The Role of SINE-VNTR-Alu (SVA) Retrotransposons in Shaping the Human Genome

Abstract

Retrotransposons can alter the regulation of genes both transcriptionally and post-transcriptionally, through mechanisms such as binding transcription factors and alternative splicing of transcripts. SINE-VNTR-Alu (SVA) retrotransposons are the most recently evolved class of retrotransposable elements, found solely in primates, including humans. SVAs are preferentially found at genic, high GC loci, and have been termed "mobile CpG islands". We hypothesise that the ability of SVAs to mobilise, and their non-random distribution across the genome, may result in differential regulation of certain pathways. We analysed SVA distribution patterns across the human reference genome and identified over-representation of SVAs at zinc finger gene clusters. Zinc finger proteins are able to bind to and repress SVA function through transcriptional and epigenetic mechanisms, and the interplay between SVAs and zinc fingers has been proposed as a major feature of genome evolution. We describe observations relating to the clustering patterns of both reference SVAs and polymorphic SVA insertions at zinc finger gene loci, suggesting that the evolution of this network may be ongoing in humans. Further, we propose a mechanism to direct future research and validation efforts, in which the interplay between zinc fingers and their epigenetic modulation of SVAs may regulate a network of zinc finger genes, with the potential for wider transcriptional consequences.

Keywords: SVA; ZNF; evolution; polymorphism; primate; retrotransposon; zinc finger protein.

Figure 1

Canonical SINE-VNTR-Alu (SVA) structure. Canonical SVAs typically contain five distinct regions; a (CCCTCT)_n hexamer repeat at the 5′ end, an Alu-like domain, a variable number tandem repeat (VNTR), a SINE-derived region, and a poly(A) tail. The SVA F1 subfamily deviates from this typical structure as the (CCCTCT)_n hexamer has been replaced by a 5′ transduction of the first exon of the MAST2 gene. While SVAs are approximately 2 kb in length, their size can vary due to changes in copy number in their repetitive domains.

Figure 2

SVA elements preferentially cluster on chromosome 19. (a) Chromosome 19 is the only autosome that displays no correlation between SVA and transcript number per megabase (correlation coefficient = −0.055, Supplementary File 3). This apparent lack of correlation is skewed by a small number of regions which are vastly over-represented for SVAs based on their gene density (red points). (b) Mapping SVA density per megabase across the whole genome revealed regions with the highest number of SVAs. Red bars indicate regions with four or more SVA A, B, and Cs, while blue bars identify regions with six or more SVAs of the more recent D–F1 subfamilies. By overlaying SVA data for both older and younger SVA subclasses, we find that the chr19:20,000,000–24,000,000 locus is the only region in the genome at which clustering of both older and younger SVA classes is observed. This suggests sustained SVA-mediated evolution at this specific region on chromosome 19, from the evolution of the earliest SVA class to more recent human-specific changes. We further identify six regions that have high rates of SVA D–F1 subfamily insertions, including three additional zinc finger loci on chromosomes 4, 7, and 19.

Figure 3

SVAs on chromosome 19 cluster at a KRAB ZNF zinc finger locus. (a) Visualisation of the chr19:20,000,000–24,000,000 locus and flanking regions (chr19:19,000,000–25,000,000; GRCh37/hg19) reveals that SVA clustering in this area falls directly across a large zinc finger gene locus, with no SVAs in the regions flanking either side. Observation of the Multiz vertebrate genome alignment data demonstrated significant change around this zinc finger locus through primate evolution, in clear contrast to the strongly conserved flanking region to the left. The white region on the right of the Multiz alignment track shows the centromere of chromosome 19, an area with no available sequence data. (b) The ZNF91 gene resides in the chr19:20,000,000–24,000,000 SVA-rich zinc finger cluster (boxed in (a); chr19:23,464,801–23,613,281 shown in (b)), which is of interest due to its known role in suppressing SVA mobilisation. ZNF91 is a primate-specific gene and contains an SVA C and SVA D insertion in the third intron which may have influenced expression or splicing of this gene specific to primate species from gorillas through to humans. Further, we observe a second SVA C approximately 24 kb upstream of the ZNF91 transcriptional start site, and an SVA B around 67 kb downstream of the 3′ UTR. From these locations, all four SVAs have the potential to modulate ZNF91 expression through methods including binding transcription factors, altering local chromatin structure, or modulating splicing.

