The Enterococcus faecalis conjugative plasmids pAD1 and pAM373 encode a mating response to the peptide sex pheromones cAD1 and cAM373 respectively. Sequence determination of both plasmids has recently been completed with strong similarity evident over many of the structural genes related to conjugation. pAD1 has two origins of transfer, with oriT1 being located within the repA determinant, whereas the more efficiently utilized oriT2 is located between orf53 and orf57, two genes found in the present study to be essential for conjugation. We have found a similarly located oriT to be present in pAM373. oriT2 corresponds to about 285 bp based on its ability to facilitate mobilization by pAD1 when ligated to the shuttle vector pAM401; however, it was not mobilized by pAM373. In contrast, a similarly ligated fragment containing the oriT of pAM373 did not facilitate mobilization by pAD1 but was efficiently mobilized by pAM373. The oriT sites of the two plasmids each contained a homologous large inverted repeat (spanning about 140 bp) adjacent to a series of non-homologous short (6 bp) direct repeats. A hybrid construction containing the inverted repeat of pAM373 and direct repeats of pAD1 was mobilized efficiently by pAD1 but not by pAM373, indicating a significantly greater degree of specificity is associated with the direct repeats. Mutational (deletion) analyses of the pAD1 oriT2 inverted repeat structure suggested its importance in facilitating transfer or perhaps ligation of the ends of the newly transferred DNA strand. Analyses showed that Orf57 (to be called TraX) is the relaxase, which was found to induce a specific nick in the large inverted repeat inside oriT; the protein also facilitated site-specific recombination between two oriT2 sites. Orf53 (to be called TraW) exhibits certain structural similarities to TraG-like proteins, although there is little overall homology.
pAD1 (60 kb) is representative of a widely disseminated family of conjugative plasmids commonly found in clinical isolates of Enterococcus faecalis. Bacterial virulence in animal models is associated with a pAD1-encoded cytolysin (Gilmore et al., 1994) and a surface protein known as aggregation substance (AS) is associated with the size of vegetations appearing in rabbit endocarditis models (Chow et al., 1993), as well as binding to mammalian cells in culture (Kreft et al., 1992; Sussmuth et al., 2000). AS also plays a significant role in the formation of mating aggregates resulting from a response induced by a peptide sex pheromone, cAD1, produced by recipient (plasmid-free) enterococci (Dunny et al., 1979; Galli et al., 1989). Regulation of the pAD1 pheromone response has been investigated in some detail (for recent reviews, see Clewell, 1999; Clewell and Dunny, 2002), and the nucleotide sequence has recently been completed (Francia et al., 2001).
pAD1 has two origins of transfer (oriT) that are separated by about 180° on the circular element (Francia et al., 2001). oriT1 is located within repA, which is close to or may overlap the vegetative origin of replication (Weaver et al., 1993; An and Clewell, 1997), whereas oriT2 is in a region of the plasmid containing genes related to conjugation (Francia et al., 2001). When a segment containing either of the transfer origins is cloned on the vector plasmid, pAM401, the latter can be mobilized in trans by pAD1. A chimera containing oriT2 is mobilized with an efficiency several orders of magnitude greater than one containing oriT1 (Francia et al., 2001) and is probably the preferred transfer origin on pAD1, at least in intraspecies matings.
Transfer origins (oriT’s) and their role in the initiation of conjugative DNA transfer have been investigated in a number of laboratories, with the focus being mainly on Gram-negative systems (for reviews, see Lanka and Wilkins, 1995; Firth et al., 1996; Zechner et al., 2000). Initiation of plasmid transfer generally requires the assembly at the oriT site of a protein complex containing a relaxase (nickase) and accessory DNA binding proteins. The relaxase catalyses the cleavage of a specific phosphodiester bond at the nic site, within the related oriT, during which it becomes covalently linked to the 5′-end of the cleaved strand through a tyrosine residue. Single-stranded DNA is transferred to the recipient cell, and the ends are subsequently re-ligated through the cleaving– joining activity of the relaxase. The process has features of the first round of DNA synthesis during rolling circle replication (Erickson and Meyer, 1993; Pansegrau et al., 1993; Waters and Guiney, 1993; Lanka and Wilkins, 1995). ‘Relaxosomes’ consisting of plasmid DNA–protein complexes have been isolated directly from bacteria or reconstructed in vitro whereby nicking may subsequently be activated causing a relaxation of the supercoiled plasmid. ‘Relaxation complexes’ of this nature were originally identified in Escherichia coli in the case of the mobilizable plasmid ColE1 (Clewell and Helinski, 1969) and the conjugative element ColIb-P9 (Clewell and Helinski, 1970), although it was several years later that a relationship to conjugative transfer was established (Inselburg, 1977; Warren et al., 1978; Willetts and Wilkins, 1984; Pansegrau et al., 1990; Wilkins and Lanka, 1993; Lanka and Wilkins, 1995).
The recent completion of the nucleotide sequences of both pAD1 (Francia et al., 2001) and another sex pheromone-responding plasmid pAM373 (De Boever et al., 2000) did not reveal any determinants bearing signifi-cant homology with known relaxases, although these two plasmids resembled each other over many of the structural genes related to conjugation. In this communication, we report on an investigation of the oriT2 site of pAD1 including identification of the specific nick site and show that a similar site is present in pAM373. We also show in the case of pAD1 that determinants on either side of the nic site are necessary for conjugation, and one, orf57, encodes the specific nicking enzyme. In addition, Orf57 is shown to facilitate a site-specific recombinational event between oriT2 sites on two replicons, resulting in a cointegrate structure.
Bacterial strains, plasmids and oligonucleotides used in this study are listed in Table 1.
Table 1. Bacterial strains, plasmids and oligonucleotides used.
