Genome-wide detection and characterization of endogenous retroviruses in Bos taurus

Abstract

Endogenous retroviruses (ERVs) are the proviral phase of exogenous retroviruses that become integrated into a host germ line. They can play an important role in the host genome. Bioinformatic tools have been used to detect ERVs in several vertebrates, primarily primates and rodents. Less information is available regarding ERVs in other mammalian groups, and the source of this information is basically experimental. We analyzed the genome of the cow (Bos taurus) using three different methods. A BLAST-based method detected 928 possible ERVs, LTR_STRUC detected 4,487 elements flanked by long terminal repeats (LTRs), and Retrotector detected 9,698 ERVs. The ERVs were not homogeneously distributed across chromosomes; the number of ERVs was positively correlated with chromosomal size and negatively correlated with chromosomal GC content. The bovine ERVs (BoERVs) were classified into 24 putative families, with 20 of them not previously described. One of these new families, BoERV1, was the most abundant family and appeared to be specific to ruminants. An analysis of representatives of ERV families from rodents, primates, and ruminants showed a phylogenetic relationship following their hosts' relationships. This study demonstrates the importance of using multiple methods when trying to identify new ERVs and shows that the number of bovine ERV families is not as limited as previously thought.

FIG. 1.

Diagram representing the number of ERVs detected by each detection method (represented by the circle size) and common elements detected by two or more methods (included in the overlapping areas).

FIG. 2.

RT region-based phylogenetic tree of BoERVs. A total of 247 BoERVs detected by at least two methods and with the pol gene longer than 500 nucleotides from this work were included. Eight experimentally detected cow ERVs (43) and 12 sheep ERVs (17) were also included. Retroviruses used as queries and retroviruses used in previous phylogenetic studies, such as gibbon ape leukemia virus (GALV) (GenBank accession number NC_001885), MLV (accession number NC_001501), feline leukemia virus (FeLV) (accession number NC_001940), Jaagsiekte sheep retrovirus (JSRV) (accession number NC_001494), mouse mammary tumor virus (MMTV) (accession number NC_001503), bovine leukemia virus (BLV) (accession number NC_001414), human T-lymphotropic virus (HTLV) (accession number NC_001436), equine infectious anemia virus (EIAV) (accession number NC_001450), HIV (accession number NC_001803), Visna virus (accession number NC_001452), human spumaretrovirus (HSRV) (accession number NC_001795), bovine coronavirus (BCV) (accession number NC_001831), BoEV (accession number X99924), HERV-E (accession number M10976), PERV (accession number AJ293656), MPMV (accession number NC_001550), IAPM (accession number M17551), OMVV (accession number NC_001511), FeFV (accession number U78765), and MuERV-L (accession number Y12713), were used as an outgroup. Topology was based on the neighbor-joining method with a p distance of 1,000 bootstrap. The tree was rooted with the Drosophila melanogaster ZAM (accession number AJ000387) element. Above the branches, the NJ bootstrap values and ML bootstrap values are shown; below the branches, the Bayesian posterior probability is shown.

FIG. 3.

Partial amino acid sequence of the RT region of the representative ERVs from the 24 putative families. The positions of the functional motifs LPQG and YV/MDD are boxed.

FIG. 4.

RT region-based unrooted phylogenetic tree of ERVs from different species. Topology is based on Bayesian inference (10⁶ generations). In the branches on the left, the maximum likelihood bootstrap value is shown; on the right, the Bayesian posterior probability is shown. Sixteen representative sequences from human ERV families (40), 38 from chimpanzees (29), 27 from mice (2, 23), 7 from rats (2), 14 from sheep (17), 8 from pigs (27), and 24 from cows (our data) are included. Representatives of BoERV families are boxed.