Site-Specific Integration of Foreign DNA into Minimal Bacterial and Human Target Sequences Mediated by a Conjugative Relaxase

Background Bacterial conjugation is a mechanism for horizontal DNA transfer between bacteria which requires cell to cell contact, usually mediated by self-transmissible plasmids. A protein known as relaxase is responsible for the processing of DNA during bacterial conjugation. TrwC, the relaxase of conjugative plasmid R388, is also able to catalyze site-specific integration of the transferred DNA into a copy of its target, the origin of transfer (oriT), present in a recipient plasmid. This reaction confers TrwC a high biotechnological potential as a tool for genomic engineering. Methodology/Principal Findings We have characterized this reaction by conjugal mobilization of a suicide plasmid to a recipient cell with an oriT-containing plasmid, selecting for the cointegrates. Proteins TrwA and IHF enhanced integration frequency. TrwC could also catalyze integration when it is expressed from the recipient cell. Both Y18 and Y26 catalytic tyrosil residues were essential to perform the reaction, while TrwC DNA helicase activity was dispensable. The target DNA could be reduced to 17 bp encompassing TrwC nicking and binding sites. Two human genomic sequences resembling the 17 bp segment were accepted as targets for TrwC-mediated site-specific integration. TrwC could also integrate the incoming DNA molecule into an oriT copy present in the recipient chromosome. Conclusions/Significance The results support a model for TrwC-mediated site-specific integration. This reaction may allow R388 to integrate into the genome of non-permissive hosts upon conjugative transfer. Also, the ability to act on target sequences present in the human genome underscores the biotechnological potential of conjugative relaxase TrwC as a site-specific integrase for genomic modification of human cells.


Introduction
Bacterial conjugation is a mechanism for horizontal gene transfer among bacteria. By this process, a DNA molecule of any origin and length can be transferred to a recipient cell, if it contains an origin of transfer (oriT); the conjugative machinery can be provided in trans. Under laboratory conditions, conjugation to eukaryotic cells has been reported, from bacteria to yeast [1], plants [2] and mammalian cells [3].
Mechanistically, bacterial conjugation can be viewed as a plasmid DNA replication system linked to a secretion channel [4], leading to horizontal rather than vertical transmission of the plasmid. Accordingly, there are three functional modules in the conjugative machinery [5]: i) the relaxosome, a nucleoprotein complex required for plasmid DNA processing, which is related to rolling-circle replication systems. This complex includes a relaxase protein, often together with accessory proteins which assist its function, and the oriT, which is the only DNA sequence required in cis for DNA transfer.
ii) the Type IV secretion system (T4SS), a multiprotein complex which forms the transmembranal channel for substrate secretion. iii) the coupling protein (T4CP), responsible for linking the other two modules by protein-protein interactions.
The relaxase nicks the oriT in the DNA strand which is going to be transferred and remains covalently bound to the 59 end; current models for conjugative DNA transfer propose that this nucleoprotein complex is secreted to the recipient cell through the T4SS, followed by active DNA pumping by the ATPase activity of the T4CP [4]. Once in the recipient cell, the relaxase is functional [6], and presumably recircularizes the transferred DNA.
The relaxosome of the conjugative plasmid R388 comprises the oriT (oriTw), the relaxase-helicase TrwC, and two helper proteins, R388-encoded TrwA and host-encoded IHF (integration host factor) [7]. The minimal sequence of the oriTw essential for in vivo mobilization is only 17 bp in length; in vitro, it was shown that the sequence 6 nt 59 and 2 nt 39 to the nic site (6+2) is absolutely required for TrwC binding and for DNA nicking and strand transfer reactions on oligonucleotides [8]. The helper protein TrwA binds to oriTw; it acts as a transcriptional repressor of the trwABC operon and also increases TrwC nicking activity on oriTw [9], [10]. IHF is a host protein which binds specifically to oriTw. IHF has been proposed to have a negative effect on TrwC nicking activity, but not TrwC binding [7].
TrwC relaxase activity, responsible for DNA strand cleaving and transfer activities on supercoiled or single-stranded substrates, is contained in its N-terminal 293 residues (N293) [11]. This domain contains catalytic residues Y18 and Y26, which act sequentially [12]: Y18 is the only tyrosine able to act on supercoiled plasmid substrates and so it is responsible for the initial nicking reaction, while Y26 would catalyze the final strandtransfer reaction. The DNA helicase, ATPase and dimerization activities of TrwC are located in the C-terminal residues (C774) [13].