Characterization of oriT2 in pAD1 and a similar sequence in pAM373
The identification of oriT2 (Francia et al., 2001) was based on the mobilization of a pAM401 chimera (pAM8100) carrying a cloned 0.7 kb segment of pAD1 by a pAD1::Tn917 derivative (pAM714) with wild-type conjugation properties. To determine the minimal size of the mobilization-enabling region, we constructed and examined deletions of pAM8100 (Figs 1 and 2). Constructs were tested for their ability to facilitate mobilization of the vector-encoded cat determinant (Cm-resistance) from RecA-negative donors (UV202) in short (10 min) matings in broth after a 90 min exposure to sex pheromone cAD1. The smallest mobilizable fragment was that of pAM8103, which corresponded to a 285 bp segment providing a high efficiency of transfer. As seen in Fig. 2, pAM8103 contains a group of five direct repeats of 6 bp each with 5 bp spacing, followed by a large inverted repeat over a 141 bp span. (Other repeats are seen outside the critical region as shown in Figs 1 and 2.)Figure 3A illustrates a ‘folded’, single-strand view of the inverted repeat. Both the direct-repeat and the inverted repeat regions are essential as deletion of either of these (e.g. pAM8104 or pAM8105) results in loss of transfer activity (Fig. 1).
Interestingly, a site with sequences resembling the large inverted repeat is also present in the Enterococcus faecalis pheromone-responding plasmid pAM373 (De Boever et al., 2000) (see Fig. 2 (pAM8300) and Fig. 3B). To determine if the region in pAM373 also functions as an origin of transfer, a representative segment (447 bp) from this plasmid was generated by polymerase chain reaction (PCR) and cloned in pAM401 giving rise to pAM8301. The latter was readily mobilized by a pAM373::Tn917lac derivative, pAM4020, with wild-type transfer properties as shown in Fig. 1. Mobilization was specific, as a clone containing the pAM373-related oriT was not mobilized by the pAD1 system nor was the pAD1-related oriT mobilized by the pAM373 system. It is noteworthy that the large inverted repeats of the two systems bear significant sequence identity and, whereas the pAM373-sequence corresponding to the direct repeats of pAD1 is totally different, it does contain direct repeats (i.e. six hexanucleotide repeats with 6 bp spacings).
oriT2 facilitates transfer into Staphylococcus aureus
Although we have not been able to establish pAD1 in S. aureus (unpublished data), an earlier report (Jones et al., 1987) suggested transfer occurred at extremely low frequencies (10−10–10−9 per recipient). It was not certain whether pAD1 (in our hands) was not able to replicate in S. aureus or simply could not transfer into this species. Insofar as the vector pAM401 does replicate in S. aureus, we examined the ability of pAM307 (pAD1::Tn917 (non-haemolytic) with wild-type transfer properties) to mobilize pAM401 chimeras carrying oriT1 (pAM3314) or oriT2 (pAM8103). As shown in Table 2, only pAM8103 could be mobilized from E. faecalis to S. aureus in overnight filter matings, and this occurred only when synthetic pheromone (cAD1) was present. Cm-resistant transconjugants appeared at approximately 10−3 per donor, and restriction analyses of three isolates using the restriction enzyme EcoR1 showed that the pAM8103 was intact in the staphylococcal host (not shown). Er-resistant transconjugants appeared at a frequency two orders of magnitude lower, and these were also Cm-resistant. Analysis of the plasmid content in these cases showed a pattern consistent with a pAM8103::pAM307 cointegrate structure (not shown). The data are consistent with the notion that pAM307 can transfer into S. aureus but is not able to replicate; that is, maintenance requires a covalent association (cointegration) with a plasmid (pAM8103 in this case) able to replicate in this host.
Table 2. Transfer frequency E. faecalis×S. aureus, involving oriT1 or oriT2 of pAD1 and oriT of pAM373, in the presence or absence of pheromone.
The mobilization frequencies indicated for each derivative represent the average of three independent experiments and are expressed as the number of Cm-resistant or Er-resistant transconjugants per donor cell.
1.0 × 10−3
1.5 × 10−5
2.9 × 10−6
8.3 × 10−4
7.0 × 10−7
Table 2 shows similar results relating to the pAM373 system. Here pAM4020 is able to mobilize the vector chimera pAM8301 carrying the oriT of pAM373. Addition of the synthetic pheromone peptide cAM373 significantly enhanced transfer. The fact that some transfer was also observed in the absence of the added peptide is probably due to the fact that S. aureus itself produces a cAM373 activity (Clewell et al., 1985). Er-resistant transconjugants, reflecting transfer of pAM4020, occurred at a much lower frequency and was observed only when cAM373 was provided; and these were also Cm-resistant. Analysis of plasmid DNA from three independently isolated transconjugants also revealed the presence of cointegrate structures (not shown).
The data indicate that oriT2 of pAD1 and the oriT of pAM373 are utilized in the movement of these plasmids from enterococci to staphylococci and that a trans-acting product(s) from the parent system is necessary. The inability to observe oriT1 of pAD1 to facilitate mobilization of the vector chimera implies that this origin at least does not play a selective role in transfer of DNA to the different genus. The possibility that it could facilitate transfer at a frequency below that detectable here cannot be ruled out. (Recall that even in enterococcal matings oriT1 operates at frequencies several orders of magnitude lower than that of oriT2.)