Figure 4

Three of the remaining five regions enriched for reference SVA D–F1 insertions are zinc finger clusters. In addition to the chr19:20,000,000–24,000,000 locus, three of the remaining five loci with the highest rates of SVA D–F1 insertions are at zinc finger regions. A second locus was identified on chromosome 19 at chr19:53,000,000–54,000,000, and two separate loci at chr4:1–1,000,000 and chr7:64,000,000–65,000,000 (expanded loci shown as chr19:51,812,498–55,187,502, chr4:1–2,250,000, and chr7:63,000,000–66,000,000). (a) The chr19:53,000,000–54,000,000 locus contains one SVA A, four SVA Ds, and three SVA Es, as well as 25 zinc finger genes. We observe the extension of this zinc finger gene cluster outside of this specific megabase, with additional SVA Ds present at either end of the zinc finger cluster. (b) A significantly smaller zinc finger gene cluster resides within the first megabase of chromosome 4 (chr4:1–1,000,000), encompassing three SVA Ds, two SVA Es, and an SVA F, arranged over a stretch of six zinc finger genes on the tip of the chromosome and into the remainder of this megabase. We note that the two human-specific SVA Es are directly over a genome-wide associated region for Parkinson’s disease (PD) encompassing the GAK and DGKQ genes, with one SVA E within a GAK intron, and the second lying within 2 kb of the DGKQ transcriptional start site. SVAs at this region could therefore modulate expression of both zinc finger and PD-related genes at this locus uniquely in humans. (c) The final SVA-rich zinc finger gene cluster is identified at chr7:64,000,000–65,000,000 and contains one SVA B, two SVA Cs, five SVA Ds, one SVA E, and two SVA Fs, spread across six zinc finger genes. Expanding this locus shows four additional zinc finger genes adjacent to this megabase, with a sixth SVA D.

Figure 5

Composition of SVA subfamilies within 1 Mb zinc finger gene regions compared to the whole genome. (a) Calculating the percentage of each SVA subfamily across the human genome demonstrated that 34.44% of all SVAs were comprised of the evolutionary older A–C subfamilies and are therefore conserved at their respective loci within multiple primate species including humans. The remaining 65.56% is comprised of younger subfamily members D–F1. SVA D is by far the largest SVA subfamily, comprising 44.39% of all SVA elements in the genome, some of which are human-specific, with others being present in multiple primate species. 21.17% of SVAs (subfamilies E–F1) are entirely human-specific. SVA B, F, and C each represent over 10% of all SVAs, with a percentage of 16.45, 13.43, and 10.17%, respectively. The remaining subfamilies (SVA A, E, and F1) make up less than 10% of the total each. (b) The percentage of each SVA subfamily at the 1 Mb SVA clustered ZNF loci on chromosomes 4, 7, and 19 showed that older SVA subfamilies SVA A–C were reduced by 2.15-fold to 16.00% compared to the whole genome composition. This shift is due to an under-representation of SVA A and B subtypes (0.51- and 0.24-fold change, respectively), and a strong 4.86-fold increase in the SVA E subfamily = at the identified ZNF loci.

Figure 6

Both reference SVA insertions and SVA retrotransposon insertion polymorphisms (RIPs) are preferentially found at genic regions. (a) Plotting the number of reference SVAs vs. transcripts per megabase across the whole genome shows that SVAs are preferentially found at genic regions, with higher transcript number per megabase correlating with higher SVA number (correlation coefficient = 0.352, Supplementary File 3). (b) Plotting the number of SVA retrotransposon insertion polymorphisms (RIPs) vs. transcript number per megabase shows that new and polymorphic SVA insertions follow the same trend as established reference SVAs, with a preference for genic regions (correlation coefficient = 0.306, Supplementary File 3).

Figure 7

SVA retrotransposon insertion polymorphisms (RIPs) cluster at a third zinc finger gene locus on chromosome 19. Distribution analysis of SVA RIPs per megabase revealed three regions with the highest density of new and polymorphic SVA insertions in humans, with five SVA RIPs per megabase. Visualisation of these regions using the UCSC Genome Browser showed that the chr19:44,000,000–45,000,000 locus (chr19:43,000,000–46,000,000 shown above) contains an additional zinc finger gene cluster. With only three reference SVA insertions, this zinc finger cluster appears to be the target of SVA RIP insertion independently of previous reference SVA-mediated evolution, which we have shown to occur at other zinc finger gene loci. This suggests that SVA-mediated evolution of zinc finger clusters may be ongoing, with the chr19:44,000,000–45,000,000 locus in particular continuing to undergo genomic change which may modulate the expression of genes at this locus uniquely in modern humans, and differently between individuals based on the presence or absence of these insertions.