A single-copy IS5-like transposon in the genome of a bdelloid rotifer.

In the course of sequencing telomeric chromosomal regions of the bdelloid rotifer Adineta vaga, we encountered an unusual DNA transposon. Unlike other bdelloid and, more generally, eukaryotic transposable elements (TEs), it exhibits similarity to prokaryotic insertion sequences (ISs). Phylogenetic analysis indicates that this transposon, named IS5_Av, is related to the ISL2 group of the IS5 family of bacterial IS elements. Despite the apparent intactness of the single open reading frame coding for a DDE transposase and the perfect identity of its 213-bp terminal inverted repeats (TIRs), the element is present in only one copy per diploid genome. It does not exhibit any detectable levels of transcription, so that its transposase gene appears to be silent in the bdelloid host. Although horizontal transfers of TEs between kingdoms are not known to happen in nature, it appears likely that IS5_Av underwent integration into the A. vaga genome relatively recently, but was not successful in adapting to the new host and failed to increase in copy number. Alternatively, it might be the only known member of a novel eukaryotic DNA TE superfamily which is so rare that its other members, if any, have not yet been identified in eukaryotic genomes sequenced to date.

Introduction

Transposable elements (TEs) are omnipresent in all three domains of life: Bacteria, Archaea, and Eukarya. Of the two major types of TEs, that is, retrotransposons and DNA transposons, the latter are particularly prone to horizontal transmission (see Silva et al. 2004 for review). First inferred from patchy distribution of P-elements in Drosophila spp. (Kidwell 1983; Daniels et al. 1990) and mariner transposons in insects (Robertson 1993), horizontal gene transfer (HGT) was subsequently suggested to constitute the primary mode of survival for DNA transposons of this type (Robertson and Lampe 1995; Hartl et al. 1997). According to this scenario, a DNA TE enters a new host and proliferates within its genome, giving rise to multiple copies, which then undergo silencing, mutational decay, and eventually get erased from the host genome, so that the long-term persistence of a TE depends on the ability of a functional copy to escape into a new host. During such horizontal escapes, however, TEs do not typically cross domain boundaries: no cases of recent HGT between Bacteria (or Archaea) and Eukarya have been documented to date, and naturally occurring cases incompatible with vertical inheritance typically involve movements between taxonomically close groups (Daniels et al. 1990; Diao et al. 2006; Pace et al. 2008; reviewed in Feschotte and Pritham 2007). Interkingdom transfers, for example, from animals into protists and bacteria, have been accomplished in the laboratory (Gueiros-Filho and Beverley 1997; Rubin et al. 1999). In these cases, however, expression of the transposase gene involved its placement under a heterologous promoter known to be functional in the new host (e.g., Rubin et al. 1999; Zhang et al. 2000). Fundamental differences in prokaryotic and eukaryotic gene expression, such as incompatible basal promoter elements, translation start sites, or coupling between transcription and translation, may, at least in part, account for the barriers to successful interdomain transfers.

Here we describe an apparent case of horizontal transfer of an IS5-like DNA transposon, which was found during cloning and sequencing of telomeric regions in the bdelloid rotifer Adineta vaga (Gladyshev et al. 2008). As telomeric regions of bdelloid genomes contain numerous foreign genes originated in bacteria, fungi, or plants, it was therefore not as surprising as it could have been in the absence of a massive HGT that a DNA transposon of apparently bacterial origin was also able to find its way into the bdelloid genome. Interestingly, however, this TE appears to have been much less successful in adapting to the new host environment than certain bacterial protein–coding genes, such as the D-Ala-D-Ala ligase gene that is transcribed, spliced, and encodes a functional protein (Gladyshev et al. 2008). As described below, IS5_Av integration apparently represented a one-time event and did not result in proliferation in the new host, possibly due to the lack of compatibility with the transcriptional–translational apparatus or damage to the element. Our study indicates that in nature, given the opportunity, TEs highly similar to prokaryotic could incorporate into metazoan chromosomes. Alternatively, IS5_Av could represent the sole member of a hitherto unknown eukaryotic TE superfamily.

