Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
, 10, 25
eCollection

Transcription and Chromatin Determinants of De Novo DNA Methylation Timing in Oocytes

Affiliations

Transcription and Chromatin Determinants of De Novo DNA Methylation Timing in Oocytes

Lenka Gahurova et al. Epigenetics Chromatin.

Abstract

Background: Gametogenesis in mammals entails profound re-patterning of the epigenome. In the female germline, DNA methylation is acquired late in oogenesis from an essentially unmethylated baseline and is established largely as a consequence of transcription events. Molecular and functional studies have shown that imprinted genes become methylated at different times during oocyte growth; however, little is known about the kinetics of methylation gain genome wide and the reasons for asynchrony in methylation at imprinted loci.

Results: Given the predominant role of transcription, we sought to investigate whether transcription timing is rate limiting for de novo methylation and determines the asynchrony of methylation events. Therefore, we generated genome-wide methylation and transcriptome maps of size-selected, growing oocytes to capture the onset and progression of methylation. We find that most sequence elements, including most classes of transposable elements, acquire methylation at similar rates overall. However, methylation of CpG islands (CGIs) is delayed compared with the genome average and there are reproducible differences amongst CGIs in onset of methylation. Although more highly transcribed genes acquire methylation earlier, the major transitions in the oocyte transcriptome occur well before the de novo methylation phase, indicating that transcription is generally not rate limiting in conferring permissiveness to DNA methylation. Instead, CGI methylation timing negatively correlates with enrichment for histone 3 lysine 4 (H3K4) methylation and dependence on the H3K4 demethylases KDM1A and KDM1B, implicating chromatin remodelling as a major determinant of methylation timing. We also identified differential enrichment of transcription factor binding motifs in CGIs acquiring methylation early or late in oocyte growth. By combining these parameters into multiple regression models, we were able to account for about a fifth of the variation in methylation timing of CGIs. Finally, we show that establishment of non-CpG methylation, which is prevalent in fully grown oocytes, and methylation over non-transcribed regions, are later events in oogenesis.

Conclusions: These results do not support a major role for transcriptional transitions in the time of onset of DNA methylation in the oocyte, but suggest a model in which sequences least dependent on chromatin remodelling are the earliest to become permissive for methylation.

Keywords: DNA methylation; Histone modifications; Imprinting; Oocytes; Transcription.

