Catalytic domain of pAD1 TraX defines a group of relaxases related to restriction endonucleases

María Victoria Franciaa,1, Don B. Clewellb,c, Fernando de la Cruzd, and Gabriel Moncaliánd

aServicio de Microbiología, Hospital Universitario Marqués de Valdecilla e Instituto de Formación e Investigación Marqués de Valdecilla, Santander 39008, Spain; bDepartment of Biologic and Materials Sciences, University of Michigan School of Dentistry, Ann Arbor, MI 48109; cDepartment of Microbiology and Immunology, University of Michigan Medical School, Ann Arbor, MI 48109; and dDepartamento de Biología Molecular e Instituto de Biomedicina y Biotecnología de Cantabria, Universidad de Cantabria–Consejo Superior de Investigaciones Científicas–Sociedad para el Desarrollo Regional de Cantabria, Santander 39011, Spain

Edited by Roy Curtiss III, Arizona State University, Tempe, AZ, and approved July 9, 2013 (received for review May 30, 2013)

Plasmid pAD1 is a 60-kb conjugative element commonly found in known relaxases show two characteristic sequence motifs, motif I clinical isolates of Enterococcus faecalis. The relaxase TraX and the containing the catalytic Tyr residue (which covalently attaches to primary origin of transfer oriT2 are located close to each other and the 5′ end of the cleaved DNA) and motif III with the His-triad have been shown to be essential for conjugation. The oriT2 site essential for relaxase activity (facilitating the cleavage reaction contains a large inverted repeat (where the nic site is located) by activation of the catalytic Tyr). This His-triad has been used as adjacent to a series of short direct repeats. TraX does not show a relaxase diagnostic signature (12). fi any of the typical relaxase sequence motifs but is the prototype of Plasmid pAD1 relaxase, TraX, was identi ed (9) and shown to oriT2 a unique family of relaxases (MOB ). The present study focuses on cleave within . It does not share overall homology with pre- C viously known relaxases. Together with MobC, from Enterobacter the genetic, biochemical, and structural analysis of TraX, whose 3D cloacae structure could be predicted by protein threading. The structure mobilizable plasmid CloDF13 (14), they form a unique consists of two domains: (i) an N-terminal domain sharing the group of relaxase family, known as MOBC (3), which contains invariant motifs unrelated to those of His-hydrophobe-His (HUH) topology of the DNA binding domain of the MarR family of tran- ii relaxases. An in-frame deletion of such motifs (of pAD1) resulted scriptional regulators and ( ) a C-terminal catalytic domain related in complete loss of relaxase activity, suggesting an important to the PD-(D/E)XK family of restriction endonucleases. Alignment function, at or near an of the protein (9). MOBC of MOBC relaxase amino acid sequences pointed to several con- contains relaxases belonging to two well differentiated clades, served polar amino acid residues (E28, D152, E170, E172, K176, clade MOBC1 comprised of Gammaproteobacteria mobile genetic R180, Y181, and Y203) that were mutated to alanine. Functional elements and clade MOBC2 composed of relaxases from Gram- analysis of these mutants (in vivo DNA transfer and cleavage positive bacteria. MobC_CloDF13 and TraX_pAD1 are the assays) revealed the importance of these residues for relaxase respective prototypes. activity and suggests Y181 as a potential catalytic residue similarly In this paper, we present data concerning the characterization to His-hydrophobe-His relaxases. We also show that TraX binds of TraX, which include the identification of a relaxase domain specifically to dsDNA containing the oriT2 direct repeat sequences, essential for the catalytic activity of TraX. This characterization fi fi con rming their role in transfer speci city. The results provide provides a catalytic model for the MOBC family of relaxases. insights into the catalytic mechanism of MOBC relaxases, which Moreover, our study widens the general model of the DNA differs radically from that of His-hydrophobe-His relaxases. processing for .

– Results lasmid pAD1 is a conjugative sex-pheromone responding fi Pplasmid originally identified in Enterococcus faecalis (1, 2). It Identi cation of Two Structural Domains in TraX. The TraX relaxase of plasmid pAD1 belongs to the MOBC relaxase family (3). Fig. 1 is the prototype conjugative plasmid of the MOBC relaxase family (3). In the pheromone-mediated conjugative process, shows an alignment of some MOBC relaxase sequences. The plasmid-free bacteria secrete multiple pheromones (small, hy- conserved signature D-x6–17-E-x-E-(RL)-x2-K-x3-R-Y is apparent drophobic, linear peptides) that induce a process in which spe- in the alignment, as shown in Fig. 1. This signature has no re- cific plasmid-containing cells become activated by a particular lationship with the invariant motifs in HUH relaxases (3). To pheromone for adherence to potential recipients and plasmid further characterize this MOBC family of proteins, we carried transfer functions (4–8). The related conjugative DNA pro- out a structural prediction and modeling of TraX by the Protein cessing machinery includes a specific relaxase that generates Homology/analogY Recognition Engine (PHYRE) Web Server a in the strand of the plasmid to be transferred (9), which (15). Two structural domains within TraX were predicted, an ultimately results in the acquisition of the plasmid by the re- N-terminal domain, between residues 1 and 90, and a C-terminal domain encompassing the 170 C-terminal residues (residues cipient cell. When the recipient cells have acquired the plasmid, – shutdown or masking of the corresponding endogenous phero- 132 261). The homology model of TraX N-terminal domain was mone takes place (10, 11). This serves to prevent self-induction directly obtained by submission of the full-length sequence of caused by residual amounts of endogenous pheromone, thus TraX. The structural prediction of TraX C-terminal domain was ensuring that induction occurs only in the presence of recipient generated by Phyre2 One-to-one threading using the model of cells. So, although pheromone responding are highly transmissible among E. faecalis populations, plasmid transfer is highly regulated and induced only in the presence of sufficient Author contributions: M.V.F., D.B.C., F.d.l.C., and G.M. designed research; M.V.F. and G.M. pheromone concentrations that reflect the presence of nearby performed research; M.V.F., D.B.C., F.d.l.C., and G.M. analyzed data; and M.V.F., D.B.C., potential recipients. F.d.l.C., and G.M. wrote the paper. The term relaxase is used to define proteins involved in initi- The authors declare no conflict of interest. ation and termination of DNA conjugative transfer (12, 13). This article is a PNAS Direct Submission. They recognize a specific sequence called nic located within the 1To whom correspondence should be addressed. E-mail: [email protected]. origin of transfer (oriT2 in the case of pAD1) and cleave it in a This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. strand- and site-specific manner to initiate DNA transfer. Most 1073/pnas.1310037110/-/DCSupplemental.

13606–13611 | PNAS | August 13, 2013 | vol. 110 | no. 33 www.pnas.org/cgi/doi/10.1073/pnas.1310037110 Downloaded by guest on September 29, 2021 MICROBIOLOGY

Fig. 1. Alignment of representative MOBC relaxases. (Upper) Sequence alignment of representative MOBC relaxases obtained with ClustalW. Accession numbers: MobC_CloDF13, CAB62410; MobC_PI, AAP70292; MobC_Y1, CAD58576; MobC_Y2, NP_995437; Orf8_pAM373, NP_072012; and TraX_pAD1, AAL59457. Color code: red on yellow, invariant amino acids; blue on blue, strongly conserved; black on green, similar; green on white, weakly similar; black on white, not conserved. The violet bar below the alignment indicates the region deleted in the in-frame traX deletion mutant resulting in a defective relaxase

(9). The arrows refer to the conserved polar amino acid residues that were selected for site-directed mutagenesis. (Lower) Conserved in MOBC relaxases. The height of each letter is proportional to the frequency of the amino acid residue. The logo was obtained by using WebLogo (36).

the C-terminal domain of CloDF13 MobC. MobC_CloDF13 is active site residues of BamHI, suggesting a putative function for the only member of the MOBC family of relaxases biochemically this domain as the TraX catalytic domain. Moreover, the con- characterized to date (3). Its C-terminal domain model was served basic residues K176 and R180, both in α2, are located in obtained by submission of MobC C-terminal sequence to Phyre2, a similar position as the RE residues that interact with the DNA as template. Both TraX domains were modeled at >90% confi- molecule to be cleaved. To evaluate the functional importance of dence. The N-terminal domain shares the topology of the MarR such residues along with other conserved amino acid residues in – family of transcriptional regulators, including a winged helix MOBC relaxases (alignment shown in Fig. 1), site-directed mu- turn–helix DNA binding motif (16, 17). TraX N-terminal domain tagenesis of the traX structural gene was performed. The amino is predicted to be formed by four helices (α2, α3, α4, and α5) and acid residues selected for replacement are shown in Fig. 1, and a two-strand β-sheet (β1 and β2) (Fig. 2). As in other winged- all contain polar side chains. The selected residues were all helix DNA-binding proteins, the loop adjacent to the HTH motif changed to Ala, and four different approaches were applied to connects two antiparallel strands and forms like a wing that binds characterize the properties of the mutant proteins: (i) in vivo the DNA minor groove. Accordingly, the α4 helix would be re- DNA transfer assays (overnight filter matings), (ii) in vivo sponsible for the interaction with the DNA major groove. TraX nicking assays on supercoiled plasmid DNA containing pAD1 C-terminal domain shares structural homology with the PD- oriT or run-off assays (9), (iii) site-specific cleavage assays of (D/E)XK family of restriction endonucleases (REs; Fig. 2). This oligonucleotides containing the pAD1 nick region, and (iv) (of which BamHI is the prototype) requires three DNA binding assays using DNA fragments containing pAD1 + conserved acidic residues (DEE) that coordinate two Mg2 ions oriT (mobility shift assays). as cofactors to catalyze DNA hydrolysis (18, 19). The structural core of TraX catalytic C-terminal domain is predicted to be a RE Motifs Are Required for in Vivo DNA Transfer. The influence of central, three-stranded mixed β-sheet flanked by a α-helix in a the specific TraX mutations on conjugative pAD1 transfer was topology β2, β3, α2–3, and β4. The acidic residues D152 at β2 and studied by complementation assays as described by Francia and E170 and E172 at β3 appear in the same positions as the DEE Clewell in 2002 (9). In vivo complementation of plasmid