Materials and Methods

Reverse transcription polymerase chain reaction (RT-PCR) and Southern blot hybridization experiments were done as described in Gladyshev and Arkhipova (2007). The following primers were used for PCR amplification: TIR, GAGTAAAGTTTTGTGTTCACTG; RT-F1, GTCAGAATGGAGTCGCAACA; and RT-R1, TCGGTGGTTACAACAATCACA. AvDdl primers were as in Gladyshev et al. (2008). Southern blots and genomic library were probed with the 32P-labeled 1,637-bp fragment amplified by the TIR primer and with the 534-bp fragment amplified with the RT-F1–RT-R1 primer pair. The IS5 data set was assembled from 1) IS Finder database entries (www-is.biotoul.fr), including at least three representatives from each subgroup of the IS5 group, and all representatives from the ISL2 group; 2) Repbase entries (www.girinst.org/repbase/index.html), including representatives of the Harbinger/PIF and IS4EU/ISL2EU superfamilies; 3) top GenBank hits from diverse archaea and bacteria identified by Blast search; and 4) IS5-like families in protist genomes identified by Blast search. Amino acid sequences were aligned with T-Coffee (Notredame et al. 2000) and Clustalw implemented in MEGA4 (Tamura et al. 2007). Neighbor-joining and minimum evolution analyses were done with MEGA4 (p-distance or Poisson correction; pairwise deletion; 1,000 bootstrap replications). The value of the parameter gamma was determined by ProtTest (Abascal et al. 2005). Alternative start codons were analyzed using the EasyGene server (www.cbs.dtu.dk/services/EasyGene/). Search for the TSD–TIR…TIR–TSD combination (TSD, target site duplication) was performed on a 1,419-bp region upstream and a 784-bp region downstream from the hobo open reading frame (ORF; between the stop codon of the upstream NHL gene and the 5′ long terminal repeat (LTR) of a downstream TE), using a custom Perl script allowing one or two mismatches. Only cases with 8-bp TSD were considered, as hobo_Av is phylogenetically close to insect hobo elements and therefore should not deviate from the general rule specifying 8-bp TSDs for hobo-like TEs (Rubin et al. 2001). Unusual hAT elements yielding atypical 5- and 6-bp TSD (Putnam et al. 2007) belong to very distant and well-separated clades (data not shown).

Results

Structure and Copy Number

Structural organization of the A. vaga IS5-like transposon is shown in figure 1. The transposon harbors a single ORF exhibiting similarity to transposases of the IS4–IS5 families (table 1; pfam01609:Transposase_11, which includes prokaryotic transposases for IS4, IS421, IS5377, IS427, IS402, IS1355, and IS5, and is a member of the DDE megafamily of transposases/integrases), and which contains all of the highly conserved residues required for catalytic activity of these enzymes. A multiple sequence alignment of IS5-like transposases is provided as Supplementary data, Supplementary Material online. The IS5_Av transposon is flanked by 213-bp perfect TIRs, including a 2- to 6-bp ambiguity in the outermost nucleotides CATATG…CATATG, all or part of which could also be regarded as a 2-, 4-, or 6-bp TSD (TA, ATAT, or CATATG). Such ambiguity can usually be resolved by comparing different flanking sequences from additional genomic TE copies. To our surprise, we were unable to identify any such copies, neither by exhaustive genomic library screens (data not shown) nor in Southern analyses using two different restriction enzyme combinations (fig. 2). The single band on a Southern blot of A. vaga genomic DNA digested with PvuII, which has no recognition sites within IS5_Av, demonstrates that this is the only copy present in the genome. The size of the band (7.0 kb) coincides with the expected size of the PvuII fragment harboring IS5_Av on the sequenced telomeric fosmid, ruling out possible contamination or misassembly in the course of random shotgun sequencing of fosmid subclones. Similarly, XhoI–HpaI digest of genomic DNA yields an expected 4-kb band. In addition, sequencing of cloned and total PCR products obtained with the TIR primer, as expected, revealed no nucleotide sequence polymorphisms, which might have been observed if different full-length copies existed in the genome but were not present in the genomic library (data not shown).

F

Structure (A) and genomic environment (B) of IS5_Av. (A) Within the transposase (TPase) ORF, the positions 788; 884; 1,040; and 1,714 denote the TTG, GTG, ATG, and TGA codons, respectively. Also shown are the positions of the DDE catalytic residues, and the primers used for PCR amplification of the internal region between TIRs and for RT-PCR shown in figure 2 (RT-F1, RT-R1) (half arrows). A small gray box denotes a putative 63-aa ORF extension in case of a −1 frameshift. Hypothetical 10-bp secondary TIRs are shown by tiny arrowheads. Primers for the outermost part of TIRs (unnamed half arrows) failed to yield PCR products, apparently due to PCR interference with the DNA secondary structure. Scale bar, 100 bp. (B) A 40-kb region (region 63135…103135 from GenBank accession no. EU643477) near the Adineta vaga telomere K_A (designated by the letter T), with nine TE insertions (table 2) and five protein-coding ORFs, of which histidine-ammonia lyase (HAL, 3′ truncated) is presumably of bacterial origin, acyl-CoA synthetase (3′ truncated) is of metazoan origin, and the rest is of uncertain origin. Coding sequences with transcriptional orientation toward the telomere are shown above the line, whereas those with transcriptional orientation away from the telomere are shown below the line. Two AATAAA signals in the hobo 3′ UTR are denoted by letters A. The intergenic region scanned for TSD–TIR combinations is denoted by a thick line. Scale bar, 1 kb.

