Intruder (DD38E), a recently evolved sibling family of DD34E/Tc1 transposons in animals

A family of Tc1/mariner transposons with a characteristic DD38E triad of catalytic amino acid residues, named Intruder (IT), was previously discovered in sturgeon genomes, but their evolutionary landscapes remain largely unknown. Here, we comprehensively investigated the evolutionary profiles of ITs, and evaluated their cut-and-paste activities in cells. ITs exhibited a narrow taxonomic distribution pattern in the animal kingdom, with invasions into two invertebrate phyla (Arthropoda and Cnidaria) and three vertebrate lineages (Actinopterygii, Agnatha, and Anura): very similar to that of the DD36E/IC family. Some animal orders and species seem to be more hospitable to Tc1/mariner transposons, one order of Amphibia and seven Actinopterygian orders are the most common orders with horizontal transfer events and have been invaded by all four families (DD38E/IT, DD35E/TR, DD36E/IC and DD37E/TRT) of Tc1/mariner transposons, and eight Actinopterygii species were identified as the major hosts of these families. Intact ITs have a total length of 1.5–1.7 kb containing a transposase gene flanked by terminal inverted repeats (TIRs). The phylogenetic tree and sequence identity showed that IT transposases were most closely related to DD34E/Tc1. ITs have been involved in multiple events of horizontal transfer in vertebrates and have invaded most lineages recently (< 5 million years ago) based on insertion age analysis. Accordingly, ITs presented high average sequence identity (86–95%) across most vertebrate species, suggesting that some are putatively active. ITs can transpose in human HeLa cells, and the transposition efficiency of consensus TIRs was higher than that of the TIRs of natural isolates. We conclude that DD38E/IT originated from DD34E/Tc1 and can be detected in two invertebrate phyla (Arthropoda and Cnidaria), and in three vertebrate lineages (Actinopterygii, Agnatha and Anura). IT has experienced multiple HT events in animals, dominated by recent amplifications in most species and has high identity among vertebrate taxa. Our reconstructed IT transposon vector designed according to the sequence from the “cat” genome showed high cut-and-paste activity. The data suggest that IT has been acquired recently and is active in many species. This study is meaningful for understanding the evolution of the Tc1/mariner superfamily members and their hosts.


Introduction
The mobilome is defined as the entire set of mobile (transposable) elements of a genome, which can be categorized into four classes: self-splicing molecular parasites, plasmids, bacteriophages, and transposons [1]. Transposons, which can move about or propagate within the genome, are the major constituents of the mobilome, and are distributed extensively in prokaryotic and eukaryotic genomes [2]. These were once regarded as "junk" DNA, but increasing evidence indicates that they play significant roles in genomic evolution as well as genes [3,4], and are major determinants of genome size in vertebrates [5,6]. They can alter the genomic landscape by horizontal transfer (HT) between non-mating species [7][8][9], and are a key cause of genetic polymorphisms and mutations linked with genomic rearrangements and distinctive characteristics of chromosomes [10], which are increasingly known as major factors in eukaryotic genomic evolution [11,12]. Transposons can also evolve into new genes by undergoing "molecular domestication", where the transposons are incorporated into host genes and evolve new functions [13,14]. In addition, transposons contribute to cis-regulatory DNA components and transcription network modifications [10,15]. Transposons are typically classified into two classes according to their transposition mode: RNA and DNA transposons, RNA transposons also named as retrotransposons, which need RNA intermediate in their development cycle. While DNA transposons generally do not need RNA intermediate for their transposition, but mostly they form a rather heterogeneous group, composed of cut-and-paste transposons, polintons, and helitrons [2]. Cut-and-paste transposons are the most diverse and abundant category of DNA transposons and comprise at least 17 superfamilies [16]. The bestcharacterized cut-and-paste DNA transposon is the Tc1/ mariner superfamily, which was named based on the first Tc1 (Transposon Caenorhabditis elegans number 1) element identified in C. elegans [17] and the first mariner element detected from Drosophila mauritiana [18]. Most of the Tc1/mariner transposon elements vary from 1.3 to 2.4 kb in length and comprise a lone gene encoding a polypeptide surrounded by terminal inverted repeats (TIRs) defining their borders between 5′ and 3′. They seek "TA" sequences to be inserted into the host genome, as a consequence "TA" target site duplications flanking the inserted transposon [19]. At least four families of Tc1/mariner transposons, including DD41D/VS, DD37E/TRT, DD36E/IC, and DD35E/TR have been well-described recently [20][21][22][23]. While DD34E/Tc1 [17], DD × D/pogo [24], and DD34D/mariner [18] were discovered very early and have been reported extensively, and DD × D/pogo has been suggested as a separate superfamily of IS630-Tc1-mariner transposon group very recently [25]. Twelve elements (Tc1, Tc3, Famar1, Minos, Mos1, Osmar5, ISY100, Mboumar-9, Fot1, Impala, Thm3, and Passport) of this superfamily are known to be active in their natural form [26][27][28], and half of them, including Tc1 [17], Tc3 [29], Impala [30], Minos [31], and Passport [26], and two artificially reconstructed Tc1/mariner transposons, including Sleeping Beauty and Frog Prince [32], are from the DD34E/Tc1 family, while DD35E/TR [22] and DD36E/IC [21] were discovered as new families in our previous studies: they are strongly linked to DD34E/Tc1 phylogenetically but form distinct sibling clades from DD34E/Tc1 and seem to have originated from this family. DD36E/IC is distributed in both vertebrates and invertebrates, including insects, arachnids, jawless fish, ray-finned fish, frogs, and bats [21], while DD35E/TR displays a restricted taxonomic distribution in the animal kingdom, and has only been detected in three classes (ray-finned fish, Anura, and Squamata) and 91 species of vertebrates [22].
Previously, a Tc1/mariner transposon family (DD38E), here named Intruder (IT), was identified in sturgeon (Acipenseridae) genomes [33]; however, the evolutionary landscape of this family, particularly its phylogenetic relationship with other families of the Tc1/mariner group, remains largely unknown. Here we describe the evolutionary profile of IT, including its distribution, phylogenetic position, structural organization, and HT in eukaryotic organisms; we have also functionally characterized the transpositional activity of a naturally occurring, intact IT sequence.

