Findings

In contrast to the vertical transmission of the genetic material from parents to offspring, the horizontal transfer (HT) is a process in which new genetic information is transmitted between different, sometimes distant, species[1, 2]. HT is likely to be one of the factors leading to the persistence of transposable elements (TEs) in eukaryotes[3–5], and complicating the evolutionary trees.

The detection of HT is mostly inferential, mainly based on the combination of two types of evidence: unusually high similarity between TE sequences from species that have long diverged from each other, and a limited distribution of one particular TE family within a group of species[6]. To date, numerous HTs have been detected in eukaryotes[6–10], but of particular interest are HTs across distant branches. A recent example of such a rare event is HT of hAT DNA transposon families between vertebrate and invertebrate species[11].

Here, we report two families of Mariner-type DNA transposons that have possibly undergone HT from insects to mammals. The first family, called Mariner_Tbel, was originally identified in the tree shrew (Tupaia belangeri), but families nearly identical to Mariner_Tbel were also found in the genome of another mammal, European hedgehog (Erinaceus europaeus), and in two ant species: red harvester ant (Pogonomyrmex barbatus) and Jerdon's jumping ant (Harpegnathos saltator) (Tables1 and2). Although the copy numbers and divergence vary between the families, the family consensus sequences reconstructed in each genome show a high level of identity to each other throughout the entire length (approximately 1.3 kb) (Table1). The lowest identity is found between the two families in E. europaeus and ant P. barbatus (95.84%), and the highest identity is found between the two ant families (98.45%). Therefore, unless otherwise stated, the four families in the genomes are referred collectively to as Mariner_Tbel families. Given the long divergence time between insects and mammals (approximately 1 billion years)[12–14], this high identity strongly indicates that HT took place during the evolutionary history of Mariner_Tbel families. This notion is consistent with the fact that mammal Mariner_Tbel sequences were found only in two distantly related mammalian species, even though over 30 mammalian genomes were sequenced to date.

Table 1 Pairwise identities (%) between the Mariner_Tbel consensus from mammals ( Tupaia belangeri , Erinaceus europaeus ) and insects ( Pogonomyrmex barbatus , Harpegnathos saltator )
Full size table
Table 2 Divergence of Mariner transposable element (TE) families in mammalian and insect genomes
Full size table

We then estimated the approximate ages of the four Mariner_Tbel families in each genome. In mammals, we compared the sequence divergences of Mariner_Tbel to an older Mariner-type family (TIGGER1), relatively common in the mammalian genomes (Table2). TIGGER1 elements are present in multiple copies in eutherian mammals, but only one or two degenerated copies were found in marsupial genomes, including Macropus eugenii, Monodelphis domestica, and Sarcophilus harrisii (Figure1A). Therefore, mammalian TIGGER1 families likely expanded after the split of marsupials and placentals (190 million years ago (MYA)), but before the placental radiation (approximately 100 MYA)[15]. In the genome of the tree shrew and European hedgehog the divergence of the TIGGER1 family is 21.2 ± 2% and 28.0 ± 3%, respectively (Table2). Therefore, based on the divergence of Mariner_Tbel in the two mammal genomes (15.0 ± 2% and 19.4 ± 3%, respectively), the ages of Mariner_Tbel in the two mammals were calculated to be approximately 134 to 70 MYA and approximately 131 to 69 MYA, respectively. Because it is unlikely that the mammalian Mariner_Tbel expanded in the common ancestor of placental mammals before 100 MY, we adjusted the ages to be 100 to 70 MYA and 100 to 69 MYA, respectively (see Figure1A and[15]).