Table 1

BlastP Similarity Scores for the IS5_Av Query Used to Search the Nonredundant GenBank Database (A) and the IS Finder Database (http://www-is.biotoul.fr/) (B)

Sequences Producing Significant Alignments:			Score (Bits)	E Value
(A) Database: All nonredundant GenBank CDS translations + PDB + SwissProt + PIR + PRF excluding environmental samples from WGS projects (7,031,513 sequences; 2,427,977,331 total letters, Query = IS5_Av, Length = 310)
ref | YP_594226.1 | transposase, IS4 [Deinococcus geothermalis D…			81.3	1e−13
ref | ZP_02737947.1 | transposase, IS4 [Gemmata obscuriglobus UQ…			77.8	1e−12
ref | ZP_02732909.1 | transposase, IS4 [Gemmata obscuriglobus UQ…			76.6	2e−12
ref | ZP_03176005.1 | putative transposase [Streptomyces sp. SPB…			75.1	7e−12
ref | ZP_03143215.1 | transposase IS4 family protein [Cyanothece…			70.5	1e−10
ref | ZP_03141860.1 | transposase IS4 family protein [Cyanothece…			69.3	3e−10
ref | ZP_03178672.1 | putative transposase [Streptomyces sp. SPB…			68.6	6e−10
ref | ZP_03143523.1 | transposase IS4 family protein [Cyanothece…			68.6	6e−10
ref | ZP_02942299.1 | transposase IS4 family protein [Cyanothece…			68.6	7e−10
ref | ZP_02939447.1 | transposase IS4 family protein [Cyanothece…			68.2	7e−10
ref | ZP_01731099.1 | hypothetical protein CY0110_01550 [Cyanoth…			68.2	9e−10
ref | ZP_02942406.1 | transposase IS4 family protein [Cyanothece…			67.4	1e−19
(B) Database: IS protein database (3,402 sequences; 1,042,201 total letters, Query= IS5_Av, 310 letters)
Sequences Producing Significant Alignments	IS Family	Group	Origin			Score (Bits)	E (Value)
ISDge6	IS5	ISL2	Deinococcus geothermalis			77	1e−15
IS1515	IS5	ISL2	Streptococcus pneumoniae I41R			62	3e−11
IS702	IS5	ISL2	Calothrix sp. PCC7601			62	4e−11
ISL2A	IS5	ISL2	Lactobacillus helveticus LH27			54	1e−08
ISL2	IS5	ISL2	L. helveticus LH28			54	1e−08
ISMae4	IS5	ISL2	Microcystis aeruginosa			53	2e−08
IS493	IS5	ISL2	Streptomyces lividans CT2			45	5e−06
IS1381	IS5	ISL2	S. pneumoniae			42	4e−05
IS470	IS5	ISL2	Streptomyces coelicolor A3(2) M145			42	5e−05
IS1373	IS5	ISL2	S. lividans 66 1326.32			40	2e−04
IS1381A [V]	IS5	ISL2	Streptococcus agalactiae A909			39	3e−04
IS112	IS5	ISL2	Streptomyces albus G J1147			39	4e−04

NOTE.—Two IS5-derived ORFs from Naegleria gruberi yield the same top hits as IS5_Av. Note that most of the GenBank entries are misannotated as IS4 transposases.

F

Copy number (A) and expression (B) of IS5_Av. (A) Ethidium bromide staining (left panels) and Southern blot hybridization with the 32P-labeled IS5_Av probe (right panels) of the Adineta vaga genomic DNA digested with PvuII or XhoI/HpaI, as indicated. (B) RT-PCR with IS5_Av (left) and Ddl_Av (right) forward and reverse primers. Shown are the sizes of the corresponding PCR products, which for the Ddl gene differ by 50 bp due to splicing. RT-, no reverse transcriptase added; RT+, addition of SuperScriptII; DNA, control genomic DNA amplification. M, 1 kb+ ladder (Invitrogen).

Table 2

Top BlastP Database Hits for Eight TEs from the Adineta vaga Telomere K_A (EU643477) Shown in figure 1