DD38E/IT is distributed among invertebrates and vertebrates
To assess the distribution of IT among species, a TBLASTN search against all the available prokaryotic (archaea and bacteria) and eukaryotic (chromista, plantae, animalia, protozoa, and fungi) genomes placed at the NCBI database was performed using the sturgeon IT transposase sequence as the query. This revealed that IT has a restricted taxonomic distribution compared with the families of DD41D/VS [23] and DD37E/TRT [20] of Tc1/mariner transposons, for which the taxonomic breadth has been well defined. Considerable hits encoding the preserved DD38E motifs were identified only in the animalia among eukaryotes, where ITs were present in 40 species of Arthropoda and one of the Cnidaria in invertebrates, one Agnathan species, one Anuran species, and 98 species in the Actinopterygii in vertebrates (Fig. 1a, b). Although IT elements are found as truncated copies in several species, intact copies of ITs with putative HT activity were also detected in many species across multiple lineages, indicating that they might be active. In ray-finned fish (Actinopterygii), IT elements were detected in 30 orders, and more than half of the species (55/98) contained IT elements flanked with TIRs, designated as full-length IT, but in only 26 species do they encode intact transposases, here designated as intact ITs. In the phylum Arthropoda, IT elements are present in 40 species of seven orders, with 22 species containing full-length IT copies, and 15 species harbouring intact IT elements. Intact copies of IT were also detected in the Anura and Cnidaria, but in the Agnatha, the ITs are present as full-length copies but code a truncated transposase (303 amino acids, aa; Fig. 1b and Supplementary Table S1).
In addition, about 20 IT copies (> 80% identity and 40% coverage) were also detected in the assembled genome of the domesticated cat (Supplementary Figure S1). However, all these copies were characterized in small contigs (1014-1588 kb) and most were found to be flanked by TA repeat regions (1000-2000 bp; Supplementary Figure S2), indicating possible sequence contamination. DNA samples from the genome of the Abyssinian cat breed and other domesticated breeds were also used to amplify the IT copies by PCR, followed by TA cloning and sequencing; however the PCR product bands were not as specific as expected (Supplementary Figure S3), and we did not obtain any positive results by sequencing over 20 clones, which confirmed again that IT in the assembled cat genome likely represents sequence contamination. By using BLAST at NCBI database to search against the nucleotide collection (nr/nt), we found these IT copies display very high sequence similarity to Tc1 transposons identified in the sturgeon genome (Supplementary Figure S4), indicating that this might be the source of contamination.

