Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2018 Aug 20;9(1):3331.
doi: 10.1038/s41467-018-05841-x.

LTR Retrotransposons Transcribed in Oocytes Drive Species-Specific and Heritable Changes in DNA Methylation

Affiliations
Free PMC article

LTR Retrotransposons Transcribed in Oocytes Drive Species-Specific and Heritable Changes in DNA Methylation

Julie Brind'Amour et al. Nat Commun. .
Free PMC article

Abstract

De novo DNA methylation (DNAme) during mouse oogenesis occurs within transcribed regions enriched for H3K36me3. As many oocyte transcripts originate in long terminal repeats (LTRs), which are heterogeneous even between closely related mammals, we examined whether species-specific LTR-initiated transcription units (LITs) shape the oocyte methylome. Here we identify thousands of syntenic regions in mouse, rat, and human that show divergent DNAme associated with private LITs, many of which initiate in lineage-specific LTR retrotransposons. Furthermore, CpG island (CGI) promoters methylated in mouse and/or rat, but not human oocytes, are embedded within rodent-specific LITs and vice versa. Notably, at a subset of such CGI promoters, DNAme persists on the maternal genome in fertilized and parthenogenetic mouse blastocysts or in human placenta, indicative of species-specific epigenetic inheritance. Polymorphic LITs are also responsible for disparate DNAme at promoter CGIs in distantly related mouse strains, revealing that LITs also promote intra-species divergence in CGI DNAme.

Conflict of interest statement

The authors declare no competing interests.