Francia et al. PNAS | August 13, 2013 | vol. 110 | no. 33 | 13607 Downloaded by guest on September 29, 2021 difference compared with WT TraX), indicating mutation Y203A is not essential in the conditions used for the experiment. DNA transfer was partially recovered with mutants TraX E28A, D152A, E170A, E172A, K176A, and R180A, but transfer values were severely reduced (>10,000-fold decrease), suggesting the functional importance of them all for relaxase activity. Comple- mentation experiments of WT TraX pAD1 derivative plasmid, pAM307, showed none of the assayed relaxase mutants had a dominant-negative effect. However, at least a 100-fold decrease of the transfer frequency was observed when WT TraX was also expressed from the complementing plasmid, suggesting the ne- cessity of a strictly controlled TraX expression for full DNA processing activity.

nic-Cleavage Reaction Is Abolished in TraX Transfer Mutants. To Fig. 2. Modeled structure of the DNA-binding and catalytic domains of check the cleavage activity of the selected TraX mutants, run-off DNA synthesis analyses were performed in cells TraX. Ribbon representation of the structural prediction analysis of the N- oriT terminal domain (Left) and the C-terminal domain (Right) of TraX bound to cotransformed with pAD1 and TraX derivatives containing DNA. The secondary structure elements are numbered. Catalytic residues are plasmids upon IPTG induction (similarly as shown in ref. 9). As shown as red sticks and the Mg+2 cations are represented as green spheres. shown in Fig. 3, none of the TraX mutants supported detectable nicking activity in vivo, in contrast to WT TraX. This result highlights the essential role of these conserved residues. pAM8130 (a traX in-frame deletion mutant of pAD1 showing To check the in vitro cleavage ability of TraX and the selected undetectable relaxase activity) conjugation was analyzed in the mutants, their codifying genes were cloned into the pET30b presence of the TraX mutants cloned in the E. faecalis non- expression vector (Table S1). To facilitate purification, these mobilizable vector pMSP3545Sp. The results of the comple- genes were expressed as His-tagged C-terminal fusion pro- fi mentation assays are shown in Table 1. Complementation of the teins. All proteins were partially puri ed in a one-step pro- fi different TraX mutants resulted in variable transfer frequencies cedure by using af nity chromatography on Ni-columns. As most known relaxases reversely hydrolyze oligonucleotides containing the that, in all cases, were reduced compared with WT TraX. In- + specific nick site in the presence of Mg 2 ions, WT TraX (or their terestingly, mutation Y181A resulted in a protein incapable of fi complementation and thus nonfunctional (undetectable DNA corresponding mutants) site-speci c cleavage activity was tested on oligonucleotides containing pAD1 nic (Table S2). Oligonucleotides transfer), indicating that Y181 is essential for TraX activity. On + containing the pAD1 nick region were incubated with TraX or the the contrary, the tra phenotype was almost fully recovered when derivative mutants as described in Materials and Methods. Cleavage TraXY203A mutant was analyzed (two orders of magnitude products could be separated from uncut oligonucleotides by elec- trophoresis in sequencing gels. TraX was unable to cut any oligonucleotide, suggesting that a different kind of substrate or Table 1. Conjugation frequencies of pAM8130 (encoding a additional pAD1 gene products are needed for TraX in vitro defective relaxase) and pAM307 (encoding a WT relaxase) traX cleavage, in contrast to the results obtained for other relaxases complemented with WT or the indicated mutants (13). All these results taken together emphasize the wide dif- Donor/plasmid in trans Transfer frequency ferences in the biochemical behavior of TraX compared with pre- viously characterized relaxases, like R388_TrwC (20), F_TraI (21), UV202 (pAM8130) − RP4_TraI (22), RSF1010_MobA (23), and pMV158_MobM (24). WT traX 8.7 × 10 1 − E28A 8.5 × 10 5 TraX Binds the Double-Stranded Direct Repeat Region but Not the − D152A 4.5 × 10 7 nic Region. Because of the significant differences in TraX bio- − E170A 3.2 × 10 6 chemical activities, we decided to ascertain its DNA binding − E172A 2.2 × 10 7 sites, which could give the clue as to its biochemical proper- K176A 1.4 × 10−6 ties. The DNA binding activity of TraX and its mutants (in- − R180A 4.7 × 10 4 dicative of relaxosome formation ability) was assayed by gel − Y181A <10 8 mobility shift assays (Fig. 4). pAD1 oriT-containing fragments Y203A 1 × 10−2 were labeled and incubated at 30 °C in binding buffer with +2 Empty vector <10−8 purified TraX or mutant proteins. No Mg is present in the UV202 (pAM307) assays, preventing DNA cleavage from interfering with binding. As − WT traX 2 × 10 3 a negative control, extracts corresponding to bacteria containing × −1 the empty expression vector pET30b were used. Poly(dI-dC) was E28A 9 10 fi D152A ND added as nonspeci c competitor DNA, and the reaction mixtures E170A 1.14 were subjected to 5% (wt/vol) polyacrylamide electrophoresis. Under these conditions, protein–DNA complexes remain bound E172A 1.11 and migrate through the gel slower than the corresponding free- K176A 1.05 A × −1 DNA fragments. As shown in Fig. 4 , TraX was able to form R180A 9.6 10 complexes with both the entire oriT (oriT2) and the oriT-direct Y181A 1.25 repeats containing DNA fragments (DR), but did not show similar × −2 Y203A 1.06 10 complexes in the case of the inverted repeat DNA fragment where Empty vector 1.25 the nic site is located (IR), implying that the TraX relaxase spe- fi oriT Filter matings were conducted overnight using E. faecalis UV202 (rif, fus, ci cally binds the -direct repeats. The DNA binding activity of − + B oriT recA ) as donors and E. faecalis OG1SS (str, spc, gel ) as the recipient, with the TraX mutants is shown in Fig. 4 . Migration of the pAD1 selection for transconjugants on plates containing streptomycin and eryth- DNA fragments was shifted in the presence of all mutants (with romycin. The transfer frequencies indicated for each derivative represent the the exception of mutant E28A), indicating that oriT-specific rec- average of at least two independent experiments and are expressed as the ognition and binding (and therefore relaxosome formation) was number of erythromycin-resistant transconjugants per donor cell. possible with all TraX mutants affecting the catalytic domain.