Similar distribution patterns of IT and IC transposon sequences
Comparison across the IT, TR, IC and TRT sequence families revealed that IT and IC display a very similar taxonomic distribution pattern (Fig. 2a). Over 90% of orders with IT detected (31/34) overlap with the orders with IC detected; only three are IT-specific, and 58% of the species with IT detected (83/143) overlap with the species with IC. Both TR and TRT elements were found to be distributed in the Anura, Squamata, and Actinopterygii, but the similarity was relatively low at order and species levels compared with that between IT and IC, with about 30% of orders with TR detected (11/33) overlapping with the orders with TRT, or about 61% (11/18) overlapping with these of TR (Fig. 2a-c). On the other hand, the taxonomic distributions of Tc1/mariner families (IT, TR, IC, and TRT) in animals also revealed that they share some common hosts. Thus, some orders and species seem to be more hospitable to Tc1/mariner transposons than others, which has also been noted in a recent study [23]. The Actinopterygii and Anura tend to be more susceptible to the invasions of Tc1/mariner transposons, as all well-defined close sibling families of DD34E/Tc1 (IT, TR, IC and TRT) were detected in these lineages. One order of Amphibia (Anura) and seven Actinopterygian orders (Characiformes, Cichliformes, Cypriniformes, Cyprinodontiformes, Esociformes, Perciformes and Salmoniformes) are the most common orders with HT events and have been invaded by all four families (IT, TR, IC and TRT) of Tc1/mariner transposons (Fig. 2b). Fifteen other orders are also very common reservoirs of Tc1/mariner transposons and have been invaded by at least three families. Eight species in the Actinopterygii (Astyanax mexicanus, Cyprinodon variegatus, Dicentrarchus labrax, Esox Lucius, Larimichthys crocea, Nothobranchius furzeri, Salmo salar, and Stegastes partitus) were identified as the hosts of most HT events and have been invaded by all four families (IT, TR, IC and TRT) of Tc1/mariner transposons (Fig. 2c).

DD38E/IT might originate from DD34E/Tc1
The phylogenetic trees generated from the alignments of the DDE domains proved that all identified elements belong to the DD38E/IT family, which was more intimately linked to the DD34E/Tc1 and DD35E/TR, DD36E/IC and DD37E/TRT ( Fig. 3a and Supplementary Figure S5). The sequence identity matrix also showed that DD38E/ IT transposases were more intimately linked to DD34E/ Tc1 than other families by an average percentage identity of 35% (Fig. 3b [19], were determined in the IT transposases by in silico forecast, and the DBD domain and DDE signature and its distances in the DDE domain appeared to be strongly conserved throughout the IT family (Fig. 4b, c, and Supplementary Figure S6).

Evidence of multiple HT events of IT transposons in vertebrates
IT elements were further classified into six major clusters based on the alignment of the full-length transposon nucleotide sequences: vertebrate species mainly distributed in five clusters (1-5), Clusters 1, 4 and 5 were detected in 15, 3 and 7 species of ray-finned fish, respectively; cluster 2 was identified in one species of jawless fish, 36 species of ray-finned fish, and one species of Anura; cluster 3 was present in five species of ray-finned fish and one species of Arthropoda; while cluster 6-most common in invertebrate species-was present in one species of Cnidaria and 20 species of Arthropoda ( Fig. 5a and Supplementary Figure S7). The overall topology of this phylogenetic tree was quite distinct from the established phylogeny of these species from which it is extracted, this may mean the exposure of IT elements to several HT events. To test this assumption, pairwise distances among all consensus sequences or representative sequences of IT transposons and RAG1 coding sequences were calculated. Indeed, for most pairwise comparisons (561/629), the distances  Table S2), which are typically used to predict HT incidents of transposons in vertebrates [34].
Meanwhile, most of species included in these IT pairwise distances engaged a last common ancestor more than 166 million years ago (Supplementary Table S2). Together, these results clearly indicate that the existence of IT in several of the main vertebrate lineages tested here The green arrows represent TIRs, the black rectangles represent HTH motifs, the black triangles represent GRPR sequences, the yellow circle represents the NLS, the orange rectangles represent catalytic domains, and the grey regions represent transposases. The dotted box represents the portion of the transposases that might be deleted in a particular species. b IT in the "cat" genome. The grey areas at the top and bottom represent IT. We selected copies 1 and 2 to mark the locations in this genome. In the middle is a schematic diagram of the complete IT structure. The red arrows represent TIRs, the green rectangle represents the DNA-binding domain, and the orange rectangle represents the catalytic domain. c Motifs prediction for IT transposases. This analysis was performed using multiple alignment with Bioedit and with modifications in Genedoc. results from HT incidents that occurred after these lineages diverged from each other.