Figure 1
figure 1

(A) Distribution of five Mariner families ( Mariner_Tbel , Mariner1_BT , TIGGER1 , Mariner-28_SIn , and Mariner-N1_CPe ) in placental mammals, marsupials, and insects. The phylogeny is adopted from published trees of insects and mammals[12–15]. Marsupials are represented by three species (Monodelphis domestica, Macropus eugenii, Sarcophilus harrisii). The scale bar on the bottom left indicates time of evolutionary branching in the tree. Species with genomic sequences available from public databases are highlighted in orange. (B) Phylogenetic position of the two horizontally transferred Mariners (Mariner_Tbel and Mariner1_BT: red color), relative to other closely related Mariners. Most of the families are from insects with color-coded branches: ants (green), bees (orange), flies (pink) and other insects (blue). The remaining branches (black) are from mammals and planarian (SMAR7). Except for a few individual sequence segments (with accession numbers), all other families are represented by consensus sequences deposited in Repbase (except for highly similar copies). (C) Structural relationship between Mariner1_BT and Mariner-N1_CPe family. The species abbreviations are: ACe (Atta cephalotes), AEc (Acromyrmex echinatior), AFl (Apis florea), AMe (Apis mellifera), BTe (Bombus terrestris), BT (Bos taurus), CA (Chymomyza amoena), CFl (Camponotus floridanus), Del (Drosophila elegans), DEr (Drosophila erecta), DF (Drosophila ficusphila), EEu (Erinaceus europaeus), HSal (Harpegnathos saltator), LHu (Linepithema humile), MRo (Megachile rotundata), PBa (Pogonomyrmex barbatus), SIn (Solenopsis invicta), SMAR7 (Schmidtea mediterranea), Tbel (Tupaia belangeri), CPe (Carollia perspicillata).

Full size image

In the ant genomes, no Mariner family was yet identified as unambiguously present in the common ancestor of all ant species. Among potential candidates are the oldest known Mariner families present in some of the ant genomes (for example, Mariner-28_SIn or Mariner-94_HSal; Figure1). These small families may have expanded in the common ancestor of all ant species (140 MYA)[13], assuming that they were lost in some ant species. Alternatively, these old families might have expanded in some ant species after they split from their common ancestor. Under either scenario, the outermost ages when the two ant Mariner_Tbel families expanded could be still estimated by comparing their diversities with the diversity of Mariner-28_SIn (Table2). Based on that, Mariner_Tbel family in the red harvester ant (P. barbatus) and Jerdon's jumping ant (H. saltator) expanded at most approximately 43 and approximately 50 million years ago, respectively.

The above age estimates suggest that the two ant Mariner_Tbel families are possibly younger than the mammalian Mariner_Tbel families. However, the history of Mariner_Tbel can be traced further back in ants and their insect relatives than in mammals. Individual Mariner_Tbel-like elements from distinct families, such as AEAQ01009575, AEAB01001421 and AFJA01006902 (Figure1B), were also found in the genomes of two other ants (Solenopsis invicta and Camponotus floridanus) as well as in the alfalfa leafcutting bee (Megachile rotundata). These Mariner_Tbel-like sequences and Mariner_Tbel sequences form a single lineage in the phylogenetic tree (Figure1B), with the bee sequences in a more ancestral position (Figure1B). The topology of this particular lineage mirrors the evolutionary history of the ant and bee species (Figure1A). Furthermore, Figure1B indicates that the Mariner_Tbel family and many other similar Mariner families in ants and other insects shared a common ancestral sequence. These observations suggest the ancestor of ants Mariner_Tbel may have been present in some ant or other insect species very long time ago, probably as far back as the common ancestor of bees and ants (approximately 150 MYA)[14]. Thus, the mammalian Mariner_Tbel families probably originated from HTs from insects to mammals through some unknown vectors. Given that the two mammals belong to two distinct lineages, Mariner_Tbel in tree shrew and hedgehog may represent two independent HTs (Figure1A). Notably, we cannot rule out the possibility that the Mariner_Tbel families in one of the two ant species, or both, also originated by HTs. This possibility is suggested by two facts: (a) the relatively young ages (at most approximately 43 to 50 MYA) of the two families, (b) the high identity (98.5%) between the two family consensus sequences, even H. saltator and P. barbatus diverged from each other approximately 100 million years ago[13]. Among insect species, frequent HTs have been documented in flies[16]. Alternatively, Mariner_Tbel sequences could have survived for a very long time in either of the two ant genomes before the most recent family expansions.

