Complete fusion of a transposon and herpesvirus created the Teratorn mobile element in medaka fish

Mobile genetic elements (e.g., transposable elements and viruses) display significant diversity with various life cycles, but how novel elements emerge remains obscure. Here, we report a giant (180-kb long) transposon, Teratorn, originally identified in the genome of medaka, Oryzias latipes. Teratorn belongs to the piggyBac superfamily and retains the transposition activity. Remarkably, Teratorn is largely derived from a herpesvirus of the Alloherpesviridae family that could infect fish and amphibians. Genomic survey of Teratorn-like elements reveals that some of them exist as a fused form between piggyBac transposon and herpesvirus genome in teleosts, implying the generality of transposon-herpesvirus fusion. We propose that Teratorn was created by a unique fusion of DNA transposon and herpesvirus, leading to life cycle shift. Our study supports the idea that recombination is the key event in generation of novel mobile genetic elements.

Mobile genetic elements (MGE) are major players in genome evolution. Here, the authors report a novel MGE called Teratorn in the genome of the medaka. Teratorn is the product of a fusion between a herpesvirus and a piggybac transposase. The authors characterize two subtypes of Teratorn in the genome of the Hd-Rr medaka strain, demonstrate that the element is capable of both excision and integration in HEK293 cells and that it is present in all four local populations of medaka in varying copy numbers.
It was a pleasure to read this paper, the discovery of this new element is definitely exciting and intriguing. Overall, I found that the analyses performed by the authors are robust and that the results support most of the conclusions. Below is a list of comments that should however be addressed: 1-1) -based on the presence of Teratorn in all four populations of medaka, the authors conclude that the element was present in the ancestor of these populations and that it has been transmitted vertically over 25 million years. In my opinion, this conclusion is weak because it is only based on indirect evidence. The congruence between the phylogeny of Teratorn and that of the medaka gene could be due to the fact that Teratorn is transmitted horizontally (in addition to vertically) on a regular basis, and that such transfers tend to occur between medaka individuals that are closely related both phylogenetically and geographically.
Such horizontal transmission is plausible given that genes of viral origin that are involved in viral particle formation and virus replication are intact (devoid of non-sense mutations).
Teratorn may well be inherited both vertically and horizontally and its presence in the medaka species may be much more recent than 25 myrs.
A direct evidence demonstrating the proposed antiquity of Teratorn would be to identify orthologous Teratorn copies between the various medaka strains, i.e., copies that are present at the same locus in two, three or even four strains. More specifically, a definitive conclusion on the presence of Teratorn in the medaka genome for 25 myrs requires the characterization of at least one copy shared at an orthologous locus between the HSQK or Nilan populations and at least one of the three other populations. Such a search for orthologous insertions requires a better characterization and comparison of the regions flanking each Teratorn copy in the four strains.
1-2) -Based on the presence of a Teratorn element showing the same gene order in both the javanicus and latipes species group of the genus Oryzias, the authors conclude that this element originated in the common ancestor of these species and that it was inherited vertically during 60 Myrs. It has been shown that once integrated in a host genome, copies of mobile elements transmitted vertically in host populations typically evolve neutrally (e.g. Lampe et al. 2003 MBE). If Teratorn is present in Oryzias spp since more than 60 Myrs, many old and highly degraded copies should be found in the genome of these species and the average pairwise distance between copies within a given genome should be rather high, contrary to what is observed (Page 9, Line 3). Furthermore, if Teratorn is old, as proposed by the authors, some copies should be shared at orthologous locus between species from the latipes and javanicus groups. I thus reiterate my request of performing a search for such orthologous copies.
Without such direct evidence for vertical inheritance, the authors cannot conclude without doubt that the element was present in the ancestor of the Oryzias genus 60 Myrs ago. Much like in the first paper, this point should be further explored. How many stop codons are expected during 60 Myrs under neutral evolution in teleosts? Are copies of Teratorn showing this expected number of stop codons observed? Would an alternative hypothesis that would imply a more recent age for Teratorn and involve a mix of vertical and horizontal transfer not be more plausible?
2 -Teratorn, like other known genetic elements such as retroviruses and polintons, is at the boundary between viruses and transposable elements. Its transposase and terminal inverted repeats allow it to integrate and multiple itself into host genomes and its genes of viral origin may allow him to form infectious viral particles. In the discussion, the authors first conclude that Teratorn is a DNA transposon (p19, line 5) and then propose that it is a novel herpesvirus endogenized in the medaka genome (p21, line 7). This is confusing and may really impede the understanding of readers who are not familiar with the complexity of mobile genetic elements.
The potentially dual nature of Teratorn should be more explicitly described and the authors should not attempt to provide a definitive conclusion on whether it is a virus or a transposon.
They should also clearly mention that to be confirmed, the putative viral fonctions of Teratorn await observation of viral particles.
3 -The capacity of Teratorn to reactivate and form viral particles is tested by the authors through treatment of medaka embryonic cells with chemicals that are known to reactivate latent human herpesviruses. No significant reactivation is observed. In a recent study, Fischer & Hackl (2016) show that the endogenous virophage Mavirus is reactivated by the presence of the Cafeteria roenbergensis virus, with which it shares promoter motifs. I was wondering whether the authors could have access to a strain of alloherpesvirus that they could use to test whether a similar interaction between endogenous and exogenous herpesvirus exists. 4 -Page 9, line 9: there seems to be a word missing in this sentence. This is a very interesting, and provocative manuscript describing a novel transposon subfamily in medaka fish. The peculiar feature of this transposon is that it apparently evolved out of a piggyBac element that received an ancient insertion of a herpesvirus.