TE Superfamily (Position)	Top BlastP Hit, Species	E-Value	% Identity (Similarity)	CD Hits	In-Frame Stops/Frameshifts/Indels
mariner (66428…68107)	Transposase, Pachygrapsus marmoratus (coastal crab)	5e−61	36% (54%)	Transposase_1	2/1/0
piggyBac (68182…70303)	Transposase, Acyrthosiphon pisum (pea aphid)	9e−89	42% (63%)	none	1/0/0
mariner (71477…71555, 72009…72645)	Avmar1 transposase, Adineta vaga (bdelloid rotifer)	1e−78	78% (83%)	Transposase_1	1/1/2
hobo (77307…78830, 80417…81399)	Transposase, Bactrocera tryoni (Queensland fruit fly)	2e−39	24% (44%)	hATC superfamily	1/1/1
mariner (78833…80416)	Transposase, Caenorhabditis elegans (roundworm)	2e−40	31% (50%)	none	0/0/0
IS5 (81508…83527)	Transposase, Deinococcus geothermalis (radio-resistant micrococci)	1e−13	25% (44%)	Transposase_11	0/0/0
LTR retrotransposon (gag; pol; env) (84198…93163)	Retrotransposon gag protein, Asparagus officinalis (Liliopsida); pol polyprotein, Danio rerio (zebrafish); env-like transmembrane glycoprotein, coronavirus	1e−06; 0.0; 3e−01	21%(44%); 39% (57%) 27% (47%)	Retrotrans_gag; RVP superfamily; RT_LTR (RT-like superfamily); rve superfamily	0/0/0; 0/2/0; 0/0/0
piggyBac (100845…102137)	Transposase, Nasonia vitripennis (jewel wasp)	6e−18	31% (54%)	none	6/0/1

NOTE.—Also listed are hits to the CD database, and disruptions in ORFs, if any.

Genomic Environment

IS5_Av is located in a subterminal region of the A. vaga telomere K_A (Gladyshev et al. 2008), in an environment rich in eukaryotic TEs that are typical of bdelloid rotifers (Arkhipova and Meselson 2005; Gladyshev et al. 2007). This particular region has no homologous partner among A. vaga chromosomes, because collinearity between homologous telomeres K_A and K_B begins ∼50 kb proximally to IS5_Av (see fig. S1 in Gladyshev et al. 2008), and therefore, IS5_Av exists as the only copy per diploid genome. In figure 1, it may be seen that this element is located in an exceptionally TE-rich environment: It is apparently inserted into the 3′ untranslated region (UTR) of a hobo-like transposon and is surrounded by three mariner-like, one retrovirus-like, and two piggyBac-like transposons. The distance from IS5_Av to the chromosome end is 19.6 kb. The hobo element has several defects in its ORF, and therefore likely represents an ancient insertion; other TEs also carry in-frame stop codons, frameshifts, or indels (table 2). Interestingly, each mariner or piggyBac copy belongs to a different subfamily within the corresponding superfamily, underscoring the considerable diversity of DNA TEs in bdelloids.

In the absence of other copies similar in sequence to the hobo element in figure 1, we sought to identify its exact boundaries by scanning the immediate flanks (the regions between the hobo ORF and the next adjacent element) for the presence of an 8-bp TSD combined with a short inverted repeat sequence. The scanned region contained only one potential TSD–TIR–TIR–TSD combination (8-bp TSD + 7-bp TIRs with one mismatch, aattgattAAATAAT…ATcATTTaattgatt) (fig. 1). This boundary yields a hobo ORF framed by approximately 600-bp UTRs, which is very similar to the UTR length in other known rotifer hAT hobo families (Arkhipova and Meselson 2005 and unpublished data), thereby placing IS5_Av into the hobo 3′ UTR within 108 bp from its ORF end. Alternatively, one of the TIRs could have undergone deletion, in which case the boundaries would be difficult to define.

One could also consider a scenario under which the 213-bp TIRs belong to a different nonautonomous foldback-like element, with IS5_Av subsequently inserted between these TIRs. We searched for an additional, shorter internal TIR-TSD combination and could not find any putative TIRs longer than 10 bp (GCGACTGAAT…ATTCAGTCGC; fig. 1); if these internal TIRs are the true TIRs of IS5_Av, they are unusually short and are not surrounded by TSD.

Phylogenetic Placement

In contrast to the neighboring elements (and all other TEs previously identified in bdelloid rotifers), which are typical of eukaryotes and yield significant Blast hits to other eukaryotic TEs from the corresponding superfamilies (hAT, piggyBac, or mariner/Tc; table 2), IS5_Av appears quite different: its top database hits come from bacteria and not from eukaryotes (tables 1 and 2). The putatively bacterial origin indicated by similarity scores is also suggested by phylogenetic analysis (fig. 3). It may be seen that IS5_Av from A. vaga clusters with a specific subgroup of IS5-like TEs, namely, the ISL2 group, which is characterized by 15- to 40-bp TIRs, and TSD ranging between 0, 2, 3 and 7 bp, with preference for TA- or TNA-containing targets (Chandler and Mahillon 2002; www-is.biotoul.fr). There are two known eukaryotic DNA TE superfamilies that are related to the IS5 group of bacterial TEs: Harbinger/PIF (containing Transposase_11 converted domain, CD; 3-bp TSD) and IS4EU/ISL2EU (yielding no transposase-related CD hits; 2-bp TSD) (Kapitonov and Jurka 2004, 2007; Zhang et al. 2004), which form distinct eukaryotic clades in figure 3. The A. vaga IS5-like element, however, does not fall into any of these clades and neither does it contain any additional ORFs, which represent one of the defining characteristics of Harbinger/PIF and IS4EU/ISL2EU superfamilies. Interestingly, all members of the ISL2 subgroup from the IS Finder database fall into two clades, rather than into a single ISL2 clade. We therefore designated the clade not containing ISL2 as the IS493 group, by the name of its first described representative.