Figures

Fig. 1
Fig. 1
Rates of de novo DNA methylation of different sequence features in growing oocytes. a Screenshot of a 2.3-Mb interval of chromosome 1 depicting methylation in NGOs, 60–65 µm and GV oocytes. Vertical bars represent mean methylation of 2-kb windows, with 1-kb steps, height and colour denoting % methylation. The horizontal lines are set at 50% methylation, with higher levels of methylation above the line and lower levels below the line and shaded according to the colour scale on the left. The 60–65 µm data are from PBAT from the current manuscript; NGO and GV data are from [5, 11]. b Violin plots showing distribution of CpG methylation values in all hypermethylated domains, transcribed hypermethylated domains (≥90% of the length of the domain covered by transcript, domains ≥5 kb), transcriptionally silent hypermethylated domains (≤10% of the length of the domain covered by transcript, domains ≥5 kb), CGIs, LINE L1s and SINE-B2s in NGO, 60–65 µm and GV oocytes. Shape of the violin plot represents Kernel density estimation, i.e. probability density of the data at the different values. White dots correspond to the median, yellow dots to the average, bold lines the interquartile range and thin lines adjacent values, i.e. minimum and maximum values within the ×1.5 interquartile range from the first and third quartile, respectively. c Box whisker plots reporting CpG density and GC content of 2-kb genomic regions that are fully methylated in GV oocytes (≥75% DNA methylation) categorised according to their % DNA methylation in 60–65 µm oocytes (x axis). Boxes, interquartile range, with bar as median and whiskers as ×1.5 interquartile range, outliers not shown. Between 3619 and 30320 2-kb intervals were analysed in each methylation category. d Violin plots showing methylation levels of Cs in CpG, CHG and CHH contexts in NGOs and 60–65 µm oocytes of Cs that are fully methylated (≥75%) in GV oocytes
Fig. 2
Fig. 2
Properties of ongoing de novo DNA methylation in growing oocytes. a Average proportion of neighbouring CpG pairs where both CpGs are methylated (M–M fraction) by distance of CpG pairs in NGO, 60–65 µm and GV oocytes. The value of the M–M fraction was quantified for each possible distance between two neighbouring CpGs on the same sequencing read using formula M–M pairs/(M–M + M–U pairs), where M–U pairs represent CpG pairs where upstream CpG is methylated and downstream unmethylated. Only reads mapping to chromosome 1 were analysed. The horizontal lines represent the genomic average methylation level of each stage. b The distant-dependent correlation of methylation between CpG pairs in 60–65 µm oocytes, compared with random-shuffled data
Fig. 3
Fig. 3
CpG islands gain DNA methylation at different rates in growing oocytes. a Barchart of CGI methylation in the oocyte size populations from the RRBS and PBAT datasets. The number of CGIs covered in each dataset is given in Additional file 1: Table S1. b Methylation of gDMRs in RRBS datasets, displaying the basal level in 40–45 µm oocytes, and the increases in methylation to the subsequent size populations. gDMRs are ordered according to their methylation level in 60–65 µm oocytes, which is comparable with PBAT data (see Additional file 3: Fig. S2A). c Validation of CGI methylation in different oocyte size populations. Heatmap shows methylation progression at CGIs that become methylated between 40 and 45 µm and MII oocytes (data from published GV and MII RRBS datasets). Five early-methylating CGIs and five late-methylating CGIs were selected, and their methylation in 50–55 µm oocytes (early-methylating CGIs) or both 50–55 and 60–65 µm oocytes was confirmed by locus-specific bisulphite sequencing. White dots represent unmethylated CpGs and black dots methylated CpGs
Fig. 4
Fig. 4
Transcription dynamics in growing oocytes. a Barchart showing time of first detection of genes in growing oocytes, according to classification as reference gene, from canonical TSS (w/o upTSS) or novel upstream TSS (upTSS), or novel multi- or monoexonic gene. The total numbers of genes classified as expressed in each RNA-seq datasets are given in Additional file 6: Table S5. b Browser screenshots of representative early-methylating (Igf2r, Zac1) and late-methylating (Cdh15, Nnat) gDMRs in relation to RNA-seq data from different stages of the oocyte growth and DNA methylation acquired in 60–65 µm oocytes. In the RNA annotation track, red gene structures are transcribed from left to right and blue gene structures from right to left, with arrows showing the most upstream TSSs and direction of transcription. RNA-seq data show that transcriptional pattern is established prior to DNA methylation establishment
Fig. 5
Fig. 5
Gene body and CpG island methylation kinetics in relation to transcription. a Cumulative distribution plot of methylation level of reference and novel genes (≥4 kb in length and ≥2 FPKM) in 60–65 µm oocytes (PBAT dataset). The numbers of reference and novel genes satisfying the criteria for analysis were 105 and 32, respectively. b Box whisker plots of methylation of gene bodies of reference (1396) and novel (373) genes in relation to expression level in 60–65 µm oocytes. c Box whisker plot showing methylation level of CGIs in 60–65 µm oocytes grouped according to the stage in oocyte growth that expression of overlapping gene attained the threshold of >1 FPKM in the RNA-seq datasets. d Box whisker plot showing the corresponding data from expression level in 60–65 µm oocytes. The numbers of genes in (c) and (d) are: 1013 for 10–30 µm oocytes, 76 for 40–45 µm, 70 for 50–55 µm, 47 for 60–65 µm, 57 for GV and never 289. e Methylation level of intragenic CGIs (CGIs fully methylated in GV oocytes) in relation to expression level of the corresponding gene in 60–65 µm oocytes. The numbers of CGIs analysed in each methylation category (from lowest to highest) are: 269, 210, 281 and 50. f Methylation in 60–65 µm oocytes of CGIs (CGIs fully methylated in GV oocytes) according to prior activity as TSS as determined in e18.5 oocytes: 112 TSS-CGIs and 1229 non-TSS-CGIs. Asterisks denote p values of Student’s t test: *0.01–0.001, **0.001–0.0001, ***<0.0001
Fig. 6
Fig. 6
Motifs differentially enriched in early- and late-methylating CpG islands. a Summary of results of DREME analysis identifying motifs differentially enriched in CGIs that become fully methylated in GV oocytes grouped according to methylation level in 60–65 µm oocytes. Codes 0–25, 25–50, 50–75 and 75–100 represent CGIs with corresponding percentage methylation in 60–65 µm oocytes. The numbers of CGIs in each category are 470, 329, 384 and 63, respectively. b DREME motifs significantly enriched in CGIs methylated ≤25% in oocytes compared with ≥75% methylated CGIs that correspond to binding site motifs for known TFs. In motif sequence, B = C/G/T and M = C/A. P value and E values are as defined by DREME and binding sites as identified by Tomtom
Fig. 7
Fig. 7
CpG island methylation kinetics in relation to chromatin parameters. a Box whisker plots showing enrichment (log-transformed corrected read count) of H3K4me2, H3K4me3 and H3K36me3 at CGIs in relation to DNA methylation in 60–65 µm oocytes (PBAT data). The ChIP-seq data shown are from p10 oocytes; similar trends were observed in ChIP-seq data from e18.5 oocytes. Pearson’s correlation coefficients are: −0.293 for H3K4me2, −0.173 for H3K4me3, 0.240 for H3K36me3. The numbers of CGIs analysed in each methylation category (from lowest to highest) were: 464, 327, 382 and 63. b Box whisker plots showing the degree of DNA methylation change at CGIs in Kdm1a- and Kdm1b-null MII oocytes in relation to methylation in 60–65 µm oocytes. Pearson’s correlation coefficients are: −0.296 for Kdm1a and −0.357 for Kdm1b. The numbers of CGIs analysed in each methylation category (from lowest to highest) were: 244, 185, 255 and 28 for Kdm1a, and 270, 199, 268 and 31 for Kdm1b. c Browser screenshots of a representative early-methylating and late-methylating CGI (84.2 and 12.2% methylation in 60–65 µm oocytes, respectively) in relation to p10 H3K4me2 enrichment and DNA methylation attained in wild-type (WT) or Kdm1b-null MII oocytes
Fig. 8
Fig. 8
Modelling factors determining rate of CpG island methylation. Lasso regression model plot showing the effect of nine independent variables on variability of CGI methylation in 60–65 µm oocytes. Each line represents one of the variables. The earlier the line deviates from the horizontal line with coefficient 0.0, the more the corresponding variable contributes to the variability of the response variable, and the steeper the slope the greater the effect. If the steepness of the slope of one variable already in the model changes when a new variable comes into the model, it is a sign of correlation between two independent variables