-
I have the following observations: 1. The requirement for the internal TIR for transposition may suggest that the cleavage reactions by the transposase might actually occur at the internal TIR. Supplementary Fig. 8 shows the 5' end of the integrated transposon, but does that allow identification of which 5' TIR was actually used? In other words, transposition could take place at the external 5' TIR or at the internal 5' TIR. If transposition takes place at the internal TIR, then the entire plasmid is expected to integrate. This could be investigated by using Southern and a plasmid backbone specific probe, and/or by PCR, and/or by direct sequence analysis of integrants.
2. The most intriguing aspect of this study is unfortunately the least documented. Is the herpesvirus still "alive"? As the manuscript stands, the reader, although intrigued, is left with mixed feelings. It could well be, as authors describe in Discussion, that genomic integration of this herpesvirus was helpful in the distant past to generate tolerance against new herpesvirus infections. In the absence of further biological experiments, it could also be that the herpesvirus integration in this piggyBac transposon was a chance event that never had biological significance, and in that sense this herpesvirus genome is a mere passenger of piggyBac transposition.
Reviewer #3 (Remarks to the Author): Very interesting paper that reports the largest eukaryotic transposons so far discovered. The fusion of a PiggyBac transposon with a complete Alloherpesvirus genome to form these Teratorn transposons certainly is a novel, remarkable phenomenon. Although preliminary results have been reported previously (Ref. 12), this paper presents definitive data on genome structure and transposase activity of Teratorn.
The main question that remains unanswered is: can a herpesvirus be activated from a Teratorn element? An experimental demonstration of such activation could be difficult to achieve and is probably unreasonable to expect as part of this paper. However, there are two amendments related to this point that the authors could and I think should implement to improve the paper.
First, the sequences of the capsid proteins and proteins involved in morphogenesis (terminase, protease) should be analyzed in greater detail, to determine whether their sequences are compatible with virion formation (conservation of structural elements in the capsid proteins and the catalyitc sites in the enzymes). Second, although the authors discuss the analogy with polintons, they seem to miss the key point, namely that most of the polintons encode two capsid proteins along with the ATPase and protease required for virus formation, even though virions so far have not been discovered experimentally. Thus, the analogy between the polintons and Teratorn is actually quite complete and I think should be discussed along these lines. Furthermore, following the same theme, the data on virophage integration probably

