Human endogenous retrovirus HERV-K14 families: status, variants, evolution, and mobilization of other cellular sequences

Abstract

The human genome harbors many distinct families of human endogenous retroviruses (HERVs) that stem from exogenous retroviruses that infected the germ line millions of years ago. Many HERV families remain to be investigated. We report in the present study the detailed characterization of the HERV-K14I and HERV-K14CI families as they are represented in the human genome. Most of the 68 HERV-K14I and 23 HERV-K14CI proviruses are severely mutated, frequently displaying uniform deletions of retroviral genes and long terminal repeats (LTRs). Both HERV families entered the germ line approximately 39 million years ago, as evidenced by homologous sequences in hominoids and Old World primates and calculation of evolutionary ages based on a molecular clock. Proviruses of both families were formed during a brief period. A majority of HERV-K14CI proviruses on the Y chromosome mimic a higher evolutionary age, showing that LTR-LTR divergence data can indicate false ages. Fully translatable consensus sequences encoding major retroviral proteins were generated. Most HERV-K14I loci lack an env gene and are structurally reminiscent of LTR retrotransposons. A minority of HERV-K14I variants display an env gene. HERV-K14I proviruses are associated with three distinct LTR families, while HERV-K14CI is associated with a single LTR family. Hybrid proviruses consisting of HERV-K14I and HERV-W sequences that appear to have produced provirus progeny in the genome were detected. Several HERV-K14I proviruses harbor TRPC6 mRNA portions, exemplifying mobilization of cellular transcripts by HERVs. Our analysis contributes essential information on two more HERV families and on the biology of HERV sequences in general.

FIG. 1.

Maps of HERV-K14 ORFs as generated from fully translatable consensus sequences. (Top) HERV-K14I ORF maps. The variant without env is depicted in the upper part, while the variant with a putative env gene is shown in the lower part. (Bottom) HERV-K14CI ORF map. The locations of retroviral gag, pro, pol, and env reading frames are shown, as well as the locations of typical protein domains. LTR14C, associated with HERV-K14CI, is included in the bottom panel. LTRs were omitted in the top panel, as HERV-K14I is associated with three distinct LTR families (see the text). The scales are in kilobases.

FIG. 2.

Dot matrix comparison of HERV-K14I and associated LTR variants with different env gene regions. (Left) Comparison of an HERV-K14I provirus with a putative env gene with the HERV-K14I reference sequence (lacking LTRs) in Repbase (bottom) and with the HERV-K14CI reference sequence (flanked by LTR14C) (top). The locations of proviral LTRs and genes are indicated. (Right) Comparison of variant LTR14Bv sequences that flank HERV-K14I env⁺ proviruses with the LTR14B (bottom) and LTR14A (top) reference sequences. Note that the LTR14C sequence is much less similar to the three LTR sequences shown. The dot matrix comparisons were generated with MacVector, employing a window size of 30 nt and a minimum similarity of 60%. The scales are in kilobases.

FIG. 3.

Dot matrix comparison exemplifying a hybrid provirus between HERV-K14I and HERV17 (HERV-W) that was identified multiple times in the human genome. Provirus number 52 (see Table SA in the supplemental material) is compared to a sequence consisting of HERV-K14I (lacking env; see the text) flanked by LTR14B (bottom) and to a sequence consisting of the Repbase HERV17 sequence flanked by LTR17 (top). The locations of proviral LTRs and genes are indicated. The locations of proviral genes in HERV-17 were estimated from TBLASTN database search results. Parameters of comparisons are given in Fig. 2. The scales are in kilobases.

FIG. 4.

Presence of HERV-K14-homologous sequences in various primate species, as revealed by PCR. (Top) Presence of HERV-K14I homologues in hominoids and Old World primates but not in New World primates and prosimians. Specific PCR products of ∼740 bp were amplified from positive species. PCR products in New World primates and prosimians proved to be nonretroviral after being sequenced. The lower PCR yields in lanes Pha and Msp are probably due to suboptimal genomic DNA quality for those species. (Bottom) Similar species distributions of HERV-K14CI homologues, as revealed by a specific PCR product of ∼620 bp. Species abbreviations are as follows. Hominoidea: Hsa (Homo sapiens), Ptr (Pan troglodytes), Ppa (Pan paniscus), Ggo (Gorilla gorilla), and Ppy (Pongo pygmaeus). Old World primates (OWP): Hla (Hylobates lar), Pha (Papio hamadryas), Cgu (Colobus guereza), Mfa (Macaca fascicularis), Mmu (Macaca mulatta), Msp (Mandrillus sphinx). New World primates (NWP): Ase (Alouatta seniculus), Cja (Callithrix jacchus), Ssc (Saimiri sciureus). Prosimian (Pros.): Nyc (Nycticebus coucang).

FIG. 5.

Neighbor-joining analysis of HERV-K14I sequences for the proviral intergenic and gag gene regions. The provirus numbers correspond to the numbers in Table SA in the supplemental material. When possible, each provirus is followed by information regarding the LTR14 family flanking the provirus (LTR14A [A], LTR14B [B], or LTR14Bv [Bv]) and the presence of a true env gene. Note that a subgroup of HERV-K14I sequences, supported by a bootstrap value of 75% in 1,000 replicates, is almost exclusively flanked by LTR14A. Consensus sequences, as generated in this study and as given in Repbase, were also included in the analysis.

FIG. 6.

Dot matrix comparisons of HERV-K14I proviruses associated with TRPC6 mRNA portions. Two proviruses in which the sizes of HERV-K14I and TRPC6 sequence portions differed are depicted. The long variant (left) contains additional HERV-K14I and fewer TRPC6 portions than the short variant (right). Each variant is compared to a sequence consisting of HERV-K14I flanked by LTR14B (bottom) and to a recently reported full-length TRPC6 mRNA (11) (top). The locations of proviral LTRs and genes, as well as the open reading frame within the TRPC6 mRNA, are indicated. The parameters of comparisons are identical to the ones given for Fig. 2. The scales are in kilobases.