Cointegrate formation involves recombination between oriT regions
The above results dealing with transfer of pAD1 DNA (pAM307) from E. faecalis to S. aureus also implied that cointegrate formation was dependent on oriT2, as neither the vector alone nor the oriT1 clone (pAM3314) resulted in Er-resistant transconjugants. This raised the question of whether cointegrate formation involves a site-specific recombination between the oriT2 sites of the chimera (pAM8103) and pAM307 that was consistent with the restriction analyses of the transconjugants (above). To further examine this notion we attempted to generate PCR products using primers flanking each of the two expected junctions. An illustrative scheme is shown in Fig. 4A. Figure 4B shows products corresponding to ‘left’ and ‘right’ junctions of DNA obtained from three independent Er-resistant transconjugants. Such products were not observed in the case where pAM8103 or pAM307 alone were used as template DNA, and sequence analysis (not shown) confirmed the presence of oriT2 in each junction (PCR product). Similar PCR products were also obtained from various enterococcal donors (no recipients present) harbouring pAM8103 and a pAD1 derivative and grown in the presence or absence of pheromone. An example is shown in Fig. 4C. Furthermore, the amount of product reflecting each cointegrate junction was significantly greater after exposure of the cells to synthetic cAD1 (compare lane 1 with lane 2, and lane 3 with lane 4). The control amplification products corresponding to the oriT2 region of the pAD1 derivative (i.e. using primers #15 and #21) were similar regardless of whether or not the cells were exposed to cAD1 (lanes 5 and 6). This indicates that the pheromone-related differences seen for the cointegrate junctions (lanes 1 through 4) were not due to differences in plasmid recovery under the two conditions. Insofar as pheromone is known to induce the synthesis of a number of proteins necessary for conjugation (Clewell, 1993a; b), it is likely that this includes production of one or more products necessary for recognition of oriT2. The low level observed in the non-induced donors may reflect the previously identified phase variation phenomena, which reversibly switches on constitutive expression of conjugation functions at a frequency of 10−4–10−3 per cell per generation (Pontius and Clewell, 1991; Heath et al., 1995). In the case of the pAM373 system, cointegrate formation probably arose by a similar recombination, as no transfer was observed (Table 2) without the presence of the oriT segment within the vector and as suggested by the restriction analyses of DNA from Er-resistant transconjugants; however, we did not further address that system.
Transfer of pAD1 requires genes flanking oriT2
Open reading frames (ORFs) previously noted as orf53 and orf57 (Francia et al., 2001) are located on either side of oriT2 as indicated in Fig. 1. Interestingly, both are among the few ORFs of the pAD1 conjugation-related proteins that show significant differences when compared with their counterparts in pAM373. Although most of the conjugation genes show >95% identity compared with the pAM373 counterpart, Orf53 and Orf57 are 82% and 59% identical to Orf4 and Orf8 of pAM373 respectively. This and the results presented above suggest a possible relationship of these products to transfer specificity. The orf53 gene encodes a protein of 747 amino acids (MW 85 562) with a pI of 6.1; the SMART program (Schultz et al., 2000) predicts it to have three transmembrane segments in its amino-terminal region, the first of which corresponds to a signal sequence. The predicted Orf53 also has motifs corresponding to ATP-binding sites (Walker et al., 1982) and a putative FtsK–SpoIIIE domain (BLAST program). These characteristics are common to ‘TraG-like’ proteins that are essential for conjugal transfer (Cabezon et al., 1997; Gomis-Ruth et al., 2001). Database searches revealed homologous ORFs from plasmids in Streptococcus thermophilus, Streptococcus mutans, Pseudomonas putida and Salmonella enterica; and, interestingly, several of them are cited as hypothetical TraG proteins. To determine if Orf53 plays a role in transfer, we generated an in-frame deletion removing the complete Walker motif A (Fig. 5). The mutation was placed in pAM307, a pAD1::Tn917 derivative defective in production of cytolysin but with wild-type transfer properties. As shown in Table 3, the mutated plasmid pAM8131 completely abolished the ability to transfer and was not able to mobilize pAM8103 (pAM401 vector bearing oriT2). Only when the complete Orf53 was provided in trans via pAM8138 could transfer be partially restored, indicating that this product is essential for pAD1 conjugation.
Table 3. Conjugation and mobilization frequencies of pAM307 and its orf53 and orf57 mutants in the presence or absence of the complementing proteins.
The transfer frequencies indicated for each derivative represent the average of at least two independent experiments and are expressed as the number of Cm-resistant or Er-resistant transconjugants per donor cell.
pAM307 (parent plasmid)
9.0 × 10−2
pAM8130 (orf57 mutant)
pAM8131 (orf53 mutant)
6.0 × 10−1
4.1 × 10−2
6.0 × 10−4
The product of orf57 has a predicted size of 262 amino acids (MW 31 517) and a pI of 8.12. Database searches revealed no homology to known proteins. Interestingly, an internal 21-amino-acid segment is very similar to a portion of the CloDF13 protein MobC (Nunez and de la Cruz, 2001); however, when the total lengths of the proteins are compared there is not a strong resemblance (Fig. 5). MobC has been reported to be a relaxase without significant similarity to other known proteins including previously published relaxases (Nunez and de la Cruz, 2001). An in-frame deletion was generated in pAM307 which eliminated 15 amino acids of the segment of Orf57 common to MobC (Fig. 5) giving rise to pAM8130. The latter was unable to transfer nor could it mobilize pAM8103 (Table 3); however, transfer was almost fully restored when Orf57 was supplied via complementation (pAM8134). Thus Orf57 appears to be essential for pAD1 conjugation.
Genetic analyses of oriT2
As noted above, the large inverted repeat in oriT2 of pAD1 and the oriT of pAM373 exhibit interesting similarities. As seen in Fig. 3, single-stranded forms could take on similar folded structures; however, differences mainly situated in the loops are clearly evident. To determine if these differences could have a role in transfer specificity, a DNA segment representing a hybrid containing the inverted repeat of the pAM373 origin and the direct repeats of the pAD1 oriT2 was generated (see Experimental procedures) and placed in the pAM401 vector, resulting in pAM8106 (Fig. 6). As indicated in Table 4, pAM8106 was mobilized by pAM714 almost as well as (within an order of magnitude) the control with oriT2 (pAM8103). However, when the pAM373 system was used (pAM4020), transfer of the hybrid origin was very low (reduced by five orders of magnitude). The data imply that transfer specificity mainly resides in the region containing the direct repeats and that the inverted repeats may be interchangeable with only minor loss of efficiency.