In addition to mammalian Mariner_Tbel families, Mariner1_BT DNA families might also have originated by HT from insects (Figure1A). We were able to obtain high quality consensus of Mariner1_BT from bovine species (Bos taurus, Bos indicus, Bos grunniens, Bubalus bubalis) and bottlenosed dolphin Tursiops truncatus (Table2). All the derived consensus sequences show similar lengths (approximately 1,280 bp), and high pairwise identities throughout the entire length (>98%). Blast screening against National Center for Biotechnology Information (NCBI) databases using Mariner1_BT consensus sequence as query also detected this family in several other mammalian species, including one bat species, Carollia perspicillata (Seba's short-tailed bat), additional ruminants, and whale (Table3). These BlastN hits show similar score and query coverage (>80% identity to the consensus and >90% coverage). In summary, Mariner1_BT type TEs were found only in three taxonomic groups to date: ruminants, tooth whales (Odontoceti), and New World leaf-nosed bats (Phyllostomidae) (Figure1A). Notably, in C. perspicillata (short-tailed fruit bat) and Desmodus rotundus (vampire bat), we also detected a family of non-autonomous DNA transposon, called Mariner-N1_CPe, which was likely derived from the bat Mariner1_BT family (Figure1C).

Table 3 Mariner1_BT sequences detected in mammals
Full size table

Remarkably, the other closest relatives of Mariner1_BT are all found in ant species: Mariner1_BT coclusters significantly (bootstrap = 83) with three other ant Mariner families (Mariner-5_ACe, Mariner-28_SIn and Mariner-35_HSal) (Figure1B). Given the vast diversity of Mariners found in insects (Figure1B), and the confined distribution of Mariner1_BT in mammals, we propose Mariner1_BT family could also originate from a horizontally transferred insect-like element. Using a similar method above, that is, based on the family divergence and mammalian phylogeny (Table2 and Figure1A), we estimated the ages of bovine Mariner1_BT to be 90 to 85 MYA, and 90 to 63 MYA for dophin T. truncatus Mariner1_BT family. The age of Mariner1_BT in bat could not be estimated due to insufficient data. We also could not determine if HT happened in mammals more than once, because the three taxonomic groups that include Mariner1_BT are relatively close.

In summary, this is the first report of two cases of horizontally transferred Mariner elements (Mariner_Tbel and Mariner1_BT) between insects and mammals. Previously, four families of DNA transposons from the hAT superfamily were also found to be involved in multiple waves of HT between insects and other vertebrates including mammals[11]. This could partially be attributed to the fact that insects are the largest and the most diverse group of invertebrate animals on earth. While insects are the most likely source of the horizontally transferred transposons, the original source or possible intermediaries, such as parasitic insects[11] or viruses, remain unclear. This is complicated by the possibility that recurrent HTs of related Mariner elements are likely to take place between different insects[16]. The role of viruses in HT proposed some time ago[17] still remains to be understood. As more genome sequence data become available, more mechanistic details on HT between mammals and insects are likely to emerge.

Methods

Mariner transposable elements from Repbase (http://www.girinst.org/repbase/) were used as an initial query to screen Mariners in diverse genomes available at NCBI (National Center for Biotechnology Information:http://www.ncbi.nlm.nih.gov/). Family consensus sequences were constructed whenever possible. The copy numbers in each family were determined by BLASTN using consensus sequences as queries. Sequence divergence within each family was assessed by the average pairwise k-distance (Kimura two-parameter model) between individual insertions and the corresponding consensus sequences. The k-distance was calculated using the software MEGA 4[18]. For a given family, individual sequences used in k-distance calculation were randomly chosen from the family members; in most cases individual sequences matched >70% of the consensus length.

We used Mariner_Tbel and Mariner1_BT as BLASTN queries against Repbase to select top scoring TE entries for phylogeny analysis. Individual sequences selected from GenBank were also used in the tree if Repbase consensus sequences were not available. The sequence alignments are shown in Additional file1. The alignments were done using the online MAFFT server (http://mafft.cbrc.jp/alignment/software/). The phylogeny tree was inferred using MEGA 4[18], using the neighbor joining (NJ) method and k-distances. Branch support was estimated using 1,000 bootstrap replicates.