Figures

Fig. 1
Fig. 1
Syntenic intergenic regions show divergent DNA methylation in rodent and human oocytes. a Evolutionary distance between rat, human, and the two mouse strains used in this study (adapted from http://timetreebeta.igem.temple.edu/). b Density plots depicting the distribution of DNAme in mouse, rat, or human oocytes. The percentage of each genome with low (<30%) or high (>70%) DNAme is indicated in red. DNAme domains were identified using Changepoint analysis. c Proportion of hypermethylated (>70% DNAme) genome-wide 1 kb bins in genic or intergenic regions. Total number of bins with sufficient coverage: mouse: 934,621 genic/1,132,257 intergenic; rat: 794,069 genic/1,195,429 intergenic; human: 1,024,005 genic/1,132,257 intergenic. d Heat map of the correlation between DNAme patterns in oocytes or sperm over 433,111 syntenic genomic regions (1 kb bins with >5× coverage over >5 CpGs in all 3 species). e Venn diagrams showing the overlap in hypermethylated (>70% DNAme) syntenic genomic regions (1 kb bins). Methylated bins in syntenic intergenic regions are more divergent than those overlapping an annotated gene (mm10, rn6, or hg19 Ensembl annotation) in all three species. f Genome browser screenshot of the Pik3c3 locus in mouse oocytes. Note the presence of de novo DNAme and H3K36me3 coincident with the predicted intergenic transcription units. Mouse and human WGBS datasets (described in the data summary table) are from refs. ,,
Fig. 2
Fig. 2
LTR-initiated transcription impacts species-specific DNAme in oocytes. a Bar chart of the relative contribution of LTR retrotransposon classes driving all LITs in mouse, rat, and human oocytes. b Screenshot of genic and intergenic LITs resulting in two hypermethylated domains in mouse oocytes: the canonical TSS of the C430002N11Rik gene is embedded in an MTD, and a second intergenic transcript initiates in an MTC. Both LTR promoters are enriched for H3K4me3 while the downstream de novo methylated regions are enriched for H3K36me3 in GVO. ce Heat maps of the correlation between transcription and downstream (+strand) DNAme at 2154 LITs (TSS ±20 kb) in mouse oocytes, 1020 LITs in rat oocytes, and 620 forward strand LITs in human oocytes. NGO non-growing oocytes, GVO germinal vesicle oocytes, MII metaphase II oocytes, 10 wk PGCs 10-week female PGCs (human). f Cartoon illustrating a DNA methylated genomic region that is overlapped by an LIT. g Pie charts showing the proportion of hypermethylated regions (normalized to the domain size) that are overlapped by an LIT and/or an annotated Ensembl gene in mouse, rat, or human oocytes. h Venn diagrams showing syntenic 1 kb bins that are hypermethylated (>70% DNAme) in mouse, rat, and/or human oocytes. Syntenic regions that overlap with an LIT (right) show significantly greater divergence in DNAme between species than the rest of the genome (left). The percentage of total syntenic genomic regions (433, 111 1 kb bins with >5 CpGs with >5× coverage) is indicated above each Venn diagram. Mouse and human WGBS datasets analyzed from refs. ,,,, human RNA-seq data from ref. , and mouse H3K4me3 from ref.
Fig. 3
Fig. 3
LTR-initiated transcription impacts species-specific gene transcription and gene body DNAme in oocytes. a Pearson correlation between mouse, rat, and human oocyte transcript levels over 11,186 syntenic Ensembl annotated genes. aC57BL/6, bCast/Ei, cSprague Dawley, dWistar Han. b Number of annotated Ensembl genes with one or more overlapping LIT(s). s: sense overlap, splicing into a genic exon and >10% contribution to all of the gene’s isoforms, as antisense overlap. c Comparison of gene transcription between mouse and rat GVOs (gray). Highlighted are genes with an LTR-driven isoform that contributes to at least 10% of the gene transcripts in mouse (red), rat (blue), or in both species (black). d Comparison of gene transcription between mouse and human GVOs (gray). Highlighted are genes with an LTR-driven isoform that contributes to at least 10% of the gene transcripts in mouse (red), human (blue), or in both species (black). e Screenshot of the Bmp5/BMP5 locus. In mouse oocytes, transcription initiates from an intragenic MTA element, and gene body H3K36me3 and DNAme are observed only downstream of the MTA. Note that the ortholog is not transcribed in rat or human oocytes. f Screenshot of the Sirt2/SIRT2 locus. In rat oocytes, transcription initiates in an upstream RLTR31B2 element, and DNAme downstream of this alternative promoter encompasses the canonical Sirt2 promoter. In both mouse and human oocytes, the orthologous gene is transcribed from the canonical (hypomethylated) promoter. g Screenshot of the Zfp90/ZFP90 locus. A human-specific isoform initiates in an LTR12C upstream of the canonical CGI promoter, which is hypermethylated exclusively in human oocytes. Zfp90 is transcribed from the canonical TSS in mouse and rat oocytes. h Screenshot of the Th locus showing an unannotated isoform that initiates in an RMER19A LTR- upstream of the canonical promoter exclusively in mouse and rat oocytes. DNAme extends downstream of the RMER19A TSS in both species, coincident with H3K36me3 in mouse, and encompasses the annotated Th TSS. Mouse and human WGBS datasets analyzed from refs. , and human RNA-seq data from ref.
Fig. 4
Fig. 4
LTR transcription leads to species- and rodent-specific CpG island (CGI) methylation. a Proportion of CGIs (>5 CpGs covered by WGBS) with low (<30%), intermediate (30–70%), or high (>70%) DNAme levels in mouse, rat, and human oocytes. The proportion and number of hypermethylated CGIs embedded within an LIT is depicted in the accompanying bar chart (black). b Identification of syntenic CGIs between two species by calculation of identity. CGIs with an identity >0.5 between two species were included in our analyses. c Proportion of CGIs overlapping with an annotated TSS (±500 bp; promoter CGIs), gene body (intragenic), or intergenic region in human (H), mouse (M), rat (R), mouse + rat (MR), and mouse + rat + human (MRH). Total number of CGIs with >5 CpGs >5× WGBS coverage in each species and subset of CGIs syntenic (identity >0.5) between mouse (M) and rat (R) or all three species are shown. The percentage of CGIs overlapping a TSS is also shown for each. d Venn diagram showing the overlap in DNAme at all syntenic CGIs hypermethylated in in at least one of mouse, rat, or human oocytes (384). e Syntenic promoter CGIs hypermethylated in mouse and/or in rat oocytes (gray boxes). The proportion of meCGIs that overlap with a transcript initiated in an MTA, a non-MTA MT element (MTB, MTC, MTD, or MTE), or a non-MT LTR element is depicted in the adjacent pie charts. f Genome browser screenshots of the Zfp64 CGI promoter. In mouse and rat oocytes, an MTC-driven antisense LIT overlapping the canonical promoter CGI appears to be responsible for DNAme, consistent with the H3K36me3 profile in mouse oocytes. g Screenshot of the Piwil1 locus illustrating transcription and DNAme in mouse, rat, and human oocytes. H3K36me3 in oocytes and DNAme in PG blastocysts and E7.5 embryos are also shown for mouse. A mouse-specific Piwil1 LIT initiates in an MTD element situated upstream of the canonical TSS, and only the mouse CGI promoter is hypermethylated. Mouse and human WGBS datasets analyzed from refs. , and human RNA-seq data from ref.
Fig. 5
Fig. 5
Persistence of LIT-associated oocyte DNAme in mouse and human following fertilization. a Density plot depicting the distribution of DNAme in mouse blastocyst, PG blastocyst, ICM (C57BL/6×DBA F1 cross) and ICM (maternal allele). DNAme domains were identified using Changepoint analysis. b Density plots comparing oocyte DNAme to PG blastocysts over genome-wide 2 kb bins (left) or 2 kb bins overlapping an LIT (right). c. Scatter plots of oocyte vs PG blastocyst (mouse) or blastocyst (human) CGI promoter DNAme. Promoter CGIs overlapped by an LIT in oocytes are highlighted in black. d Scatter plots of oocyte vs E7.5 epiblast CGI promoter DNAme. Promoter CGIs overlapped by an LIT in oocytes are highlighted in black. Sperm DNAme is indicated as a color gradient. Note that the rat Piwil1 gene is oriented in the reverse direction. e DNAme in WT or Tet TKO E7.5 embryos over 24 promoter CGIs that gain DNAme as the result of LTR-initiated transcription in oocytes. Yellow highlight: genes showing apparent retention of maternal DNAme in E7.5 embryos. Black arrow: genes identified as TET targets in Dai et al.). Gray background: LITs present in C57BL/6 and Cast/Ei meCGIs; pink background: C57BL/6 private LITs. Note that the genes E330021D16Rik and 1700014NO6Rik also gain DNAme on the paternal genome in the peri-implantation stage (see d). f Bar chart including 87 CGIs methylated on the maternal allele in human placenta, categorized here via LIONS or manual inspection of de novo transcripts (*), as LIT or non-LIT (other) associated in oocytes. The 16 LIT-associated DMRs are further subcategorized based on annotated 5′ LTR. Bold: DMRs associated with paternal-specific gene transcription in placenta (ref. ). Mouse and human WGBS datasets analyzed from refs. ,,,,
Fig. 6
Fig. 6
LTR polymorphisms lead to strain-specific transcripts and CGI DNAme in mouse oocytes. a Density plot of DNAme over 1,101,575 genome-wide 2 kb bins (>3 CpGs >1× coverage in Cast/Ei GVOs) in C57BL/6 or Cast/Ei GVOs. Regions hypermethylated (>70%) in C57BL/6 and/or Cast/Ei GVOs are highlighted. The mean percentage of DNAme in C57BL/6 parthenogenetic blastocysts (PG Bl.) is indicated as a color gradient. b Proportion of 2 kb bins hypermethylated in C57BL/6 and/or Cast/Ei GVOs overlapping with an LIT driven by an MTA element, another subfamily of MT element (MTB, C, D, or E) or another type of LTR. c Scatter plot of differential DNAme (Δ DNAme) and H3K36me3 enrichment (Δ H3K36me3) over 11,030 promoter CGIs in C57BL/6 and Cast/Ei GVOs. Differentially methylated CGIs are highlighted in green boxes. d Bar chart depicting each type of transcript overlapping differentially methylated promoter CGIs identified in c. MTA, MT, or LTR-initiated transcripts were identified by LIONs and/or manual inspection. None: no transcript or non-LTR initiated transcript. e Screenshot of the Slc36a4 CGI promoter. The Slc36a4 transcript initiates in a polymorphic MTA element (insertion in the Cast/Ei genome) in Cast/Ei GVOs. f Screenshot of the Zfp113 CGI promoter. The Zfp113 transcript initiates in an upstream MTA element in C57BL/6 GVOs but from the canonical TSS in Cast/Ei GVOs. g Heat map of all hypermethylated promoter CGIs (69) embedded within an LIT in C57BL/6 GVOs. DNAme levels in Cast/Ei GVOs, C57BL/6 GVOs, and C57BL/6 PG blastocysts is shown (columns), and CGIs embedded within LITs that are present in both strains or private to C57BL/6 oocytes are clustered (rows). Genes with CGI promoters retaining >45% DNAme in PG blastocysts are labeled. Bold: genes previously identified as TET targets at the implantation stage (see Fig. 5e). C57BL/6 GVO WGBS and ICM datasets from refs. ,