Table 4. Mobilization frequencies for the derivatives containing base substitutions or deletions in the inverted repeat of the minimal pAD1 oriT site along with the hybrid oriT site (represented in Fig. 6), into E. faecalis OG1SS recipients.
The mobilization frequencies indicated for each derivative represent the average of at least two independent experiments and are expressed as the number of Cm-resistant or Er-resistant transconjugants per donor cell.
9.5 × 10−2
1.3 × 10−1
3.0 × 10−1
6.9 × 10−1
2.5 × 10−1
1.3 × 10−1
1.4 × 10−1
2.3 × 10−1
2.0 × 10−1
2.8 × 10−2
3.9 × 10−4
1.2 × 10−4
1.0 × 10−7
2.2 × 10−1
0.7 × 10−1
4.0 × 10−6
2.2 × 10−1
1.2 × 10−1
Several vector-oriT2 derivatives with deletions or base substitutions in the inverted repeat sequences (pAM8107–pAM8114; see Fig. 6 and Table 1) were then examined with respect to their mobilization in the presence of pAM714. The derivatives pAM8107–pAM8110 involved changes in only 2–4 bp, presumably affecting folded conformations and, as seen in Table 4, the slight reductions in transfer frequency do not appear significant. Even in the case of pAM8111, which contains a 52 nt deletion in the arm of the inverted repeat distal to the direct repeats, mobilization was still relatively efficient (Table 4). When larger deletions of the distal arm of this repeat were introduced (pAM8112, 66 nt deletion; and pAM8113, 87 nt deletion), a much greater reduction, about three to four orders of magnitude, was evident; although this level (on the order of 10−4 per donor) still represents significant transfer (e.g. compared with non-pheromone-responding plasmids). In the case of pAM8114, however, which contained an additional 22 nt deletion (compared with pAM8111) affecting the proximal arm of the inverted repeat (see Fig. 6), transfer was nearly abolished. The data suggest that the ability to form a hairpin structure involving the two repeats contributes significantly to the ability to transfer, and at least one specific segment of the inverted repeat proximal to the direct repeats is absolutely essential to the process.
In the cases where plasmid chimeras had structures that still allowed for folding identical to that shown in Fig. 3A, mobilization was functionally ‘wild-type’ (e.g. pAM8103 and pAM8107–pAM8110), and 100% of the transconjugants analysed contained an intact version of the donor plasmid, including the oriT site. (Thirty transconjugants of each were examined by restriction analysis, and PCR and sequencing was conducted on several representatives.) However, in cases where the distal arm of the inverted repeat was partially or totally missing (i.e. pAM1112 and pAM1113), 90% of the transconjugants appeared devoid of oriT2, judging by the negative PCR results obtained with specific primers. Also, different restriction patterns could be observed suggesting that some reorganization was occurring in the recipient, probably related to difficulty in aligning the 3′- and 5′-ends of the transferred strand for ligation. This percentage was reduced to just 10% in the case of pAM8111. The data are consistent with a role for the inverted repeat in assuring the recircularization of the transferred strand in the recipient.
Identification of nic site within the oriT2 and the involvement of Orf57
In an effort to determine the location of the transfer initiation site (or nic site) of pAD1, we first took advantage of the selection for cointegrate plasmids appearing in S. aureus and the site-specific recombination at oriT that produced such structures. In addition, we made use of the oriT hybrid (pAM8106) (see Fig. 6) which contained the large inverted repeat of the pAM373 origin and the direct repeats of the pAD1 oriT2 (the differences between the inverted repeat of pAD1 and the one of pAM373 are noted in Fig. 3). Er-resistant trans-conjugants appearing in S. aureus using E. faecalis UV202 donors contained cointegrates with distinguishable oriT sequences at the plasmid junctions (not shown). Importantly, the sequences of both junctions implied that site-specific recombination occurred inside the proximal arm of the inverted repeat, in the 37 nt region indicated in Fig. 6. This also overlaps partially with the 22 nt segment that was deleted in the non-mobilizable pAM8114 (see above section). The inability of pAM714 to mobilize pAM8114 is consistent with the possibility that a nic site necessary for transfer is located within the region that was deleted. And, assuming that the site-specific recombination occurs in the nic site as known to be the case for other conjugative systems, it should be located in the 13 nt overlap between positions 36585 and 36597 in Fig. 6.
To determine if a nic site is indeed located in this region, we performed ‘run-off’ DNA synthesis analyses, a sen-sitive technique detecting DNA strand discontinuities generated in vivo. E. faecalis cells carrying pAM307 (pAD1::Tn917 with wild-type mating properties) that had been induced or not with cAD1 and mixed with recipient bacteria were used. (The utilized primer was P.E./5.2.) As shown in Fig. 7A (lane 6), a band corresponding to 78 nt, which terminates at a specific site within the inverted repeat, is clearly evident in the case for the cAD1-induced cells. No band is present when the cells were not exposed to cAD1 (lane 2). The Taq DNA polymerase used is known to display terminal transferase activity, adding an additional nucleotide once it reaches the end of the template; thus the cleavage should be located between the T and G as indicated in Fig. 7A. The same site of interruption was observed when a different primer (8100/5) was used (not shown). A primer (8103/3) designed to detect a terminus on the opposite strand did not reveal a run-off product. Parallel experiments were conducted examining cells harbouring pAM8130 (mutated orf57) or pAM8131 (mutated orf53); and as shown in Fig. 7A, a similar termination site was observed in the case of pAM8131 (orf53 mutant) but not for pAM8130 (orf57 mutant) (lanes 7 and 8 respectively). The data imply that Orf57 has specific nicking activity and that the strand cleaved is on the strand complementary to the reading frames of orf53 and orf57.
