Paleogenomic record of the extinction of human endogenous retrovirus ERV9

Abstract

An outstanding question of genome evolution is what stops the invasion of a host genome by transposable elements (TEs). The human genome, harboring the remnants of many extinct TE families, offers an extraordinary opportunity to investigate this problem. ERV9 is an endogenous retrovirus repeatedly mobilized during primate evolution, 15 to 6 million years ago (MYA), which left a trace of over a hundred provirus-like copies and at least 4,000 solitary long terminal repeats (LTRs) in the human genome. Then, its proliferation ceased for unknown reasons, and the family went extinct. We have made a detailed reconstruction of its last active subfamily, ERV9_XII, by examining 115 solitary LTRs from it. These insertions were grouped into 11 sets according to shared nucleotide variants, which could be placed in a sequential order of 10 to 6 MYA. At least 75% of the subfamily was produced 8 to 6 MYA, during a stage of intense proliferation. With new analytical tools, we show that the youngest and most prolific sets may have been produced by effectively instantaneous expansions of corresponding single-sequence variants. The extinction of this family apparently was not a consequence of its slow gradual degeneration, but the outcome of the fixation of specific restrictive alleles in the human-chimpanzee ancestral population. Three species-specific insertions (two in humans and one in chimpanzees) were identified, further supporting that extinction took place when these two species were beginning to diverge. These are the only fixed differences of this kind so far observed between humans and chimpanzees, apart from those belonging to the human endogenous retrovirus K family.

FIG. 1.

Nucleotide differences among the consensus sequences of the different groups of subfamily XII. The first three rows refer to base positions relative to Fig. 1 in Costas and Naveira (; see also the supplemental material). Double- and single-underlined positions indicate sites forming part of CpG dinucleotides in the general consensus sequence of subfamily XII, with a 70% or a 90% threshold frequency, respectively (see Materials and Methods). The general consensus sequence of subfamily XI was used as a reference. Dots indicate identity with this reference sequence. Uppercase letters indicate nucleotides present in >70% of the sequences belonging to a group; lowercase letters indicate nucleotides present in 50% to 70% of the sequences in a group.

FIG. 2.

Phylogenetic relationships within ERV9_XII, based on analyses of consensus sequences of the different groups established according to shared nucleotide differences. The displayed tree was obtained by the NJ method, and it is rooted with the general consensus of subfamily XI (ERV9_XI), used as an outgroup. Values indicate the percentages of equally parsimonious trees supporting internal branches (only values of >70% are indicated; three equally most parsimonious trees were obtained). Sequence groups that did not depart significantly from a star phylogeny, after contrasting observed and expected values of Jurka's coefficient, are marked with a star symbol.

FIG. 3.

Phylogenetic reconstruction of the full set of ERV9_XII sequences (97 solitary LTR insertions) by the NJ method, after CpG positions were excluded. The tree is rooted with the consensus of ERV9_XI. Each sequence is designated by its GenBank accession number, followed by the name of the group it has been assigned to in this work.