Apart from its role in conjugation, TrwC has been shown to act as a site-specific recombinase between two oriTw copies repeated in tandem [14]. This reaction occurs in the absence of conjugation, and thus in the absence of single stranded intermediates. This ability of TrwC is not a general feature of relaxases; TraI of plasmid F, which shares a similar 3D catalytic fold, is not able to catalyze this reaction [15]. TrwC-mediated sitespecific recombination is strongly enhanced by TrwA [15]. Host factors which affect DNA topology, such as IHF or transcription through the oriTw, affect the reaction [16]. One oriTw copy could be narrowed to the core sequence of 17 bp (14 nt 59+3 nt 39of nic) and still support TrwC-mediated recombination efficiently [15]. A search for possible natural targets sequences for TrwC in the human genome demonstrated that there are at least two possible natural targets on which TrwC can act as a recombinase [17]. The recombinase domain of TrwC locates to the nucleus in human cells, and, by random mutagenesis, a TrwC mutant which enters the nucleus was also obtained [17]. Interestingly, TrwC can catalyze site-specific integration of the incoming DNA into an oriTw-containing plasmid in the recipient cell [6]. Overall, these data underscore the biotechnological potential of TrwC as a sitespecific integrase which could be introduce into human cells for genomic modification.
In this work we have modified the integration assay in order to address the DNA and protein requirements of the reaction. Our results suggest a model for TrwC-mediated site-specific integration. We have obtained TrwC-mediated site-specific integration on minimal oriTw and human target sequences and into the Escherichia coli chromosome, underlying the biological significance of this reaction and the potential of TrwC for biotechnological purposes.

Optimization of the integration assay
In earlier work, the ability of TrwC to catalyze site-specific integration was assayed by mobilization of a suicide plasmid to an oriTw-containing recipient strain, in a recA background [6]. The assay is depicted in Figure 1A. The suicide plasmid pR6K::oriTp oriTw, containing the oriTs of plasmids R388 (IncW) and RP4 (IncP), is mobilized by the conjugative system of plasmid R388 from donor strain CC118 lpir (coding for the Pir protein which allows replication of the R6K replicon) to a recipient strain which contains a plasmid harbouring another oriTw copy, where integration takes place. The suicide plasmid cannot replicate in the recipient strain. The integrants are selected in Cm plates, the resistance conferred by the suicide plasmid, and integration subsequently confirmed by PCR amplification of a specific region of the cointegrate molecule with primers P1 and P2 (Fig. 1A). The published integration frequency reached with this assay was 5.4610 26 integrants per donor. It was also reported that the reaction was TrwC-dependent, since no integration events were detected when, in a similar assay, the suicide plasmid was mobilized by the RP4 conjugative relaxase [6]. The suicide plasmid is represented with a grey line and the recipient plasmid with a black line. The nic site is indicated by an arrowhead. P1 and P2, oligonucleotides used in the PCR reaction to detect the cointegrates. B and C. Restriction analysis with enzymes that cut only once in the recipient plasmid. NdeI was used for integrants obtained in pKK:oriT (in B, to the left of the MW marker), XcmI for integrants in pSU:oriT (in B, to the right of the MW marker), and BstEII for integrants in pLA58 and pLA59 (in C, nicm and nic., respectively). The cointegrate is indicated with an arrow. DP, donor plasmid; RP, recipient plasmid. I1-I2, DNA from two independent integrants obtained on each RP. Sizes in kb of the MW marker from top of the gel: 10-8.0-6.0-5.0-4.0-3.0-2.5. doi:10.1371/journal.pone.0031047.g001 To improve the integration reaction, we used a different Pir donor strain, II1, which can be handled at 37uC since it is devoid of the l prophage. As recipient plasmids, we assayed two different replicons: pKK::oriTw (pMB8 replicon) and pSU::oriTw (p15A replicon). With either recipient plasmid, a significant improvement in the integration frequency was obtained ( Table 1): ca. 2 logs with pKK::oriTw and 1 log with pSU::oriTw, compared to earlier work. However, we observed a residual level of integration (2.6610 27 integrants/donor) in the absence of TrwC when using pKK::oriTw, for unknown reasons, while no integrants were found in the absence of TrwC when the pSU:oriTw recipient plasmid was used (Table 1). Thus, the pSU::oriTw recipient plasmid rendered the best integration frequency maintaining TrwC specificity and was used as the recipient plasmid for subsequent integration assays.
We tested the ability of RP4_TraI to promote site-specific integration under similar assay conditions, mobilizing the suicide plasmid by this relaxase to a recipient which contained pSU::oriTp. No integrant colonies were found (Table 1, last line, last column), suggesting that RP4_TraI relaxase is not able to catalyze sitespecific integration of an incoming DNA. However, transfer of TraI-RP4 to the recipient cell has not been proved experimentally so far. Thus, we cannot discard that the absence of integration could be due to the absence of TraI in the recipient cell.
DNA from the integrants obtained for both systems (pKK and pSU) was analysed by restriction analysis with an enzyme that only cuts the recipient oriT-containing plasmid. The results ( Figure 1B) show that the proportion of cointegrate molecules co-residing with the unaltered recipient plasmid varies significantly depending on the system used. When a pMB1 replicon (pKK system) is used as recipient plasmid, there is roughly 50% of each molecular species, as previously shown [6]. However, when the p15A replicon (pSU system) is used, almost all molecules are cointegrates. This difference could reflect an involvement of the plasmid replication machinery in the reaction. Alternatively, since a similar amount of cointegrate molecules is found in both cases, we reasoned that this could be due to a minimal amount of cointegrate molecules required to resist the selection with Cm25. However, the same integration frequencies and proportion of cointegrate molecules were obtained when applying Cm selection of 10, 25 or 40 mg/ml (data not shown).
It was described that the DNA strand harbouring the nic site affected the efficiency of oriT-specific recombination mediated by TrwC [15]. When the nic site of the oriT was in the lagging strand, the percentage of recombined colonies increased 5 times; this was explained by the longer exposure of ssDNA in the lagging strand, which would favour TrwC nicking reaction. In both recipient plasmids used in the integration assays, the nic site was located in the lagging strand. We assayed two plasmids with the nic site in both orientations with respect to the replication fork, pLA58 (p220::nicm), nic site in the lagging strand, and pLA59 (p220.2::nic.), nic site in the leading strand. We mobilized the suicide plasmid to a recipient cell containing either pLA58 or pLA59 and no significant differences in the integration frequency obtained were observed (4.7610 26 and 3.2610 26 integrants per donor, respectively). Restriction analysis of the DNA of the integrants showed that the percentage of cointegrate molecules was around 50% when the recipient nic site was present in the lagging strand, and around 70-80% when the nic site was in the leading strand ( Figure 1C). A plausible explanation is that TrwC is promoting the resolution of the cointegrate molecules preferentially when both nic sites lie on the lagging strand, as described in [15].

Role of relaxosomal proteins TrwA and IHF in integration
The R388 relaxosome is formed by the oriT and proteins TrwC, TrwA, and host-encoded IHF. TrwA enhances the nicking activity mediated by TrwC on scDNA, while IHF acts as an inhibitor [7]. With respect to TrwC-mediated oriT-specific recombination on scDNA substrates, TrwA increases drastically the efficiency of the reaction in the absence of conjugation, and it also enhances TrwCmediated site-specific recombination in the recipient cell by almost six times [15]. IHF was shown to inhibit recombination only in the absence of TrwA and on DNA substrates lacking one of the two TrwA and IHF binding sites [16]; in the presence of TrwA, however, the presence or absence of IHF was irrelevant.
To test the effect of TrwA and IHF in site-specific integration, the suicide plasmid (pR6K::oriTp oriTw) was mobilized to isogenic IHF+ or IHF2 recipient strains containing the recipient plasmid pSU::oriTw in the presence or absence of a plasmid which expressed trwA (pET3a::trwA). The recipient plasmid pSU::oriTp was assayed as a negative control in all experimental conditions tested. The integration frequency was compared in wild type or IHF-deficient recipient strains in the presence or absence of TrwA. Results are shown in Table 2. Both TrwA and IHF are enhancers of the integration reaction, producing an increase of about 3.5 fold and more than 30 fold, respectively, in the integration frequency. DNA of the integrants was analyzed and the same restriction pattern was obtained from IHF2 or IHF+ backgrounds (data not shown). The lack of both proteins in the recipient cell decreases by two logs the frequency of integration mediated by TrwC.

Integration catalyzed by TrwC expressed from recipient cells
The integration assay described ( Figure 1A) is based on a conjugation assay, meaning that functional oriT and TrwC are required to achieve conjugative transfer of the suicide plasmid, which prevents the study of the DNA and protein requirements of the reaction. We assayed the ability of TrwC to catalyze sitespecific integration when expressing trwC from the recipient cell as Table 1. TrwC-mediated site-specific integration into different recipient plasmids. follows. The integration reaction mediated by TrwC also takes place upon the mobilization of the suicide plasmid to a recipient cell which contains a plasmid with an oriT, but in this case, the donor plasmid is mobilized by the RP4 conjugative system and trwC is being expressed from another plasmid in the recipient cell ( Figure 2, compare A and B). When plasmid pSU1621 (pET3a:trwC) was present in the recipient, we observed an integration frequency of 1.4610 27 (Table 3). Thus, trwC expressed in the recipient cell is able to locate both oriT-containing plasmids and catalyze integration. However, the frequency of this assay is 2-3 logs less than when we mobilized the suicide plasmid attached to TrwC from the donor cell (6.5610 25 integrants per donor; Table 2).
In an attempt to improve this frequency, we also expressed trwA in the recipient cell. The integration frequency increased by 30 fold (Table 3). Since the helper plasmid coding for trwA+trwC expresses these genes from the putative trwABC promoter of R388, the increase could be due to a higher expression level of trwC rather than to the presence of trwA. To examine this possibility, we expressed trwC in the recipient cell under the inducible tac promoter. The integration assay was performed on plates supplemented with IPTG to induce the expression of trwC. The results demonstrated that an increase in the production of TrwC in the recipient cell is not correlated with higher frequencies; in fact, a slight decrease in the integration frequency was observed ( Table 3, last two columns), which could be due to toxicity of trwC overexpression. Thus, the improvement in the integration frequency when trwC+trwA are expressed in the recipient cell is due to the enhancing role of TrwA in the integration reaction.
DNA of the integrants was analyzed (Fig. 2D, top gel). The proportion of cointegrate molecules was much lower when TrwA is expressed in the recipient. So, TrwA, while enhancing TrwCmediated integration, also promotes resolution of the cointegrates in the cell. Since oriT-specific recombination in the presence of both TrwA and TrwC is more efficient than integration, probably all molecules would be resolved if selection for the cointegrates was not applied.
Since TrwC-mediated site-specific integration can be obtained when TrwC is mobilized from the donor and also when trwC is expressed in the recipient, we reasoned that the integration frequency could be improved by joining both TrwC sources, as shown in Figure 2C. We mobilized the suicide plasmid by TrwC to a recipient harbouring the oriT-containing plasmid and a helper plasmid coding for trwC. The results, however, showed a decrease in the integration frequency of about 1-log compared to integration rates with no trwC expression in the recipient ( Table 3, bottom row). When trwA was expressed together with trwC in the recipient, no recovery of this low frequency was observed (Table 3). DNA analysis of integrants showed the same pattern observed in previous assays, with most molecules in the form of cointegrates except when TrwA is present in the recipient (Fig. 2D, bottom gel).

TrwC requirements for integrase activity
The integration assay expressing trwC in the recipient cell allowed us to test TrwC mutants for their integrase activity. The following experiments were all performed mobilizing the suicide donor plasmid by the RP4 conjugative system.
We tested if the site-specific integration reaction requires the strand transferase activity of TrwC by using a set of point mutants in the catalytic tyrosyl residues: Y18F, Y26F, and the double mutant Y18FY26F. These mutants have been shown to affect conjugation and recombination reactions to different extents: Y18F strongly affects both activities, while Y26F has only a mild effect and the double mutant abolishes both processes [15]. We tested integration catalyzed by these TrwC mutants expressed from the recipient cell and no integrants were found, indicating that the reaction is dependent on both catalytic Tyr residues of the protein ( Table 4). When TrwA was also produced in the recipient, a few integrants were obtained, allowing analysis of their DNA ( Figure 3A). We observed that when Y18 was mutated, the resolution reaction was not favoured in spite of the presence of TrwA in the recipient, probably due to the fact that Y18 is the only tyrosine able to act on supercoiled DNA substrates. When only Y26 was mutated, although the integration reaction was reduced to residual levels, resolution of the cointegrates took place.
The C-terminal DNA helicase domain of TrwC is required for conjugative DNA transfer, as it supports unwinding of the DNA to be transferred. Its involvement in the integration reaction was tested using a TrwC point mutant (K502T) affecting motif I of the helicase superfamily I (Walker A box, GxGKT); this mutant is deficient in conjugation and proficient in recombination [15]. TrwC(K502T) showed a decrease of less than 3 fold in the integration frequency when compared to TrwC wt (Table 4). A similar slight decrease was observed in the recombination assay, which was attributed to the observation that TrwC(K502T) showed roughly four times less protein product when compared to wild type in a western blot [15]. Thus, the DNA helicase activity is dispensable for TrwC-mediated site-specific integration.
The minimal domain of TrwC able to promote site-specific recombination efficiently was TrwC-N600 [15]. To delimit the minimal TrwC domain sufficient to catalyze integration, we assayed N600 and N450 expressed in the recipient together with TrwA (Table 4). N450 catalyzes integration with about 2-log lower efficiency than wild-type while N600 catalyzes integration about ten times less efficiently than full-length TrwC. This efficiency decreased in about 1-log when TrwA was not expressed in the recipient ( Table 4), meaning that TrwA is helping this domain as it helps TrwC-mediated integration. DNA of the integrants was analysed by restriction analysis ( Figure 3B). In contrast to fulllength TrwC, TrwA does not promote the resolution of the cointegrates formed by N600. This result suggests that TrwA can play different roles in integration (enhancing the reaction catalyzed by both TrwC and N600) and in recombination (promoting TrwC-mediated, but not N600-mediated resolution of the cointegrates).  oriT requirements for integration We have analyzed the specificity of TrwC for its integration target, the R388 oriT (oriTw). The nicking and binding sites of TrwC ( Figure 4A) are the minimal oriTw requirements for different TrwC activities in vivo and in vitro (see Introduction). To test the requirement of the binding site we have assayed the mutIR oriT, in which the sequence of both arms of the inverted repeat IR2 to which TrwC binds was changed by another sequence maintaining the secondary structure (Fig. 4A). This mutation provokes a drastic 5-log reduction of the mobilization frequency [8]. To test the requirement of the nicking site, we used mut23-25 oriT ( Figure 4A), which has 3 nucleotides changed 59 to the nic site. This mutation almost abolishes mobilization, as this region is critical for the initial nicking reaction [8]. We tested TrwCmediated integration with recipient plasmids carrying those mutations and expressing trwC and trwA in the recipient cell, and we did not detect any integration event in any case (integration frequency ,10 28 integrants per donor), confirming that TrwC also requires its binding and nicking sites to catalyze integration. The same mutations were tested in the integration via TrwC from the donor, i.e. when TrwC enters the recipient cell covalently attached to the transferred DNA strand. The results are shown in Table 5. In this case, only a modest decrease (5 to 9fold) in the integration frequency was observed, indicating different TrwC DNA requirements at the initiation and termination steps of the integration reaction (see Discussion).
In the recombination reaction mediated by TrwC, it was found that one of the oriT target copies could be reduced to 17 bp (14+3) containing the nic site and the proximal arm of the IR2, maintaining a recombination frequency of 80% [15]. To test if TrwC was also able to promote integration into a recipient plasmid containing this minimal oriT, plasmid pSU::oriT(14+3) (coordinates 174 to 190 from ref. [18]) was used as recipient plasmid (Fig. 4A). The results ( Table 5) show that TrwC is able to promote site-specific integration on this minimal oriT; however, the reaction decreases by around 2 logs. When trwA was also expressed in the recipient cell, a modest increase in integration frequency was obtained (Table 5). DNA analysis of the integrants showed the same proportion of cointegrates when the recipient plasmid carried full-length or minimal oriT ( Figure 4B).
In a recent report, we demonstrated that TrwC can promote site-specific recombination on substrate plasmids where the oriT1 copy was replaced by human genomic DNA which contained sequences resembling the essential oriT core region (Fig. 4A): two (15+3) sequences, located in chromosomes 5 and X, with no mismatches in the essential core sequence (6+2), worked as targets with the same efficiency as with the canonical 14+3 sequence [17]. We assayed TrwC mediated site-specific integration into recipient plasmids containing these human targets (HuX 15+3 (27) and Hu5 15+3(210); Fig. 4A) and a helper plasmid which provided TrwA. TrwC catalyzed site-specific integration into both human target sequences; the efficiency of the reaction decreases by only 2-3 times compared to the minimal oriT(14+3) ( Table 5). When DNA from the integrants was analyzed, we found that 100% of the molecules were cointegrates (Fig. 4B). This suggests that the deviations from the consensus in the human sequences, lying out of the essential nicking region, do not affect significantly termination of the integration reaction, but prevent resolution of the cointegrates formed.

Integration into chromosomal DNA
We tested the ability of TrwC to catalyze site-specific integration into a chromosomal oriT copy. In the E.coli chromosome we did not find any sequences which resemble the  minimal R388 oriT sequence required for TrwC binding and nicking. Two recipient strains were constructed containing an oriTw copy plus a resistance marker in place of the chromosomal lacZ gene (see Experimental Procedures): CMS1 and CMS2, with the oriTw nic site lying in the lagging and leading strands with respect to the replication fork, respectively ( Figure 5A). The integration assay was performed by mobilizing the suicide plasmid (pR6K::oriTp oriTw) either with R388-TrwC or with RP4-TraI to recipient strains CMS1 or CMS2, or HMS174 (the isogenic oriTstrain) as a negative control. We did not obtain any integrant in any case (,10 28 integrants per donor) except when mobilizing the suicide plasmid with TrwC and using CMS1 as recipient. The integration frequency was low (1.1610 27 integrants per donor, mean of five independent assays) but the results were reproducible. DNA analysis of the integrants was done by PCR amplification of the lacZ gene ( Figure 5B). Amplification products of the expected size were obtained in all cases: 3 kb (size of the lacZ gene) in HMS174, 1.5 kb in CMS1 and CMS2 (size of the oriT-Km cassette in place of the lacZ ORF), and around 5 kb for the integrants (which contain the integrated suicide plasmid). It can also be observed that a 1.5 kb product is also visible, indicating that the integration reaction is reversible. In addition, specific amplification of a region of the cointegrate was performed with primers P A and P C ( Figure 5C). This specific 850 bp amplification product was gel-extracted and the DNA sequence was determined, confirming the expected integration structure.
The negative controls (oriT-recipient strain, or mobilization with RP4-TraI to any of the strains: ,4.3610 28 integrants per donor in all cases) indicate that the reaction is dependent on TrwC and its target R388 oriT. Thus, TrwC is able to catalyze the integration reaction into the chromosome although with 2-3 logs lower efficiency than into an oriTw-containing plasmid. Interestingly, no integration events were detected when using CMS2 as a recipient strain. This is likely due to the fact that the nic site of the oriTw lies in the leading strand of replication, meaning that it could be less accessible to TrwC. So, the exposure of single-stranded DNA, which was shown not to be a requisite in integration into an oriTw-containing plasmid, is a relevant factor in order to obtain integration into the host chromosome; in plasmid DNA, this exposure may be easier to obtain by local supercoiling than in the chromosome.

Discussion
The conjugative relaxase TrwC is the protein in charge of piloting the transferred DNA into the recipient cell during bacterial conjugation of plasmid R388. It was reported that this protein acts not only as a conjugative relaxase, but also as a sitespecific recombinase and integrase in recipient bacteria [6]. By modifying the donor bacterial strain and the recipient oriTcontaining plasmid, we have significantly improved the integration frequency reported previously (Table 1). Using different types of integration assays (Fig. 2), we have tried to elucidate the Donor cells are II1 with pCIG1077 and pR6K::oriT P oriTw, and recipient cells are DH5a containing the indicated recipient plasmid, with (+TrwA) or without (2TrwA) plasmid pET3a::trwA. Data are the mean of at least three independent assays. Frequency is given in integrants per donor. doi:10.1371/journal.pone.0031047.t005 Figure 5. Integration assay in the bacterial chromosome. A. Scheme of the expected integrants in the chromosomal oriT copy of recipient strains CMS1 or CMS2. Symbols as in Fig. 1. B and C. PCR amplification with oligonucleotides P A and P B , flanking the lacZ gene (in B), or with P A and P C , annealing only to the cointegrate (in C). C1, C2 and HMS, strains CMS1, CMS2, and HMS174, used as a negative control. Int C1, integrants obtained with recipient strain CMS1. The black arrow indicates the amplification product of the suicide plasmid integrated in the chromosomal oriT copy. HL: MW marker. doi:10.1371/journal.pone.0031047.g005 underlying mechanism of the site-specific integration reaction mediated by TrwC in bacteria. Taken together, the results obtained suggest that the reactions mediated by TrwC to accomplish integration mimic those for initiation and termination of conjugation; this assumption is based on the following evidences: i) TrwC-mediated site-specific integration was favoured by the presence of the R388 relaxosome components IHF and TrwA in the recipient cell (Table 2). This contrasts with TrwC-mediated site-specific recombination tested on plasmids containing two oriT copies, where the observed effect of TrwA (enhancer) and IHF (null or repressor under certain conditions) correlated with their described effect on TrwC nicking on supercoiled DNA. This difference probably reflects the fact that recombination is boosted by two initial nicking reactions, while integration in the recipient requires a final strand-transfer reaction. Thus, while a complete relaxosomal complex may be a reluctant substrate for nicking by TrwC, it behaves as an optimal substrate for strand transfer by a TrwC-DNA complex. ii) We have shown that the integration reaction also took place when trwC was expressed only in the recipient, which implies that TrwC is able to locate both targets to perform strand transfer reactions. Curiously, when TrwC-DNA entered from the donor cell, additional expression of trwC in the recipient caused a decrease in the integration frequency (Table 3). This could be likely due to the fact that free TrwC molecules are bound to the oriT of the recipient plasmid, preventing strand transfer from incoming TrwC-DNA complexes. iii) A double mutant in both TrwC catalytic residues (Y18FY26F) was inactive, confirming that nicking and strand transfer reactions are required for conjugation, recombination and integration. However, integration was totally dependent on both catalytic residues of TrwC, Y18 and Y26 (Table 4), while the Y26F mutation is well tolerated in recombination [15]. Y18 and Y26 are proposed to act in the initiation and termination of conjugative DNA transfer [12], [19]; it has been proposed that Y26 could have access to the nic site on supercoiled DNA once Y18 has formed the covalent complex [19]. Thus, while recombination requires two initiation nicking reactions both catalyzed by Y18, integration requires initiation and termination events. iv) It has been described that formation of the IR2 hairpin at oriT enhances Y18 activity [19]. It can be expected that TrwC requirements for oriT recognition are more strict for the initial nicking reaction than for strand-transfer into a second target once the covalent complex is formed with the first target. This model correlates with our observation that changes in the critical area affecting either the nic or the binding sites abolished TrwC integration activity when TrwC was only expressed in the recipient cell and thus, it is required to act on the supercoiled mutant oriT. In contrast, incoming TrwC-DNA complexes could transfer DNA into acceptor sites with mismatches in the core oriT sequence (Table 5). Figure 6 depicts a model for the integration reaction based on the above mentioned results. A TrwC-DNA complex enters the recipient cell, where a supercoiled oriT-containing acceptor plasmid is present. TrwC is attached to the T-strand through Y18, the residue responsible for the initial nicking event on supercoiled DNA (Fig. 6A). The relaxosome formed at the recipient oriT with TrwA and IHF would be the preferred Figure 6. A model for TrwC-mediated site-specific integration. A. TrwC (blue oval) arrives to the recipient cell covalently bound to the suicide plasmid (dashed line) through the Y18 residue. The recipient contains the recipient plasmid (red lines). TrwA (green spheres) and IHF (yellow ovals) sit on the oriTw forming a relaxosome conformation which increase the exposure of ssDNA and the nic site (blue triangle). B. The incoming TrwC-DNA complex has a free Y26 residue. Y26 nicks (orange star) and binds covalently to the recipient nic site. C. Strand-transfer reactions are produced by the attack of the free -OH groups generated to Y18 covalently bound to the suicide plasmid and to Y26 attached to the recipient plasmid. As a result, the transferred DNA strand is integrated into the recipient plasmid. D. The host replication machinery duplicates the integrated DNA. E. Two molecules are obtained, the cointegrate and the recipient plasmid (nic sites indicated by triangles). The cointegrate can be resolved by TrwC-mediated sitespecific recombination (enhanced by TrwA), producing the two initial molecules. The suicide plasmid is lost, since it cannot be replicated in the recipient cell. doi:10.1371/journal.pone.0031047.g006 substrate for the free Y26 residue in the TrwC-(Y18)-DNA complex, which nicks the recipient oriT and forms a covalent intermediate (Fig. 6B); free TrwC molecules, however, would sit on the oriT copy and inhibit integration of incoming TrwC-DNA complexes. In this intermediate, there are two free 39-OH ends which could attack each of the covalent Y-DNA complexes; as a result, the T-DNA strand would be integrated into the recipient oriT (Fig. 6C). Integration requires an additional strand-transfer reaction compared to conjugative DNA transfer, which may explain why the Y26F mutation affects integration more than conjugation. The resolution of the covalent intermediate (Fig. 6D) is probably mediated by the host replication machinery, as previously suggested for site-specific recombination [15]. Finally, in the presence of TrwC and TrwA, the cointegrate molecules would be resolved (Fig. 6E), giving rise to a higher proportion of recipient plasmids, as observed (Figs. 2, 3). This model is based on the action of a single TrwC monomer, as previously suggested also for conjugation [19], since the existence of two catalytic Tyr residues allows a monomer to perform both strand-transfer reactions; however, previous evidences of TrwC oligomerization in vitro [20] and DNA-independent TrwC transfer into recipients [6] leave open the possibility that TrwC acts as an oligomer both in donor and recipient cells.
Only a few relaxases have been reported to catalyze sitespecific recombination, and the only other conjugative relaxase reported to catalyze intermolecular recombination is that of the self-transfer system of the integrative and conjugative element ICEclc [21]. This difference among conjugative relaxases has no correlation with taxonomic proximity (according to [22]), or with the number of catalytic Tyr residues. Intriguingly, the relaxase of ICEclc can act on two different oriTs for ICE transfer with similar efficiencies; however, it is able to catalyze in vivo strand-exchange between two plasmids carrying oriT1 but not when carrying oriT2 [21]. This result points to differences in the interaction of the relaxase with other relaxosomal proteins or with the oriT as factors which could determine the recombinase/integrase activity of relaxases, rather than to intrinsic catalytic differences among these proteins.
No other relaxase has been reported to perform site-specific integration of incoming DNA in recipient bacteria. In addition, we show here that TrwC could integrate the incoming DNA into a chromosomal oriT copy ( Figure 5). Considering that R388 is a broad host range plasmid, the ability of TrwC to integrate the incoming DNA strand into a recipient chromosomal target may allow integration of R388 into the genome of non-permissive recipient hosts if a suitable target sequence exists in the recipient genome. We have shown that the 17 bp core of the R388 oriT was sufficient to act as a target for TrwC-mediated integration, and that DNA requirements are less stringent on the acceptor target site, broadening the possibility of finding natural TrwC targets in any recipient genome. This would confer R388 the possibility to colonize a broader range of microbial hosts in nature.
There have been reports of conjugative DNA transfer into different types of eukaryotic cells, implying that TrwC bound to a DNA molecule of any length could also have access to eukaryotic genomes. In this work, we show that TrwC catalyzed integration into two DNA sequences found in the human genome which resemble the minimal oriT (Table 5). Interestingly, integrants formed at these sites did not revert (Fig. 4B), so integration events obtained in this way would be stable, overcoming the limitation of reversible recombinases such as Cre [23]. Our results, together with previous reports on TrwC targeting to the nucleus [17], underscore the potential of TrwC as an integrase for genomic engineering of higher organisms.