Orf57 appears sufficient to generate cleavage in vivo in E. coli
To determine if Orf57 was sufficient for generating the nick within oriT2, an examination using E. coli was conducted. pAM8155 represents the His-tag expression vector pET30b carrying orf57. It was placed together with the pAM8151 plasmid that carries oriT2 in E. coli BL21. Exponentially growing cells preinduced with IPTG were then examined using a run-off DNA synthesis assay in the same manner as described above for E. faecalis. Figure 7B, lane 2, shows a band representing the same interruption that was noted for the fully conjugative pAD1 derivative (pAM307) in E. faecalis; the control experiment performed with the empty vector plus pAM8151 did not exhibit such a band (lane 1). Western blotting assays (anti His-tag) carried out in parallel showed that only in the case where pAM8155 was used, was Orf57 indeed present (not shown). The data imply that the pAD1 Orf57 protein on its own was able to specifically nick within the oriT2 site. It should be kept in mind that the excess of Orf57 in the E. coli system may suppress the need for accessory proteins that could be necessary for nicking to occur in the original E. faecalis system under normal physiological conditions.
A characteristic known thus far to occur exclusively in enterococci is the involvement of sex pheromones in the transfer of certain conjugative plasmids (Dunny et al., 1978; 1979). Such plasmids, which can transfer at frequencies approaching 100% under optimal conditions, are ubiquitous in E. faecalis; and there is evidence that they facilitate the mobilization of resident non-conjugative plasmids as well as enhance the transfer of other self-mobilizing elements such as conjugative transposons (Franke and Clewell, 1981; Clewell and Gawron-Burke, 1986; Clewell, 1990; 1999). pAD1 has been a useful model system for studying the pheromone response and is now yielding significant information directly relating to plasmid DNA transfer. In the work presented here, we have localized the second oriT identified in pAD1 (oriT2) to a 285 bp segment between orf53 and orf57, two genes required for plasmid transfer and located near the downstream end of a large group of similarly oriented structural genes relating to conjugation. We identified a structurally similar oriT site in pAM373 between orf4 and orf8, homologues of orf53 and orf57 respectively. Both sites were shown to facilitate transfer to S. aureus, and analyses of the pAD1 oriT2 sequence revealed the precise nicking site. Orf57 was found to be the relaxase that cleaves within the nic site, whereas Orf53 was found to be absolutely required for transfer.
The oriT sites in both pAD1 and pAM373 possessed multiple repeats (Fig. 2), a characteristic also found in other bacterial oriT regions (Lanka and Wilkins, 1995). A large inverted repeat sequence as well as nearby direct repeats within oriT2 were shown to be important for conjugation and constituted the minimal oriT site; and a similar organization was evident in pAM373. Indeed, the sequences of the two systems are closely related in the inverted repeats, but quite different in the direct repeats. Interestingly, examination of an oriT2/oriT (pAD1/ pAM373) hybrid construct suggested that plasmid transfer specificity resides mainly in the direct repeats. A schematic representation of the minimal oriT is shown in Fig. 8A. A very similar sequence containing a highly homologous inverted repeat near a series of direct repeats (non-homologous) is present in a cryptic plasmid, pER371, in S. thermophilus (Solaiman and Somkuti, 1998), although its function as an origin of transfer has not been reported. Similarly, there are sequences present in two plasmid systems in E. faecalis V583 (complete genome sequenced by TIGR) that are highly homologous and are likely to serve as oriT sites. (Based on extensive homology outside these sites, the V583 plasmids appear closely related to pheromone-responding elements.) Although these five systems exhibit strong similarity in their large inverted repeat, particularly around the nic site (oriT2), significant homology with other known oriT sites from both Gram-positive and -negative systems is not evident. However, a similar arrangement to the one shown in Fig. 8A has been recently reported for the non-conjugative, but mobilizable, Gram-negative plasmid CloDF13 oriT (Nunez and de la Cruz, 2001). Only a small region close to the nic site of CloDF13 exhibits some sequence similarity to the nic sequence of interest in the present study (Fig. 8B).
It is interesting that a sequence organization very similar to the oriT structures identified in the present study is present in the double-stranded replication origins (dso) of the pMV158 family of rolling circle replication plasmids (del Solar et al., 1998). The dso of these elements consists of two or three direct repeats (called the bind region), a spacer region, and an inverted repeat called the nic region or hairpin I. The distance between the conserved nic site and the non-conserved bind locus ranges between 14 and 95 nt among the different replicons of the family. In the case of pMV158, RepB protein binds in vitro to a dsDNA fragment containing the direct repeats. These sequences appear to be essential for plasmid replication in vivo, but not for in vitro relaxation of supercoiled DNA. Hairpin I of pMV158 is sufficient for the nicking-closing reaction mediated by RepB, which is also able to recognize in vitro the nic regions of other plasmids of the family (del Solar et al., 1998).
A reasonable hypothesis consistent with the data from the pAD1 and pAM373 systems is that the direct repeats serve as specific binding sites for the protein involved in the nicking reaction providing the observed specificity, whereas the nick site would be located in the inverted repeat. Also, it is possible that the inverted repeat could assume a hairpin structure that may be essential to the initiation of DNA transfer. This would be consistent with the 1000-fold drop in the transfer frequency when the distal arm of the inverted repeat was deleted. However, the frequent appearance of aberrant structures among the reduced number of transconjugants that do arise may relate to a role for the inverted repeat in termination of DNA transfer by recircularizing the newly transferred DNA strand. Extensive studies of the oriT sites of other plasmids have shown that initiation and termination of transfer are processes requiring different sequence characteristics (Bhattacharjee et al., 1992; Zhang and Meyer, 1995; Becker and Meyer, 2000; Furuya and Komano, 2000), reflecting the generally accepted model maintaining that initiation requires supercoiled dsDNA whereas termination involves ssDNA (Lanka and Wilkins, 1995;de la Cruz and Lanka, 1998). Indeed, termination usually requires an inverted repeat commonly found in the oriT in the last portion of the plasmid to be transferred (Bhattacharjee et al., 1992; Furuya and Komano, 2000). Thus, the 1000-fold reduction we observed in transfer when the distal arm of the inverted repeat was missing may relate to the inability to terminate efficiently in the recipient.