Minor issues
The authors habitually use 'sequence homology', a common but wrong terminology. Should be 'sequence similarity' (observation) and 'homology' (conclusion).
The 'helicase' that is repatedly mentioned in the text and Replies to the Reviewer 1 Comments: We thank all Reviewers for giving us valuable comments and suggestions. We were pleased to know that all reviewers found our papers interesting. We have revised the manuscript following their comments and suggestions. Major revised portions are underlined in the text.
1-1) based on the presence of Teratorn in all four populations of medaka, the authors conclude that the element was present in the ancestor of these populations and that it has been transmitted vertically over 25 million years. In my opinion, this conclusion is weak because it is only based on indirect evidence. The congruence between the phylogeny of Teratorn and that of the medaka gene could be due to the fact that Teratorn is transmitted horizontally (in addition to vertically) on a regular basis, and that such transfers tend to occur between medaka individuals that are closely related both phylogenetically and geographically. Such horizontal transmission is plausible given that genes of viral origin that are involved in viral particle formation and virus replication are intact (devoid of non-sense mutations). Teratorn may well be inherited both vertically and horizontally and its presence in the medaka species may be much more recent than 25 myrs.
A direct evidence demonstrating the proposed antiquity of Teratorn would be to identify orthologous Teratorn copies between the various medaka strains, i.e., copies that are present at the same locus in two, three or even four strains. More specifically, a definitive conclusion on the presence of Teratorn in the medaka genome for 25 myrs requires the characterization of at least one copy shared at an orthologous locus between the HSQK or Nilan populations and at least one of the three other populations.

Such a search for orthologous insertions requires a better characterization and
comparison of the regions flanking each Teratorn copy in the four strains.
> Following the comment, we compared integration sites of Teratorn between the three medaka inbred strains (Hd-rR, HNI, HSOK), using the recently published genome data (http://utgenome.org/medaka_v2/#!Top.md). However, we failed to identify any orthologous copy in any pairs of the three strains (see Fig. 1  Cyprinid herpesvirus 3 is the most intensively studied alloherpesvirus species because of the importance for marine fishery industry. As suggested by Reviewer 3, instead of experimental approaches, we analyzed the sequences of capsid proteins and proteins involved in virion morphogenesis (capsid maturation protease and DNA packaging terminase) to test whether their sequences are compatible with virion formation. We found that catalytic residues of the virion morphogenesis enzymes are conserved in Teratorn (data are included in revised Fig. 4a and Supplementary Fig. 9). In addition, we found the clear sequence >We have corrected these minor points in the revised manuscript.
****************************** Replies to the Reviewer 2 Comments: We thank all Reviewers for giving us valuable comments and suggestions. We were pleased to know that all reviewers found our papers interesting. We have revised the manuscript following their comments and suggestions. Major revised portions are underlined in the text.
1) The requirement for the internal TIR for transposition may suggest that the cleavage reactions by the transposase might actually occur at the internal TIR. Supplementary Fig.   8 shows the 5' end of the integrated transposon, but does that allow identification of which 5' TIR was actually used? In other words, transposition could take place at the external 5' TIR or at the internal 5' TIR. If transposition takes place at the internal TIR, then the entire plasmid is expected to integrate. This could be investigated by using Southern and a plasmid backbone specific probe, and/or by PCR, and/or by direct sequence analysis of integrants.
>We think that, in this assay, chromosomal integration took place mainly via internal TIRs. In Supplementary Fig.7c, specific bands of 2.3-kb region for subtype 1 and 5.0-kb region for subtype 2 were detected in all colonies.
These bands were derived from a DNA fragment that covers the right TIR and plasmid backbone produced by HindIII cut (Supplementary Fig. 7b, double-headed arrows). This data indicates the chromosomal integration via internal TIRs. Furthermore, in reply to this comment, we conducted inverse PCR using primers that specifically amplify integration sites mediated by either external or internal TIRs. We again identified integrated copies mediated by internal TIRs, but failed to obtain the evidence that supports integration via external TIRs. Thus, it is highly likely that internal TIRs were mainly used in this integration assay. We have no idea of why internal TIRs were mainly used for integration. The high frequency of internal TIR-mediated transposition could be due to the artificial circular configuration of the indicator plasmid. In any case, the aim of this in vitro assay was to test the activity of transposase and so we did not change the text in the revised manuscript.
2) The most intriguing aspect of this study is unfortunately the least documented. Is the herpesvirus still "alive"? As the manuscript stands, the reader, although intrigued, is left with mixed feelings. It could well be, as authors describe in Discussion, that genomic integration of this herpesvirus was helpful in the distant past to generate tolerance against new herpesvirus infections. In the absence of further biological experiments, it could also be that the herpesvirus integration in this piggyBac transposon was a chance event that never had biological significance, and in that sense this herpesvirus genome is a mere passenger of piggyBac transposition.
> Regarding the activity of herpesvirus, we analyzed the sequences of capsid proteins and proteins involved in virion morphogenesis (capsid maturation protease and DNA packaging terminase) to test whether their sequences are compatible with virion formation, as suggested by Reviewer 3. We found that catalytic residues of the virion morphogenesis enzymes are conserved in Teratorn (data are included in revised Fig. 4a and Supplementary Fig. 9). In addition, we found the clear sequence similarity Regarding the biological significance, we don't have any evidence to support the idea that Teratorn has some biological benefits to host organisms (medaka). However, we think that the fusion event has some biological significance for Teratorn itself; i.e. the fusion event enabled Teratorn to undergo intragenomic propagation. We do not think that Teratorn was created by an integration of herpesvirus into a piggyBac transposon and that the virus is a mere passenger of piggyBac transposition. If Teratorn was created by a chance event of integration as the reviewer suggests, copies of the same piggyBac transposon should exist without herpesvirus sequences. However, as presented in Supplementary   Fig. 5, there is no such copy in the medaka genome. ******************************