13608 | www.pnas.org/cgi/doi/10.1073/pnas.1310037110 Francia et al. Downloaded by guest on September 29, 2021 predict that Y181 could be the catalytic that remains covalently attached to the 5′ end of the nic site after cleavage. In fact, Y181 is essential for pAD1 conjugation (Table 1). More- over, the relative locations of the acidic residues and the catalytic tyrosine are similar to the organization found in typical relaxases (i.e., HUH relaxases). HUH relaxases are related to rolling-circle replication proteins and a specific type of transposases. RE homologues have also been found to be involved in rolling-circle replication [Rep of plasmid pUA140 (25)] and transposition [TnsA transposase (26)]. Therefore, conjugative relaxases come from HUH replication proteins or REs. This structural similarity suggests an adaptive functional convergence of these two com- pletely separate protein folds. However, no covalent protein– DNA intermediate could be found in CloDF13 (14), and no catalytic Tyr is required in the restriction in which the is cleaved by hydrolysis. The presence of a stable covalent intermediate has not been observed for other replication initiator proteins such as RepB (27) and gpII (28). However, later observations concluded that indeed there is a co- valent complex in filamentous phage that dissociated rapidly Fig. 3. Nicking ability of TraX and derivatives by runoff assays. E. coli BL21 cells unless gpII was inactivated (29). Thus, gpII possesses activities to containing pAM8151 (pAD1 oriT cloned into pSU18) and pET30b expressing form the covalent bond and to break it. RepB was also predicted TraX or its derivatives were preinduced with 1 mM IPTG and processed as to form an unstable covalent complex by the study on the chirality described by Francia and Clewell (9). Lane 1, E28A; lane 2, WT TraX; lanes of phosphorothioate in the strand transfer reaction (30). There- 3 to 9, D152A, E170A, E172A, K176A, R180A, Y181A, and Y203A. The “T” fore, so far, all the characterized RCR replicases or conjugative and “C” lanes indicate the corresponding sequencing reactions using the relaxases require a covalent intermediate. This phosphotyrosine fmol sequencing kit, pAM8151 as template DNA, and the 32P-labeled covalent complex is necessary to have the 5′ phosphate activated runoff primer PE/5.2. The arrow on the left represents the location of the at the recipient cell so that the second reaction could be per- fi speci c nick site within the pAD1 oriT sequence. formed. We cannot envision how conjugative transfer could ter- MICROBIOLOGY minate without the relaxase being covalently bound to the DNA after the first cleavage reaction. DNA binding capability demonstrated for all these mutants also PD-(D/E)XK restriction enzymes have been found to have two indicate that no point mutation introduced in TraX resulted in metal ions at the active sites. BamHI active site contains two severe effects on protein secondary structure (that might have divalent cations at the active site that are separated by approx- been responsible of the aforementioned DNA transfer results). In imately 4 Å and are coordinated by Glu-77, Asp-94, and Glu-111 contrast to other relaxases, the TraX recognition and (31). PD-(D/E)XK hydrolyze the phosphodiester bond at pAD1 oriT is more than 70 bp away from the nic site, which by direct inline nucleophile attack, resulting in the inversion of explains the negative results in the oligonucleotide cleavage assays configuration at the phosphorous atom (32). When the location shown earlier. The binding site for other well known relaxases is an of the TraX conserved catalytic residues was compared with inverted repeat adjacent to the nic (13), whereas for TraX appears BamHI, we observed that they lay in the same spatial position. to be a series of five direct repeats shown in Fig. 4A and Fig. S1. Thus, we propose the catalytic reaction to proceed in TraX as Discussion depicted in Fig. 5 by comparison with the BamHI catalytic model proposed in a previous work (31). According to this model, two In the work presented in this article, the pAD1 relaxase structure divalent cations are also involved in the TraX reaction. The was modeled, showing to contain two structural domains. The C- terminal domain belongs to the PD-(D/E)XK family of REs. The acidic residues D152, E170, and E172 occur in a similar position as the DEE active site residues of BamHI, suggesting a role for + them in the coordination of two Mg 2 cations that activate the nucleophilic attack of the phosphodiester bond at the nic site (18). Actually, site-directed mutagenesis of these residues in TraX_pAD1 abolished DNA transfer as well as in vivo nic cleavage. The basic residues K176 and R180 are also conserved in the BamHI-like REs. In fact, site-directed mutagenesis of both residues showed undetectable nicking activity and a drastic re- duction in conjugative transfer. At the C-terminal end of β4, we find Y203, a conserved residue in the MOBC family of proteins (Figs. 1 and 2). This polar amino acid is near the active site. It could be involved in the conformation of the active site or the interaction with the DNA. TraX Y203A was not able to cata- lyze in vivo cleavage, although TraX Y203A mutant was able to fi catalyze DNA transfer in vivo. As the approach used to check Fig. 4. Gel mobility-shift assays showing TraX or derivatives speci c binding to oriT2 dsDNA. (A) Labeled dsDNA PCR fragments (lane 1) containing dif- cleavage in vivo relies on primer extension on relaxosome fi complexes isolated from cells, it might not be sensitive enough ferent oriT2 regions (as indicated at the top of the gure and below the corresponding gel shifts, oriT2, DR, IR) were incubated in the absence (lane to detect a potential residual cleavage activity of TraX Y203A. A fi fi 2) or presence (lane 3) of puri ed TraX. Empty vector (pET30b)-derived reduced cleavage ef ciency showed by this mutant would contribute control protein extracts were used in lane 2. Free DNA forms and protein– to its lower conjugation frequency (a 100-fold lower transfer fre- DNA complexes are indicated. (B) Gel mobility-shift assays with labeled quency compared with WT TraX). dsDNA PCR fragments containing oriT2 DR sequence. Lane 1, dsDNA; lane 2, An interesting residue is Y181, in which a mutation to Ala dsDNA plus BL21(pET30b) control protein extract; lanes 3 to 11, dsDNA plus abolished pAD1 conjugal transfer. By comparison with other WT TraX, E28A, E170A, E172A, K176A, R180A, Y181A, Y203A, or D152A, known conjugative relaxases of the MOBF or MOBP systems, we respectively. Free DNA forms and protein–DNA complexes are indicated.