Five families of oriT sites have been noted (Guzman and Espinosa, 1997; Zechner et al., 2000) upon comparison of a large variety of origins. Previously studied Gram-positive (conjugative and mobilizable) plasmids can be included in one or another of these five families by sequence similarities close to the nick site. Thus, pIP501 (Wang and Macrina, 1995), pGO1 (Climo et al., 1996), pSK41 (Berg et al., 1998) and pMRC101 (Dougherty et al., 1998) oriT sites exhibit similarities to IncQ oriT´s, whereas pNZ4000 oriT (van Kranenburg and de Vos, 1998) and oriT1 of pAD1 (An and Clewell, 1997) are highly homologous to the IncP type. The pMV158 oriT gave rise to the formation of a new family, that now includes mainly Gram-positive mobilizable RCR plasmids (Guzman and Espinosa, 1997); and several recently identified Gram-negative members (Smith and Parker, 1998; Szpirer et al., 2001) are also associated with this family. oriT2 of pAD1 and oriT of pAM373 are not similar to any of these groups. Figure 8B shows a comparison to the oriT consensus sequence for all five oriT families and to the consensus sequence of the dso of RCR plasmids (Zechner et al., 2000). The only similarities relate to the centrally located TG site and an A, located four residues away, that interestingly, represent the highest conserved nucleotides when compared with the nic sites in both oriT´s and dso’s (Fig. 8B). It would appear that the group consisting of pAD1, pAM373, pER371 and CloDF13 represents a new family of oriT sites, also reflecting the existence of specific relaxases.
To date, conjugative relaxases have been classified in four groups, according to conservation of their amino acid sequences and of their target nic sites (Zechner et al., 2000). Based on the high degree of homology with Orf57 of pAD1, Orf8 of pAM373 is also likely to represent the specific relaxase for that plasmid. Orf57 and Orf8 have no significant homologues in the database other than with respective determinants on plasmids in the recently sequenced E. faecalis V583 genome. They do not have the ‘3-histidine motif’ characteristic of other relaxases and apparently essential for relaxase activity, nor does MobC, the functional relaxase of CloDF13 (Nunez and de la Cruz, 2001). Interestingly, when they are compared with MobC, there is a 21-amino-acid segment that is conserved but, whereas the sizes of Orf57 and MobC are very similar, significant homology over the entire reading frames is not evident. In addition, their hydrophilicity profiles do not closely resemble each other outside the 21-residue segment they have in common. The present study showed that an in-frame deletion (15 amino acids, including a tyrosine residue) in this region eliminated the nicking function of Orf57. This possibly suggests an important function for these amino acids at or near an active site of the protein, although we cannot rule out that the mutation inactivated the protein via a major conformational change. The ability of Orf57 to facilitate site-specific recombination between oriT sites, as revealed in the present work by the cointegration phenomena associated with transfer from E. faecalis to S. aureus, is not unique in that other relaxases are known to exhibit such an activity (Broome-Smith, 1980; Llosa et al., 1994).
Although Orf53 was found to be required for pAD1 transfer, it was not necessary for nicking; and it is likely that this is the case for Orf4 of pAM373. The TraG-like characteristics present in these proteins suggest they could be the ‘coupling proteins’ that interact with both DNA processing (relaxosome) and mating-pair-formation functions specific for each system, similar to the situation observed for other well characterized conjugative plasmids (Balzer et al., 1994; Cabezon et al., 1997; Errington et al., 2001; Gomis-Ruth et al., 2001). It is conceivable therefore that Orf53 interacts with Orf57 or accessory proteins forming the pAD1 relaxosome at some point during DNA transfer. It is interesting that MobB of CloDF13, the proposed TraG-like protein in that system, is required for nicking at oriT, suggesting that the nature of such interactions in the two systems may be different (Nunez and de la Cruz, 2001). The fact that the Orf53 in-frame deletion mutant was not fully complemented in trans may imply that the protein functions as a multimer and that the mutant protein had a dominant-negative effect. A similar behaviour was observed in the complementation studies of TrwB, the TraG-like protein of the Gram-negative plasmid R388 (Moncalian et al., 1999). According to its recently solved crystal structure (Gomis-Ruth et al., 2001), six equivalent protein monomers would associate to form an almost spherical quaternary structure, with a central channel that would be able to accommodate a single DNA strand.
Our studies provide evidence that when oriT2 is utilized, pAD1 transfers from donor to recipient as a single strand of DNA conforming to the model established for the conjugative transfer of other Gram-negative and -positive plasmids (Lanka and Wilkins, 1995; Zechner et al., 2000), except for Streptomyces in which double-strand DNA was recently shown to transfer (Possoz et al., 2001). Assuming that the direction of transfer is 5′ to 3′, as in the other systems described so far (Zechner et al., 2000), then the orf53 determinant would enter the recipient first, whereas the relaxase determinant would enter last. Transfer specificity appears to relate to the cognate relaxase as well as its specific recognition site, which is consistent with our data involving the direct repeats in the transfer specificity; however, the studies with the hybrid origin suggest that an additional component is also involved. This relates to the finding that the transferable pAM373 derivative (pAM4020) was able to mobilize the hybrid oriT (pAM8106), albeit at a greatly reduced efficiency compared with pAD1, but not the oriT2 (pAM8103) of pAD1 (Table 4). Preliminary negative results (not shown) from complementation experiments in which Orf57 was supplied in trans to attempt to rescue the mobilization of pAM8103 by pAM4020, are also consistent with the involvement of an additional specificity factor(s). Conceivably Orf53 and/or maybe another protein forming part of the pAD1 relaxosome, will contribute in this regard, as it is the case for other well studied conjugation systems (Cabezon et al., 1997; Fekete and Frost, 2000; Hamilton et al., 2000). In the future, orf53 and orf57 will be referred to as traW and traX respectively.