Very recent invasions of ITs in vertebrates
To investigate the evolutionary dynamics of ITs in animals, we also compared the insertion ages and sequence identities of ITs in these species across clusters, which revealed differential evolutionary dynamics. Most species have experienced recent and sharp peak activities (less than 5 million years ago) of ITs, suggesting recent invasions. Some species, such as Myaetiola destructor, Spodoptera litura, Spodoptera frugiperda, Danaus plexippus   Fig. 6 and Supplementary Figure S8). In addition, the overall mean sequence identity (50.66 ± 22.21%) of ITs across species is similar to that for DD36E/IC (52.48 ± 19.19%), but lower than for DD35E/TR (82.33 ± 10.01%; Supplementary Figure S9A), but most species in clusters 1, 2, 4 and 5 in vertebrates display very high sequence identities, ranging from 86 to 95% (Fig. 5a and Supplementary Figure S9B-E), indicating very recent HT events of ITs in these species. The discovery of intact ITs in multiple lineages of animals and high sequence identities in vertebrates, combining recent and sharp peak activities in most animal lineages, suggest that this family is a recently evolved clade of Tc1/mariner transposons that might still be active in some of these lineages.

IT is transpositionally active in mammalian cells
The dual plasmid-based assay [35,36] was applied to test the transpositional activity of the particular IT element identified in the "cat" genome assembly. The donor plasmids, harbouring a PGK promoter and a neomycin phosphotransferase cassette flanked by TIRs can confer G418 resistance in mammalian cells upon transposition into chromosomes. The helper plasmid has an expression cassette of transposases driven by a CMV promoter. The original 210 bp 5′TIRo and 3′TIRo and 1023 bp transposase (341 aa) of the intact copy of IT in "cat" Fig. 6 Insertion ages of ITs. This taxonomic tree represents the distribution of the species identified in the animal kingdom, and each colour represents a phylogenetic tree cluster. Insert age analysis was calculated by using the RepeatMasker program. The phylogenetic relationships were taken from the TimeTree database genome (AANG04004031.1|:3377-4964), and the consensus sequences of IT TIRs (5′TIRc and 3′TIRc), which are 97.62 and 100% identical to the 5′TIRo and 3′TIRo (Supplementary Figure S10), were used to construct vectors. The SB transposon together with the SB100X hyperactive transposase [35,37] was applied as a positive control. A schematic of the donor and helper plasmids is shown in Fig. 7a. The donor plasmid was then mixed with the helper plasmid 1:1 and co-transfected into HeLa cells with subsequent G418 selection. We found that both the IT and SB transposases displayed substantial transposition activity in human cells (Fig. 7b). The frequency of IT transposition was about 45% of SB in HeLa cells by measuring the numbers of neomycin- Fig. 7 IT is transpositionally active in mammalian cells. a IT and Sleeping Beauty (SB) transposon vectors for in vitro transposition activity assays. SB was used as a positive control. The three TIR vectors-pITo-Neo, pITc-Neo and pSB-Neo-have the same vector frame, and the TIR elements on both sides are the only differences. pITo-Neo is composed of two original TIR sequences of IT, pITc-Neo is composed of one consistent TIR and the other original TIR of IT, and pSB-Neo is composed of two SB TIRs. The two transposase vectors pCMV-itTPase and pCMV-SB100X also have the same frame. b-e HeLa cells were stably transfected with 1 μg of mixed plasmids (donor and helper plasmids at a 1:1 ratio. For selection, the transfected cells were reseeded onto 10-cm dishes (b and c, 1:100 plating; d and e, 1:10 plating). Selection was performed using 1000 mg/ml G418 for 14 days in DMEM. To determine the relative rates of transgenesis, foci of neomycin-resistant cell colonies that remained on each dish were counted after being fixed in 4% paraformaldehyde and stained with methylene blue. Bars represent the mean neomycin-resistant colonies ± standard deviations from three independent experiments resistant colonies (Fig. 7c). In addition, there were more colonies in the cell group transfected with pITc-Neo plus transposase than with pITo-Neo plus transposase (1% replating selection), indicating that the transposition activity of consensus TIR sequences of IT may be higher than that of the original TIR, which was confirmed by using 10% replating selection, where we found a significant difference (P < 0.01) between pITc-Neo and pITo-Neo groups (Fig. 7d, e). The integration sites of IT in the human genome were recovered by ligation-mediated PCR as described in the methodology and confirmed by Sanger sequencing. These data indicate that IT sequences are still functionally active and can potentially undergo HT in mammalian cells.

Intra-family diversity of DD34E/Tc1
DD34E/Tc1 is a classical family of Tc1/mariner transposons, first found in C. elegans in the 1980s [17], and displays extensive distribution in nature [28]. Diverse DD34E/Tc1 transposons were discovered in multiple lineages of animals, including Topi and Quetzal in mosquitoes [48,60], Impala in fungi [30], Minos, S elements and Bari-3 in fruit flies [49,50,53], Frog Prince in frogs [32], and Passport in fish [26]. The DD34E/Tc1 group was then divided into at least five distinct clusters based on the DD34E/Tc1 elements identified from six teleost species in our earlier study, combined with previously reported elements of this family from other laboratories [58]; however, the intra-family diversity is still ambiguous. In addition, one distinct family (Gambol) also has the DD34E domain, but is far away from the DD34E/Tc1 sequence in phylogenetic position, and is close to the DD35E/IS630 family [61].
Recently, two new families (DD35E/TR and DD36E/ IC) were discovered, and are close to Tc1 in phylogenetic position [21]. Here, we identified a third new family (DD38E/IT) with a varying DDE domain (DD38E), but still retaining a close polygenic relationship with DD34E/Tc1. Generally, these three families (DD35E/TR, DD36E/IC, and DD38E/IT) display similar evolutionary profiles with restricted distribution, relatively recent invasion history in most species, and high sequence identity across species [21]. They seem to have originated from the Tc1 family and have evolved into new clades with varied DDE domains. Thus, these groups might represent subfamilies of Tc1, not sperate families of Tc1/mariner transposons. Such results demonstrate that the Tc1 family has high diversity. Systematically characterizing the taxonomic distribution of homologous Tc1 elements and defining the whole evolutionary landscape of this family will be very interesting and might help to illustrate the family structure of Tc1.

Major reservoir hosts of Tc1/mariner transposons
Tc1/mariner families, with well-defined evolutionary profiles, including the DD35E/TR, DD36E/IC and DD37E/TRT families discovered recently, and the DD38E/IT family reported here, repeatedly invaded the arthropod phylum of invertebrates and/or the class Actinopterygii of vertebrates, suggested that arthropods and ray-finned fish might be Tc1/mariner transposons major reservoir hosts. Furthermore, the evolutionary landscape of transposable elements (TEs) in vertebrates revealed that almost all main kinds of eukaryotic TEs exist in ray-finned fish, which display the highest TE diversity across vertebrate groups [5,62]. The evolutionary profile of TEs in arthropods also demonstrated that almost all known TEs have been identified in this phylum. This could be explained either because these lineages might be more prone to the exchange of genetic material or are more congenial to them than others, or because they comprise a great diversity of species. Previous reports suggested that some taxa, such as the bats (Chiroptera) are more prone to the invasion of DNA transposons than other types and have experienced multiple invasion events of the main DNA transposon superfamilies (hAT, piggyBac, Tc1/mariner and helitron). Bats are also suggested as the main reservoir hosts of many high-impact viruses that cause severe human diseases [63]. By contrast, arthropods have been suggested as major reservoir hosts for many of the negative-sense RNA viruses [64]. These results suggest that some lineages are more susceptible to the invasion of genetic materials than others, although the mechanisms remain largely unknown. Additionally, and in numerical terms, ray-finned fish (Actinopterygii) are the dominant class of vertebrates, comprising half of all living vertebrate species and approximately 32,000 species are recognized within this class [65]; while the phylum of arthropods accounts for more than 80% of all recognized animal species, with values of the number of arthropod species being 5-10 million [66]. Both ray-finned fish (Actinopterygii) and Arthropod lineages have great species diversity, so might serve as the major reservoir of most eukaryotic TEs. On the other hand, it is often believed that TEs facilitates diversification or biological and genomic distinction between organisms [67,68], TE activity is positively corresponding to the speciation rate in mammals [69] and lineages harbouring recently acquired TE families are also linked with latest speciations [70], suggesting that TE activity might play roles in the facilitation of reproductive isolation, and ultimately in speciation.

