Specificity determinants of conjugative DNA processing in the Enterococcus faecalis plasmid pCF10 and the Lactococcus lactis plasmid pRS01

Summary

The DNA-processing region of the Enterococcus faecalis pheromone-responsive plasmid pCF10 is highly similar to that of the otherwise unrelated plasmid pRS01 from Lactococcus lactis. A transfer-proficient pRS01 derivative was unable to mobilize plasmids containing the pCF10 origin of transfer, oriT. In contrast, pRS01 oriT-containing plasmids could be mobilized by pCF10 at a low frequency. Relaxases PcfG and LtrB were both capable of binding to single-stranded oriT DNAs; LtrB was highly specific for its cognate oriT, whereas PcfG could recognize both pCF10 and pRS01 oriT. However, pcfG was unable to complement an ltrB insertion mutation. Genetic analysis showed that pcfF of pCF10 and ltrF of pRS01 are also essential for plasmid transfer. Purified PcfF and LtrF possess double-stranded DNA binding activities for the inverted repeat within either oriT sequence. PcfG and LtrB were recruited into their cognate F—oriT DNA complex through direct interactions with their cognate accessory protein. PcfG also could interact with LtrF when pCF10 oriT was present. In vivo cross-complementation analysis showed that ltrF partially restored the pCF10ΔpcfF mutant transfer ability when provided in trans, whereas pcfF failed to complement an ltrF mutation. Specificity of conjugative DNA processing in these plasmids involves both DNA—protein and protein—protein interactions.

Introduction

Conjugation is the process of unidirectional DNA transfer from a donor to a recipient bacterium through direct cell—cell contact. The DNA-processing reaction at the origin of transfer (oriT) requires formation of a multiprotein complex (relaxosome) containing relaxase and accessory proteins (Lanka and Wilkins, 1995; Zechner et al., 2000). Relaxase cleaves site- and strand-specifically at the nic site in the oriT via transesterification, leading to production of a single-stranded transfer DNA (Byrd and Matson, 1997). After the cleavage reaction, relaxase covalently binds to the 5′-end of the cleaved strand and subsequently the nucleoprotein complex is transferred into the recipient through the mating pore (Byrd and Matson, 1997). Almost all characterized relaxases are capable of binding and cleaving oriT containing single-stranded oligonucleotides (ssDNA), independent of other relaxosome proteins. However, plasmid or host-encoded accessory proteins are often required when double-stranded supercoiled DNA is used as a substrate (Zechner et al., 2000).

The tetracycline-resistance plasmid pCF10 represents a family of highly mobile elements found in Enterococcus faecalis, where expression of conjugation functions in plasmid-containing donor cells is induced by peptide mating pheromones produced by recipient cells (Clewell and Dunny, 2002; Clewell et al., 2002). In addition to promoting transmission of the antibiotic resistance and virulence determinants they frequently carry, the pheromone-inducible plasmids can mobilize other elements to both enterococcal and non-enterococcal recipients (Clewell and Dunny, 2002; Staddon et al., 2006), further contributing to the dissemination of medically important traits among bacterial pathogens. While the pheromone-controlled regulatory circuits of these plasmids and their virulence genes have been studied extensively, less attention has been paid to the mechanism by which conjugative transfer occurs, and to the genes involved in conjugative DNA processing and in mating pair formation. Molecular and genetic analyses of several pheromone-inducible plasmids suggest that the genes encoding pheromone sensing and response are generally quite similar, and that they probably arose from a common ancestor, even though different plasmids determine response to different peptide pheromones (Francia et al., 2001; Clewell and Dunny, 2002; Hirt et al., 2005). Recent sequencing studies suggest that this conservation does not extend into the pheromone-inducible transfer genes believed to encode DNA processing and mating pair formation (Francia et al., 2001; Hirt et al., 2005); rather the grouping of the three functions required for pheromone-inducible conjugation (pheromone response, mating pair formation, and DNA processing) seem to have arisen on multiple occasions by linkage of modules encoding each function that appear to have different origins in different plasmids (Hirt et al., 2005). For example, in the case of pCF10, the sequences encoding putative DNA-processing enzymes and the oriT target are most closely related to those of the Lactococcus lactis element pRS01 (Mills et al., 1995; 1998; Staddon et al., 2006), and are members of a large superfamily of conjugative mobilization systems that include the staphylococcal plasmids pC221 and pC223 (Caryl et al., 2004; Caryl and Thomas, 2006). The pCF10 relaxase protein PcfG shares 50% sequence identity with the pRS01 relaxase LtrB, and pcfG contains a functional target for the insertion of the mobile Group II intron Ll.ltrB (Staddon et al., 2004) The cognate oriT regions are over 70% identical (Staddon et al., 2006); experiments involving insertion of conditionally non-splicing derivatives of the intron into both relaxase genes provided important genetic evidence that these genes were essential for transfer (Mills et al., 1996; Staddon et al., 2006). In contrast, the DNA-processing regions of the pheromone-responsive plasmids pAD1 and pAM373 show no sequence similarity to pCF10 (Francia et al., 2001; Francia and Clewell, 2002), and the relaxase proteins of these plasmids appear to be members of a new class of proteins that includes CloDF13 of Gram-negative origin; these proteins lack a ‘3-Histidine’ motif common in many other relaxases including PcfG and LtrB (Nunez and De La Cruz, 2001).

