• Loading metrics

Flipping chromosomes in deep-sea archaea

  • Matteo Cossu,

    Affiliation Institute for Integrative Biology of the Cell (I2BC), Microbiology Department, CEA, CNRS, Univ. Paris‐Sud, Université Paris‐Saclay, Gif‐sur‐Yvette, France


  • Catherine Badel,

    Affiliation Institute for Integrative Biology of the Cell (I2BC), Microbiology Department, CEA, CNRS, Univ. Paris‐Sud, Université Paris‐Saclay, Gif‐sur‐Yvette, France


  • Ryan Catchpole,

    Affiliation Institute for Integrative Biology of the Cell (I2BC), Microbiology Department, CEA, CNRS, Univ. Paris‐Sud, Université Paris‐Saclay, Gif‐sur‐Yvette, France

  • Danièle Gadelle,

    Affiliation Institute for Integrative Biology of the Cell (I2BC), Microbiology Department, CEA, CNRS, Univ. Paris‐Sud, Université Paris‐Saclay, Gif‐sur‐Yvette, France

  • Evelyne Marguet,

    Affiliation Institute for Integrative Biology of the Cell (I2BC), Microbiology Department, CEA, CNRS, Univ. Paris‐Sud, Université Paris‐Saclay, Gif‐sur‐Yvette, France

  • Valérie Barbe,

    Affiliation Genoscope, Laboratoire de Biologie Moléculaire pour l'Etude des Génomes C.E.A., Institut de Génomique - 2 rue Gaston Crémieux, EVRY, France

  • Patrick Forterre,

    Affiliation Institute for Integrative Biology of the Cell (I2BC), Microbiology Department, CEA, CNRS, Univ. Paris‐Sud, Université Paris‐Saclay, Gif‐sur‐Yvette, France

  • Jacques Oberto

    Affiliation Institute for Integrative Biology of the Cell (I2BC), Microbiology Department, CEA, CNRS, Univ. Paris‐Sud, Université Paris‐Saclay, Gif‐sur‐Yvette, France


Flipping chromosomes in deep-sea archaea

  • Matteo Cossu, 
  • Catherine Badel, 
  • Ryan Catchpole, 
  • Danièle Gadelle, 
  • Evelyne Marguet, 
  • Valérie Barbe, 
  • Patrick Forterre, 
  • Jacques Oberto

This is an uncorrected proof.


One of the major mechanisms driving the evolution of all organisms is genomic rearrangement. In hyperthermophilic Archaea of the order Thermococcales, large chromosomal inversions occur so frequently that even closely related genomes are difficult to align. Clearly not resulting from the native homologous recombination machinery, the causative agent of these inversions has remained elusive. We present a model in which genomic inversions are catalyzed by the integrase enzyme encoded by a family of mobile genetic elements. We characterized the integrase from Thermococcus nautili plasmid pTN3 and showed that besides canonical site-specific reactions, it catalyzes low sequence specificity recombination reactions with the same outcome as homologous recombination events on DNA segments as short as 104bp both in vitro and in vivo, in contrast to other known tyrosine recombinases. Through serial culturing, we showed that the integrase-mediated divergence of T. nautili strains occurs at an astonishing rate, with at least four large-scale genomic inversions appearing within 60 generations. Our results and the ubiquitous distribution of pTN3-like integrated elements suggest that a major mechanism of evolution of an entire order of Archaea results from the activity of a selfish mobile genetic element.

Author summary

Mobile elements (MEs) such as viruses, plasmids and transposons infect most living organisms and often encode recombinases promoting their insertion into cellular genomes. These insertions alter the genome of their host according to two main mechanisms. First, MEs provide new functions to the cell by integrating their own genetic information into the DNA of the host, at one or more locations. Secondly, cellular homologous recombination will act upon multiple integrated copies and produce a variety of large-scale chromosomal rearrangements. If such modifications are advantageous, they will spread into the population by natural selection. Typically, enzymes involved in cellular homologous recombination and the integration of MEs are distinct. We describe here a novel plasmid-encoded archaeal integrase which in addition to site-specific recombination can catalyze low sequence specificity recombination reactions akin to homologous recombination.


Large-scale genomic rearrangements allow organisms to evolve much more rapidly than through random mutation alone. Rearrangements can result in the movement of genes within genomes, changes in coding strand use, loss of nonessential functions and the incorporation of foreign DNA. As a result, the organization, content and processing of genetic information can be deeply altered. In all three domains of life, chromosomal reorganization is mainly promoted by recombination between homologous sequences, for example between redundant ribosomal operons [1,2] or integrated copies of mobile elements (ME) such as prophages [3,4], transposons [5,6] and insertion sequences (IS) [7]. Such recombination can result in the DNA inversions readily observed in closely related genomes [8,9]. In addition to homologous recombination, chromosomes can undergo rearrangement through retrotransposon-associated non-homologous recombination [10]. Other elements like integrons confer rapid adaptation to bacteria in changing environments by shuffling cassette arrays encoding a variety of functions, a process involving a site-specific recombinase and two types of attachment sites [11]. Further genomic rearrangement/reorganization can occur through the acquisition of new genetic material, predominantly by lateral gene transfer. Such gene transfer occurs in all organisms through infection by mobile elements such as viruses or plasmids, or through the uptake of free or encapsulated DNA from the environment [12,13]. Genomes can acquire novel genes in a fashion ranging from transient to permanent depending on the type of element and the physiological conditions of the host. When ME succeed in stably inserting their genome, the inserted DNA is then replicated as part of the host chromosome. The transactions between ME DNA and host genome are catalyzed by recombinases typically encoded by the elements themselves. These recombinases rank in different classes based on their enzymatic activity and the specificity of their DNA targets. The smallest ME are insertion sequences (IS) composed of a short DNA segment encoding only the enzymes involved in their transposition which can occur at many different genomic locations [14]. The related transposons are larger DNA segments which can be transposed by two flanking IS and frequently carry additional genes such as antibiotic resistance determinants [15]. The most frequent IS recombinases are DDE transposases which do not form covalent transposase-DNA intermediates during transposition [16]. Other and typically larger ME such as plasmids and viruses encode recombinases promoting DNA transactions with a stronger DNA sequence specificity. Such site-specific recombination is not only used for mobile element integration and excision in bacteria but also in the spread of antibiotic resistance by transposable elements, the control of plasmid copy number, regulation of gene expression and the resolution of concatenated chromosomes [17]. Site-specific recombinases can be categorized into the serine recombinases and tyrosine recombinases (Y-recombinases); which, in contrast to DDE transposases, form covalent enzyme-DNA intermediates during recombination, albeit with markedly different mechanisms of action. Before religation of the two recombining DNA strands, serine recombinases generate breaks in all strands while Y-recombinases produce two sequential single-strand breaks [17]. As a rule, site-specific integration/excision reactions promoted by Y-recombinases occur via a synaptic complex composed of two DNA duplexes carrying the specific sites bound by four recombinase protomers [17]. The two-recombinase pairs are activated sequentially, allowing one strand from each duplex to be exchanged at a time via two consecutive and symmetrical Holliday junctions. A notable exception is Vibrio cholerae phage CTX. Not only does this phage integrate into its host genome in single stranded form where two sites fold into a hairpin structure, mimicking a recombination target for the cellular XerCD chromosome resolvase; but also only requires XerC for integration [18].

One of the best-studied Y-recombinases is the integrase of phage λ. The primary function of this enzyme is the integration of phage DNA into the chromosome of its bacterial host (and its excision). This function is achieved by promoting site-specific recombination between the phage attachment site attP and its chromosomal counterpart attB [19]. Under particular circumstances, the integrase of the lambdoid phage HK022 is capable of generating inversions between attP and a secondary attachment site in the HK022 left operon [3]. Similarly, the primary function of the yeast FLP protein is the control of the 2μ plasmid copy number [20] by DNA inversion between two divergent 34bp FRT sites located on the plasmid [21]. FLP recombinase activity has also been successfully used for integration and excision of synthetic DNA in mammalian genomes [22]. The recombination activities of both λ integrase and FLP recombinase are summarized as shown in S1 Fig. Historically, this reciprocal and conservative recombination between two stringently defined double-stranded DNA sequences in each chromosome was denominated the Campbell model [23].

The sequences of a considerable number of Y-recombinases have been compared to reveal the position of conserved residues and infer the location of the catalytic active site [24]. They share in their C-terminal moiety a rather well conserved region of ~120 amino acids containing up to six nearly invariant amino acids R..K..HxxR..[W/H]..Y forming the active site [25,26]. A small number of Y-recombinases have been characterized biochemically in Archaea, for example the XerA recombinase of the hyperthermophilic euryarchaeon Pyrococcus abyssi which exhibits a perfect active site consensus [27]. Sequence alignments have revealed that other archaeal active sites diverge slightly from the bacterial consensus R..HxxR..Y [28]. The integrases of viruses SSV1 isolated from the hyperthermophilic crenarchaeon Sulfolobus shibatae [29] and SSV2 from Sulfolobus islandicus [30] share the consensus R..KxxR..Y while the plasmidic integrase of Sulfolobus sp. NOB8H2 displays R..YxxR..Y [28].

