Latent regulatory potential of human-specific repetitive elements

Abstract

At least half of the human genome is derived from repetitive elements, which are often lineage specific and silenced by a variety of genetic and epigenetic mechanisms. Using a transchromosomic mouse strain that transmits an almost complete single copy of human chromosome 21 via the female germline, we show that a heterologous regulatory environment can transcriptionally activate transposon-derived human regulatory regions. In the mouse nucleus, hundreds of locations on human chromosome 21 newly associate with activating histone modifications in both somatic and germline tissues, and influence the gene expression of nearby transcripts. These regions are enriched with primate and human lineage-specific transposable elements, and their activation corresponds to changes in DNA methylation at CpG dinucleotides. This study reveals the latent regulatory potential of the repetitive human genome and illustrates the species specificity of mechanisms that control it.

Figure 1

In a Mouse Carrying Human Chromosome 21, Most Locations of Human Chromosome 21 Are Transcribed in Liver Largely as in Human Tissues, Yet Specific Loci Show Differences (A) High-throughput sequencing of chromatin immunoprecipitations and poly(A) mRNA enrichment revealed that the CSTB locus is occupied by RNA polymerase II, enriched for H3K4me3, and transcribed into RNA in both human and Tc1 mouse liver. (B) The DSCR4/8 locus on HsChr21 shows similar evidence of transcription in Tc1 mouse liver tissue, which is not evident in normal human liver.

Figure 2

Tc1-Specific Transcription Initiation Occurs in Somatic and Germline Tissues and Influences Gene Expression (A–C) Chromosome-wide identification of all regions enriched for H3K4me3 on HsChr21 in mouse and human liver, kidney, and testes; heatmap is sorted by descending signal strength and by species specificity. (D) Similar sets of loci on HsChr21 are enriched for H3K4me3 in both mouse and human, and these are largely a subset of loci found in testes in both species. In liver and kidney, Tc1-specific loci enriched for H3K4me3 were found in similar locations, which were distinct from those found in Tc1 mouse testes. (E) H3K4me3-enriched sites in the Shared and Tc1-specific categories in liver were associated with the nearest transcription start site and the expression levels of genes on Tc1-HsChr21 determined relative to human genes (Shared, n = 86; Tc1 specific, n = 17) (^∗∗∗p ≤ 0.0005, one-sided Mann-Whitney U test).

Figure 3

Tc1-Specific Locations of Transcription Initiation Are Enriched for Young, Lineage-Specific Repetitive Elements (A) Fraction of repeat elements within a 10 kb window around the H3K4me3 peak summit in Shared (black) and Tc1-specific (red) events in liver. Shown is age of the repeats in Tc1-specific and Shared sites as determined by nucleotide substitution rates of repeat instances at H3K4me3 peak summits. (B) As in (A) for kidney. (C) As in (A) for testes. (D) Number and lineage of repeat elements at H3K4me3 peak summits in liver, kidney, and testes. Pie charts are scaled relative to the total number of H3K4me3 binding events in the Shared category where the purple proportion represents the fraction of H3K4me3 events that have a repeat element at the H3K4me3 peak summit. Bar charts represent the lineage in which the repeat elements originated; the numbers of human-specific (and primate-specific for Shared testes sites) are reported beside the bar charts. The composition of repeat lineages on HsChr21 is shown in the bar graph on the right panel (^∗p ≤ 0.05, ^∗∗p ≤ 0.005, Wilcoxon rank-sum test).

Figure 4

Repetitive Elements Contain Latent Transcriptional Regulator Binding Sites (A) Individual CEBPA, HNF4A, and CTCF binding sites can be carried by specific repeat elements. An LTR12C repetitive element upstream of the SOD1 locus reveals a latent CTCF binding site in Tc1 mouse liver. Upstream of the Shared H3K4me3 site at the CSTB gene, there is a Shared HNF4A and CEBPA binding event, while Alu-associated CEBPA and HNF4A sites are revealed in the Tc1 mouse. (B) Heatmap representation of repeats enriched in Tc1-specific events (red) or enriched in Shared (black) at H3K4me3 and transcription factor peak summits in liver, kidney, and testes. p values as calculated by chi-square test are presented in –log¹⁰ scale. Only repeat names that are significant in at least one data set are shown (p = 0.05).

Figure 5

Tc1-Specific Sites of Transcription Initiation Are Depleted in DNA Methylation (A) Fraction of methylated DNA at H3K4me3-enriched sites Shared between HsChr21 (blue) and Tc1-HsChr21 (red) in liver. (B) Fraction of methylated DNA at Tc1-specific H3K4me3-enriched sites. (C) Fraction of methylated DNA at regions where there is no H3K4me3 enrichment in Human or Tc1 mouse. Interrogated CpG sites are shown within LTR and LINE repetitive elements (purple) or genes (gray). Each experiment was performed using three biological replicates.

Figure 6

Changes in Transcription Status between Human and Tc1 Mouse Are Associated with Changes in the Repressive Histone Mark H3K9me3 (A) The genome-wide occupancy of H3K9me3 was determined in human liver (blue) and Tc1 mouse liver (red) by ChIP-seq and then compared with the transcriptional activation status of these regions. The vertical axis shows the log2 normalized read counts for H3K9me3 ChIP experiments averaged across three individuals from each species. (^∗p ≤ 0.05, ^∗∗∗p ≤ 0.0005, Wilcoxon-matched pairs test). (B) The occupancy of H3K9me3, H3K4me3, and Pol II is shown for the BAGE gene on HsChr21 located in human liver (blue) and Tc1 mouse liver (red). Human liver shows enrichment of the repressive histone mark H3K9me3 over the BAGE gene, which is missing in Tc1 mouse liver. Conversely, the mouse liver shows hallmarks of transcription (H3K4me3 and Pol II occupancies) that are absent from human.