Replies to the Reviewer 3 Comments:
We thank all Reviewers for giving us valuable comments and suggestions. We were pleased to know that all reviewers found our papers interesting. We have revised the manuscript following their comments and suggestions. Major revised portions are underlined in the text.
1) The main question that remains unanswered is: can a herpesvirus be activated from a Teratorn element? An experimental demonstration of such activation could be difficult to achieve and is probably unreasonable to expect as part of this paper. However, there are two amendments related to this point that the authors could and I think should implement to improve the paper.
1-1) First, the sequences of the capsid proteins and proteins involved in morphogenesis (terminase, protease) should be analyzed in greater detail, to determine whether their sequences are compatible with virion formation (conservation of structural elements in the capsid proteins and the catalyitc sites in the enzymes).
> Following the suggestion, we analyzed the sequences of capsid proteins and proteins involved in virion morphogenesis (capsid maturation protease and DNA packaging terminase) to test whether their sequences are compatible with virion formation. We found that catalytic residues of the virion morphogenesis enzymes are conserved in Teratorn (data are included in revised Fig. 4a and Supplementary Fig. 9). In addition, we found the clear sequence similarity of major capsid protein and subunit 2 capsid triplex protein between Teratorn and exogenous alloherpesvirus species (data are included in new Supplementary Fig. 10). Thus, although virus function of Teratorn awaits virion detection, Teratorn could possibly be a "bona-fide" virus. We thus added a new sentence to the revised manuscript, "further experimental efforts to detect virions will be needed to understand the life cycle of Teratorn and the biological significance of the existence of Teratorn in the medaka genome (p.27. line 5~)".
1-2) Second, although the authors discuss the analogy with polintons, they seem to miss the key point, namely that most of the polintons encode two capsid proteins along with the ATPase and protease required for virus formation, even though virions so far have not been discovered experimentally. Thus, the analogy between the polintons and Teratorn is actually quite complete and I think should be discussed along these lines.
> Following the suggestion, we added description on the analogy between the two mobile elements in the paragraph of the discussion part, pointing out the potential to produce virus particles (p.26. line 4~). Thus, we cited these two papers in the paragraph of the discussion part, in the light of its biological significance for their hosts (p.26. line 18~).

1-4) Finally, the authors repeatedly state that most transposons only contain 1-3 genes
which is somewhat disingenuous given that the widespread polintons are much larger.
> Following this comment, we deleted this sentence from our manuscript.
Minor issues 2-1) The authors habitually use 'sequence homology', a common but wrong terminology.
> Following this comment we corrected the wording of the two terms.