Similar articles

See all similar articles

Cited by 9 articles

See all "Cited by" articles

References

    1. Mouse Genome Sequencing Consortium, et al. Initial sequencing and comparative analysis of the mouse genome. Nature. 2002;420:520–562. doi: 10.1038/nature01262. - DOI - PubMed
    1. Leung DC, Lorincz MC. Silencing of endogenous retroviruses: when and why do histone marks predominate? Trends Biochem. Sci. 2012;37:127–133. doi: 10.1016/j.tibs.2011.11.006. - DOI - PubMed
    1. Faulkner GJ, et al. The regulated retrotransposon transcriptome of mammalian cells. Nat. Genet. 2009;41:563–571. doi: 10.1038/ng.368. - DOI - PubMed
    1. Thompson PJ, Macfarlan TS, Lorincz MC. Long terminal repeats: from parasitic elements to building blocks of the transcriptional regulatory repertoire. Mol. Cell. 2016;62:766–776. doi: 10.1016/j.molcel.2016.03.029. - DOI - PMC - PubMed
    1. Peaston AE, et al. Retrotransposons regulate host genes in mouse oocytes and preimplantation embryos. Dev. Cell. 2004;7:597–606. doi: 10.1016/j.devcel.2004.09.004. - DOI - PubMed

Publication types

MeSH terms

LinkOut - more resources

Feedback