DNA methylation in mice is influenced by genetics as well as sex and life experience

Abstract

DNA methylation is an essential epigenetic process in mammals, intimately involved in gene regulation. Here we address the extent to which genetics, sex, and pregnancy influence genomic DNA methylation by intercrossing 2 inbred mouse strains, C57BL/6N and C3H/HeN, and analyzing DNA methylation in parents and offspring using whole-genome bisulfite sequencing. Differential methylation across genotype is detected at thousands of loci and is preserved on parental alleles in offspring. In comparison of autosomal DNA methylation patterns across sex, hundreds of differentially methylated regions are detected. Comparison of animals with different histories of pregnancy within our study reveals a CpG methylation pattern that is restricted to female animals that had borne offspring. Collectively, our results demonstrate the stability of CpG methylation across generations, clarify the interplay of epigenetics with genetics and sex, and suggest that CpG methylation may serve as an epigenetic record of life events in somatic tissues at loci whose expression is linked to the relevant biology.

Fig. 1

Variability in methylation observed across animals. a. Standard deviation of methylation level among biological replicates. CpG sites are binned according to site depth, where all replicates (three males and three females) must have depth within the specified range (e.g., 1–10 and 11-20). Totally, 20,000 CpG sites per replicate group (genotype) per depth bin were randomly selected for the boxplot. b Standard deviation of methylation level among all animals shown as a cumulative distribution function. Per-site calculations include only animals with depth at least 10; CpG sites are excluded if most animals lack sufficient depth (at least 18 of the 24 study animals must have depth > 10 at the CpG site). The top fifth percentile (most variable sites) is indicated by a dashed gray line. c Distribution of selected sites in 500 nt tiled windows (left), methylation level of selected sites (middle), and overlap of selected sites with CpG islands (right). Variable refer to sites among top 5% of standard deviation among all animals as in panel B (N = 871,166); random refer to another 5% of sites (N = 871,166) meeting the same depth criteria. Color scale in left panel indicates the number of CpG sites per 500nt tiled window. In the box-and-whisker plots, the box depicts the 25th to 75th percentiles, the black dot is the median, the whiskers extend to data points up to 1.5*IQR beyond the box, and open gray circles are data points outside the whisker range

Fig. 2

DMRs between B6 (n = 6) and C3 (n = 6) parental animals, identified by DSS. a Genome browser view of example B6 > C3 DMR (red) and C3 > B6 DMR (green). b Distribution of DMR size, N = 6380. c Distribution of distance between DMR and nearest TSS (based on RefSeq gene models as of Feb 8, 2016), N = 6380. d Distribution of methylation level in DSS DMRs, by strain and DMR polarity. N = 3166 B6 > C3 DMRs and 3214 C3 > B6 DMRs. The box depicts the 25th to 75th percentiles (color by genotype where B6 is red and C3 is green), the black dot is the median, the whiskers extend to data points up to 1.5*IQR beyond the box, and open gray circles are data points outside the whisker range. e Distribution of CpG density at DSS DMRs (N = 6380), compared to size-matched random genomic regions (N = 6380). f Overlap of DSS DMRs (N = 6380) with DNase hypersensitive sites (UCSC Genome Browser) and H3K4me1 enriched regions (ENCODE peaks ENCFF001XXZ), compared to the average overlap observed for 1000 iterations of size-matched random genomic regions

Fig. 3

DNA methylation patterns at DSS genotype–DMRs are recapitulated on parental alleles in F1 progeny. a Heatmap view of weighted methylation scores per B6-vs.-C3 DSS DMR (N = 6380) per animal. Hierarchical clustering of animals performed by R package “amap” (hcluster with method = euclidean and link = average). DMRs were split by direction (B6 > C3: red bar, or C3 > B6: green bar) then sorted by average methylation score over all 24 animals. b Distribution of average weighted methylation scores, with DMRs (N = 6380) split into quartiles according to decreasing methylation score of the hypermethylated parental genome. The comparison between B6 and F1 or between C3 and F1 are significant at p < 1e−80 (Mann–Whitney) in all quartiles. c Methylation in F1 progeny at the read level for an exemplar DSS DMR (B6 > C3). Each box represents the collection of read fragments (for a given F1 genotype) that could be unambiguously assigned as originating from either the B6 parent (left side) or the C3 parent (right side) based on the presence of a diagnostic SNV. Each row within a box and column represents one sequenced fragment, with each CpG site indicated by a circle colored according to the following scheme: yellow = unmethylated cytosine, blue = methylated cytosine, red = noncytosine, and empty/gray = no mapped base at given CpG site in given fragment. d Number of SNVs local to DMRs (N = 6380), compared to size-matched random genomic regions (N = 6380). p < 1e−300 (Mann–Whitney test). e Distribution of distance to the nearest SNV for DMRs (N = 6380) or size-matched random genomic regions (N = 6380). Distance = 0 indicates that at least one SNV is found within a region of a given type. In the box-and-whisker plots, the box depicts the 25th to 75th percentiles, the black dot is the median, the whiskers extend to data points up to 1.5*IQR beyond the box, and open gray circles are data points outside the whisker range