F

Phylogenetic placement of IS5_Av. Shown is a Neighbor-Joining phylogram including representatives of different groups of the IS5 megafamily, which contains six known groups of the bacterial IS5 family (IS5, IS903, ISH1, IS1031, IS427, and ISL2) and two known eukaryotic IS5-like superfamilies (Harbinger/PIF and IS4EU/ISL2EU). Also shown are the newly designated groups ISL2PR (from various protists) and IS493 (members of which are assigned to ISL2 in the IS Finder database). Bootstrap support values for the major named groups, obtained by Neighbor-joining and minimum evolution methods, respectively, are as follows: ISL2, 48/58; IS493, 90/89; IS5, 100/100; IS903, 100/100; ISH1, 100/100; IS1031, 100/100; IS427, 100/100; ISL2PR, 34/43; IS4EU/ISL2EU, 100/100; and Harbinger/PIF, 51/47. The bacterial IS5, IS903, ISH1, IS1031, and IS427 groups can be joined into a well supported (97/87) supergroup. Support for IS5_Av/Naegleria branch is 51/64. Eukaryotic species are in boldface, and branches leading to them are shown by thick lines. Scale bar, 0.1 amino acid substitutions per site. For multiple sequence alignment, see supplementary data, Supplementary Material online.

Several IS5-like (Transposase_11 CD-containing) ORFs from Trichomonas vaginalis and certain other protists (Entamoeba invadens, Phytophthora spp., Aphanomyces euteiches, and Hyaloperonospora parasitica), instead of grouping with other eukaryotic clades, apparently form their own clade, albeit poorly supported (fig. 3). One of T. vaginalis ORFs (TVAG_485520) is single copy and is not framed by TIRs, suggesting domestication of this transposase-derived gene, whereas four others represent typical DNA TEs with a single ORF and 10–20 highly similar copies per T. vaginalis genome. In the Trichomonas families containing TVAG_135750 and TVAG_517480, the transposase domain is fused to the ubiquitin hydrolase-like cysteine peptidase (clan CA, family C19), whereas in the families containing TVAG_148970 and TVAG_413280, the transposase domain is not fused to any other domains. Despite the fact that these two groups of Trichomonas IS5-like families differ substantially in their amino acid sequence and their monophyly is not well supported, they share a peculiar feature: Their relatively long TIRs (180–290 bp) contain shorter imperfect hairpin regions, ultimately resulting in formation of an imperfect direct repeat embedded into each of the inverted repeats. It is conceivable that this entire group (designated ISL2PR) evolved upon invasion of the ancestral protistan genome by an IS5-like transposon. However, IS5_Av does not fall into this group either. Instead, it occupies a very basal position in the ISL2 clade, together with two IS5-derived genes from the heterolobosean Naegleria gruberi. Because this clade is not well supported, it is difficult to say whether an IS5-like element was transferred from bacteria into both A. vaga and N. gruberi or between a Naegleria-like protist and A. vaga.

Expression Analysis

Although the transposase-encoding IS5_Av ORF contains no in-frame stop codons or frameshifts that might indicate its nonfunctionality, it does exhibit an apparent deficiency: The first methionine is found at a position that would yield a protein that is 60–100 amino acid (aa) shorter than other transposases of the IS5 superfamily, and would not include the first region of similarity shared between all transposases at the N-terminus. As the use of splicing or ribosomal frameshifting could in principle lead to production of a full-length transposase in the apparent absence of a correctly positioned ATG codon, we decided to examine the transcriptional activity of the element. RT-PCR analysis of A. vaga RNA (fig. 2) demonstrates that the level of IS5_Av transcripts, if present, is below the detection limits of the technique, whereas the positive control, the A. vaga Ddl gene (Gladyshev et al. 2008), as expected, yields a band corresponding to a transcribed and spliced message. The lack of transcriptional activity may account for IS5 inability to give rise to additional copies, perhaps due to the incompatibility of the promoter sequences with the host transcriptional machinery. Alternatively, the element may have been inactivated by a deletion interfering with RNA stability. Finally, IS5_Av may have been inactivated by a frameshift due to a replication slippage in a T8 stretch at pos. 782, whereby deletion of a T would yield a 317-aa ORF, adding 7 aa to the uninterrupted 310-aa polypeptide sequence, or addition of a T would yield a 373-aa ORF. Such an extended ORF, however, does not exhibit additional similarity to the N-termini of any known elements, and there is no apparent reason for transcriptional inactivation due to a single frameshift.

