Differential enrichment of H3K9me3 at annotated satellite DNA repeats in human cell lines and during fetal development in mouse

Abstract

Background: Trimethylation of histone H3 on lysine 9 (H3K9me3) at satellite DNA sequences has been primarily studied at (peri)centromeric regions, where its level shows differences associated with various processes such as development and malignant transformation. However, the dynamics of H3K9me3 at distal satellite DNA repeats has not been thoroughly investigated.

Results: We exploit the sets of publicly available data derived from chromatin immunoprecipitation combined with massively parallel DNA sequencing (ChIP-Seq), produced by the The Encyclopedia of DNA Elements (ENCODE) project, to analyze H3K9me3 at assembled satellite DNA repeats in genomes of human cell lines and during mouse fetal development. We show that annotated satellite elements are generally enriched for H3K9me3, but its level in cancer cell lines is on average lower than in normal cell lines. We find 407 satellite DNA instances with differential H3K9me3 enrichment between cancer and normal cells including a large 115-kb cluster of GSATII elements on chromosome 12. Differentially enriched regions are not limited to satellite DNA instances, but instead encompass a wider region of flanking sequences. We found no correlation between the levels of H3K9me3 and noncoding RNA at corresponding satellite DNA loci. The analysis of data derived from multiple tissues identified 864 instances of satellite DNA sequences in the mouse reference genome that are differentially enriched between fetal developmental stages.

Conclusions: Our study reveals significant differences in H3K9me3 level at a subset of satellite repeats between biological states and as such contributes to understanding of the role of satellite DNA repeats in epigenetic regulation during development and carcinogenesis.

Keywords: Cell lines; ChIP-Seq; Development; Epigenetics; H3K9me3; Heterochromatin; Histone marks; Human genome; Mouse genome; Satellite DNA.

Fig. 1

Cancer cells are characterized by reduced H3K9me3 level at satellite elements compared to normal cells. a Two-dimensional PCA plot of cell lines based on H3K9me3 enrichment at autosomal satellite elements. Ellipses define 95% confidence intervals around group mean. b Heatmap (based on scaled log₂-transformed FC values) representing H3K9me3 enrichment of 382 differentially enriched satellite elements on autosomal chromosomes shows clustering of normal versus cancer cells, with an exception of PBMC and Dnd41. Names of normal cell lines are in blue; cancer cells are in red. c Density plot of calculated fold change (ChIP over input) for normal and cancer cell lines at 407 differentially enriched satellite elements (p-value < 0.05; Welch two-sample t-test). Averages are shown by dashed lines. d Density plot showing fraction of zero quality mapping reads at satellite elements with differential enrichment of H3K9me3 between normal and cancer cell lines. Density plots are shown for two replicates of HMEC and A549 cell lines

Fig. 2

Enrichment of H3K9me3 at a cluster of GSATII elements on human chromosome 12 pericentromere. The region coordinates are divided into 500 non-overlapping windows and fold enrichment over input (log₂-transformed) is calculated and plotted for each window. Highlighted in yellow is the region (chr12:34439944-34555157) that shows differential enrichment between normal (titled in blue) and cancer cell lines (titled in red). Regions on the blocklist are highlighted in pink. RepeatMasker track is shown at the bottom (retrieved from UCSC Genome Browser). Lc denotes regions of low complexity. The red line in the chromosome ideogram corresponds to the region that is shown enlarged in the tracks below

Fig. 3

Cumulative distribution of expression from regions (that have RNA evidence) overlapping satellite elements; for long polyadenylated RNA (left), long non-polyadenylated RNA (middle) and short RNA (right). BPKM (bases per kilobase per million mapped bases) values represent units of transcript expression for long RNA and RPKM (reads per kilobase per million mapped bases) for short RNA

Fig. 4

Differential enrichment of H3K9me3 at satellite elements during fetal development in mouse. a Two-dimensional PCA plot of mouse samples from diverse tissues across fetal development stages based on H3K9me3 enrichment at 28,937 autosomal satellite elements. Ellipses define 95% confidence intervals around group mean. Samples are colored by developmental stage (days after conception) as indicated by the legend. Days_0 denotes samples collected after birth. b Heatmap shows Z-score of FC values at 864 differentially enriched satellite elements on autosomal chromosomes. Stages are indicated by color above the heatmap. Tissues are shown per column and labeled in colored font by developmental stage (EFP—embryonic facial prominence). Satellite DNA instances are shown by row. Satellite families are color-coded on the left of the heatmap, as indicated in the legend. c distribution of FC per sample calculated for all annotated satellite DNA elements (red; N = 28,937) and elements with differentially enriched H3K9me3 (green; N = 864). Points denote median values and lines represent interquartile ranges