Mobile elements therefore contribute to genome evolution through both site-specific and homologous recombination, which usually operate by distinct mechanisms and enzymatic activities. Homologous recombination is also known to occur frequently between multiple IS copies resulting in large scale archaeal genomic rearrangements, as observed in both Crenarchaeota e.g. Sulfolobus islandicus [31] and Euryarchaota e.g. Pyrococcus abyssi [32]. The distribution of archaeal ISs is patchy not only at the phylum level but also at genus level [9]. Interestingly, genome shuffling occurs in Thermococcus [33] even if ISs are seldom found in this genus suggesting that alternative recombination mechanisms are capable of producing large-scale genomic rearrangements.

If site-specific recombination only requires specific nucleotide sequences targeted by a dedicated recombinase, homologous recombination on the other hand is a much more complex process. In all organisms, homologous recombination constitutes one of several pathways to repair double-strand breaks. In addition to DNA synthesis, it requires dedicated recombinases and their accessory factors which act on stretches of near-sequence-identical DNA. In eukaryotic and bacterial cells, the enzymes and pathways involved in homologous recombination have been extensively studied (see [34,35] for reviews), whereas archaeal homologous recombination is still an active field of investigation. It is known that the initial resectioning step after double-strand break involves the Rad50–Mre11–HerA–NurA complex to generate 3’ single-strand substrates [36,37]. The RecA paralog RadA and its accessory functions associate with this ssDNA to constitute the presynaptic filament, which will scan and pair with homologous sequences [38]. In the archaeon Thermococcus kodakarensis, homologous recombination has been detected experimentally between stretches of identical DNA sequences equal to or greater than 500bp [39].

To our knowledge, a direct overlap between site-specific and homologous recombination processes has not been described so far. In the present work, we report the discovery and characterization of a new integrase from the hyperthermophilic archaeon Thermococcus nautili [40,41] capable of catalyzing both site-specific recombination and low sequence specificity recombination reactions mimicking homologous recombination. The wide distribution of this particular Y-recombinase among the Thermococcus genus provides a valid rationale for the observed genomic rearrangements in these Archaea.


Dotplot comparisons identify synteny breakpoints in Thermococcus chromosomes

We compared the chromosomes of the 13 completely sequenced Thermococcus species available to date by dotplot analysis and observed high levels of genome scrambling as shown in Fig 1A. Strikingly, comparison of T. onnurineus and T. sp. 4557 chromosomes by this approach revealed only two large inversions of 139/143Kb and 102/74Kb respectively (Fig 1B & 1C). This relatively small number of inversions facilitated the investigation of the synteny breakpoints bordering both inversions. Using the SyntTax web tool [42], a composite representation was obtained as shown in Fig 1C. Gene order is conserved immediately upstream and downstream of each inversion border and was used to identify the synteny breakpoints. For each inversion, the breakpoints are located within tRNA gene pairs, transcribed in opposite orientations. Interestingly, T. nautili plasmid pTN3 integrates in the tRNALeu gene BD01_0018 [41,43] (S2 Fig) and this gene displays over 97% sequence identity with tRNALeu (GQS_t10759), which borders a large chromosomal inversion between T. onnurineus and T. sp. 4557 (Fig 1B). The concordance between the chromosomal attachment site of the pTN3 integrase (IntpTN3) and the recombination targets bordering each inversion (in opposite orientations) led us to define a working model to explain the formation of genomic inversions observed in the Thermococcus genus. We hypothesize that the frequent genomic inversions observed in the evolution of the Thermococcales order are a result of enzymatic activity of the integrase encoded by horizontally mobile elements, such as pTN3.

Fig 1. Genomic dotplots and synteny analysis.

Genomic dotplots (A) between T. kodakarensis and T. nautili and (B) between T. onnurineus and T. sp. 4557. All genomes are centered on their putative predicted origin of replication [33]. C. The two synteny breaks in the genomic alignment between T. onnurineus and T. sp. 4557 (Panel B) were further analyzed. Gene order conservation and recombination endpoints of the two major inversions were identified using composite images generated by the SyntTax web tool. Inversion “1” occurred between tRNALeu (GQS_t10759) and tRNAThr (GQS_t10745) genes; T. sp. 4557 GQS_t10759 gene is orthologous to the T. nautili tRNALeu gene (BD01_0018) which corresponds to the chromosomal attachment site of plasmid pTN3. Inversion “2” (Panel B) occurred between tRNALeu (GQS_t10807) and tRNAGly (GQS_t10803) genes.

IntpTN3 is a bona fide tyrosine recombinase

The integrase of pTN3 shares significant sequence similarity with canonical Y-recombinases and its predicted active site can be defined as R..K..AxxR..Y which only slightly diverges from the consensus (S3A Fig). In addition, IntpTN3 displays a high degree of conservation with two biochemically characterized hyperthermophilic Y-recombinases, the archaeal IntSSV1 [44] and IntSSV2 [30] (S3B Fig). Thus, it seemed worthwhile to compare the enzymatic activities of IntpTN3 to those of other enzymes of the same family such as phage λ integrase and Saccharomyces cerevisiae 2μ plasmid FLP protein and to validate them against the canonical Y-recombinase model.

IntpTN3 is an active site-specific tyrosine recombinase

In order to characterize the activities of IntpTN3, it was necessary to over-produce and purify the enzyme (S4 Fig) and to construct DNA substrates carrying appropriate attachment sites (as determined by sequential deletions (S5 Fig). An integrase variant (IntpTN3Y428A) in which the catalytic tyrosine is substituted with an alanine was constructed, purified and tested (S6 Fig). We used these proteins and DNA components in a series of in vitro and in vivo experiments, detailed below, to ascertain the properties of IntpTN3.

IntpTN3 catalyzes attP-attB integration.

In order to measure the activity of purified IntpTN3, we initially developed a simple test in which integrase-catalyzed integration of one plasmid-encoded attB site in an identical site on a second plasmid results in formation of a plasmid-plasmid dimer (S1A Fig), which can be detected by gel electrophoresis. In accordance with our identification of tRNALeu as a potential attB site, we generated a supercoiled DNA template carrying a quasi-full-length T. nautili tRNALeu gene, Leu2-88 (see below). We observed the formation of dimeric DNA molecules only with DNA templates carrying attB tRNALeu, and only in the presence of IntpTN3 (Fig 2). Thus, the IntpTN3 is able to catalyze the site-specific recombination of one att site with another.

Fig 2. Dimer formation.

Supercoiled (SC) plasmids pUC18 and pJO322 carrying the Leu2-88 fragment (S5A Fig) were incubated with IntpTN3 in a standard reaction (see Materials and methods) and compared with linearized pJO322 by agarose gel electrophoresis. The integrase has no effect on pUC18 with the exception of the production of a faint linear species (indicated by an arrow). The integrase increases considerably the formation of plasmid pJO322 dimers and to a lower extent that of multimers. No increase in the formation of open circular (OC) form was observed.

IntpTN3 catalyzes attL-attR excision.

The capacity of IntpTN3 to catalyze the inverse reaction i.e. the excision of a DNA segment located between attL and attR sites was tested using the template pMC479, which carries a Leu2-88 site and a minimal Leu2-44 site in the same orientation, separated by a 762bp segment. In the presence of IntpTN3, the restriction digestion pattern revealed the presence of two bands of 2358 and 849bp, consistent with the excision of a circular DNA species between two attB sites (Fig 3). The recombination reaction also generated an additional band of 4056bp, explainable by the integration of the 849bp circular product into the initial pMC478 template. This demonstrates that IntpTN3 is able to efficiently catalyze both DNA integration and excision reactions.

Fig 3. Intptn3 excision and integration.

Plasmid pMC479 carries two copies of tRNALeu cloned in direct orientation and separated by a 762bp spacer fragment (see Material and methods). The direct repeats consist of the minimal tRNALeu 2–44 and the longer tRNALeu 2–88, both proficient in dimerization reactions. Plasmid pMC479 was incubated with IntpTN3 in a standard reaction (see Materials and methods). The NdeI restriction enzyme generates two fragments of 3207 and 1366bp respectively in pMC479. Upon incubation with IntpTN3, NdeI digestion generates additional fragments of 2358bp corresponding to recombined pMC479* and 849bp corresponding to the circularized spacer and recombined att site. Both constitute the products of the excision reaction. A larger 4056bp fragment is generated as well and corresponds to the recombination product generated by integration of the 3207 and 849bp species. The relative intensity of the bands is compatible with an expected equilibrium reaction.

IntpTN3 can re-activate related integrated mobile elements.

The species T. kodakarensis carries in its genome the stably integrated element TKV4 [45], which is closely related to pTN3 of T. nautili. As shown for pTN3 (S2 Fig), the integration of TKV4 into the T. kodakarensis genome has disrupted the gene encoding IntTKV4, rendering TKV4 incapable of spontaneous chromosomal excision. Considering that IntpTN3 and IntTKV4 display extensive sequence similarity (S3 Fig) and promote integration in orthologous tRNALeu genes [45], we investigated the capacity of IntpTN3 to excise TKV4 in vitro. Excision and circularization of a DNA molecule is detectable by PCR amplification using suitably oriented primers (Fig 4A). Treatment of T. kodakarensis genomic DNA with purified IntpTN3 resulted in products consistent with TKV4 circularization (Fig 4B), demonstrating that IntpTN3 could excise, and hence re-activate this dormant mobile element. In light of this in vitro activity, we endeavored to test this TKV4 resurrection reaction in vivo. This experiment involved the construction of specialized T. kodakarensis expression vectors pRC524 and pRC526 expressing wild type IntpTN3 and mutant IntpTN3Y428A respectively (Fig 4C) (see Material and methods). Surprisingly, both IntpTN3 and the active site mutant IntpTN3Y428A were able to revive TKV4 in vivo (Fig 4D). Not only does this result demonstrate the ability of pTN3 to excise, and therefore re-activate integrated mobile elements, it also strongly suggests that the activity of mutated IntpTN3Y428A could be complemented by the truncated IntTKV4 encoded by the integrated element, whereas both variants are inactive on their own. A similar phenomenon of complementation has been reported between a DNA-binding impaired mutant and a catalytic tyrosine residue mutant of IntSSV1 [44].

Fig 4. TKV4 excision in vitro and in vivo.

A PCR amplification assay was designed to assert artificial IntpTN3-mediated TKV4 circularization (Panel A). The assay was first performed in vitro on four samples of purified T. kodakarensis genomic DNA incubated with wild type IntpTN3 or inactive IntpTN3 Y428A mutated enzyme in a standard reaction analyzed by agarose gel electrophoresis (see Materials and methods). Only reactions using wild-type enzyme generated a 1710bp band of the expected excision size (Panel B). The same TKV4 excision reaction was tested in vivo by transforming T. kodakarensis KUW1 with shuttle plasmids pRC524 (expressing wild type integrase) and pRC526 (expressing mutated IntpTN3Y428A) or with the vector alone (Panel C). Total DNA was extracted from the transformants and amplified as described above. In this in vivo experiment, both enzymes were TKV4 excision-proficient (Panel D).

IntpTN3 catalyzes DNA inversion between att sites.

The ability of IntpTN3 to catalyze the inversion of DNA sequences is key in our model of large-scale integrase-mediated chromosomal rearrangements in the Thermococcus genus. To test the IntpTN3 invertase activity, we constructed a plasmid (pMC478) with two attachment sites in inverted orientation: the full-length tRNALeu gene and the minimal Leu2-44. The restriction digestion pattern showed the presence of two new bands corresponding to the inversion of the DNA segment between the attB sites only when DNA was treated with the integrase (Fig 5). This result indicates that, like the S. cerevisiae FLP recombinase, IntpTN3 is capable of efficiently performing all three canonical reactions characteristic of site-specific Y-recombinases: integration, excision and inversion. No recombination products could be observed in inversion reactions performed with the inactivated integrase variant IntpTN3Y428A (S6 Fig).

Fig 5. Intptn3 inversion.

Plasmid pMC477 carries two copies of tRNALeu cloned in inverted orientation and separated by a 892bp spacer fragment (see Material and methods). The inverted repeats consist of the minimal tRNALeu 2–44 and the longer tRNALeu 2–88, both proficient in dimerization reactions. Plasmid pMC473 carries tRNALeu 2–44 and tRNAThr GQS_t10745, in inverted orientation as well. Both plasmids were incubated with IntpTN3 in a standard reaction (see Materials and methods). The NdeI restriction enzyme generates in each case two fragments of 2796 and 1777bp. Upon incubation with IntpTN3, NdeI digestion of pMC477 generates additional fragments of 2358 and 2215bp corresponding to the recombinant pMC477*. As for the integration/excision reactions, the relative intensity of the bands is compatible with an expected equilibrium reaction. We could not detect any inversion between tRNALeu and tRNAThr in plasmid pMC473.

Synteny analysis of the inversion endpoints observed between T. sp. 4557 and T. onnurineus indicates that recombination may have occurred between different tRNA genes, namely between tRNALeu (GQS_t10759) and tRNAThr (GQS_t10745) as well as between tRNALeu (GQS_t10807) and tRNAGly (GQS_t10803). Interestingly, inversion templates combining tRNALeu and tRNAThr failed to produce recombination products (Fig 5).

Thermococcus nautili undergoes rapid genomic rearrangement under laboratory conditions

The large-scale genomic inversions observed between T. sp. 4557 and T. onnurineus display minor gene order rearrangements near the recombination endpoints indicating that these events are not recent and might have undergone remodeling (Fig 1C). In order to identify more recent rearrangements, we investigated whether large-scale genomic inversions could occur spontaneously under laboratory conditions. T. nautili carrying its natural plasmids was sub-cultured in two independent experiments for 60 and 66 generations (therefore termed T. nautili 60G and 66G) in rich liquid medium with intermittent storage at 4°C and the metagenomes of the resulting populations were completely re-sequenced. We observed in both T. nautili 60G and 66G sub-cultures a high proportion of a novel rearranged genome exhibiting four new large-scale chromosomal inversions when compared to the original published T. nautili genome (GenBank accession NZ_CP007264) [41] (Fig 6A). By mapping the frequency of the Illumina reads around the four inversion sites, we measured the incidence of the rearranged genome in the T. nautili 66G population, which was found in most cases to exceed that of the original genome (S3 Table). Both T. nautili 60G and 66G rearranged chromosomes were remarkably similar when compared by dotplot analysis (S7 Fig). Additionally, plasmid pTN3 was largely underrepresented in the T. nautili 66G sub-culture (S3 Table), whereas the smaller pTN1 and pTN2 were conserved. The chromosomally-integrated pTN3 copy carrying the disrupted integrase gene was also retained. The chain of nested inversion events leading to these new recombined genomes could be reconstructed (Fig 6C) and allowed us to analyze and precisely map the recombination endpoints. Each of the four genomic inversions occurred between paralogous gene pairs: between tRNAGly genes BD01_1557 and BD01_1976, between methyl accepting chemotaxis genes BD01_1166 and BD01_1584, between transposase genes BD01_1317 and BD01_1763 and finally between UDP-glucose-6 dehydrogenase genes BD01_1333 and BD01_1481. For each pair of paralogous genes, the inversion events always occurred between two inverted segments of DNA sharing extensive sequence identity (S8 Fig). However, we could not detect significant similarity between inverted DNA segments corresponding to different pairs of paralogous genes using BLAST (e-value ≥ 0.075). Furthermore, none of these sequences could be aligned with the original pTN3 attachment site, tRNALeu (e-value ≥ 10). In a control experiment, in contrast to T. nautili, the genome of a closely related organism, the plasmid-less Thermococcus sp. 5–4 (GenBank accession CP021848) remained stable when sub-cultured for 36 or 66 generations in two separate experiments (Fig 6B and S7 Fig).

Fig 6. Laboratory inversions events.

A. Dotplot analysis of the original isolate of T. nautili (GenBank accession NZ_CP007264) and the same organism after 66 generations (S2 Dataset). B. Dotplot analysis of the original isolate of T. 5–4 (GenBank accession CP021848) and the same organism after 66 generations (S4 Dataset). C. One of the possible sequential inversion scenarios leading to T. nautili 66G (Panel A), drawn to scale. The arrows direction reflects the chromosomal segment orientation in the original T. nautili strain. Genomic coordinates are indicated and the identifiers of the genes bordering each inversion are boxed.

IntpTN3 also catalyzes DNA inversion between non-att sites on the archaeal chromosome

The remarkable differences in the outcome of T. nautili and T. sp 5–4 sub-culturing experiments and the observation that tRNAGly genes could recombine in these conditions suggested a causal link between IntpTN3 and genome shuffling. To ascertain if the new recombinations in T. nautili 60G and 66G could have been indeed generated by IntpTN3, we decided to test whether this integrase was able to catalyze in vitro inversions using the sequences detected at the borders of these recombination events. New inversion templates pCB548 and pCB552 were thus constructed respectively carrying sequences encompassing tRNAGly genes BD01_1557 and BD01_1976 or sequence fragments from chemotaxis genes BD01_1166 and BD01_1584 (S8 Fig). To limit the number and size of generated fragments, an in vitro inversion assay was conducted on linear fragments originating from these plasmids and compared to a linear fragment carrying inverted attP sites derived from pCB524. Inversions could be detected with all three templates albeit with significantly longer incubation times or higher IntpTN3 concentrations for pCB548 and pCB552-derived templates as compared to pCB524 (Fig 7). To confirm this recombination event, one of the products of the pCB548 template inversion reaction was further characterized by DNA sequencing and corresponded to a bona fide cross-over between BD01_1557 and BD01_1976 (S9 Fig). We conclude that IntpTN3 is able to catalyze low sequence specificity recombination reactions between sites that differ in sequence from its cognate att site, with the same outcome as homologous recombination events. It is to be noted that IntpTN3 catalyzes these two types of reactions with a different efficiency. Site-specific recombination reactions reach the equilibrium within 30 minutes whereas several hours and higher enzyme concentrations are required to detect all low sequence specificity recombinations.

Fig 7. IntpTN3-promoted low sequence specificity reactions on archaeal sequences.

IntpTN3 catalyzes inversion on linear DNA substrates between archaeal gene pairs separated by a Kanamycin resistance determinant. White arrowheads refer to original fragments and black arrowheads indicate inversions products. A. Inversion between two identical copies of tRNAleu gene GQS_t10759 from T. sp. 4557. B. Inversion between tRNAGly genes BD01_1557 and BD01_1976 from T. nautili. C. Inversion between chemotaxis genes BD01_1166 and BD01_1584 from T. nautili. IntpTN3 concentration multipliers refer to the standard assay described in Materials and Methods. The detailed DNA sequences involved in these reactions are illustrated in S8 Fig.

IntpTN3 catalyzes low sequence specificity recombination reactions mimicking homologous recombination between any DNA sequence pairs

The absence of inter-pair DNA similarity observed in T. nautili 60G and 66G chromosomal inversions prompted us to test whether IntpTN3 could catalyze recombination between homologous non-archaeal sequences. The simplest experiment consisted of the incubation of cloning vector pBR322 DNA with the integrase in the same conditions as described above. This recombination reaction promoted by IntpTN3 yielded a ladder of plasmid multimers produced by sequential integration, which could be readily observed by eletrophoretic migration whereas no homologous integration reaction was detected with the mutated IntpTN3Y428A (Fig 8A). Surprisingly, IntpTN3 generated also a double-strand cut at the pBR322 ColE1 origin of replication for which we have no explanation at this stage (S10 Fig). This cleavage does not constitute an intermediate step in the recombination reaction since none of IntpTN3 linear substrates shown in Fig 7 carries the ColE1 origin. In addition to the homologous integration reaction, we investigated the capacity of IntpTN3 to promote inversions between homologous sequences of bacterial origin. Short DNA segments of decreasing length (250, 175 and 100bp, see S11 Fig) originating from the E. coli lacZ gene were cloned in opposite orientations respective to the lacZα gene of pUC18 to generate plasmids pCB574, pCB571 and pCB558, respectively. These templates were linearized, incubated with IntpTN3 and tested by subsequent restriction analysis. In each case, IntpTN3 generated additional bands consistent with homologous inversion reactions displaying efficiencies proportional to the extent of DNA identity (Fig 8B).

Fig 8. IntpTN3-promoted low sequence specificity reactions on exogenous sequences.

A. Low sequence specificity reactions mimicking homologous DNA integration are visualized by the accumulation of multimers of increasing size only when the reaction occurs in the presence of wild-type IntpTN3. A linear pBR322 species generated by IntpTN3-generated double-strand cleavage is visible and migrates close to a control plasmid digested by the EcoRI endocnuclease. OC and SC refer to the open circle and supercoiled DNA forms, respectively. B. IntpTN3 catalyzes inversion on linear DNA substrates between two inverted E. coli lacZ gene segments of varying sizes separated by a Kanamycin resistance determinant. The sequence identity between the inverted segment amounts to 250, 175 and 100bp respectively in plasmids pCB574, pCB572 and pCB538 (see Materials and methods). White arrowheads refer to original fragments and black arrowheads indicate inversions products. IntpTN3 concentration multipliers refer to the standard assay described in Materials and Methods.


The major mechanism producing chromosomal rearrangements is recombinational exchange between homologous sequences [46]. These rearrangements often consist of DNA inversions between IS elements [9,46,47]. The observation that, in the Thermococcus genus, large chromosomal inversions occur even in the absence of IS elements prompted us to investigate the molecular mechanism behind these rearrangements. The presence of tRNA genes at recombination endpoints in genomes as diverse as plant chloroplasts [48,49] and Thermococcales [9], combined with the fact that integrases often target tRNA genes [50], lead us to propose a precise molecular model involving IntpTN3 to explain large-scale genomic rearrangements. Using a combination of comparative genomics, in vitro analyses, and serial culturing experiments, we uncovered a mechanism and enzymatic activity responsible for the shuffling-driven chromosomal evolution in Thermococcales. By means of deep comparative genomic analyses, we were able to correlate genome scrambling with the presence of a mobile element. This mobile element has been identified as plasmid pTN3, naturally present in T. nautili both as an episome and integrated in the genome [41,43]. Plasmid pTN3 encodes the IntpTN3 integrase of the Y-recombinase superfamily capable of promoting its site-specific plasmid integration at a tRNALeu gene of its host. Due to perfect DNA conservation between attB and attP attachment sites (S2B Fig), an intact and presumably expressed tRNALeu is reconstituted upon pTN3 chromosomal integration. We successfully reproduced, with high efficiency in a purified in vitro system, the canonical DNA reactions of integration and excision expected from a bona fide integrase. Site-specific mutation of the active site tyrosine to alanine abolished these activities. A positive excision reaction was also obtained in vivo by expressing wild-type IntpTN3 and the catalytic tyrosine mutant IntpTN3Y428A in T. kodakarensis KOD1 cells. The genome of this strain carries the integrated episome TKV4 [45] which is remarkably similar to pTN3 (Fig 9). Surprisingly, both wild-type and mutant forms of the integrase excised TKV4 in circular form. This suggests that a truncated C-terminal IntTKV4, presumably impaired in DNA-binding but carrying the catalytic tyrosine, can complement IntpTN3Y428A. A plausible explanation invokes the participation of integrase dimers in the recombination reaction. In this case, only the heterodimeric form would possess an active catalytic site where Tyr428 is provided by the first monomer while the second monomer contributes the remaining conserved residues. This cleavage in trans was initially reported for the FLP recombinase [51,52]. Similarly, the complementation of activity between a DNA-binding impaired mutant and a catalytic tyrosine residue mutant has been described for another archaeal integrase, IntSSV1 [44].

Fig 9. pTN3-like integrated elements in Thermococcales.

The presence of pTN3-like integrated elements was investigated in all completely sequenced Thermococcales genomes by synteny analysis using the SyntTax web server [42]. In addition to T. nautili, the genomes of T. guaymasensis DSM11113, T. eurythermalis A501, T. kodakarensis KOD1, T. barophilus CH5, and T. cleftensis CL1 carry an extensive genomic region corresponding to plasmid pTN3 shown on top. Each arrow corresponds to an individual gene numbered according to GenBank annotations. The consistent gene color code illustrates orthology across organisms while white color indicates its absence. As indicated by a blue dotted line, conservation of synteny is clearly visible on the right border and limited by the gene encoding pTN3 C-ter integrase and its remnants. Truncated N-terminal-encoding integrase genes constitute pseudogenes lacking a stop codon and are therefore not annotated. Genetic divergence appears stronger on the left border.

The peculiar location of tRNALeu GQS_t10759 at the exact border of a large DNA inversion observed between the genomes of T. onnurineus and T. sp. 4557 suggested that this inversion could have occurred by the recombinase activity of IntpTN3. In our purified system, we could obtain highly efficient DNA inversions between two inverted copies of GQS_t10759. Paradoxically, we were unable to promote inversion between tRNALeu GQS_t10759 and tRNAThr GQS_t10745 contrary to what the genomic comparisons between T. onnurineus and T. sp. 4557 suggested. An experiment of prolonged T. nautili cultivation was instrumental in elucidating the large-scale inversion mechanism in Thermococcus. The strain carrying its natural plasmids was cultivated during 60 or 66 generations; total DNA was extracted from this population and sequenced in a manner similar to a metagenome. We observed the high incidence in the resulting populations of a particular recombined genome with four large chromosomal inversions and a very low copy number of plasmid pTN3 encoding active IntpTN3 (< 2/chromosome) (S3 Table). This plasmid loss could have contributed to the higher fitness and spread of a particular clone in the population. The four large-scale inversions occurred between four pairs of naturally occurring paralogous genes sharing at least 104bp of sequence identity in inverted orientation (S8 Fig). No significant sequence conservation could be detected between the four pairs. We did not observe chromosomal rearrangements after prolonged incubation of Thermococcus sp. 5–4, which does not carry plasmids. The potential causal link between pTN3 and a number of unrelated sequence pairs involved in large scale genomic shuffling in T. nautili was difficult to conciliate with the classical site-specific recombination properties we described for IntpTN3. Remarkably, by in vitro assays with this integrase, we succeeded in producing inversions between several pairs of inverted paralogous genes detected in our T. nautili sub-culturing experiments. These results suggested that the recombination properties of IntpTN3 could be extended to virtually any homologous pair of DNA sequences. Using exogenous pBR322 plasmid DNA or genes segments from bacterial origin, we demonstrated in vitro that IntpTN3 actively promotes low sequence specificity reactions mimicking homologous integration and inversion of any sequence pair as short as 100bp. The catalytic site mutation in variant IntpTN3Y428A abolishes this particular recombination reaction as well. Interestingly, cellular homologous recombination in Archaea operates according to a different pathway with dedicated enzymes [36,37] and in Thermococcus kodakarensis has only been reported between DNA segments of 500bp or more [39].

These reactions unveiled a specific IntpTN3-generated double-strand cut at the ColE1 origin of replication carried by pBR322 and its derivatives (S10 Fig). At this moment, we do not have a precise rationale to explain this observation other than a potential distant secondary structure similarity between the small RNAI and RNAII encoded by the ColE1 origin and the tRNALeu encoded by IntpTN3 attB substrate. Biological interactions between tRNAs and ColE1 RNAs have been reported [53]. Clearly, this double-strand cleavage does not participate in any recombination reaction since we demonstrated all in vitro IntpTN3 inversions on linear DNA segments devoid of ColE1 origin.

The positive in vitro IntpTN3-promoted low sequence specificity recombination results explain the failure of this enzyme to promote inversion between tRNALeu GQS_t10759 and tRNAThr GQS_t10745. These sites were initially thought to constitute inversion endpoints between the genomes of T. onnurineus and T. sp. 4557 but do not share sufficient sequence similarity to be efficiently recombined in vitro. The particular positioning of these sequences in opposite orientations could have occurred through previous overlapping inversions between a different set of paralogs or by less frequent native homologous recombination. We observed a similar situation in the sequence of the T. nautili 60G and 66G populations. In several cases, homologous segments were in direct orientation in the original genome but became opposed due to a previous overlapping inversion therefore indicating that T. nautili 60G and 66G inversions occurred sequentially.

In order to investigate whether pTN3 could account for large-scale rearrangements in the Thermococcus genus, we examined by synteny analysis the distribution of pTN3-like integrated element among completely sequenced Thermococcales. Out of 17 sequenced Thermococcus, and in addition to the previously reported T. kodakarensis TKV4 element [45], five isolates were found to harbor a pTN3-related element (Fig 9). The natural competence for DNA uptake of some Thermococcales such as T. kodakarensis [39] and the capacity of pTN3 to be transferred between cells using membrane vesicles [43] could explain the ubiquitous presence of this mobile element.

Protein sequence and structural comparisons between IntpTN3 and other hyperthermophilic archaeal integrases such as that of crenarchaeal virus SSV1 indicate that these proteins are clearly related. However, IntpTN3 possesses several additional interspersed domains relative to SSV1 (S2 and S12 Figs). We surmise that these additional domains contribute to the low sequence specificity recombination reactions akin to homologous recombination events that we have observed.

By summing up all direct and indirect evidence reported here, it is very likely that the integrase encoded by pTN3-like plasmids can account for the genomic shuffling observed in the Thermococcus genus. Plasmids of the pTN3 class are genetically closely related to viruses as they encode a capsid protein and a DNA packaging ATPase [43] but pTN3 virions have not be observed to date. It is not clear at this stage whether plasmids or viruses equipped with an IntpTN3-like integrase have a better fitness either due to provirus maintenance or by virion spreading. An integrase mimicking homologous recombination could promote viral integration into the host genome only if both viral and cellular chromosomes share significant DNA similarity. This enzyme however, could facilitate integration of a virus into the genome of a closely related provirus.

The question arises whether an enzyme promoting genome shuffling using very short repeated segments as substrates, would be beneficial for a cellular organism. On one hand, ‘wrongly’ recombined genomes would result in suboptimal gene expression programs and cells carrying scrambled genomes would display a reduced fitness and clearly be counter-selected in the population. Interestingly, the presence of a pTN3-specific spacer in a T. nautili CRISPR locus strongly suggests that the presence of this plasmid is deleterious [41]. On the other hand, it is also possible to envision situations where high-level genome shuffling by inversion could be advantageous. Alternate gene expression patterns could increase, for instance, adaptation to rapid environmental changes. In addition, for organisms such as Thermococcales where highly-expressed essential housekeeping genes maintain invariable positions [33], genome scrambling could be beneficial by relocating “less desirable” integrated elements to chromosomal areas of reduced gene expression, therefore minimizing their impact on cellular physiology.

Materials and methods

Bacterial, archaeal strains, plasmids and media

Escherichia coli strain XL1-Blue was used for cloning, plasmid amplification and site-directed mutagenesis. Overexpression of recombinant wild-type or mutant IntpTN3 was carried out in strain BL21 (DE3) (Novagen). All E. coli strains were grown in Luria-Bertani medium supplemented with 100μg/mL ampicillin or/and 50μg/mL kanamycin when necessary. T. kodakarensis KUW1 (ΔpyrF ΔtrpE) was grown anaerobically in ASW-YT medium [54] at 85°C. Long term Thermococcus sub-culturing experiments were carried out in the same conditions by sequential 50x dilutions of stationary phase cultures into fresh media. The number of generations was assessed statistically at each dilution step using a Thoma cell counting chamber under 400x magnification. The plasmids used or constructed in this work are listed in S1 Table. Transformation with pRC524 and pRC526 plasmids (see below) was performed following standard protocols [55]. Plasmid-containing KUW1 strains were grown in ASW-CH medium [54] supplemented with uracil (10 μg/mL). T. nautili sp. 30–1 (CP007264) was grown anaerobically at 85°C in Zillig’s broth [56].

Bioinformatics and sequencing

Genomic sequences were compared and aligned by dotplot analysis using Gepard [57]. Conservation of gene order was assessed by synteny analysis using Absynte [58] and SyntTax [42]. The original genome of Thermococcus 5–4 JCM31817 (GenBank accession CP021848) and the genomes of sub-cultured T. nautili 60G and 66G and T. sp. 5–4 36G and 66G were sequenced by Genoscope (Centre National de Séquençage, France), using Illumina MiSeq. Reads were assembled with Newbler (release 2.9) and gap closure was performed by PCR, Sanger sequencing and Oxford Nanopore MinION. The primary genomic sequences of rearranged T. nautili 60G, 66G and T. 5–4 36G, 66G are available in S1, S2, S3 and S4 Datasets, respectively. These genomic sequences are compared by dotplot analysis in S7 Fig.

Metagenome analysis

Genomic regions corresponding to ~2000bp upstream and downstream of inversion break-points were extracted from both the ancestral T. nautili sequence, and the sub-cultured T. nautili 66G sequence. Illumina sequencing reads were mapped to the ancestral sequence, and the pool of unmapped reads were mapped to the 66G sequence (Geneious 6.1.8). Two positions close to the break-point which differ in base composition between ancestral and 66G sequences were chosen to classify reads as resulting from original or inverted genome sequences. Bases were enumerated at these positions, and the percentage of reads corresponding to original sequences or inversions were calculated. The prevalence of pTN3 in the population was determined by comparing read depth across the entire T. nautili 66G genome (excluding the integrated pTN3 region) to that of pTN3 (S3 Table).

Recombinant protein expression and purification

The gene encoding the integrase of the plasmid pTN3 of T. nautili 30–1, (gene ID: 17125032) was codon-optimized for expression in E. coli and synthesized by GenScript. The synthetic gene contained a Strep-Tag at the 5’ end and was cloned into pET26b+ expression vector (Novagen) to yield pJO344. Plasmid pJO496 carrying the mutated IntpTN3Y428A was obtained by site directed mutagenesis of pJO344 with primers Int_A and Int_B (S2 Table) using the Agilent QuikChange Lightning Site-Directed Mutagenesis Kit. Wild-type IntpTN3 and mutated IntpTN3Y428A were purified from E. coli BL21 (DE3) strain (Novagen) harboring respectively pJO344 or pJO496 by affinity chromatography and gel filtration (S4 Fig). All integrase enzymatic assays were conducted with strep-tagged protein derivatives.

Integrase plasmid substrates

Plasmids used for the integrase dimerization assays were constructed as follows. EcoRI and BamHI restriction sites were added respectively at the 5’ and 3’ end of the various oligonucleotides shown in S5 Fig. Each oligonucleotide (Sigma-Aldrich) was annealed to its complementary sequence and the resulting double-stranded segments were cloned between the corresponding restrictions sites of pUC18. To generate plasmid pMC451, the Leu2-88 fragment was cloned in pBR322 instead of pUC18. Plasmids pMC477 and pMC479 used respectively for att integration/excision and inversion assays were constructed using pMC451 as backbone. The insertion fragment was amplified with primers Leu43scaI_fw and Leu43scaI_rev using pMC449 plasmid DNA as template. It contains tRNALeu gene (2-44bp) and lacZa gene for blue-white screening. This amplified region was cloned in pMC451 in both possible orientation using ScaI and NruI blunt sites. Plasmid pCB538 was obtained by amplifying with primers LacZ100-Sac1-For and KanR-Xba1-Rev (S2 Table) a 1364bp fragment from pUC4K and subsequent cloning between the XbaI-SacI sites of pUC18. The other plasmids: pCB548, pCB552, pCB572 and pCB574 used for non-att inversion assays were generated by Gibson Assembly [59]. Briefly, for pCB548, the genomic region corresponding to -80 to +245 of BD01_1557 (T. nautili) was amplified by PCR (Phusion Polymerase, ThermoScientific) using primers 1557_fwd and 1557_rev (S2 Table); the region from –80 to +245 of BD01_1976 was amplified using primers 1976_fwd and 1976_rev. The KmR gene was amplified from plasmid pUC4K using primers KanR_fwd and KanR_rev. Fragments were assembled into EcoRI + SalI digested pUC18 using the NEBuilder HiFi DNA Assembly Master Mix (New England Biolabs) following the manufacturer’s protocols. Similarly, for pCB552, part of the genes BD01_1166 and BD01_1584 (S8 Fig) were amplified by PCR and assembled into EcoRI + SalI digested pUC18 with the KmR gene sequence. To construct pCB538, a fragment containing KmR and the beginning of the lacZ gene (lac100) was PCR-amplified from pUC4K with the primers LacZ100-Sac1-For and KanR-Xba1-Rev containing the restriction sites for SacI and XbaI, respectively, at the 5’ end. The adequately digested fragment was then ligated into a SacI-XbaI digested pUC18. For plasmids pCB572 and pCB574, part of the lacZ gene was amplified from pUC18 and the KmR gene sequence was amplified from plasmid pUC4K. The two fragments were then assembled into the EcoRI digested pUC18. Purified plasmids pCB548, pCB552, were digested using ScaI and EcoRI and plasmids pCB572 and pCB574 were digested using ScaI. The fragments containing the non att-sites were then gel purified using the kit NucleoSpin Gel and PCR Clean-up (Macherey Nagel). All plasmid constructs were confirmed by DNA sequencing (Beckman Coulter Genomics).

In vitro/in vivo integrase enzymatic assay

Standard in vitro integrase assays were performed as follows: 165ng (8.25ng/μL, 3.1pmol) purified IntpTN3 and 0.5μg (25ng/μL, 10pmol) supercoiled plasmid substrates were incubated 30 min at 65°C in a reaction buffer containing 300mM KCl, 27 mM Tris HCl pH8, 0.17mM DTT and 1mM MgSO4. Depending on the size of the plasmid substrate, the DNA/integrase molar ratio varied from 30 to 60. For substrates with non-att sites, the integrase concentration was increased up to 50pmol. To assay dimer formation, the reaction products were separated by gel electrophoresis and visualized with ethidium bromide. For the excision and inversion assays, reaction products were purified with the NucleoSpin Gel and PCR Clean-up kit (Macherey-Nagel) and digested with appropriate restriction enzymes (Thermo Scientific) prior to eletrophoretic separation. In vitro circularization of TKV4 was performed in a standard integrase assay with genomic DNA of T. kodakarensis isolated as described previously [60]. The reaction products were purified using NucleoSpin Gel and PCR Clean-up kit (Macherey-Nagel). Recircularized products were scored by amplifying a reconstituted full-length TKV4 integrase gene. PCR was performed using Phusion Polymerase (ThermoScientific) and primers TKV4_FW and TKV4_REV (S2 Table) in conditions recommended by the supplier. In vivo circularization of TKV4 was obtained using total DNA from T. kodakarensis KUW1 transformed with plasmid pRC524 or pRC526. These plasmids express constitutively wild type integrase and mutated IntpTN3Y428A from the PhmtB promoter present in parental pLC70. DNA extraction and PCR reactions was performed as per the in vitro assay described above. To generate plasmids pRC524 and pRC526, the IntpTN3 integrase gene was amplified by PCR with primers int_fwd and int_rev (S2 Table), using total T. nautili genomic DNA as a template. The amplification product was cloned into pJET1.2 using the CloneJET PCR Cloning Kit (Thermo Fischer Scientific). The Y428A mutation was introduced into the integrase gene using the QuickChange II Site Directed Mutagenesis Kit (Agilent Technologies) with primer intY428A_fwd and its reverse complement. Both the wild-type and Y428A alleles were digested from pJET1.2 using SalI and NotI and cloned into the corresponding sites of pLC70. All in vitro and in vivo recombination junctions and plasmid constructs were confirmed by DNA sequencing (Beckman Coulter Genomics).

Supporting information

S2 Table. Oligonucleotides used in this work.


S3 Table. Metagenomic reads mapping (T. nautili 66G).


S1 Fig. Classical site-specific recombination model.

A. The intermolecular site-specific integration between cognate attP and attB sites generates a co-integrate with recombined attL and attR sites in direct orientation. The reverse reaction of excision regenerates the original components. B. In the intramolecular site-specific inversion reaction, the att sites are in opposite orientation. This reaction is reversible as well.


S2 Fig. pTN3 integration.

A. The comparison between the replicative and the chromosomal integrated forms of plasmid pTN3 enabled us to reconstitute the integration event. A stretch of 41bp is shared by both attP and attB sites. The nucleotides corresponding to the leucine anticodon are underlined. Upon integration, the integrase gene is disrupted and a full length tRNALeu gene is reconstituted although separated from its original promoter. An excision event would regenerate the original recombination partners. B. DNA sequence alignment between the integrase gene of pTN3 (black) and the tRNALeu gene (red). The start and stop codons of the integrase open reading frame are boxed in blue. The integration sites attP and attB as defined by Krupovic & Bamford [5] are boxed in their respective color.


S3 Fig. Tyrosine recombinases sequence comparison.

A. Alignment of IntpTN3 with tyrosine recombinases from the three domains of life. The protein sequence of IntpTN3 (WP_022547007.1) is aligned using Praline [6] with the reconstituted integrase from T. kodakarensis TKV4 and other previously characterized tyrosine recombinases from the three domains of life. These recombinases consist of the integrases from Sulfolobus Spindle Viruses SSV1 (P20214.1) and SSV2 (NP_944456.1), phage λ integrase (ALA45781.1), phage HP1 integrase (NP_043466.1), XerD resolvase from Escherichia coli (NP_417370.1) and FLP recombinase from Saccharomyces cerevisiae 2μ plasmid (P03870.1). The region corresponding to the catalytic signatures (BoxI, Kβ, BoxII) of crystallized tyrosine recombinases are boxed in light gray. The predicted residues composing IntpTN3 catalytic site are shown (R..K..AxxR..Y) and the catalytic tyrosine residue is indicated by a black arrow. The color code refers to the extent of residue conservation at each position as show in the color scale. B. Alignment of IntpTN3 with IntTKV4 and the hyperthermophilic tyrosine recombinases IntSSV1 and IntSSV2. Global protein sequence similarities were computed with the Needleman-Wunsch algorithm (Needle EMBOSS, IntpTN3-IntTKV4: 93.6%; IntpTN3-IntSSV1: 33.0% and IntpTN3-IntSSV2: 31.2%.


S4 Fig. Intptn3 overexpression and purification.

A. Protein expression was induced with 1mM IPTG in 1L of LB medium; cells harvested by centrifugation, and lysed by sonication. The soluble fraction of the sonicate was heated at 65°C for 10 minutes, and denatured proteins removed by centrifugation and by passing through a 0.45 μm filter. Strep-tagged proteins were purified by affinity fractionation using a Strep-Tactin column (IBA Lifesciences) as recommended by the supplier. B. Strep-Tactin fractions 4 and 5 were pooled and submitted to gel filtration (Superdex 200 16/600, GE Healthcare). C. Gel filtration fractions 21 to 31 were pooled and the purified protein was concentrated with an Amicon 3kDa cutoff concentrator (Millipore), aliquoted and stored at -80°C.


S5 Fig. AttB nested deletions.

The Integrase dimerization test was used to determine the minimal site required for IntpTN3 tRNALeu × tRNALeu recombination on nested deletions carried by plasmid templates. A. DNA sequence of the nested deletions. DNA segments corresponding to theses sequences were annealed and cloned directionally in pUC18. B. The resulting supercoiled plasmids were incubated with purified IntpTN3 in a standard reaction and scored for dimer formation by agarose gel electrophoresis where only relevant reactions are shown. The dimerization-proficient sequences in Panel A are marked as positive. It is noteworthy that the Leu41 site, a site corresponding to the 41bp of sequence identity shared by both attP and attB is not a sufficient substrate for this reaction. Therefore, the minimal site for efficient dimerization is Leu2-44 with a size of 43bp. The asterisks indicate the extent of sequence identity between chromosomal attB and pTN3 attP. The leucine CAA anticodon is underlined.


S6 Fig. Mutated IntY428A assay.

Increasing amounts of wild type IntpTN3 and mutated IntpTN3Y428A enzymes were incubated with plasmid pMC477 as substrate to analyze the inversion properties. The experimental conditions are those of the standard integrase assay (see Material and methods) except that increasing amounts of enzyme were used: 0.5, 1, 1.5, 2.5 and 5μg, respectively. No inversion is detectable with IntpTN3Y428A.


S7 Fig. Subcultures genome comparisons.

Dotplot alignment of the prominent genomes obtained after T. nautili 60G and 66G subculturing (left) and T. 5–4 36G and 66G (right).


S8 Fig. Detailed mapping of the IntpTN3-promoted in vivo inversions between four pairs of T. nautili paralogs.

The sequences corresponding to the four genomic crossovers observed in T. nautili 60G and 66G were identified each time in pairs of paralogous genes shown aligned here. The sequences blocked in grey throughout the figure refer to perfectly conserved DNA segments in each paralogous pair where recombination occurred. Short sequences boxed in red refer to open reading frames start and stop codons when applicable (see also Fig 7 for throughout consistent color-coding). Panel A shows the alignment between segments overlapping tRNAGly genes BD01_1557 and BD01_1976. The precise regions corresponding to both tRNAGly genes are boxed in black. DNA segments cloned in pCB548 indicated by green blocks refer to BD01_1557-related sequences while red blocks correspond to BD01_1976-related sequences. The BD01-1976 nucleotide highlighted in black corrects a sequencing error in the original T. nautili genome sequence. A 176bp segment (grayed) is perfectly conserved between BD01_1557 and BD01_1976. Gly anticodons are boxed in yellow color. Panel B displays the alignment between methyl accepting chemotaxis genes BD01_1166 and BD01_1584. DNA segments cloned in pCB552 indicated by yellow blocks refer to BD01_1166-related sequences while blue blocks correspond to BD01_1594-related sequences A 176bp segment (grayed) is perfectly conserved between BD01_1166 and BD01_1584. Panel C displays the alignment between UDP-glucose-6 dehydrogenase genes BD01_1333 and BD01_1481. The two separate regions of extended sequence identity (I and II) are found between these genes respectively 284 and 620bp long (greyed). The presence of gene conversion in the interval between these two regions suggests that both were presumably involved in distinct crossover events. Panel D shows the alignment between transposase genes BD01_1317 and BD01_1763. The shortest recombination segment (104bp, grayed) is shared between these two paralogous genes.