We also considered the possibility that IS5_Av may lack an appropriately positioned ATG codon because of utilization of an alternative start codon in the previous host. To check the likelihood of this scenario, we attempted to evaluate the possible usage of alternative initiation codons based on gene prediction models. We scanned the IS5_Av sequence with EasyGene (Larsen and Krogh 2003; Nielsen and Krogh 2005), which uses a high-quality training set of genes coding for known conserved proteins from each genome to estimate hidden Markov models (HMM) of gene prediction for that particular genome. Of the 138 species with HMM models in the database, 48 did not yield a predicted gene in the 2,020-bp IS5_Av sequence; the first ATG codon at pos. 1,040 was predicted as the optimal start site in 46 species, and as a suboptimal start site in 5 species; the TTG codon at pos. 788 (yielding a 308-aa transposase) was found to be optimal in 24 species and suboptimal in 2 species, and the GTG codon at pos. 884 (yielding a 276-aa transposase) was found to be optimal in 3 species and suboptimal in 2 species (see table 3 for a list of alternative codons, and supplementary table S1, Supplementary Material online, for a complete list of predicted start codons). Thus, for approximately one-third of all database species that yielded gene predictions, the alternative start codons were predicted to be optimal. Although we do not yet know the identity of the putative donor species, it is worth noting that the genera such as Bacteroides, Burkholderia, and Pseudomonas, all of which are represented in table 3, have already been identified as putative donors of foreign genes to bdelloids (Gladyshev et al. 2008).

Table 3

List of Putative Donor Species Predicted by EasyGene to Yield a Full-Length Transposase Using an Alternative Start Codon as an Optimal

Model	Feature	Start	End	Score	Startc	Species	Taxonomy
SSW02	CDS	788	1714	8.60e−27	TTG	Synechococcus sp. WH 8102	Cyanobacteria; Chroococcales
PMMI02	CDS	788	1714	9.81e−22	TTG	Prochlorococcus marinus str. MIT 9313	Cyanobacteria; Prochlorales
NM02	CDS	884	1714	3.96e−11	GTG	Neisseria meningitidis serogroup B MC58	Proteobacteria; Betaproteobacteria; Neisseriales
SE02	CDS	788	1714	7.16e−11	TTG	Salmonella enterica subsp. enterica	Proteobacteria; Gammaproteobacteria; Enterobacteriales
NO02	CDS	788	1714	1.78e−10	TTG	Nostoc sp. PCC 7120	Cyanobacteria; Nostocales
PL01	CDS	788	1714	4.16e−10	TTG	Photorhabdus luminescens subsp. laumondii	Proteobacteria; Gammaproteobacteria; Enterobacteriales
LP02	CDS	788	1714	5.54e−09	TTG	Lactobacillus plantarum WCFS1	Firmicutes; Bacilli; Lactobacillales; Lactobacillaceae
ECC02	CDS	788	1714	1.66e−08	TTG	Escherichia coli CFT073	Proteobacteria; Gammaproteobacteria; Enterobacteriales
STT02	CDS	788	1714	1.93e−08	TTG	S. enterica subsp. enterica	Proteobacteria; Gammaproteobacteria; Enterobacteriales
STY02	CDS	788	1714	4.07e−08	TTG	Salmonella typhimurium LT2	Proteobacteria; Gammaproteobacteria; Enterobacteriales
ECE03	CDS	788	1714	8.15e−08	TTG	E. coli O157:H7 EDL933	Proteobacteria; Gammaproteobacteria; Enterobacteriales
WDM01	CDS	788	1714	1.12e−07	TTG	Wolbachia endosymbiont of Drosophila melanogaster	Proteobacteria; Alphaproteobacteria; Rickettsiales
LM02	CDSsub	884	1714	1.21e−07	GTG	Listeria monocytogenes EGD	Firmicutes; Bacilli; Bacillales
CDI01	CDS	788	1714	1.54e−07	TTG	Corynebacterium diphtheriae	Actinobacteria; Actinobacteridae; Actinomycetales
BS03	CDS	788	1714	4.32e−07	TTG	Bacillus subtilis	Firmicutes; Bacilli; Bacillales
CT02	CDS	788	1714	4.96e−07	TTG	Chlamydia trachomatis	Chlamydiae/Verrucomicrobia; Chlamydiae; Chlamydiales
ECO02	CDS	788	1714	5.17e−07	TTG	E. coli O157:H7	Proteobacteria; Gammaproteobacteria; Enterobacteriales
XC02	CDS	788	1714	6.53e−07	TTG	Xanthomonas campestris pv. campestris	Proteobacteria; Gammaproteobacteria; Xanthomonadales
BT02	CDS	788	1714	8.86e−07	TTG	Bacteroides thetaiotaomicron VPI-5482	Bacteroidetes/Chlorobi group; Bacteroidetes; Bacteroidales
VC02	CDS	788	1714	1.41e−06	TTG	Vibrio cholerae	Proteobacteria; Gammaproteobacteria; Vibrionales
BS03	CDSsub	884	1714	1.49e−06	GTG	B. subtilis	Firmicutes; Bacilli; Bacillales
CB02	CDSsub	788	1714	2.72e−06	TTG	Coxiella burnetii RSA 493	Proteobacteria; Gammaproteobacteria; Legionellales
NE02	CDS	788	1714	3.23e−06	TTG	Nitrosomonas europaea	Proteobacteria; Betaproteobacteria; Nitrosomonadales
MBU01	CDS	788	1714	7.20e−06	TTG	Methanococcoides burtonii DSM 6242	Euryarchaeota; Methanomicrobia; Methanosarcinales
CB02	CDS	884	1714	7.76e−06	GTG	Coxiella burnetii RSA 493	Proteobacteria; Gammaproteobacteria; Legionellales
SO02	CDS	788	1714	8.70e−06	TTG	Shewanella oneidensis MR-1	Proteobacteria; Gammaproteobacteria; Alteromonadales
XF02	CDSsub	788	1714	1.83e−05	TTG	Xylella fastidiosa	Proteobacteria; Gammaproteobacteria; Xanthomonadales
BPS01	CDS	788	1714	0.000366	TTG	Burkholderia pseudomallei K96243	Proteobacteria; Betaproteobacteria; Burkholderiales
PS02	CDS	788	1714	0.000449	TTG	Pseudomonas syringae pv. tomato	Proteobacteria; Gammaproteobacteria; Pseudomonadales
BBA01	CDS	884	1714	0.000589	GTG	Bdellovibrio bacteriovorus	Proteobacteria; Deltaproteobacteria; Bdellovibrionales
MLO03	CDS	788	1714	0.170785	TTG	Mesorhizobium loti	Proteobacteria; Alphaproteobacteria; Rhizobiales