Plasmids and plasmid constructions
Published plasmids used in this work are listed in Table S1, and plasmids constructed for this work are detailed in Table S2. Plasmids were constructed using standard cloning procedures [29]. Restriction enzymes, Shrimp Alkaline Phosphatase, and T4 DNA ligase were purchased form Fermentas. For PCR amplification, high fidelity Vent DNA polymerase (New England BioLabs) was used. Primers used for PCR were obtained from Sigma-Aldrich. DNA sequences of all cloned PCR segments were determined.

Strain constructions
CMS1 and CMS2 are isogenic strains carrying an R388 oriT copy and a kanamycin resistance marker in place of the chromosomal lacZ ORF of E. coli strain HMS174. These strains were used as recipients to assay integration into the chromosome. The oriT-km R cassette was introduced in the chromosome by the method of Dantsenko and Wanner [30] disrupting the lacZ gene. Oligonucleotides used for PCR reactions were 59-ATGACCAT-GATTACGGATTCACTGGCCGTCGTTTTACAACA-CAGCTATGACCATGATTAC-39, containing 40 bases of the 59 region of the lacZ gene and annealing towards the oriT, and 59-GACACCAGACCAACTGGTAATGGTAGCGACCGGCGC-CAGCTGCTAAAGGAAGCGGAACA-39, with 40 bases homologous to the 39 end of the lacZ gene and annealing towards the nptII gene. Template DNAs were pCMS17 (oriTwm-Km) and pCMS18 (oriTw.-Km) (Table S2). One hundred nanograms of each PCR product were transformed into arabinose-induced HMS174 cells harbouring plasmid pKD20, coding for an Larabinose-inducible lRed recombinase. Transformants were grown at 30uC and plated on LB+Km. Km resistant colonies were confirmed by their white colour on plates supplemented with X-gal and by PCR analysis using oligonucleotides P A (59-ATGACCATGATTACGGATTCA-'3) and P B (59-GACACCA-GACCAACTGGT-'3), annealing to the 59and 39 ends of the lacZ gene, respectively.

Mating assays
Standard mating assays were performed as described [12], incubating the mating mixture for 4 hours at 37uC. Conjugation frequencies are expressed as number of transconjugants per donor cell.

Integration assays
TrwC-mediated site-specific integration assay described in [6] was optimized as follows. Matings were done using II1 as donor strain containing a plasmid harbouring an oriTw, an oriTp and an R6K replicon (only replicates in strains expressing pir), used as a suicide plasmid for mobilization into strains lacking the Pir protein. The suicide plasmid was mobilized from II1 by a non mobilizable plasmid (pCIG1077) coding for the transfer region of plasmid R388 except for the oriT. DH5a was used as a recipient harbouring a plasmid with R388 oriT (oriTw) or RP4 oriT (oriTp), as a negative control. Integrants were selected in Cm plates, as it is the resistance provided by the suicide plasmid upon integration. The frequency is reported as the number of integrants per number of donor cells. Conjugative transfer controls were performed in all experiments using pir strains as recipients, and transfer efficiency was always close to 100% transconjugants/donor. As a negative control for integration, the suicide plasmid (containing oriTp and oriTw) was mobilized from S17.1 lpir by RP4_TraI to the same recipient as in the test. Thus, TrwC is not present in the reaction, but the oriTw enters in a single-stranded form, which allow us to rule out other causes of recombination in the recipient cell. In the integration assay via TrwC from the recipient, trwA and/or trwC were expressed in recipient cells under the control of the T7 promoter, to avoid possible toxic effects of overexpression.
Integration events were also analysed at the molecular level. Plasmid DNA was analysed by restriction analysis with enzymes which cut only once in the recipient plasmid and do not cut the donor plasmid (BstEII, NdeI or XcmI, as appropriate), to estimate the proportion of cointegrates. For PCR analysis, DNA was isolated from the colonies with Instagene (Bio-Rad). Primers P1 (59-AGCGGATAACAATTTCACACAGGA-39), annealing to the recipient plasmid, and P2 (59-GCAGGATCCGC-TAAGCTTTGTCGGTCATTTCGA-39), annealing to the do-nor plasmid downstream oriTw, were used to obtain an amplicon with a size of 1.2 kb only expected for the cointegrate molecules. As a positive control we used pKK::oriT-Km [6], which contains 826 bp of the suicide plasmid, mimicking a possible cointegrate molecule.
The analysis of the integrants obtained by TrwC-mediated sitespecific integration into the chromosome were analyzed as follows. When using CMS1 as recipient strain, primers P A and P C (59-GCCTCAAAATGTTCTTTACGA-39), annealing to 39 end of the cat gene, were used to amplify a specific region (850 bp) of the integrant. To verify integrants obtained when using CMS2 as recipient cell, primers P C and P B were used to amplify a band of 1800 bp specific of this type of integrant.