2-2)
The 'helicase' that is repatedly mentioned in the text and Figure 2 -which one is this? UL9 homolog? Line 4, p. 39: > As pointed out, the helicase gene in Teratorn is UL9 homolog. To avoid confusion, we modified our text : "...such as DNA replication (DNA polymerase, primase and UL21 homolog DNA helicase), ..." (p.9. line 1).  (glyt, sh3px3, rag1, ptch1, tbr, myh6, zic1, plagl2). Since the divergence time and evolutionary rate is ambiguous in the genus Oryzias, we instead estimated evolutionary distances between those species from this tree, assuming that 3rd codon positions undergo neutral evolution. b) Maximum-likelihood trees of DNA polymerase and major capsid protein used for inference of ancestral   claim that "We think that, in this assay, chromosomal integration took place mainly via internal

2-3)
TIRs." They also state: "Furthermore, in reply to this comment, we conducted inverse PCR using primers that specifically amplify integration sites mediated by either external or internal TIRs.
We again identified integrated copies mediated by internal TIRs, but failed to obtain the evidence that supports integration via external TIRs. Thus, it is highly likely that internal TIRs were mainly used in this integration assay." These statements are in clear contradiction with data presented in Fig. 3. Please clarify.
2) In the rebuttal letter, in response to my question authors argue: "We do not think that  Fig. 3a,b clearly indicates catalytic transposase activity at the external TIRs. Yet, in their rebuttal letter authors claim that "We think that, in this assay, chromosomal integration took place mainly via internal TIRs." They also state: "Furthermore, in reply to this comment, we conducted inverse PCR using primers that specifically amplify integration sites mediated by either external or internal TIRs. We again identified integrated copies mediated by internal TIRs, but failed to obtain the evidence that supports integration via external TIRs. Thus, it is highly likely that internal TIRs were mainly used in this integration assay." These statements are in clear contradiction with data presented in Fig. 3 Anyway, we revised our manuscript as clearly as possible to explain the relationship between the presence / absence of internal TIRs and the result of the excision / integration assay (p.12. line 221~).
2) In the rebuttal letter, in response to my question authors argue: "We do not think that Teratorn was created by anintegration of herpesvirus into a piggyBac transposon and that the virus is a mere passenger of piggyBac transposition. If Teratorn was created by a chance event of integration as the reviewer suggests, copies of the same piggyBac transposon should exist without herpesvirus sequences. However, as presented in Supplementary Fig. 5, there is no such copy in the medaka genome." Just because the medaka genome does not contain "empty" piggyBac transposons does not rule out that those elements exist somewhere else. Authors now discuss a scenario, in which horizontal gene transfer played a role in distributing Teratorn elements in medaka species. Thus, it may well be that the fusion of a piggyBac transposon and the herpesvirus genome took place in another, unknown genome and the resulting element has been horizontally transferred to medaka, followed by several rounds of transposition thereby generating extra copies. I believe this is a plausible scenario. Also, for the sake of argumentation, what was the acquisition of the herpesvirus genome by the piggyBac transposon if not a chance event? Please discuss this in a concise and clear manner.
> Not clearly stated in the text, we agree that the first fusion event happened by chance, probably in somewhere else other than medaka, and that the invasion of medaka Teratorn was the result of horizontal transfer. What happened at the very early stage of the formation of Teratorn is largely unknown. We think that there are at least two possibilities; one is the integration of the herpesvirus genome into a piggyBac, as pointed out by Reviewer 2, while the other possibility is the integration of the piggyBac transposon into a latently infected herpesvirus genome floating in the nucleus. We added the above scenario in 'Discussion' (p.24. line 436~).
3) Authors sate in Discussion: "For example, several studies reported the insertion of an insect transposon into a baculovirus genome. Indeed, all viruses have the potential to shift into the intragenomic life cycle, if they acquire an integration system from other sources." This is a confusing argument. If a transposon integrates into a virus, then the transposon might gain the ability to spread within and between species through the infectious potential of the virus.
And just the other way around, if a virus integrates into a transposon (like in Teratorns), then the virus might become endogenized. This has to be clearly discussed.
> We agree with this comment. The sentence "For example, several studies reported the insertion of an insect transposon into a baculovirus genome" is confusing and not appropriate here, because this example only tells the case that transposons jumped into virus genomes. Our speculation is that the fusion event took place by integration of the piggyBac transposon into a latently infected herpesvirus genome in the nucleus, but the possibility that the integration of the herpesvirus genome into a piggyBac can not be ruled out, as described above. We therefore deleted this sentence from 'Discussion' (p.28. line 500). This does not affect our points in 'Discussion'.