High activity of Tc1/mariner transposons in ray-finned fish
TEs display drastically different evolutionary dynamics across vertebrate groups; thus, recently active DNA transposons are more frequent in ray-finned fish genomes than in birds or mammals [6,62,71]. Although multiple mammalian lineages (galagos, murine rodents, opossums, tenrecs, bats and primates) have been invaded by DNA transposons, most of them appear as truncated copies in these genomes, and have lost transpositional activity, except for piggyBac domains in bats, which have been reported as being functionally active copies [72]. Assesses of the evolutionary dynamics of DD35E/TR, DD36E/IC, and DD37E/TRT also indicate that many Tc1/mariner transposons in ray-finned fish species are recently acquired elements with intact copies and tend to be functionally active. Here, we reconfirmed that the IT motifs seem to be highly active in the ray-finned fish lineage with many species containing recent IT insertions. In addition, two active transposons in their native form, including Passport from flatfish [26] and Thm3 from silver carp [27], were also identified from the class of ray-finned fish. Here, we have shown that a third Tc1/mariner transposon (DD38E/IT) encodes functional components required for cut-andpaste transposition in human cells. IT displayed a rate of transpositional activity corresponding to up to half the level we observed for the highly active SB transposon [37]. Thus, these data reinforce that active Tc1/mariner transposons in ray-finned fish are common, and these active TEs might have played key roles in driving the genomic evolution of this group and in their speciation.

Identification of IT transposons
The sturgeon IT transposase sequences [33] were employed as a query to investigate genomes available for organisms, including eukaryotes and prokaryotes at the National Center for Biotechnology Information (NCBI) by using the cut-off value of TBLASTN of 1e − 100 . Alongside 2 kb of flanking sequences, the top uncorrelated hits were retrieved and then used against the host genome to BLAST. All hits with more than 80% identity and 40% coverage were downloaded and aligned to define transposon boundaries using the MAFFT software; the TIRs of IT sequences with low genome copies were determined manually. The consensus sequences of IT transposons were reconstructed using multiple alignments of IT copies from every genome. The new sequences identified were then used as queries to recognize more IT elements. In addition, the flanking regions of all ITs with very low copy numbers located on short contigs in the genome were checked to ensure they were not sequence contaminations. The copy number of ITs in each genome was estimated by using BLAST (40% coverage and 80% identity) with the consensus sequences or representative sequences of ITs.

Sequence analysis and phylogenetic inference
Predictions of the secondary structure of proteins were created using the software PSIPRED (http://bioinf.cs.ucl. ac.uk/psipred/). Putative nuclear localization signal (NLS) motifs were predicted using WOLFPSORT (https://wolfpsort.hgc.jp/). Multiple alignments of full transposases and DDE domains were performed using the MAFFT program [73]. The DDE domains were detected by using the profile hidden Markov Models for the hmmscan website (https:// www.ebi.ac.uk/Tools/hmmer/search/hmmscan). The phylogenetic trees were predicted by the IQ-TREE software (v. 1.6.1) using the Maximum Likelihood with ultrafast bootstrap approach (1000 replicates), and the best-fit model was chosen using ModelFinder incorporated in IQ-TREE [74]. The TP36/Zator clade, which is close to IS630-Tc1mariner group, but forms a separate superfamily [75], was used as an outgroup. Their accession numbers or genome coordinates of Tc1/mariner reference elements are listed in Supplementary Table S3. The possible open reading frame (ORF) of Intruder identified here and sequences of RAG1 protein were predicted by GENSCAN website. (http://genes. mit.edu/GENSCAN.html).

Pairwise distances between IT and RAG1 sequences
Pairwise distances between the various vertebrate organisms used in this study were determined for IT and RAG1 sequences with the purpose of testing the hypothesis of HT by using MEGA7 (highest composite probability and deletion of pairwise) depending on two multiple alignments [76]. The multiple alignments of IT consensus sequences derived for each species or representative sequences and RAG1 coding sequences used to calculate these distances are provided in supplementary files (Text S1 and Text S2), and the access number or genome coordinates of RAG1s are listed in Supplementary Table S4.

