Endogenous retroviruses (ERV) make up 5~8% of human genome. These ERV are ancient retroviruses which infected germ cells and therefore became parts of our genomes. However, unless the ERV can increase the survival chance to its host, it wouldn’t be able to expand its frequency in our genetic pool. So there must be some mechanisms that expand the copy number of ERV. The infection of germ cells can happen more than once. Alternatively, ERV can reactivate at some point of the development of the host and make new copies in the genome. Considering that the sites of integration of retroviruses are highly random, even the same ERV can appear in different loci or different chromosomes. Have large scale genome sequencing revealed a heterogenic map of ERV?
Of the papers I reviewed to answer this, Variation in proviral content among human genomes mediated by LTR recombination addresses the question most directly. I recommend you read it if you have access. To summarize, most ERV in the human genome integrated before the separation of human and chimpanzee lineages, and some before the emergence of placental mammals. Among these, there is relatively little variation:
an HERV-H element on chromosome 1 (1q25.3_H3) was shown to exist as proviral and solo LTR alleles in two related individuals
"solo LTR" allele means that the viral genes have been deleted by homologous recombination of the Long Terminal Repeats (LTR) that flank the intact provirus, leaving only one LTR. These don't code for any protein, but can still be "active" by affecting the expression of nearby regions in the host genome. The paper I cited above verified four other HERV-H variations and identified 67 potential variations.
We were intrigued to discover a dimorphic element for the HERV-W family (18q21.1_W2). This element is represented as a solo LTR in the reference genome, but our data clearly show that it also occurs as a provirus segregating in South Asian populations (Fig. 3a) and likely in other diverse populations (our pipeline predicted a provirus allele in 194 out of 279 individuals surveyed, Additional file 2). To the best of our knowledge, this is the first HERV-W locus reported to show any type of dimorphism. This particular HERV-W insertion must have occurred between 18 and 25 million years ago because a provirus is found at orthologous position in all other ape genomes including gibbon, but is absent in Old and New World monkeys . Our discovery illustrates the potential of LTR recombination to alter genome structure long after a proviral insertion has occurred.
The HERV-K family, however, integrated into the human germline relatively recently (estimates span 3 million - 250 thousand years ago).
Among the diverse assemblage of HERV families in our genome, a single subfamily known as HERV-K(HML2) has been reported to exhibit insertional polymorphism in humans [17, 28, 29, 36–47]. Thus far, approximately 50 HERV-K(HML2) proviral loci are known to occur as empty (pre-integration) and/or solo LTR alleles segregating in the human population [17, 43, 45, 46], but more may be expected to segregate at low frequency [39, 48]. These observations are consistent with the notion that HERV-K(HML2) is the most recently active HERV subfamily in the human genome [49–53].
The 2020 review article Human Endogenous Retrovirus K (HML-2) in Health and Disease gives a quantified overview of the variability of the HERV-K family.
It is known that there are more than 1,000 HERV-K (HML-2) loci in the human genome. Most of them reside in the genome as solo-LTRs generated by homologous recombination between the LTRs of a single HERV-K (HML-2), resulting in the deletion of the internal sequence (Hughes and Coffin, 2004). Solo-LTRs are approximately 10-fold more abundant than their full-length or nearly full-length proviral integrations (Subramanian et al., 2011). In total, there are 1098 reported loci whose insertional elements possess functional elements, including 1061 reference (hg19) insertions and 37 non-reference insertions (Subramanian et al., 2011; Lee et al., 2012; Marchi et al., 2014; Wildschutte et al., 2016; Wallace et al., 2018; Xue et al., 2020).
My summary: Most of the ERVs in the human genome are present only as remnant solo LTR alleles. Since virus reemergence is harmful, deletion by homologous recombination is likely selected for. This kind of deletion has been happening continuously along the human lineage and accounts for most of HERV differences between human genomes. Only the HERV-K family shows differences in insertions between humans. Across humans, roughly 95% of HERVs are identical for everyone, and roughly 5% are absent in some people. All of the HERV families are considered extinct. That is, there are no retroviruses circulating currently that correspond to any of the integrated HERVs.
I didn't see much discussion about smaller-scale insertions, deletions, or point mutations within HERVs, except to note that intact provirus could be dated by the amount of sequence difference in the LTRs at either end. That is, the LTRs are by definition identical when the retrovirus genome is inserted, but will accumulate mutations independently afterward, so the difference lets you estimate the number of mutations and thus how long ago integration occurred. If the LTRs of a provirus diverge enough, they will no longer be able to homologously recombine, and the virus insertion will become permanent. As a caveat, genomics is not my specialty; I may have combined information from different sources incorrectly.
some other notable references: