Retrotransposons Create Regulatory Elements

Retrotransposons are a rich source of regulatory elements, including promoters, en-hancers, and transcription factor binding sites (TFBSs) [52]. As retrotransposons have spread, their regulatory regions have been adopted by host genomes and used to engineer transcriptional activity, including the creation of new regulatory net-works, the modification of existing ones, and the regulation of new transcriptional units. This can occur when a new RT copy inserts in or near an existing gene such that the RT regulatory elements can influence the gene’s expression pattern.

While this may be deleterious, in some cases it has led to beneficial effects for the host genome, and the new regulatory network has been retained.

There is overwhelming evidence for this, to the extent that there is now a catalogue of genes affected by transposable elements [94]. This catalogue lists 124 experimentally validated cases of genes influenced by transposable element-derived regulatory elements in humans, and 48 examples in mouse [94]. Summary statistics show that SINEs are responsible for 112 cases (about 50%), with LINEs and LTRs accounting for the remainder, except for the 7 (∼3%) due to DNA transposons.

About 75% of the regulatory effects are due to promoters (primary or alternative), alternative splicing, and alternative polyadenylation signals. When broken down by transposable element type, LTRs are primarily responsible for new promoters, while SINEs cause alternative splicing. Alternative splice sites and promoters ac-count for 50% and 20% of LINE effects. In addition to the validated examples, genome wide studies have identified many thousands of possible sites where trans-posable elements influence genes [94–96]. Below, I describe some examples that

illustrate how retrotransposon regulatory regions can influence transcription.

The LTR regions of ERVs contain promoters used to initiate their own tran-scription as the first step in retrotransposition. If an ERV inserts near an existing gene, the LTR promoter can act as an alternative promoter for that gene, or can become the sole promoter. An example of the former is found in the human p63 gene. This gene was known to eliminate oocytes that had suffered from DNA dam-age, but an equivalent activity in the male germline had not been identified [97].

Beyer et al. identified a novel p63 transcript expressed specifically in testis, with a transcriptional start site (TSS) inside an upstream LTR. Hence, the insertion of a retrotransposon upstream of a gene provided an alternative promoter, leading to a novel transcript with new tissue-specificity [97].

A particularly well-studied and interesting example of an LTR promoter ef-fect is the Agouti viable yellow gene in mouse, reviewed in [98]. The insertion of an intracisternal A particle (IAP), a mouse-specific ERV, upstream of the Agouti gene has provided an alternative promoter. If this promoter is unmethylated, the Agouti gene is ectopically expressed, leading to a different coat colour phenotype (Figure 1.12). If methylated, normal expression occurs and the usual phenotype is observed. This particular example also demonstrates the ability of retrotrans-posons to create metastable alleles: genes that create different phenotypes depend-ing on the epigenetic state of a particular locus, in genetically identical individuals.

There are also entire gene networks that have been found to rely on retrotrans-poson regulatory elements. Chuong et al. found that the interferon pathway in humans, part of the innate immune system, relies on TFBSs derived from a par-ticular family of ERVs. Deleting these ERVs led to a reduced immune response to viral infection. They also found evidence for a similar exaptation of ERVs

Figure 1.12: The Agouti viable yellow (Avy) example of a retrotransposon influ-encing gene expression. (A) The insertion of an intracisternal A particle (IAP), a mouse-specific ERV, upstream of the Avy gene leads to ectopic expression if the ERV is unmethylated. This leads to a metastable epiallele. (B) Genetically identical individuals with different epigenetic states leading to distinct phenotypes.

Figure from [98].

in other primates [99]. A similar example is found in gene networks involved in pregnancy. In humans, progesterone triggers differentiation of endometrial cells to

form the maternal component of the placenta (the decidua). This process depends on the activation of a gene network that uses MER20 ERVs as binding sites for progesterone-responsive signalling molecules [100].

As noted above, retrotransposons can introduce new alternative splice sites, potentially leading to new isoforms of that gene. The primate-specific Alu SINE has been particularly well-studied in this context, with several studies showing that nearly all Alu-containing exons in humans are alternatively spliced [101–104]. Shen et al. found a particular enrichment for Alu-induced exons in the zinc finger pro-teins of primates and humans [104]. In some cases, the introduction of splice sites has led to tissue-specific isoforms, as observed in the selenoprotein N 1 (SEPN1) gene [105]. In humans, SEPN1 has a muscle-specific isoform resulting from an Alu exon, which is not present in macaque and chimpanzee. Lin et al. hypothesised that the introduction of Alu led to beneficial new tissue specificity [105].

Retrotransposons can also contain alternative polyadenylation sites, influenc-ing the post-transcriptional processinfluenc-ing of expressed isoforms [93]. The attractin (ATRN) gene in humans illustrates how a retrotransposon insertion can influence transcript diversity and function. An L1 insertion in an intron of ATRN causes cleavage and polyadenylation at an alternative site for some ATRN transcripts, while others splice out the L1. The former transcript encodes a soluble form of ATRN, whereas the latter encodes a membrane-bound protein [106].