Insertion age estimation
To estimate the age of IT invasion in each genome, the Kimura two-parameter distance was determined using RepeatMasker's calcDivergenceFromAlign.pl package from RepeatMasker software [77]. The insertion time of every component was calculated by the eq. T = k/2r [78], where T represents the time of insertion in millions of years, k represents the number of substitutions for nucleotides at each site, and r represents the neutral mutation rates in species lineages. We used the neutral mutation rates for Actinopterygii ( Figure S3.

Gene synthesis and vector construction
To test the transpositioning activity of IT in the domesticated "cat" genome, the TIRs and transposase of the intact IT copy (AANG04004031.1|:3377-4964) were cloned for a dual (donor and helper) plasmid-based assay. We synthesized the original sequences from the intact copy of cat IT, including an ORF of the transposase, the original 5′-terminal repeat sequence (named 5′TIRo) and 3′-terminal repeat sequence (named 3′TIRo). We also obtained the consensus sequence of TIR by alignment, which was 100% identical to 3′TIRo, and 97.62% identical to 5′TIRo. The synthesized ORF sequences of transposases and TIRs are listed in Supplementary Table S5. The 5′TIRo and 3′TIRo sequences were cloned into a pLB vector in turn using restriction enzyme sites. The consensus TIRs (named 5′TIRc and 3′TIRc) were cloned using high fidelity polymerase chain reaction (PCR) amplification from the synthesized 3′TIRo. Then, the neomycin phosphotransferase expression cassette was subcloned from the vector (PB-SB-PGK-Neo-bpA) [36] and inserted between the 5′TIRs and 3′TIRs, and the resulting donor plasmids were named pITo-Neo and pITc-Neo. The synthesized transposase ORF was subcloned into the modified pcDNA3.0 vector with the SV40-Neomycin cassette deleted, and the resulting helper plasmid was named pCMV-itTPase.
To construct the two-plasmid system of the Sleeping Beauty (SB) transposon with the same backbone as IT, the neomycin phosphotransferase expression cassette flanked by SB TIRs [36] and the SB100X transposase ORF [37] were inserted into the pLB vector and modified pcDNA3.0 vector used by IT, respectively. The resulted plasmids named as pSB-Neo and pCMV-SB100X as positive control of transposition activity. The primers for cloning are listed in Supplementary  Table S6.

Cell culture and transposition assays
HeLa and HepG2 cells (American Type Culture Collection, Manassas, VA, USA) were cultured in Dulbecco's modified Eagle's medium (DMEM) supplied with 10% foetal bovine serum and 1% penicillin-streptomycin at 37 o under 5% CO 2 in humidified air. For transposition assays, the cells were planted on six-well plates at 3 × 10 5 /well the day before transfection. 1 μg DNA (donor plasmid and helper plasmid at 1:1 ratio) with 2 μL transfection reagent of Transal-LT1 (Mirus Bio LLC, Madison, WI, USA) was applied for each well. The empty modified vector pcDNA3.0 was used to fill up in negative controls. At 48 h post-transfection, cells were replated on 10 cm dishes and trypsinized (10% or 1% plating). After selection with G418 (1 mg/mL for HeLa cells) for 14 days, cells were 4% paraformaldehyde-fixed, 0.2% methylene blue-stained and blue colonies-counted.

IT insertion libraries for Sanger sequencing
Following G418 selection (1 mg/mL) for 21 days, HepG2 cells were picked, and genomic DNA was prepared using DNEasy Blood and Tissue kits (Qiagen, Hilden, Germany). Libraries for the integration were built using linker ligation-mediated PCR as described [84]. Briefly, 10 μg aliquots of DNA fragment were digested overnight with DpnI, isolated by 1% agarose gel electrophoresis to extract the largest fragment, and ligated to True-seq linkers after sonication, end repair and dA tailing. Under strict conditions, two rounds of PCR were performed using end-specific primers complementing transposon sequences and linker-specific primers complementing the DNA linker. The second round of PCR products were isolated in 1% agarose gels, and fragments ranging from 200 to 500 bp were recovered using MiniBEST Agarose Gel DNA Extraction kits (TakaRa Bio Inc., Kyoto, Japan) as an insertion library and cloned for Sanger sequencing. Primers applied for insertion libraries are listed in Supplementary Table S7.