NOTE.—Four species predicted to utilize alternative start codons as suboptimal (CDSsub) are also included.

Discussion

In this study, we describe an unusual DNA transposon IS5_Av from the DDE megafamily of transposases/integrases, which was found in the genome of a multicellular animal, but appears more similar to prokaryotic than to eukaryotic counterparts. It is framed by perfect 213-bp TIRs and contains a single ORF coding for an IS5-like transposase. Interestingly, the host contains only a single copy of this element per diploid genome. Single-copy TEs are quite rare in eukaryotes, and whenever one TE copy per genome is reported, it is usually identified in a search of genome databases, which are far from being complete, especially when it comes to repetitive regions that are often left unassembled and are missing from most databases. We, however, verified the single-copy status of IS5_Av by Southern blot hybridization and library screening, and are confident that no other copy of this element is present elsewhere in the A. vaga genome.

In a standard life cycle of DNA TEs, there are two stages with a single-copy status (fig. 4): 1) at the time of entry of a single invading copy and 2) just before elimination of the TE from the genome, when all of its copies but the last one have already been lost. (The third possibility, i.e., indefinite maintenance of a single domesticated copy in the genome by purifying selection due to acquisition of a cellular function, usually involves loss of TIRs and is therefore highly unlikely.) At which of these two time points did we find IS5_Av? If it were about to become lost from the A. vaga genome after having peaked in copy number, it may be expected to have undergone extensive ORF-damaging mutational decay, often accompanied by secondary insertions, so that the coding region would require molecular reconstruction and the TIRs would have accumulated differences. The perfect identity of the TIRs and the absence of interruptions in the ORF make the recent entry hypothesis more plausible. The lack of expression from a single genomic copy is indicative of its pseudogene nature, and the integrity of a ∼300-aa pseudogene would be typically compromised by mutational decay in a few million years (Lynch and Conery 2000, 2003). IS5_Av resides within an ancient hobo-like element, which underwent insertions of two TEs and, in addition, carries defects in its ORF (a frameshift and an in-frame stop codon). Although the hobo itself is rather decayed, IS5_Av is most likely inserted into the 3′ UTR of hobo, because the distance between the hobo stop codon and the IS5_Av insertion site is only 108 bp and does not include polyadenylation signals (which are found downstream from the IS5_Av insertion). A search for TSD–TIR for hobo revealed only one putative 8 bp-TIR–TIR-8 bp combination, fully consistent with IS5_Av insertion into the hobo 3′ UTR. Thus, it appears that IS5_Av underwent integration into the A. vaga chromosome on its own and not as a component of delivery vehicles such as phages or viruses, fragments of which would have been detectable in the adjacent genomic environment (but see further discussion below).

F

Copy number dynamics of DNA TEs (modified after Hartl et al. 1997). The shape of the curve is arbitrary, as is the location of branches giving rise to domesticated or horizontally escaped copies.