S9 Fig. Detailed characterization of IntpTN3-promoted in vitro inversion event by DNA sequencing.

Specific sequences surrounding tRNA gene BD01_1976 are blocked in red while specific sequences surrounding tRNA gene BD01_1557 are blocked in green. Relevant anticodon sequences are boxed in yellow color. Two nucleotide mismatches between these tRNA genes are blocked in black. The tripartite composition of these DNA segments is further highlighted by blocking in grey color the stretch of identical sequenced shared by the DNA fragments carrying BD01_1976 and BD01_1557. Panel A depicts the sequence of steps involved in generating a suitable recombinant fragment for DNA sequencing. Plasmid pCB548 carries DNA segments containing T. nautili tRNAGly-encoding genes BD01_1976 and BD01_1557 in inverted orientation and separated by a Kanamycin resistance determinant originating from pUC4K. The exact sequence of the cloned DNA segments encompassing BD01_1976 & BD01_1557 is displayed in S8A Fig. The inversion reaction was performed as shown in Fig 7B: an EcoRI-ScaI fragment originating from pCB548 was incubated with IntpTN3 after which the 601bp EcoRI-NruI fragment generated by IntpTN3 recombination was gel-purified, PCR-amplified with the forward primer 5’-ccgtttaatcgtcgcgcggaagc-3’ targeting the upstream sequence of the tRNAGly gene BD01_1976 and the reverse primer 5’-cccgttgaatatggctcataacaccc-3’ targeting the beginning of the KanR cassette. The resulting fragment was submitted to Sanger DNA sequencing using the forward primer mentioned above. Panels B and C display also the alignment between the 5’ half of both tRNA genes and the minimal Leu2-44 segment involved in IntpTN3 site-specific recombination. Panel D shows the result of the DNA sequencing reaction. The crossover point in the recombination reaction occurred precisely downstream of the two nucleotide mismatches mentioned above, in the sequence blocked in grey corresponding to the 3’ half of the tRNA genes and strictly conserved sequences immediately following. The sequences boxed in black in Panels B,C and D correspond to the exact extents of tRNAsGly.


S10 Fig. Integrase-promoted double-strand cut at ori ColE1.

Circular plasmid pCB548 (4675bp) treated with IntpTN3 and digested with XhoI-NdeI endonucleases generates bands of 2966 and 1709bp due to integrase-promoted low sequence specificity recombination (white arrowheads). The original larger 3896bp XhoI-NdeI fragment undergoes an additional double-stranded cut at the plasmid ColE1 origin of replication to generate fragments of ~2400 and ~1500bp (black arrowheads). IntpTN3 concentration multipliers refer to the standard assay described in Materials and Methods.


S11 Fig. LacZ gene segments used for low sequence specificity reactions mimicking homologous recombination.

DNA sequence of the lacZ gene segments cloned in plasmids pCB538 (lac100), pCB572 (lac175) and pCB574 (lac250) (Fig 8B).


S12 Fig. Integrase structure comparisons.

The catalytic domain of IntpTN3 (B) was modeled using Phyre2 [7] and compared using PyMol (The PyMOL Molecular Graphics System, Version 1.8 Schrödinger, LLC.) with the tridimensional structure of the integrase of Sulfolobus solfataricus virus SSV1 (PDB 3VCF) (A) determined by Zhan et al [8]. The IntpTN3 catalytic tyrosine residue is highlighted.


S1 Dataset. Thermococcus nautili 60G nucleotide sequence.

Predominant T. nautili chromosome sequence obtained after sub-culturing for 60 generations.


S2 Dataset. Thermococcus nautili 66G nucleotide sequence.

Predominant T. nautili chromosome sequence obtained after sub-culturing for 66 generations.


S3 Dataset. Thermococcus 5–4 36G nucleotide sequence.

Predominant T. 5–4 chromosome sequence obtained after sub-culturing for 36 generations.


S4 Dataset. Thermococcus 5–4 66G nucleotide sequence.

Predominant T. 5–4 chromosome sequence obtained after sub-culturing for 66 generations.


S1 Text. Supporting information references.


Author Contributions

  1. Conceptualization: JO.
  2. Data curation: VB.
  3. Funding acquisition: PF.
  4. Investigation: MC CB RC DG.
  5. Resources: EM.
  6. Supervision: JO.
  7. Visualization: JO.
  8. Writing – original draft: JO.
  9. Writing – review & editing: MC CB RC DG EM VB PF JO.