Francia et al. PNAS | August 13, 2013 | vol. 110 | no. 33 | 13609 Downloaded by guest on September 29, 2021 metal ion A is coordinated by E170, D152, and the carbonyl and cleave both strands of a palindromic sequence. In fact, the oxygen of I171, whereas the metal B could be bound by D152 RCR initiation protein RepB is also formed by a helix-turn-helix and D138. Strikingly, when the position of the TraX Tyr181 was (HTH) DNA binding domain that recognizes three direct compared, we observed that the hydroxyl group is located where repeats and a HUH endonuclease domain to exert the specific the attacking water is found in BamHI. We propose that no ssDNA cleavage. Interestingly, although it has the catalytic Tyr water is involved in the nucleophilic attack by TraX. Instead, the and HUH motif, no covalent intermediate has been directly hydroxyl group of Y181 is perfectly positioned for the , found in RepB, as previously discussed. After the binding of the resulting in a covalent bond between the phosphate group of the N-terminal domain to the DR region, the RE-like binding domain cleaved DNA bond and the Y181. In our model, E172 could be would recognize and bind the ssDNA nic region catalyzing the the general base that activates the hydroxyl group of Y181 by hydrolysis of the nic site, which would initiate DNA transfer from proton abstraction (Fig. 5). donor to recipient (Fig. S2). Although there is no experimental When the purified mutant proteins were checked for dsDNA evidence of TraX covalently bound to the 5′ end of the DNA to binding activity, we observed (Fig. 4B) that all mutant proteins be transferred, we propose that, like in the other characterized were able to bind dsDNA, except mutant E28A. According to conjugative systems, TraX will be transferred to the recipient the prediction, all the conserved residues that were mutated are cell by the T4SS piloting the DNA. Then, after one cycle of located in the ssDNA binding and catalytic domain of TraX, RCR, a TraX-catalyzed second cleavage reaction in the donor whereas the E28 is located in the N-terminal domain, explaining cell will create the free 3′OH to recircularize the plasmid in the this difference. The N-terminal domain of TraX is predicted to recipient cell. have approximately 110 aa with a topology related to the DNA According to our results, not all the relaxases involved in binding domain of MarR (16, 17). E28A occurs at the C-terminal plasmid conjugation converged into the same structural fold. end of α2 helix. The main interactions with the DNA are pre- MOBF, MOBP, MOBQ, and MOBV family showed the HUH dicted to be in the N-terminal end of α4 and even α2 and the fold (3, 33), but MOBC present the RE fold. We think the dif- loop connecting β1 and β2 . However, the conserved Glu28 could ferent fold mainly results from the different triggering mecha- be involved in the stabilization of the winged-helix DNA binding nism. In HUH relaxases, conjugative transfer is triggered by the domain, and this is why this mutant is not able to bind the DR binding of the recipient cell to the donor cell. Until this cell-to- sequences. Consequently, TraX E28A was not able to promote cell interaction occurs, the HUH relaxase is bound to the oriT, conjugation in the complementation experiment. Interestingly but the nic-cleavage and joining activities are in a tight equilib- we have found that the three systems most closely related to rium in such a way that the plasmid is not relaxed unless the pAD1, namely pAM373, TX1341, and VRS11, are conserved conjugation signal is received. In the case of RE relaxases like in the nic regionbutnotinthedirectrepeatregion(Fig. S1). pAD1_TraX, conjugation is triggered by the pheromones pro- This conservation in the binding sites perfectly correlates with duced by the recipient cell and detected by the donor cell. When conservation in the protein domains. C-terminal domains are the pheromone has reached the donor cell, relaxase expression is well conserved in the four systems, whereas the N-terminal induced along with its ability to introduce the nic-specific domains and, more specifically, the DNA interacting α-helices, cleavage generating the plasmid relaxation and releasing a free 3′ are quite different. OH that is used by the cellular polymerase to produce by RCR After nic-cleavage in the absence of recipient cells, typical HUH the ssDNA strand that is transferred. Any reaction performed by relaxases remain bound to both sides of the nic site and the 3′OH is the RE relaxase will produce a relaxed plasmid with the relaxase not free. In the case of pAD1, TraX catalyses cleavage of the nic covalently bound. If this covalent product is not transferred to site in a process strictly controlled by the presence of pheromone the recipient cell, the plasmid could be unstable. The quick hy- (cAD1). In a previous work, we demonstrated that nic cleavage was drolysis of the covalent complex could be the mechanism to onlyvisiblewhencellswereexposedtocAD1(9).Alsointhis protect the plasmid DNA unless the transfer of the relaxases regard, preliminary experiments show that relaxase expression is produces a conformational change that avoids the hydrolysis. detected only in the presence of cAD1. Therefore, there is evidence The conjugation of some HUH relaxase-containing plasmids, that pAD1 DNA processing for transfer is triggered when in close such as pCF10 (34), is also induced by pheromones, but the need proximity to potential recipient cells (Fig. S2). Only then is enough for accessory proteins could prevent the relaxation of the plas- pheromone sensed by the donor cells, which, in response, activate mid in the absence of pheromone. the plasmid mating specific functions, including the relaxase. In summary, the results shown in this article defineauniquetype We have reported here that TraX is able to bind dsDNA frag- of conjugative relaxases. DNA processing by RE relaxases is, in ments containing the full oriT2 region or just the region containing general terms, similar to processing by HUH relaxases. Thus, both the five DRs, but not the nic region (Fig. 4A). Moreover, TraX- relaxases should covalently bind the plasmid DNA to be transferred specific ssDNA binding was not observed. Thus, we think to the recipient cell to perform the termination reaction. The MOBC relaxases are not able to bind the double-stranded nic catalytic mechanism is also similar, although the structure of the region unless the binding of the N-terminal winged-helix DNA relaxase is different. This is a beautiful case of adaptive conver- binding domain to the DRs at oriT2 allows the creation of a partially gence, whereby two radically different proteins (HUH and RE) melted region around the nic site. This domain could be necessary functionally converge to create similar constellation of residues in these relaxases to recognize and cleave only one of the DNA within the catalytic center and thus become able to catalyze the strands, whereas dimeric REs without this second domain recognize same biological reaction.