Bacterial strains, plasmids, oligonucleotides and reagents
Bacterial strains, plasmids and oligonucleotides used in this study are listed in Table 1. E. faecalis strains were grown in Todd–Hewitt broth (THB) (Difco Laboratories) at 37°C, unless otherwise noted. E. coli strains were grown in Luria–Bertani (LB) broth (Sambrook et al., 1989). Plating was on THB agar. The following antibiotics were used at the indicated concentrations when using E. faecalis: erythromycin, 20 μg ml−1; streptomycin, 500 μg ml−1; kanamycin, 500 μg ml−1; spectinomycin, 500 μg ml−1; chloramphenicol, 20 μg ml−1; tetracycline, 10 μg ml−1; rifampin, 25 μg ml−1; and fusidic acid, 25 μg ml−1. When using E. coli, concentrations were: ampicillin 100 μg ml−1; kanamycin, 50 μg ml−1; chloramphenicol, 25 μg ml−1; spectinomycin, 50 μg ml−1; erythromycin, 200 μg ml−1; and nalidixic acid, 20 μg ml−1. All antibiotics were obtained from Sigma Chemical Co. Xgal (5-bromo-4-chloro-3-indolyl-D-galactopyranoside) and IPTG were from Invitrogen and were used at concentrations of 40 μg ml−1 and 1 mM respectively. Synthetic cAD1 peptide was prepared at the University of Michigan peptide synthesis core facility.
Recombinant plasmids were generated in E. coli DH5α. Introduction of plasmid DNA into bacterial cells was by transformation as described previously (Hanahan, 1983; Dower et al., 1988). Electrotransformation of E. faecalis was as described by Flannagan and Clewell (Flannagan and Clewell, 1991). Plasmid DNA was purified from E. coli using established techniques described elsewhere (Sambrook et al., 1989). Isolation of plasmid DNA from E. faecalis was also as previously described (Weaver and Clewell, 1988). When necessary, DNA fragments were purified with silica gel as described by Boyle and Lew (Boyle and Lew, 1995). Recombinant DNA methodology as well as analyses of plasmid DNA using restriction enzymes and agarose gel electrophoresis involved procedures described by Sambrook and colleagues (Sambrook et al., 1989). Restriction enzymes were purchased from Invitrogen, and reactions were carried out under the conditions recommended. Polymerase chain reaction (PCR) was performed with a Perkin-Elmer Cetus apparatus under conditions recommended by the manufacturer. Specific primers were purchased from Invitrogen and Taq DNA polymerase from Roche. PCR-generated fragments were purified by using QIAquick-spin columns (Qiagen). Ligations made use of T4 DNA ligase from New England Biolabs. Nucleotide sequence analyses were carried at the University of Michigan sequencing core facility or using the fmol DNA cycle sequencing system as specified by the manufacturer (Promega).
Filter matings were performed as previously described (Clewell et al., 1985). Broth matings (in THB) were for 20 min, unless otherwise indicated, after induction of donors for 90 min with synthetic pheromone (5 ng ml−1 to 5 μg ml−1, depending on the experiment). Transfer frequencies are expressed as the number of transconjugants per donor cell (at the end of the mating). The plasmid content of transconjugants was verified electrophoretically when appropriate.
Genetic analyses of oriT
Segments of pAD1 containing oriT2 with various amounts of flanking DNA were amplified by PCR using the oligonucleotides indicated in Table 1 and cloned into pTAd via TA cloning. The same strategy was used to construct the deletion derivatives and clones containing the pAM373 oriT (see Table 1 for related primers used for each construction). From here XbaI–BamHI fragments were then subcloned into the shuttle plasmid pAM401. The derivatives with point mutations were picked up as ‘unexpected’ variants noticed upon sequencing PCR products as was the deletion relating to pAM8114.
The clone representing a hybrid of the oriT´s of pAD1 and pAM373 was constructed as follows. The 128 nt PCR fragment containing the direct repeats of pAD1 was generated using primers 8100/5 and 8106/B and pAM714 as template DNA. The 240 nt fragment containing the inverted repeat of pAM373 was generated using 8106/C and 8301/3 as primers and pAM373 as template. Both fragments were purified using QIAquick-spin columns (Qiagen), diluted 1:1000, mixed and used as a template for a new PCR reaction using 8100/5 and 8301/3 primers; the resulting PCR product (334 bp) was purified and ligated to pTAd. And, as before, a XbaI–BamHI fragment was cloned into pAM401 obtaining pAM8106.
All clones were confirmed by sequencing.
Generation of orf53 and orf57 mutants and complementation experiments
The plasmid pAM88 was used as a suicide plasmid in E. faecalis to generate the orf53 and orf57 pAD1 mutants. It was constructed as follows. A 1590 bp fragment containing the cat determinant of pAM401, was amplified by PCR (primers used shown in Table 1) and cloned into the pTAd vector, via TA cloning. The 1.6 kb EcoRI fragment from this plasmid was gel-extracted and ligated into the EcoRI site of pSU18 (Bartolome et al., 1991), resulting in pAM88.
A mutant of orf53 containing a 36 bp in-frame deletion substituted with a SalI restriction site was generated as follows. A PCR fragment of 705 bp was generated using primers 8121/5 and 8121/3 (contains added SalI site) and cloned into pTAd (no SalI site in vector) obtaining pAM8121. Another PCR fragment, of 774 bp, was generated using primers 8122/5 (contains added SalI site) and 8122/3 and cloned into pTAd, obtaining pAM8122A. This plasmid was digested with SalI–XbaI restriction enzymes and the 0.8 kb segment was cloned into the same sites of pAM8121, obtaining pAM8122B. The 1.5 kb XbaI–HindIII fragment derived from the digestion of this plasmid was then cloned into the vector pAM88 resulting in pAM8123 with a 36-bp deletion in orf53.