  1. 1. Lim K, Furuta Y, Kobayashi I (2012) Large Variations in Bacterial Ribosomal RNA Genes. Molecular Biology and Evolution 29: 2937–2948. pmid:22446745
  2. 2. Anderson P, Roth J (1981) Spontaneous tandem genetic duplications in Salmonella typhimurium arise by unequal recombination between rRNA (rrn) cistrons. Proceedings of the National Academy of Sciences of the United States of America 78: 3113–3117. pmid:6789329
  3. 3. Dorgai L, Oberto J, Weisberg RA (1993) Xis and Fis proteins prevent site-specific DNA inversion in lysogens of phage HK022. J Bacteriol 175: 693–700. pmid:8423145
  4. 4. Iguchi A, Iyoda S, Terajima J, Watanabe H, Osawa R (2006) Spontaneous recombination between homologous prophage regions causes large-scale inversions within the Escherichia coli O157: H7 chromosome. Gene 372: 199–207. pmid:16516407
  5. 5. Busseau I, Pelisson A, Bucheton A (1989) I-Elements of Drosophila-Melanogaster Generate Specific Chromosomal Rearrangements during Transposition. Molecular & General Genetics 218: 222–228.
  6. 6. Cui LZ, Neoh HM, Iwamoto A, Hiramatsu K (2012) Coordinated phenotype switching with large-scale chromosome flip-flop inversion observed in bacteria. Proceedings of the National Academy of Sciences of the United States of America 109: E1647–E1656. pmid:22645353
  7. 7. Daveran-Mingot ML, Campo N, Ritzenthaler P, Le Bourgeois P (1998) A natural large chromosomal inversion in Lactococcus lactis is mediated by homologous recombination between two insertion sequences. Journal of Bacteriology 180: 4834–4842. pmid:9733685
  8. 8. Schindler D, Echols H (1981) Retroregulation of the int gene of bacteriophage lambda: control of translation completion. Proceedings of the National Academy of Sciences of the United States of America 78: 4475–4479. pmid:6457302
  9. 9. Zivanovic Y, Lopez P, Philippe H, Forterre P (2002) Pyrococcus genome comparison evidences chromosome shuffling-driven evolution. Nucleic Acids Res 30: 1902–1910. pmid:11972326
  10. 10. Carbone L, Harris RA, Gnerre S, Veeramah KR, Lorente-Galdos B, et al. (2014) Gibbon genome and the fast karyotype evolution of small apes. Nature 513: 195–201. pmid:25209798
  11. 11. Escudero JA, Loot C, Nivina A, Mazel D (2015) The Integron: Adaptation On Demand. Microbiology Spectrum 3: MDNA3-0019-2014
  12. 12. Skippington E, Ragan MA (2011) Lateral genetic transfer and the construction of genetic exchange communities. Fems Microbiology Reviews 35: 707–735. pmid:21223321
  13. 13. Dupressoir A, Vernochet C, Bawa O, Harper F, Pierron G, et al. (2009) Syncytin-A knockout mice demonstrate the critical role in placentation of a fusogenic, endogenous retrovirus-derived, envelope gene. Proceedings of the National Academy of Sciences of the United States of America 106: 12127–12132. pmid:19564597
  14. 14. Siguier P, Gourbeyre E, Varani A, Bao TH, Chandler M (2015) Everyman's Guide to Bacterial Insertion Sequences. Microbiology Spectrum 3: MDNA3-0030-2014.
  15. 15. So M, Heffron F, McCarthy BJ (1979) The E. coli gene encoding heat stable toxin is a bacterial transposon flanked by inverted repeats of IS1. Nature 277: 453–456. pmid:368646
  16. 16. Hickman AB, Chandler M, Dyda F (2010) Integrating prokaryotes and eukaryotes: DNA transposases in light of structure. Critical Reviews in Biochemistry and Molecular Biology 45: 50–69. pmid:20067338
  17. 17. Grindley NDF, Whiteson KL, Rice PA (2006) Mechanisms of site-specific recombination. Annual Review of Biochemistry 75: 567–605. pmid:16756503
  18. 18. Val ME, Bouvier M, Campos J, Sherratt D, Cornet F, et al. (2005) The single-stranded genome of phage CTX is the form used for integration into the genome of Vibrio cholerae. Molecular Cell 19: 559–566. pmid:16109379
  19. 19. Landy A (2015) The lambda Integrase Site-specific Recombination Pathway. Microbiology spectrum 3: MDNA3-0051-2014.
  20. 20. Yen Ting L, Sau S, Ma C-H, Kachroo AH, Rowley PA, et al. (2014) The partitioning and copy number control systems of the selfish yeast plasmid: an optimized molecular design for stable persistence in host cells. Microbiology spectrum 2: PLAS-0003-2013.
  21. 21. Mcleod M, Craft S, Broach JR (1986) Identification of the Crossover Site during Flp-Mediated Recombination in the Saccharomyces-Cerevisiae Plasmid 2-Mu-M Circle. Molecular and Cellular Biology 6: 3357–3367. pmid:3540590
  22. 22. Dymecki SM (1996) Flp recombinase promotes site-specific DNA recombination in embryonic stem cells and transgenic mice. Proceedings of the National Academy of Sciences of the United States of America 93: 6191–6196. pmid:8650242
  23. 23. Campbell AM (1963) Episomes. Advances in Genetics 11: 101–145.
  24. 24. Esposito D, Scocca JJ (1997) The integrase family of tyrosine recombinases: evolution of a conserved active site domain. Nucleic Acids Research 25: 3605–3614. pmid:9278480
  25. 25. Yang W, Mizuuchi K (1997) Site-specific recombination in plane view. Structure 5: 1401–1406. pmid:9384556
  26. 26. Grainge I, Jayaram M (1999) The integrase family of recombinases: organization and function of the active site. Molecular Microbiology 33: 449–456. pmid:10577069
  27. 27. Cortez D, Quevillon-Cheruel S, Gribaldo S, Desnoues N, Sezonov G, et al. (2010) Evidence for a Xer/dif system for chromosome resolution in archaea. PLoS Genet 6: e1001166. pmid:20975945
  28. 28. She Q, Chen B, Chen L (2004) Archaeal integrases and mechanisms of gene capture. Biochemical Society Transactions 32: 222–226. pmid:15046576
  29. 29. Serre MC, Letzelter C, Garel JR, Duguet M (2002) Cleavage properties of an archaeal site-specific recombinase, the SSV1 integrase. Journal of Biological Chemistry 277: 16758–16767. pmid:11875075
  30. 30. Zhan ZY, Zhou J, Huang L (2015) Site-Specific Recombination by SSV2 Integrase: Substrate Requirement and Domain Functions. Journal of Virology 89: 10934–10944. pmid:26292330
  31. 31. Jaubert C, Danioux C, Oberto J, Cortez D, Bize A, et al. (2013) Genomics and genetics of Sulfolobus islandicus LAL14/1, a model hyperthermophilic archaeon. Open Biol 3: 130010. pmid:23594878
  32. 32. Bridger SL, Lancaster WA, Poole FL 2nd, Schut GJ, Adams MW (2012) Genome sequencing of a genetically tractable Pyrococcus furiosus strain reveals a highly dynamic genome. J Bacteriol 194: 4097–4106. pmid:22636780
  33. 33. Cossu M, Da Cunha V, Toffano-Nioche C, Forterre P, Oberto J (2015) Comparative genomics reveals conserved positioning of essential genomic clusters in highly rearranged Thermococcales chromosomes. Biochimie 118: 313–321. pmid:26166067
  34. 34. Filippo JS, Sung P, Klein H (2008) Mechanism of eukaryotic homologous recombination. Annual Review of Biochemistry 77: 229–257. pmid:18275380
  35. 35. Michel B, Leach D (2012) Homologous Recombination-Enzymes and Pathways. EcoSal Plus 5.
  36. 36. White MF (2011) Homologous recombination in the archaea: the means justify the ends. Biochemical Society Transactions 39: 15–19. pmid:21265740
  37. 37. Constantinesco F, Forterre P, Elie C (2002) NurA, a novel 5 '-3 ' nuclease gene linked to rad50 and mre11 homologs of thermophilic Archaea. Embo Reports 3: 537–542. pmid:12052775
  38. 38. Graham WJt, Rolfsmeier ML, Haseltine CA (2013) An archaeal RadA paralog influences presynaptic filament formation. DNA Repair (Amst) 12: 403–413.
  39. 39. Sato T, Fukui T, Atomi H, Imanaka T (2005) Improved and versatile transformation system allowing multiple genetic manipulations of the hyperthermophilic archaeon Thermococcus kodakaraensis. Applied and Environmental Microbiology 71: 3889–3899. pmid:16000802
  40. 40. Gorlas A, Croce O, Oberto J, Gauliard E, Forterre P, et al. (2014) Thermococcus nautili sp. nov., a hyperthermophilic archaeon isolated from a hydrothermal deep sea vent (East Pacific Ridge). Int J Syst Evol Microbiol 64: 1802–1810. pmid:24556637
  41. 41. Oberto J, Gaudin M, Cossu M, Gorlas A, Slesarev A, et al. (2014) Genome Sequence of a Hyperthermophilic Archaeon, Thermococcus nautili 30–1, That Produces Viral Vesicles. Genome Announc 2: e00243–00214. pmid:24675865
  42. 42. Oberto J (2013) SyntTax: a web server linking synteny to prokaryotic taxonomy. BMC Bioinformatics 14: 4–13. pmid:23323735
  43. 43. Gaudin M, Krupovic M, Marguet E, Gauliard E, Cvirkaite-Krupovic V, et al. (2013) Extracellular membrane vesicles harbouring viral genomes. Environ Microbiol 16: 1167–1175. pmid:24034793
  44. 44. Letzelter C, Duguet M, Serre MC (2004) Mutational analysis of the archaeal tyrosine recombinase SSV1 integrase suggests a mechanism of DNA cleavage in trans. Journal of Biological Chemistry 279: 28936–28944. pmid:15123675
  45. 45. Krupovic M, Bamford DH (2008) Archaeal proviruses TKV4 and MVV extend the PRD1-adenovirus lineage to the phylum Euryarchaeota. Virology 375: 292–300. pmid:18308362
  46. 46. Raeside C, Gaffe J, Deatherage DE, Tenaillon O, Briska AM, et al. (2014) Large Chromosomal Rearrangements during a Long-Term Evolution Experiment with Escherichia coli. Mbio 5: e01377–01314. pmid:25205090
  47. 47. Eisen JA, Heidelberg JF, White O, Salzberg SL (2000) Evidence for symmetric chromosomal inversions around the replication origin in bacteria. Genome Biology 1: research0011.0011–0011.0019.
  48. 48. Haberle RC, Fourcade HM, Boore JL, Jansen RK (2008) Extensive rearrangements in the chloroplast genome of Trachelium caeruleum are associated with repeats and tRNA genes. Journal of Molecular Evolution 66: 350–361. pmid:18330485
  49. 49. Hiratsuka J, Shimada H, Whittier R, Ishibashi T, Sakamoto M, et al. (1989) The Complete Sequence of the Rice (Oryza-Sativa) Chloroplast Genome—Intermolecular Recombination between Distinct Transfer-Rna Genes Accounts for a Major Plastid DNA Inversion during the Evolution of the Cereals. Molecular & General Genetics 217: 185–194.
  50. 50. Williams KP (2002) Integration sites for genetic elements in prokaryotic tRNA and tmRNA genes: sublocation preference of integrase subfamilies. Nucleic Acids Research 30: 866–875. pmid:11842097
  51. 51. Chen JW, Lee J, Jayaram M (1992) DNA Cleavage in Trans by the Active-Site Tyrosine during Flp Recombination—Switching Protein Partners before Exchanging Strands. Cell 69: 647–658. pmid:1586945
  52. 52. Lee J, Jayaram M, Grainge I (1999) Wild-type Flp recombinase cleaves DNA in trans. Embo Journal 18: 784–791. pmid:9927438
  53. 53. Wang ZJ, Le GW, Shi YH, Wegrzyn G, Wrobel B (2002) A model for regulation of colE1-like plasmid replication by uncharged tRNAs in amino acid-starved Escherichia coli cells. Plasmid 47: 69–78. pmid:11982328
  54. 54. Santangelo TJ, Cubonova L, Reeve JN (2008) Shuttle vector expression in Thermococcus kodakaraensis: contributions of cis elements to protein synthesis in a hyperthermophilic archaeon. Appl Environ Microbiol 74: 3099–3104. pmid:18378640
  55. 55. Marguet E, Gaudin M, Gauliard E, Fourquaux I, le Blond du Plouy S, et al. (2013) Membrane vesicles, nanopods and/or nanotubes produced by hyperthermophilic archaea of the genus Thermococcus. Biochem Soc Trans 41: 436–442. pmid:23356325
  56. 56. Lepage E, Marguet E, Geslin C, Matte-Tailliez O, Zillig W, et al. (2004) Molecular diversity of new Thermococcales isolates from a single area of hydrothermal deep-sea vents as revealed by randomly amplified polymorphic DNA fingerprinting and 16S rRNA gene sequence analysis. Appl Environ Microbiol 70: 1277–1286. pmid:15006744
  57. 57. Krumsiek J, Arnold R, Rattei T (2007) Gepard: a rapid and sensitive tool for creating dotplots on genome scale. Bioinformatics 23: 1026–1028. pmid:17309896
  58. 58. Despalins A, Marsit S, Oberto J (2011) Absynte: a web tool to analyze the evolution of orthologous archaeal and bacterial gene clusters. Bioinformatics 27: 2905–2906. pmid:21840875
  59. 59. Gibson DG, Young L, Chuang RY, Venter JC, Hutchison CA, et al. (2009) Enzymatic assembly of DNA molecules up to several hundred kilobases. Nature Methods 6: 343–U341. pmid:19363495
  60. 60. Sato T, Fukui T, Atomi H, Imanaka T (2003) Targeted gene disruption by homologous recombination in the hyperthermophilic archaeon Thermococcus kodakaraensis KOD1. Journal of Bacteriology 185: 210–220. pmid:12486058