Fig. 4

SNVs at DMRs affect DNA-binding activity of a tissue-specific transcription factor. a Top 10 enriched TF motifs in B6 > C3 DSS DMRs (after expanding DMR size to a minimum of 401 bp) according to HOMER v4.9.1. b Top 10 enriched TF motifs in C3 > B6 DSS DMRs (after expanding DMR size to a minimum of 401 bp) according to HOMER v4.9.1 c Protein purification of FOXA1 DNA-binding domain (DBD). Purified FOXA1 DBD was analyzed by SDS-PAGE. The black arrow indicates FOXA1 DBD. d FOXA1 ChIP-seq signal at SNV-containing FOXA1 motifs. Occurrences of FOXA1 motifs (as defined by HOMER motifs FOXA1.LNCAP or FOXA1.MCF7) within 100 nt of a DMR were categorized by whether the B6 or C3 allele was preferred according to the HOMER position weight matrix [PWM]. FOXA1 signal in the B6 and C3 animals is reported as input-subtracted ChIP-seq read count at 100 nt regions centered on the motif occurrences. The box depicts the 25th to 75th percentiles, the black dot is the median, the whiskers extend to data points up to 1.5*IQR beyond the box, and open gray circles are data points outside the whisker range. Reported p values are from Mann–Whitney test. e, f DNA-binding analyses of FOXA1 DBD with SNV-containing DNAs. The genomic positions and DNA substrates are indicated at the top of each panel with a representative FOXA1 motif. The position of the SNV is highlighted with light blue boxes. The band intensities were calculated using ImageJ software. Totally, 20 bp double-stranded DNAs were incubated with the FOXA1 DBD. The concentration of FOXA1 DBD was as follows: 0 μM, lanes 1,6; 0.15 μM lanes 2,7; 0.3 μM, lanes 3,8; 0.6 μM, lanes 4,9; 1.2 μM, lanes 5,10. The protein–DNA complex was separated by native-polyacrylamide gel electrophoresis. Browser tracks of ChIP-seq data for each locus is shown below the DNA-binding data

Fig. 5

DMRs between male (n = 6) and female (n = 6) parental animals, identified by DSS. a Genome browser view of example M > F DMR (blue, top) and F > M DMR (pink, bottom). b Distribution of DMR size, N = 1575. c Distribution of distance between DMR and nearest TSS (based on RefSeq gene models as of Feb 8, 2016), N = 1575. d Distribution of methylation level in DSS DMRs, by strain and DMR polarity. N = 1041 F > M DMRs and 534 M > F DMRs. The box depicts the 25th to 75th percentiles (color by sex where male is blue and female is pink), the black dot is the median, the whiskers extend to data points up to 1.5*IQR beyond the box, and open gray circles are data points outside the whisker range. e Distribution of CpG density at DSS DMRs (N = 1575), compared to size-matched random genomic regions (N = 1575). f Overlap of DSS DMRs (N = 1575) with DNase hypersensitive sites (UCSC Genome Browser) and H3K4me1 enriched regions (ENCODE peaks ENCFF001XXZ), compared to the average overlap observed for 1000 iterations of size-matched random genomic regions

Fig. 6

Recapitulation of DNA methylation patterns at DSS sex-DMRs in F1 progeny varies by DMR polarity. a Heatmap view of weighted methylation scores per male-vs.-female DSS DMR (N = 1575) per animal. Hierarchical clustering of animals performed by R package “amap” (hcluster with method = euclidean and link = average). DMRs were split by direction (M > F: blue bar, or F > M: pink bar) then sorted by average methylation score over all 24 animals. b Distribution of average weighted methylation scores, with DMRs split into quartiles according to decreasing methylation score of the hypermethylated parental sex. Animals are grouped according to both sex (M, F) and genotype group (P = parental B6 or C3, F1 = offspring B6C3F1 or C3B6F1). c Methylation in F1 progeny at the read level for an exemplar DSS DMR, organized as described for Fig3C. d Number of SNVs local to DMRs (N = 1575), compared to size-matched random genomic regions (N = 1575). e Distribution of distance to the nearest SNV for DMRs (N = 1575) or size-matched random genomic regions (N = 1575). Distance = 0 indicates that at least one SNV is found within a region of a given type. In the box-and-whisker plots, the box depicts the 25th to 75th percentiles, the black dot is the median, the whiskers extend to data points up to 1.5*IQR beyond the box, and open gray circles are data points outside the whisker range

Fig. 7

DMRs hypomethylated in females suggest link to pregnancy and lactation. a Top ten biological processed reported by GREAT v3.0.0 for DSS male > female DMRs. Bold text indicates significance scores. b Genome browser view of prolactin receptor gene locus, with zoomed panels showing CpG methylation level in parental males, parental females, and F1 females at exemplar male > female DMRs. c Top five enriched TF motifs in male > female DSS DMRs (after expanding DMR size to a minimum of 401 bp) according to HOMER v4.9.1. d Top five enriched TF motifs in female > male Metilene DSS (after expanding DMR size to a minimum of 401 bp) according to HOMER v4.9.1