Fig. 5. Catalytic mechanism of BamHI, TraX, and TrwC. Schematic diagram illustrating the proposed catalytic mechanism for the DNA cleavage reaction performed by BamHI (Left), TraX (Center), or TrwC (Right)(Discussion).

13610 | www.pnas.org/cgi/doi/10.1073/pnas.1310037110 Francia et al. Downloaded by guest on September 29, 2021 Materials and Methods In Vitro Nicking Assays. Oligonucleotides containing the pAD1 nic region Structural Modeling. The 3D structures of the N-terminal and C-terminal [Nick/oriT2, GGGTTTTAACCCACGTTGAGCGAAAAT/GGTCAGGAATTTTGC- domains of TraX were predicted by homology modeling using the Phyre2 AGGCGGAGAAAGCC; and Nick2, ACCCACGTTGAGCGAAAAT/GGTCAG- server (15). Images of the resulting 3D models were generated using Pymol GAATTTTGCAGGCGGA, with the slash marks indicating the pAD1 nic site (DeLano Scientific). according to a previous report (9)] or the corresponding complementary strand as a negative control (Nick2c, TCCGCCTGCAAAATTCCTGACCAT- Cloning and Mutagenesis of TraX for Complementation Assays. TraX was TTTCGCTCAACGTGGGT) were labeled at the 5′ endwith[γ-32P]ATP and T4 cloned into plasmid pMSP3535Sp (11). TraX point mutants were constructed polynucleotide kinase. Cleavage reactions were carried out as described pre- by PCR as described in SI Materials and Methods and Tables S1 and S2. viously (35), with the modifications indicated in SI Materials and Methods.

