Conservation and divergence of canonical and non-canonical imprinting in murids

Abstract

Background: Genomic imprinting affects gene expression in a parent-of-origin manner and has a profound impact on complex traits including growth and behavior. While the rat is widely used to model human pathophysiology, few imprinted genes have been identified in this murid. To systematically identify imprinted genes and genomic imprints in the rat, we use low input methods for genome-wide analyses of gene expression and DNA methylation to profile embryonic and extraembryonic tissues at allele-specific resolution.

Results: We identify 14 and 26 imprinted genes in these tissues, respectively, with 10 of these genes imprinted in both tissues. Comparative analyses with mouse reveal that orthologous imprinted gene expression and associated canonical DNA methylation imprints are conserved in the embryo proper of the Muridae family. However, only 3 paternally expressed imprinted genes are conserved in the extraembryonic tissue of murids, all of which are associated with non-canonical H3K27me3 imprints. The discovery of 8 novel non-canonical imprinted genes unique to the rat is consistent with more rapid evolution of extraembryonic imprinting. Meta-analysis of novel imprinted genes reveals multiple mechanisms by which species-specific imprinted expression may be established, including H3K27me3 deposition in the oocyte, the appearance of ZFP57 binding motifs, and the insertion of endogenous retroviral promoters.

Conclusions: In summary, we provide an expanded list of imprinted loci in the rat, reveal the extent of conservation of imprinted gene expression, and identify potential mechanisms responsible for the evolution of species-specific imprinting.

Keywords: Allele-specific; DNA methylation; F1 hybrid; Genomic imprinting; H3K27me3; Human; Mouse; Non-canonical imprinting; Rat.

Fig. 1

Imprinted gene expression in rat embryonic and extraembryonic cells. a Experimental design. Two distinct reciprocal crosses of rat strains (BN/CrlCrlj, “B” and WKY/NCrlCrlj, “W” and F344/NSlc, “F”) were conducted and cells from the E8.5 epiblast (Epi) and the ectoplacental cone (EPC) were collected. RNAseq was performed on all samples (in duplicate or triplicate), and WGBS was performed in duplicate on BW/WB matings. The maternal (red) strain is listed first in cross names. b Scatterplot of paternal expression ratios in rat EPCs. The paternal expression ratio was averaged over 11 samples, and expressed transcripts (RPKM ≥1) with sufficient allelic coverage (RPM ≥0.5) in at least 6 samples are shown (n=16,642). Transcripts showing parent-of-origin imprinted gene expression (Student’s t test, Bonferroni-adjusted p-val <0.05) are colored red (maternally expressed) or blue (paternally expressed). c Rat genome browser screenshots of the maternally expressed imprinted gene H19 and paternally expressed imprinted gene Igf2. For each cross, duplicates or triplicates were merged and the mean expression level is displayed in reads per million (RPM). A subset of all read alignments (gray) are highlighted if they originated from a maternal (red) or paternal (blue) alleles. The genomic position of known Refseq genes and CpG islands (CGIs) are included. d Ideogram karyotype summary of imprinted gene expression identified using a combination of T-test and Limma in rat Epi and EPC cells. Genes that show imprinted gene expression in human or mouse are indicated with an asterisk and tilde, respectively. LOC103691708, Itga1, and genes showing maternal imprinted expression uniquely in EPC cells and normally expressed in adult rat blood (RPKM ≥1) are not shown due to space limitations (see Additional file 2: Table S1 for a full list of imprinted genes)

Fig. 2

Conservation of genomic imprinting in rat and mouse. a Heatmap of parental expression ratios in rat and mouse Epi and EPC cells. Genes previously identified as imprinted in human are indicated with an asterisk. b Heatmap of allele-specific DNAme levels over DMRs associated with imprinted genes in a. The relative position of DMRs is included. DMRs that also show parent-specific methylation in human are indicated with an asterisk. c–d Rat and mouse genome browser screenshots of the Kcnq1 locus. Figure legend as in Fig. 1c, with the addition of DNA methylation information. The average DNAme level over each CpG is shown as individual bar plots, and CpGs covered by at least 1 allele-specific read (Epi, EPC) or 5 reads (gametes) are shown. Pink indicates methylation of both alleles. Rat DMRs are included. In rat, a paternally expressed unannotated antisense ncRNA is expressed from the gene body of Kcnq1. The position of the putative Kcnq1ot1 imprinted CGI promoter is indicated by a dashed box. d The mouse Kcnq1 locus is shown for comparison

Fig. 3

De novo genomic imprinting in rat. a,b Heatmaps of parental expression ratios and allele-specific DNAme levels in rat and mouse Epi and EPC cells as in Fig. 2a,b. Imprinted genes in rat that are not imprinted in mouse are shown. c,d Rat and mouse genome browser screenshots of the Zfp64 and LOC108350526 loci, two rat-specific imprinted genes. Browser tracks are as shown in Fig. 2c,d. The rat-specific transcript LOC108350526 and syntenic mouse region is highlighted in green. The location of LTR retrotransposons (RepeatMasker) is included

Fig. 4

Species-specific H3K27me3 in oocytes is associated with species-specific non-canonical imprinting Zfp64 and Zfp516, and transient imprinting of Bbx. a 2D scatterplot showing genome-wide H3K36me3 and H3K27me3 levels over 1-kb bins in rat oocytes. Datapoints are colored by average oocyte DNAme levels. A random subset of 10,000 bins is shown. The density of all data points (n=1,164,268) is summarized by a contour plot. b Cross-species Spearman correlation values between H3K4me3, H3K27me3, and H3K36me3, and DNAme levels over syntenic 1-kb bins in rat and mouse oocytes (n=823,926). c Scatterplot showing differences in rat and mouse oocyte DNAme, H3K27me3, and H3K36me3 levels over syntenic 1-kb bins. A random set of 10,000 bins are shown, and all bins (n=823,926) are summarized by a contour plot. The number of bins with a delta H3K27me3 >0.5 RPKM and DNAme >75% is indicated by a dashed box. d–f Rat and mouse genome browser screenshots of the Zfp64-LOC108350526, Zfp516, and Bbx loci. Replicates of CUT&RUN data were merged and mean levels are shown as counts per million aligned reads (CPM). Promoters are indicated by a dashed line

Fig. 5

Rat-specific non-canonical imprinted gene Slc38a1. a Rat and mouse genome browser screenshots of the Slc38a1 locus. Slc38a1 is paternally expressed in rat and biallelically expressed in mouse EPCs. Tracks are presented as in Fig. 2d. b Ensembl Region Comparison screenshot of the Slc38a1 locus in rat (rn6), mouse (mm10), and human (hg19). Syntenic regions are shown in green. The locations Refseq genes, LTRs, CGIs (purple), and ZFP57 binding motifs (orange) are shown. Rat EPC DMRs are included. The CGI promoter and intronic DMR are highlighted in yellow and magnified (right panel) for clarity

Fig. 6

X chromosome inactivation (XCI) in rat and mouse. a Rat and mouse genome browser screenshots of the Xist locus. Refseq genes and CGIs are included. CUT&Run (rat) and ChIP-seq (mouse) data are shown in counts per million (CPM). Xist is highlighted in green. b Violin plots showing the distribution of paternal expression ratios of autosomal (top) and X chromosome (bottom) transcripts in rat and mouse epiblast and EPC samples. Replicate 1 of mouse cross “CJ” was omitted due to an XO genotype. Differences in XCI skewing between CJ and JC samples (bottom right) are likely due to differences in genome quality between the parental strains, with reads preferentially aligning to the reference “C” genome