Evolution and gene capture in ancient endogenous retroviruses - insights from the crocodilian genomes

Abstract

Background: Crocodilians are thought to be hosts to a diverse and divergent complement of endogenous retroviruses (ERVs) but a comprehensive investigation is yet to be performed. The recent sequencing of three crocodilian genomes provides an opportunity for a more detailed and accurate representation of the ERV diversity that is present in these species. Here we investigate the diversity, distribution and evolution of ERVs from the genomes of three key crocodilian species, and outline the key processes driving crocodilian ERV proliferation and evolution.

Results: ERVs and ERV related sequences make up less than 2% of crocodilian genomes. We recovered and described 45 ERV groups within the three crocodilian genomes, many of which are species specific. We have also revealed a new class of ERV, ERV4, which appears to be common to crocodilians and turtles, and currently has no characterised exogenous counterpart. For the first time, we formally describe the characteristics of this ERV class and its classification relative to other recognised ERV and retroviral classes. This class shares some sequence similarity and sequence characteristics with ERV3, although it is phylogenetically distinct from the other ERV classes. We have also identified two instances of gene capture by crocodilian ERVs, one of which, the capture of a host KIT-ligand mRNA has occurred without the loss of an ERV domain.

Conclusions: This study indicates that crocodilian ERVs comprise a wide variety of lineages, many of which appear to reflect ancient infections. In particular, ERV4 appears to have a limited host range, with current data suggesting that it is confined to crocodilians and some lineages of turtles. Also of interest are two ERV groups that demonstrate evidence of host gene capture. This study provides a framework to facilitate further studies into non-mammalian vertebrates and highlights the need for further studies into such species.

Figure 1

Classification and likely relationships between the CrocERV groups and ERVs from other species. Maximum likelihood phylogenies were created from the RT domain of the crocodilian consensus sequences and a selection of sequences deposited in Repbase and published sequences. Part (a) is the entire RT tree, while (b) and (c) are expanded versions of ERV3 and 4, and ERV1 respectively. The complete version of (a) including sequence IDs is presented as Additional file 6: Figure S2. Symbols represent the taxa from which the sequences were derived. The numbers of CrocERV groups are shown outside of corresponding consensus sequences. Major ERV and retroviral groups, and the Gypsy elements, are indicated by brackets. Numbers within the phylogeny indicate aLRT values greater than 90%. The scale bar indicates branch length.

Figure 2

Presence and absence of dUTPase and env is variable between lineages within ERV3. Maximum likelihood phylogenies were created from the RT domain of Crocodilian ERV3 and ERV3 sequences deposited in Repbase. The ERV3 lineage encoding dUTPase is shaded in grey. Elements encoding a recognisable env are indicated by boxes. Symbols represent the taxa from which the sequences were derived. The numbers of CrocERV groups are shown outside of corresponding consensus sequences. Major retroviral groups are indicated brackets. Numbers within the phylogeny indicate aLRT values greater than 90%. The scale bar indicates branch length.

Figure 3

Reconstructed KIT-ligand proteins from CrocERV29 and a simplified diagram of the provirus. Maximum Likelihood phylogenies and sequence alignments were created from the reconstructed KIT-ligand proteins encoded by one lineage within CrocERV29. ERV IDs are provided within the tree and the alignment (see also Additional file 7). Four letter sequence names indicate the crocodilian species (‘Amis’ , A. mississippiensis; ‘Asin’ , A. sinensis; ‘Cpor’ , C. porosus; ‘Ggan’ , G. gangeticus). Shaded columns indicate the positions of indels within the alignment. Numbers within the tree indicate statistical support for the branches and the scale bars indicate branch length. Proviral structure is not to scale.

Figure 4

Reconstructed nectin3 proteins from CrocERV31 and a simplified diagram of the provirus. Maximum Likelihood phylogenies and sequence alignments were created from the reconstructed nectin3 proteins encoded by one lineage within CrocERV31. ERV IDs are provided within the tree and the alignment (see also Additional file 8). ‘Amis’ stands for A. mississippiensis and ‘Asin’ for A. sinensis. Shaded columns indicate the positions of indels within the alignment. Numbers within the tree indicate statistical support for the branches and the scale bars indicate branch length. Proviral structure is not to scale.