Identification and characterization of novel human endogenous retrovirus families by phylogenetic screening of the human genome mapping project database

Abstract

Human endogenous retroviruses (HERVs) were first identified almost 20 years ago, and since then numerous families have been described. It has, however, been difficult to obtain a good estimate of both the total number of independently derived families and their relationship to each other as well as to other members of the family Retroviridae. In this study, I used sequence data derived from over 150 novel HERVs, obtained from the Human Genome Mapping Project database, and a variety of recently identified nonhuman retroviruses to classify the HERVs into 22 independently acquired families. Of these, 17 families were loosely assigned to the class I HERVs, 3 to the class II HERVs and 2 to the class III HERVs. Many of these families have been identified previously, but six are described here for the first time and another four, for which only partial sequence information was previously available, were further characterized. Members of each of the 10 families are defective, and calculation of their integration dates suggested that most of them are likely to have been present within the human lineage since it diverged from the Old World monkeys more than 25 million years ago.

FIG. 1

Phylogenetic analyses of a 159-residue region of retroviral RT proteins. The trees were rooted on several gypsy LTR-retrotransposon sequences. To save space, multiple taxon names in some well-supported terminal clusters are not indicated. Instead, the number of taxa that actually clustered in that position are indicated on the taxon label. (a) Neighbor-joining tree with branch lengths proportional to the degree of divergence between the sequences. Figures on each branch represent percentage bootstrap support from 1,000 replicates; asterisks show support of at least 95%. HERVs are indicated by black circles. Novel HERV families are boxed. Previously described HERV families are shaded gray if the sequence of the pol gene was not available, and they were classified according to their closest cosmid matches (see Table 1). Elements in parentheses are likely to cluster with the adjacent family. Primer sites indicate the tRNA to which the viral PBS is most similar; these are boxed when this similarity is first reported in this study. (b) Strict consensus of 1,200 maximum-parsimony trees. The figure is labeled as in panel a except that branch lengths are not proportional to the divergence between the taxa. Also note that in contrast to panel a, maximum-parsimony data sets were pruned prior to analysis to reduce computation times. Thus, the first figure presented on some of the taxon labels represents the actual number of elements included in the analysis, whereas the number in parentheses represents the total estimated number that would cluster with the particular family if the additional elements had also been included in the data set (based on at least 95% bootstrap support by the neighbor-joining method; see Materials and Methods).

FIG. 1

Phylogenetic analyses of a 159-residue region of retroviral RT proteins. The trees were rooted on several gypsy LTR-retrotransposon sequences. To save space, multiple taxon names in some well-supported terminal clusters are not indicated. Instead, the number of taxa that actually clustered in that position are indicated on the taxon label. (a) Neighbor-joining tree with branch lengths proportional to the degree of divergence between the sequences. Figures on each branch represent percentage bootstrap support from 1,000 replicates; asterisks show support of at least 95%. HERVs are indicated by black circles. Novel HERV families are boxed. Previously described HERV families are shaded gray if the sequence of the pol gene was not available, and they were classified according to their closest cosmid matches (see Table 1). Elements in parentheses are likely to cluster with the adjacent family. Primer sites indicate the tRNA to which the viral PBS is most similar; these are boxed when this similarity is first reported in this study. (b) Strict consensus of 1,200 maximum-parsimony trees. The figure is labeled as in panel a except that branch lengths are not proportional to the divergence between the taxa. Also note that in contrast to panel a, maximum-parsimony data sets were pruned prior to analysis to reduce computation times. Thus, the first figure presented on some of the taxon labels represents the actual number of elements included in the analysis, whereas the number in parentheses represents the total estimated number that would cluster with the particular family if the additional elements had also been included in the data set (based on at least 95% bootstrap support by the neighbor-joining method; see Materials and Methods).

FIG. 2

Partial RT sequence alignment of representative members of each of the 22 independent HERV families discussed in this report. The sequence of this region is not available for the prototype members of some of the HERV families. In these cases, the cosmid from which the sequence was derived is also shown. The similarity of the cosmid sequence to the prototypic family member and its location within the cosmid are indicated in Tables 2 and 3, respectively.

FIG. 3

LTR sequences of five novel HERV families (a) and four partially characterized families (b). When both the 5′ and 3′ LTRs could be identified, they were aligned against each other, with dashes representing missing residues. The PPT (before the start of the 3′ LTR) and PBSs (after the end of the 5′ LTR) are underlined, as are the direct repeats flanking each end of the element and the inverted repeats flanking each LTR. The promoter and polyadenylation signals are boxed. In some cases, not all these features could be identified for each element (or they differ slightly from the expected sequence). This is probably due to postintegration mutation. Similarly, this is also likely to account for the observed differences between the PBS and closest-match tRNA sequence shown below each alignment. The estimated integration dates of each HERV are also shown.

FIG. 4

Alignment of conserved amino acid motifs present within the HERV families identified in this report. The figures in parentheses indicate the cosmids from which the sequences were derived. GaLV, Jaagsiekte, HSV, and MuERV.L are provided for comparative purposes. (a) Gag alignment spanning nucleotide positions 1660 to 1713 within the GaLV sequence (13); (b) Gag alignment spanning positions 2074 to 2157; (c) Pro alignment spanning positions 2266 to 2490; (d) Int alignment spanning positions 4969 to 5166; (e) Env alignment spanning positions 7220 to 7504.

FIG. 5

Alignment of the dUTPase motif within the HERV.HML5 family (derived from the element within cosmid AC004536) with a human dUTPase (34).