Complementation Assays. Plasmids encoding TraX or the point mutants gener- EMSA. dsDNA containing oriT2 fragments for binding assays were generated − ated were introduced into E. faecalis UV202 (rif, fus, recA ) containing pAM8130 by PCR as described in SI Materials and Methods and Tables S1 and S2. Gel and pAM307, respectively. Restoration or inhibition of the transfer ability was mobility shift assays were performed basically as described previously (36), subsequently checked in each case by filter matings as described previously (9). with slight modifications indicated in SI Materials and Methods. The empty vector, pMSP3535Sp (11), was used as a negative control.

fi ACKNOWLEDGMENTS. We thank Gary Dunny for providing the vector Cloning, Expression and Puri cation of WT TraX and Mutants for DNA Cleavage pMSP3535Sp, and all members of our laboratories for helpful discussions. and Binding. Recombinant TraX was expressed on E. coli BL21 pLysS (Invitrogen) This work was supported by Spanish Fondo de Investigación Sanitaria (FIS) from plasmid pAM8155 (9). TraX point mutants were constructed by PCR using and Instituto de Salud Carlos III Grant FIS PI10/01081 (to M.V.F.); Spanish FIS, pAM8155 as a template as described in SI Materials and Methods and Instituto de Salud Carlos III and Fundación Marqués de Valdecilla–Instituto Tables S1 and S2.Purification of TraX protein and mutants (Fig. S3)was de Formación e Investigación Marqués de Valdecilla Grant CES08/008 (to carried out as previously described (9), with slight modifications indicated M.V.F.); Ministerio de Sanidad y Consumo, Instituto de Salud Carlos III, Grant in SI Materials and Methods. Red Española de Investigación en Patología Infecciosa RD06/0008/0031 (to M.V.F.); European Union Sixth Framework Programme Grant LSHE-CT-2007- 037410 (to M.F.V.); Ministerio de Ciencia e Innovación (Spain) Grants In Vivo Nicking Assays. The generation of the nic site by WT TraX or the BIO2010-14809 (to G.M.) and BFU2011-26608 (to F.d.l.C.); and European corresponding mutants tested was determined by using a runoff DNA syn- Union Seventh Framework Programme Grants 248919/FP7-ICT-2009-4 and thesis assay as described previously (9). 282004/FP7-HEALTH.2011.2.3.1-2 (to F.d.l.C.).

