or (b) downstream region or (c) body is contained in the DNase I-MeDIP-SEQ fraction.

Fig. S7 Enriched GO molecular functions in genes whose introns are contained in the DNase I-MeDIP-SEQ fraction.

Fig. S8 Enriched GO molecular functions in genes whose exons are contained in the DNase I-MeDIP-SEQ fraction.

Fig. S9 Enriched GO molecular functions in genes whose 5′ region is contained in the DNase I-MeDIP-SEQ fraction.

Fig. S10 Enriched GO molecular functions in genes whose 3′ region is contained in the DNase I-MeDIP-SEQ fraction.

Table S1 Efficiency estimation of six chromatin extraction protocols

Table S4 List of genes from the DNase I-MeDIP-SEQ fraction whose expression was studied by RT-PCR

Table S5 Criteria chosen for redundancy features of genes

nph12026-sup-0002-TableS2.txtplain text document5960KTable S2 Coordinates of MeDIP reads mapped on v2.0 P. trichocarpa genome
nph12026-sup-0003-TableS3.xlsapplication/msexcel27KTable S3 List of primers used for semi-quantitative PCR or bisulfite sequencing