Naturally occurring recent interkingdom movements of TEs have not yet been reported in the literature, and our finding could represent a rare example of HGT between bacteria (or protists) and multicelluar animals, in which the TE is apparently “caught in the act” of transfer at the time when it failed to increase in copy number. Although the putative donor species is yet to be identified, and there are no other IS5_Av copies in the genome that could be used to determine the level of divergence between copies so as to estimate their arrival time, this HGT event may be regarded as relatively recent for reasons discussed above. The absence of detectable transcription from IS5_Av raises questions regarding the mechanisms responsible for its transfer and successful one-time integration into the A. vaga genome. It may be thought that DNA TEs are more prone to HGT because they can be transmitted as transpososomes–nucleoprotein complexes containing TE DNA and the element-encoded transposase, which are sufficient for integration into the target in vitro (in the presence of Mg2+) as well as in vivo. This capacity of the bacterial Tn5 transpososome, for instance, was even utilized in commercial applications (Reznikoff et al. 2004). Perhaps HGT of IS5_Av into A. vaga occurred when the transposon entered the germ line via routes operating during HGT of other foreign genes (Gladyshev et al. 2008), or possibly as a complex with the element-encoded transposase, which was able to complete the integration reaction. However, subsequent adaptation of the transcription–translation control sequences to the new host did not take place, and therefore its genomic mobility was restricted to the initial integration event, because it has never had a chance to produce transposase molecules in the new host, or perhaps yielded only misfolded and/or mislocalized molecules, even if the level of transcription (undetectable by PCR) was sufficient to produce any.

The much less likely “postamplification extinction” scenario would involve loss of the functional incoming copy plus any other transposed copies, including incomplete remnants, and retention of a single nonfunctional transposed copy, which underwent loss of the ATG codon via mutation or deletion. However, mutation of the ATG codon per se does not imply an immediate loss of promoter activity and cessation of transcription. The absence of a correctly positioned ATG codon could rather indicate that IS5_Av was using a noncanonical initiation codon, such as TTG or GTG, in its previous host. It is also formally possible that an ATG-less incoming DNA copy was bound to a functional transposase produced in trans, and entered the A. vaga genome as a transposition complex for one-time integration.

Another imaginable scenario would combine recent invasion with nearly immediate loss: The element could have arrived to a different telomere, already embedded in foreign sequences, via the same pathways which permit overall acquisition of foreign DNA (Gladyshev et al. 2008), but, because of the dynamic nature of bdelloid telomeres, the original invading copy, with these adjacent sequences, was lost upon chromosome end erosion, and what we are seeing is a daughter copy that underwent a round of transposition in A. vaga, but lost its functionality either during or soon after transposition. This explanation assumes that the invading copy was fully capable of expression and transposition in A. vaga, but was lost very rapidly without a chance to spur more than one daughter copy.

Could IS5_Av, displaying a very basal phylogenetic position in the ISL2 clade, represent a highly diverged member of a eukaryotic IS4/IS5-like TE superfamily, such as Harbinger/PIF, IS4EU/ISL2EU, or ISL2PR, which, however, artificially clusters with ISL2-like transposons as a result of homoplasy? It should be emphasized that members of both previously described eukaryotic IS4/IS5-like superfamilies typically contain another ORF in addition to transposase: In the case of Harbingers, this extra ORF is characterized by a SANT/Myb/trihelix motif (Kapitonov and Jurka 2004), whereas for IS4EU/ISL2EU, it resembles a lambda DNA exonuclease, and the transposase itself contains an N-terminal DNA-binding THAP domain (Kapitonov and Jurka 2007). Because none of these extra ORFs (and domains) could be identified in IS5_Av, it appears likely that the element belongs to a single-ORF family, as do most other bacterial TEs related to IS4 and IS5 (www-is.biotoul.fr; Chandler and Mahillon 2002). The newly identified protistan IS5-like elements also differ substantially from IS5_Av, both by phylogenetic placement and by the TIR structure. Neither does IS5_Av represent a domesticated single-copy TE (as is the case for many Harbinger-derived genes), because such domestication usually involves loss of TIRs, which are present in IS5_Av and extend for 213 bp without a single mismatch. Although bacterial TIRs in the ISL2 group usually fall within 10- to 40-bp range (http://www-is.biotoul.fr), exceptionally longer TIRs (up to 214 bp) are rare but not unprecedented in other bacterial IS groups (Kholodii et al. 2000).

Finally, one may entertain a possibility that this deep-branching element is the only known member of a novel eukaryotic IS5-related DNA TE superfamily, which is so rare that its other members, if any, have not yet been identified in eukaryotic genomes sequenced to date. Future projects aimed at sequencing previously unexplored eukaryotic genomes may be able to supply us with new representatives of this hypothetical superfamily.

Click here for additional data file.