1. Tomich PK, An FY, Damle SP, Clewell DB (1979) Plasmid-related transmissibility and 19. Steczkiewicz K, Muszewska A, Knizewski L, Rychlewski L, Ginalski K (2012) Sequence, multiple drug resistance in Streptococcus faecalis subsp. zymogenes strain DS16. structure and functional diversity of PD-(D/E)XK phosphodiesterase superfamily. MICROBIOLOGY Antimicrob Agents Chemother 15(6):828–830. Nucleic Acids Res 40(15):7016–7045. 2. Francia MV, et al. (2001) Completion of the sequence of the Enterococcus 20. Grandoso G, et al. (2000) Two active-site tyrosyl residues of protein TrwC act se- faecalis conjugative virulence plasmid pAD1 and identification of a second transfer quentially at the origin of transfer during plasmid R388 conjugation. J Mol Biol – origin. Plasmid 46(2):117–127. 295(5):1163 1172. 3. Garcillán-Barcia MP, Francia MV, de la Cruz F (2009) The diversity of conjugative re- 21. Dostál L, Shao S, Schildbach JF (2011) Tracking F plasmid TraI relaxase processing – laxases and its application in plasmid classification. FEMS Microbiol Rev 33(3):657–687. reactions provides insight into F plasmid transfer. Nucleic Acids Res 39(7):2658 2670. 4. Clewell DB, Dunny GM (2002) Conjugation and genetic exchange in enterococci. The 22. Pansegrau W, Schröder W, Lanka E (1994) Concerted action of three distinct domains in the DNA cleaving-joining reaction catalyzed by relaxase (TraI) of conjugative Enterococci: Pathogenesis, Molecular Biology and Antibiotic Resistance, ed Gilmore MS plasmid RP4. J Biol Chem 269(4):2782–2789. (ASM, Washington, DC), pp 265–300. 23. Scherzinger E, Lurz R, Otto S, Dobrinski B (1992) In vitro cleavage of double- and 5. Clewell DB (2007) Properties of Enterococcus faecalis plasmid pAD1, a member of single-stranded DNA by plasmid RSF1010-encoded mobilization proteins. Nucleic a widely disseminated family of pheromone-responding, conjugative, virulence ele- Acids Res 20(1):41–48. ments encoding cytolysin. Plasmid 58(3):205–227. 24. Guzmán LM, Espinosa M (1997) The mobilization protein, MobM, of the streptococcal 6. Chandler JR, Dunny GM (2004) Enterococcal peptide sex pheromones: Synthesis and plasmid pMV158 specifically cleaves supercoiled DNA at the plasmid oriT. J Mol Biol – control of biological activity. Peptides 25(9):1377 1388. 266(4):688–702. fi 7. Ehrenfeld EE, Kessler RE, Clewell DB (1986) Identi cation of pheromone-induced 25. Zou X, Caufield PW, Li Y, Qi F (2001) Complete nucleotide sequence and character- surface proteins in Streptococcus faecalis and evidence of a role for lipoteichoic acid ization of pUA140, a cryptic plasmid from Streptococcus mutans. Plasmid 46(2):77–85. in formation of mating aggregates. J Bacteriol 168(1):6–12. 26. Hickman AB, et al. (2000) Unexpected structural diversity in DNA recombination: The 8. Ehrenfeld EE, Clewell DB (1987) Transfer functions of the Streptococcus faecalis restriction endonuclease connection. Mol Cell 5(6):1025–1034. plasmid pAD1: Organization of plasmid DNA encoding response to sex pheromone. J 27. Moscoso M, del Solar G, Espinosa M (1995) Specific nicking-closing activity of the Bacteriol 169(8):3473–3481. initiator of replication protein RepB of plasmid pMV158 on supercoiled or single- 9. Francia MV, Clewell DB (2002) Transfer origins in the conjugative Enterococcus fae- stranded DNA. J Biol Chem 270(8):3772–3779. calis plasmids pAD1 and pAM373: Identification of the pAD1 nic site, a specificre- 28. Meyer TF, Geider K (1979) Bacteriophage fd gene II-protein. II. Specificcleavage laxase and a possible TraG-like protein. Mol Microbiol 45(2):375–395. and relaxation of supercoiled RF from filamentous phages. J Biol Chem 254(24): 10. Ike Y, Craig RA, White BA, Yagi Y, Clewell DB (1983) Modification of Streptococcus 12642–12646. faecalis sex pheromones after acquisition of plasmid DNA. Proc Natl Acad Sci USA 29. Asano S, Higashitani A, Horiuchi K (1999) Filamentous phage replication initiator ′ 80(17):5369–5373. protein gpII forms a covalent complex with the 5 end of the nick it introduced. – 11. Chandler JR, Flynn AR, Bryan EM, Dunny GM (2005) Specific control of endogenous Nucleic Acids Res 27(8):1882 1889. cCF10 pheromone by a conserved domain of the pCF10-encoded regulatory protein 30. Moscoso M, Eritja R, Espinosa M (1997) Initiation of replication of plasmid pMV158: Mechanisms of DNA strand-transfer reactions mediated by the initiator RepB protein. PrgY in Enterococcus faecalis. J Bacteriol 187(14):4830–4843. J Mol Biol 268(5):840–856. 12. Francia MV, et al. (2004) A classification scheme for mobilization regions of bacterial 31. Viadiu H, Aggarwal AK (1998) The role of metals in catalysis by the restriction en- plasmids. FEMS Microbiol Rev 28(1):79–100. donuclease BamHI. Nat Struct Biol 5(10):910–916. 13. Zechner EL, et al. (2000) Conjugative-DNA transfer processes. The Horizontal Gene 32. Pingoud A, Fuxreiter M, Pingoud V, Wende W (2005) Type II restriction endonu- Pool. Bacterial Plasmids and Gene Spread, ed Thomas CM (Harwood, Amsterdam), cleases: Structure and mechanism. Cell Mol Life Sci 62(6):685–707. pp 87–174. 33. Chandler M, et al. (2013) Breaking and joining single-stranded DNA: the HUH en- 14. Núñez B, De La Cruz F (2001) Two atypical mobilization proteins are involved in donuclease superfamily. Nat Rev Microbiol 11:525–538. – plasmid CloDF13 relaxation. Mol Microbiol 39(4):1088 1099. 34. Chen Y, Staddon JH, Dunny GM (2007) Specificity determinants of conjugative DNA 15. Kelley LA, Sternberg MJ (2009) prediction on the Web: A case study processing in the Enterococcus faecalis plasmid pCF10 and the Lactococcus lactis – using the Phyre server. Nat Protoc 4(3):363 371. plasmid pRS01. Mol Microbiol 63(5):1549–1564. 16. Hong M, Fuangthong M, Helmann JD, Brennan RG (2005) Structure of an OhrR-ohrA 35. Pérez-Mendoza D, et al. (2006) The relaxase of the Rhizobium etli symbiotic plasmid operator complex reveals the DNA binding mechanism of the MarR family. Mol Cell shows nic site cis-acting preference. J Bacteriol 188(21):7488–7499. 20(1):131–141. 36. Francia MV, Weaver KE, Goicoechea P, Tille P, Clewell DB (2007) Characterization of 17. Kumarevel T, et al. (2008) Crystal structure of the MarR family regulatory protein, an active partition system for the Enterococcus faecalis pheromone-responding ST1710, from Sulfolobus tokodaii strain 7. J Struct Biol 161(1):9–17. plasmid pAD1. J Bacteriol 189(23):8546–8555. 18. Vanamee ES, et al. (2011) Asymmetric DNA recognition by the OkrAI endonuclease, 37. Crooks GE, Hon G, Chandonia JM, Brenner SE (2004) WebLogo: A sequence logo an isoschizomer of BamHI. Nucleic Acids Res 39(2):712–719. generator. Genome Res 14(6):1188–1190.

Francia et al. PNAS | August 13, 2013 | vol. 110 | no. 33 | 13611 Downloaded by guest on September 29, 2021