Host Gene Regulation by Transposable Elements: The New, the Old and the Ugly

Abstract

The human genome has been under selective pressure to evolve in response to emerging pathogens and other environmental challenges. Genome evolution includes the acquisition of new genes or new isoforms of genes and changes to gene expression patterns. One source of genome innovation is from transposable elements (TEs), which carry their own promoters, enhancers and open reading frames and can act as 'controlling elements' for our own genes. TEs include LINE-1 elements, which can retrotranspose intracellularly and endogenous retroviruses (ERVs) that represent remnants of past retroviral germline infections. Although once pathogens, ERVs also represent an enticing source of incoming genetic material that the host can then repurpose. ERVs and other TEs have coevolved with host genes for millions of years, which has allowed them to become embedded within essential gene expression programmes. Intriguingly, these host genes are often subject to the same epigenetic control mechanisms that evolved to combat the TEs that now regulate them. Here, we illustrate the breadth of host gene regulation through TEs by focusing on examples of young (The New), ancient (The Old), and disease-causing (The Ugly) TE integrants.

Keywords: Intracisternal A-type particle elements; KRAB-associated protein 1; X chromosome inactivation; endogenous retroviruses; epigenetic repression; gene regulation; genomic imprinting; position-effect variegation; transposable elements.

Figure 1

Evolutionary map of example co-opted transposable elements (TEs) for gene regulation. Clade diagram of human and mouse evolutionary trajectories overlaid with bubble plots showing the relative prevalence of all TEs in the human (blue) and mouse (green) genomes, according to their taxonomic specificity (from the Dfam database of repetitive DNA families). The dotted line represents an arbitrary evolutionary time cutoff to classify TEs as ‘Old’ vs. ‘New’ in this review. TEs discussed in this review are annotated within their respective taxa and colour coded according to their TE class (see the key). ERV; endogenous retrovirus, LINE; long interspersed elements; SINE; short interspersed elements, SVA; SINE/VNTR/Alu elements.

Figure 2

Mechanisms by which transposable elements (TEs) regulate host genes that are discussed in this review. Heterochromatin spreading into genes is represented in (a), which can occur stochastically but often radiates from silencers bound by sequence-specific transcription factor (TF) repressors. These can recruit heterochromatin-related proteins, some of which are highlighted here. Heterochromatin spreading has been described for IAP ERVs [20,21]. An insulator function is portrayed in (b), whereby a TE can protect a host gene from heterochromatin spreading and an example is MIR elements [22]. ‘Poised’ or cryptic TE-derived enhancers and promoters are shown in (c, d), which can be poised due to their dynamic epigenetic repression as discussed in this review. TE promoters can function as alternative promoters for protein-coding genes as is the case for MERVL long terminal repeats (LTRs), which can generate chimeric transcripts with host genes expressed at the 2-cell stage of mouse development [23]. MER41 has been shown to act as a poised enhancer [24].

Figure 3

Temporal gene regulation through MERVL-derived LTR promoters. Top: Diagram of early mouse development. Middle: Median expression values (FPKM) of the transcription factors, Dux and Zscan4c (left axis) and of MERVL and its LTR promoter (MT2) as defined by ‘TEcounts’ software (right axis) are shown through development using data from [39]. Bottom: Expression of 43 2C stage-specific genes, controlled through MERVL LTRs, depicted both as the median expression in FPKM (’MERVL-controlled genes’ line graph) and as expression values of individual genes in a heatmap (values are scaled by row). A list of 2C stage-expressed genes was compiled by including previously published lists from both the following studies: [23,30]. FPKM expression values through development were extracted from [39]. Only genes that are within 10 kb downstream of MT2 LTRs that overlap a DUX binding peak (peaks extracted from [35]) and have detectable expression levels in [39] were considered here. See Table S1 (Supplementary Materials) for raw data.

Figure 4

Heterochromatin spreading from silencers embedded in IAP-type ERV elements. Circos plot showing circular visualization of the mouse genome depicting IAP elements coated with silent chromatin marks and proximal genes that they potentially regulate. Data are from C57BL/6J-derived mESCs and extracted from [21]. Track 1 corresponds to a histogram of KAP1-dependent H3K9me3 coverage across 5-Mb windows. This represents regions of H3K9me3-enrichment that are present in KAP1 control mESCs but lost in KAP1 KO mESCs. Track 2 depicts a histogram of KAP1 peak coverage in 5-Mb windows. Track 3 shows individual occurrences of IAP elements (290) that overlap H3K9me3 peaks shown in track 1, and which are less than 10 kb from genes upregulated (>2 fold) in KAP1 KO mESCs. Note that KAP1-dependent H3K9me3 peaks are used rather than KAP1 peaks, due to their increased mappability at highly repetitive (young) IAPs, resulting from their spreading into genes. Track 4 depicts the coordinates of 145 unique genes <10 kb from IAPs in track 3 and upregulated in KAP1 KO mESCs. See Table S2 (Supplementary Materials) for raw data.

Figure 5

Chromatin landscape of past co-option events vs. co-option ‘in action’ of new TEs (a) Circos plot of MER130 TEs in the mouse genome according to [119]. (1) H3K27ac peaks (ENCODE) in E14.5 whole brain which overlap MER130 elements. (2) p300 binding sites in E14.5 dorsal cerebral wall, which overlap MER130 elements. (3) MER130 elements in the mouse genome depicted in yellow, except for those MER130 instances that are conserved in 60 placental mammals, including in the human genome, which are depicted in grey. (4) Genes located near to MER130 cortical enhancers which, when perturbed, result in abnormal telencephalon morphology [119]. (b) Circos plot of IAP elements in the mouse genome under KAP1 regulation according to [21]. (1) Coordinates of KAP1-dependent H3K9me3 peaks. (2) KAP1 peaks that overlap IAPs proximal (<10 kb) to genes upregulated upon KAP1 KO. (3) IAP elements which intersect H3K9me3 peaks in (1) and overlap or are proximal (<10 kb) to genes upregulated upon KAP1 KO. IAPs conserved between 14 inbred laboratory mouse strains are depicted in grey, whereas IAPs for which there is evidence of their absence in more than one strain (exhibiting strain variability) are highlighted in yellow. Genome browser windows in the lower half of the figure depict the highlighted TE/gene intersections from the circos plot in (a) showing conservation in the mouse and human genomes or (b) an IAP integrant that is only present in C57BL/6J-strain mice vs. an IAP integrant conserved across all 14 inbred mouse genomes. See Table S3 (Supplementary Materials) for raw data.