B2 SINE Copies Serve as a Transposable Boundary of DNA Methylation and Histone Modifications in the Mouse

Abstract

More than one million copies of short interspersed elements (SINEs), a class of retrotransposons, are present in the mammalian genomes, particularly within gene-rich genomic regions. Evidence has accumulated that ancient SINE sequences have acquired new binding sites for transcription factors (TFs) through multiple mutations following retrotransposition, and as a result have rewired the host regulatory network during the course of evolution. However, it remains unclear whether currently active SINEs contribute to the expansion of TF binding sites. To study the mobility, expression, and function of SINE copies, we first identified about 2,000 insertional polymorphisms of SINE B1 and B2 families within Mus musculus. Using a novel RNA sequencing method designated as melRNA-seq, we detected the expression of SINEs in male germ cells at both the subfamily and genomic copy levels: the vast majority of B1 RNAs originated from evolutionarily young subfamilies, whereas B2 RNAs originated from both young and old subfamilies. DNA methylation and chromatin immunoprecipitation-sequencing (ChIP-seq) analyses in liver revealed that polymorphic B2 insertions served as a boundary element inhibiting the expansion of DNA hypomethylated and histone hyperacetylated regions, and decreased the expression of neighboring genes. Moreover, genomic B2 copies were enriched at the boundary of various histone modifications, and chromatin insulator protein, CCCTC-binding factor, a well-known chromatin boundary protein, bound to >100 polymorphic and >10,000 non-polymorphic B2 insertions. These results suggest that the currently active B2 copies are mobile boundary elements that can modulate chromatin modifications and gene expression, and are likely involved in epigenomic and phenotypic diversification of the mouse species.

Keywords: SINE; chromatin boundary; epigenetics; intra-specific difference; transposable elements.

Fig. 1.

B2-generated indels between B6 and MSM.(A)Statistics of the indels identified in this study. (B)A neighbor-joining tree of 1,241 B2-derived indels and 6,000 B2 copies randomly selected from the reference genome sequence. B6- and MSM-specific copies are indicated in red and light blue diamonds, respectively. (C)The alignment of the consensus sequences of B2 subfamilies. The sequences for B2_Mm1a, B2_Mm1t, and B2_Mm2 were obtained from Repbase; B2_Mm1o consensus sequence was generated in the current study. A-, B-, and TATA-box as well as the CTCF/ADNP binding motif are indicated above the sequences. Nucleotides that are different between the subfamilies are highlighted. CpG sites found in at least one consensus sequence are indicated by filled rectangle. (D)Phylogenetic tree of the B2 subfamilies. The values on the clades indicate bootstrap values. Scale bar shows divergence rate (substitutions per site). (E)Pie chart representations of the numbers for genomic (left), B6-specific (middle), and MSM-specific (right) B2 copies categorized by subfamily. (F)Divergence of genomic B2 copies from the respective consensus sequences. Color codes for panels (E) and (F) are shown on the lower right.

Fig. 2.

Expression analysis of B2 within tissues. (A)Position of the oligonucleotide probe used in panel B. (B)Northern blot results for somatic tissues and testis. The gel mobility of RNA markers (100–1,000 nt) is shown on the left, and the mobility of B2 RNA (∼190 nt) is indicated on the right; 7SL RNA (bottom) was used as an internal control.

Fig. 3.

DNA methylation analysis of B2 copies using the bisulfite PCR method.(A)Schematic representation; lollipops indicate 4–8 CpG sites in the B2 sequences analyzed; PCR primers were designed for both flanking regions. (B)DNA methylation levels of 51 genomic B2 copies in spermatogonia (Sg), spermatozoa (Sp), and liver (Lv) are shown in grayscale.

Fig. 4.

melRNA-seq analysis.(A)Schematic representation of melRNA-seq methodology. (B)Schematic representation of biosynthesis of 5.8S (top) and 5S (bottom) rRNAs. (C)The fraction (%) of 5S RNA reads in TAP-untreated (−) and -treated (+) melRNA-seq libraries prepared from testis RNA. (D)Fraction of 5S RNA reads normalized with 5.8S RNA reads in the testis. (E)5.8S RNA-normalized read counts of the SINE families in the testis library. White, TAP-untreated RNA. Black, TAP-treated RNA. F. 5.8S RNA-normalized read counts of the SINE families in the 5’-treated spermatogonia library. (G and H)Pie chart representations of read counts of B1 and B2 subfamilies, respectively, in spermatogonia. (I)Boxplots for expression levels of B2 loci categorized by the number of CpG sites and the level of methylation in spermatogonia. (J)Fraction of B2 loci that were expressed (at least one sequencing reads were mapped). The loci are categorized by the number of CpG and the level of methylation in spermatogonia. (K)Fraction of expressed loci giving an RPM value of 1 or less, 1–10, 10–100, and over 100. The loci are categorized by the number of CpG and the level of methylation in spermatogonia.

Fig. 5.

DNA methylation states surrounding B2_25 in B6, MSM, and F1 mice.Methylation status of the regions A–E in liver and spermatozoa are shown; open and closed circles represent unmethylated and methylated CpG sites, respectively. Each horizontal line represents a single PCR clone. The position of the B2_25 insertion is indicated.

Fig. 6.

Chromatin boundary formed at B2_25.(A)ChIP-seq data of H3K9 acetylation (H3K9ac) for B6 (two individuals, red), MSM (two individuals, blue), and F1 hybrids (generated by reciprocal crosses, purple). (B)Allelic frequencies in the ChIP-seq reads obtained from the F1 hybrids are shown at known SNP positions (A–E) indicated in panel A. (C)Electrophoregrams from Sanger sequencing analysis of RT-PCR products for Arcn1 in liver and testis from F1 hybrids.

Fig. 7.

Chromatin boundary formed at the B2 copy upstream of Nnt.(A)ChIP-seq data of H3K9 acetylation (H3K9ac) for B6 (two individuals, red), MSM (two individuals, blue), and F1 hybrids (generated by reciprocal crosses, purple). (B)Electrophoregrams from Sanger sequencing analysis of RT-PCR products for Nnt in liver of F1 hybrids.

Fig. 8.

Enrichment of genomic B1 and B2 copies at Chromatin boundaries.Genomic occupancies of B2 (A), B1 (B), LINE (C), and LTR elements (D) are shown in the indicated positions with respect to the boundaries of ChIP-seq peaks of various histone modifications. The color codes for histone modifications are shown in panel A. The numbers of ChIP-seq peaks are 29,230 (H3K9ac), 38,492 (H3K27ac), 77,192 (H3K4me1), 16,888 (H3K4me3), 88,353 (H3K36me3), 68,593 (H3K79me2), and 33,402 (H3K27me3).

Fig. 9.

Genomic B2 copies form CTCF, CTCFL, and ADNP binding sites.(A)The numbers ChIP-seq peaks for CTCF, CTCFL, and ADNP in liver, ESCs, and round spermatids. Peak summits within genomic B1 and B2 copies are colored purple and orange, respectively. The exact numbers and proportions (%) are shown on the right. (B)Venn diagram of CTCF peaks identified in liver and ESCs. The numbers of B1 and B2 copies overlapped with liver-specific, shared, and ESC-specific copies are shown on the bottom. (C)Statistics of DNA methylation levels for B2 copies in terms of DNA methylation and CTCF binding. Genomic B2 copies are categorized according to the number of methylated CpG sites in ESCs. (Dand E)Pie chart representation of B2 subfamilies in total, CTCF-bound, ADNP-bound genomic (D), and polymorphic copies (E).