A similar approach was used to construct an orf57 mutation with a 45 bp in-frame deletion and again containing a new SalI restriction site. A PCR fragment of 307 bp was generated using primers 8124/5 and 8124/3 (contains added SalI site) and cloned into pTAd obtaining pAM8124. The other PCR fragment of 141 bp used primers 8125/5 (contains added SalI site) and 8125/3 and was cloned in pTAd resulting in pAM8125A. This was then digested with SalI–XbaI enzymes and the resulting fragment (≈ 0.2 kb) was cloned in pAM8124 resulting in pAM8125B, which contained the deletion. The 448 bp XbaI–HindIII fragment of pAM8125B was then cloned in pAM88 resulting in pAM8126.
The plasmids used for the complementation experiments were generated as follows. The 0.3 kb BglII–NcoI fragment of pMSP3545 (contains nisin promoter) was cloned into the same sites of pET30b, obtaining pAM8132. A PCR fragment of 0.8 kb containing orf57 was generated using primers 8133/5 and 8133/3, purified, digested with NcoI–XhoI, and cloned into the same sites of pAM8132, obtaining pAM8133. This plasmid was digested with BglII–BseA1, and the resulting 1.2 kb fragment (containing the nisin promoter, the orf57 gene and the His-tag) was cloned into the BglII–XmaI sites of the pMSP3535VA E. coli-E. faecalis shuttle vector, resulting in pAM8134.
The construction of pAM8138 was as follows. A PCR fragment of 2.25 kb containing orf53 was generated, purified, digested with NotI–SalI restriction enzymes and cloned into the same sites of pASK60, generating pAM8135. The 0.2 kb fragment carrying the Bacteriocin 21 promoter was generated by PCR using pMGS100 as template, purified, digested with EcoRI, and cloned into the EcoRI site of pSU18, obtaining pAM8136. A 2.5 kb NotI–HindIII fragment carrying orf53-streptavidin tag (from pAM8135) was cloned into the EagI–HindIII sites of pAM8136, resulting in pAM8137. This plasmid was digested with EcoRI–HindIII and the 2.7 kb fragment (containing the bac promoter and orf53-streptavidin tag) was gel-extracted and cloned into the same sites of the E. coli-E. faecalis shuttle vector pDL278, obtaining pAM8138.
The plasmid constructs carrying the mutated orf53 and orf57 segments (pAM8123 and pAM8126 respectively) were introduced into E. faecalis JH2-2/pAM307 by electrotransformation, and integrants occurring via homologous recombination were selected using chloramphenicol. A representative transformant from each (pAM8128 and pAM8129) was subcultured for several passages in THB without drug and plated on medium containing erythromycin (to maintain pAM307 recombinants). Colonies were then replica-plated on medium containing chloramphenicol, and candidates (erythromycin-resistant but chloramphenicol-sensitive) were screened using PCR amplification (primers 8121/5 and 8122/3 for the orf53 mutant, and 8155/5 and 8155/3 for the orf57 mutant) to obtain DNA to be examined for the presence of the SalI cleavage site. Representative recombinants of pAM307 were designated pAM8130 (orf57 mutation) and pAM8131 (orf53 mutation) and confirmed by sequencing.
For the related complementation studies, chimeras representing orf53 (pAM8138) and orf57 (pAM8134) were introduced into the strains containing pAM8131 and pAM8130 respectively. The empty vectors, pDL278 and pMSP3535VA, respectively, were used as negative controls.
Determination of the nic location
The location of the nic site was determined using a run-off DNA synthesis assay (primer extension using Taq polymerase (Roche) coupled with amplification using a Perkin Elmer thermocycler) (Zechner et al., 1997). Parallel reac-tions providing sequence data (used as size marker) were conducted using the fmol DNA cycle sequencing system (Promega) and a DNA template containing oriT2 (pAM8103). Bacterial strains cultured overnight were diluted 1:20 for E. coli (BL21 derivatives) or 1 : 50 for E. faecalis (UV202 deriva-tives) in fresh media (5 ml) and grown to an optical density of 0.3–0.5 (600 μm). When E. faecalis was used, cAD1 (5 μg ml−1) was added, if indicated, during last 90 min of growth. E. coli was induced with IPTG (1 mM). A 10-fold excess of E. faecalis‘recipients’ (OG1SS) was added to culture, and after 5 min the cells were rapidly cooled to 0°C. Aliquots of cultures containing 10–30 × 106 colony-forming units (cfu) were collected and added to reaction mixtures as described by Zechner and colleagues (Zechner et al., 1997). The conditions for the cycle programme were chosen empirically by first optimizing the yield and specificity of DNA amplification from plasmid-carrying bacteria with the appropriate oligonucleotides. 8100/5 or P.E/5.2 primers were used for the strand that is nicked, and 8103/3 for the complementary strand (see Table 1). The cycle programme was set for 35 cycles as follows: (i) for E. faecalis/pAD1, 95°C × 40 s, 55°C × 1 min, and 72°C × 1 min; (ii) for E. coli/pAM8151, 95°C × 30 s, 60°C × 30 s, and 72°C × 30 s. The reaction products were treated as described and analysed on 6% polyacrylamide gels containing 6 M urea. Gels were visualized by autoradiography at –70°C for 1–5 d with Kodak X-Omat film and intensifying screens.
Cells were disrupted with SDS and β-mercaptoethanol and subjected to SDS–PAGE (12%), stained with Coomassie brilliant blue R-250, according to the method of Laemmli (Laemmli 1970). Western blotting was performed as described elsewhere (Sambrook et al., 1989), and the His-tag fusion protein (Orf57) detected using polyclonal anti-His antibody (#sc803, Santa Cruz Biotechnology) and the ECL Western Blotting analysis system (Amersham Pharmacia Biotech).
This work was supported by National Institutes of Health grant GM33956. M.V.F. was supported in part by a grant from NATO. We thank Gary Dunny for providing the vectors pMSP3535VA and pMSP3545, and Shuhei Fujimoto for the vector pMGS100. We thank all members of our laboratory for helpful discussions.