Similar articles

See all similar articles

Cited by 13 PubMed Central articles

See all "Cited by" articles

References

    1. Bourc’his D, Xu G-L, Lin C-S, Bollman B, Bestor TH. Dnmt3L and the establishment of maternal genomic imprints. Science. 2001;294(5551):2536–2539. doi: 10.1126/science.1065848. - DOI - PubMed
    1. Branco MR, King M, Perez-Garcia V, Bogutz AB, Caley M, Fineberg E, Lefebvre L, Cook SJ, Dean W, Hemberger M, Reik W. Maternal DNA methylation regulates early trophoblast development. Dev Cell. 2016;36(2):152–163. doi: 10.1016/j.devcel.2015.12.027. - DOI - PMC - PubMed
    1. Kaneda M, Okano M, Hata K, Sado T, Tsujimoto N, Li E, Sasaki H. Essential role for de novo DNA methyltransferase Dnmt3a in paternal and maternal imprinting. Nature. 2004;429(6994):900–903. doi: 10.1038/nature02633. - DOI - PubMed
    1. Seisenberger S, Andrews S, Krueger F, Arand J, Walter J, Santos F, Popp C, Thienpont B, Dean W, Reik W. The dynamics of genome-wide DNA methylation reprogramming in mouse primordial germ cells. Mol Cell. 2012;48(6):849–862. doi: 10.1016/j.molcel.2012.11.001. - DOI - PMC - PubMed
    1. Kobayashi H, Sakurai T, Imai M, Takahashi N, Fukuda A, Yayoi O, Sato S, Nakabayashi K, Hata K, Sotomaru Y, Suzuki Y, Kono T. Contribution of intragenic DNA methylation in mouse gametic DNA methylomes to establish oocyte-specific heritable marks. PLoS Genet. 2012;8(1):e1002440. doi: 10.1371/journal.pgen.1002440. - DOI - PMC - PubMed

Publication types

LinkOut - more resources

Feedback