The high degree of similarity between the regions of pRS01 and pCF10 encoding DNA processing prompted us to undertake a detailed comparative analysis of these systems. We identified highly similar accessory proteins LtrF and PcfF as essential components of each transfer system, in addition to the cognate relaxases and oriT sequences. We also show that, in spite of the high degree of similarity of these systems, each processing apparatus was specific for its cognate oriT, with the pRS01 system unable to promote mobilization of the pCF10 oriT and pCF10 promoting only inefficient mobilization of the pRS01 oriT. The cumulative genetic studies and biochemical analyses of the DNA binding and nicking activities of the relaxase and accessory proteins reported here suggest that the molecular basis for the specificity of the two systems is conferred by relaxase/DNA interactions, as well as interactions between the relaxases and the accessory proteins.

Results

Genetic analysis of pCF10 and pRS01 DNA-processing regions

The nucleotide sequence of pCF10 was recently completed (Hirt et al., 2005), and oriT was mapped to a sequence of 40 nucleotides (Staddon et al., 2006). Flanking the oriT, there are four open reading frames (ORFs), pcfD, -E, -F and -G, encoding putative DNA-processing proteins and sharing homology with the Tra1 and Tra2 transfer locus of pRS01. Figure 1A shows the overall gene organizations of pCF10 pcfD-pcfG and pRS01 Tra1–2 regions, and a summary of percentages of identity between related proteins encoded by the ORFs. The amino terminus of PcfD aligns with bacterial primases (COG0358) and contains a DnaG motif 5. The PcfD carboxy-terminal segment aligns with anti-restriction enzymes (COG4227) and is 49% identical to the carboxyl terminus of LtrC. LtrM and the amino terminus of LtrC do not contain any domains recognized by NCBI software, nor do they have pCF10 homologues. PcfE and LtrD are 36% identical and have no homo-logues in the database. pcfF is located between the oriT and pcfG, encoding a small basic protein that is 45% identical to LtrF. Both PcfF and LtrF are members of the MobC accessory protein family. As previous sequence analyses and genetic experiments (Mills et al., 1995; Staddon et al., 2006) indicated that ltrB and pcfG encoded the relaxases for these plasmids and that these genes were essential for transfer, we investigated the requirements for the putative accessory genes linked to the relaxase determinants.

Fig. 1.

A. Comparison of pCF10 and pRS01 DNA-processing regions. The minimal pCF10 oriT is a 40 bp sequence located in the intergenic region between pcfE-pcfF (Staddon et al., 2006). The intergenic region between ltrD-ltrE contains the functional pRS01 oriT (Mills et al., 1998), but the minimal oriT sequence was not determined previously. Dashed lines represent homology between gene products and the percentages of amino acid sequence identity are indicated. ltrM and ltrE encode proteins with no homologues in pCF10. ltrB is interrupted by the group II intron Ll.ltrB (Mills et al., 1996). The intron insertion site in ltrB is indicated by a vertical arrow.

B. Sequence alignment of the minimal pCF10 oriT with other origins of transfer (adapted from Staddon et al., 2006). Inverted repeats are shown as dashed arrows. Asterisks represent nucleotides identical to the pCF10 sequence. The nic sites in pC221, pC223 and RP4 are indicated as ⇓. The Tn5252 sequence is part of a 2 kb sequence that includes the relaxase substrate (Srinivas et al., 1997).

We generated in-frame deletion mutations in pcfD, pcfE and pcfF, respectively, to test the requirements of these genes in conjugation (Table 2). Deletion of pcfD resulted in a 17- to 25-fold reduction in plasmid transfer frequency, and deletion of pcfE reduced the frequency 50- to 70-fold. No plasmid transfer was detected in the pcfF mutant. Complementation with PcfD, PcfE or PcfF in trans fully restored the conjugation ability of these mutants (Table 2). These data suggest that pcfD and pcfE are involved in, but not essential for transfer, while pcfF is absolutely required.

Table 2.

Effects of mutations in the DNA-processing genes on pCF10 and pRS01 transfer.

Plasmids in donor	Transfer frequency (T/D)a,b	Comments
pCF10	1.7 × 10-1
pCF10ΔpcfD	7.11 × 10-3
pCF10ΔpcfD/pCY18	3.2 × 10-1	Complementation with pcfD
pCF10ΔpcfE	2.5 × 10-3
pCF10ΔpcfE/pCY17	2.4 × 10-1	Complementation with pcfE
pCF10ΔpcfF	< 10-8
pCF10ΔpcfF/pCY16	1.27 × 10-1	Complementation with pcfF
pCF10ΔpcfF/pCY23	1.97 × 10-5	Complementation with ltrF
pCF10::Ll.ltrBΔORF-Kan	< 10-8	Non-splicing intron inserted in pcfG
pCF10::Ll.ltrBΔORF-Kan/pCY19	3.5 × 10-1	Complementation with pcfG
pCF10::Ll.ltrBΔORF-Kan/pCY20	< 10-8	Complementation with ltrB
pM2036	6.3 × 10-1	Transfer-proficient pRS01 derivative
pM2036ΔltrF	< 10-8
pM2036ΔltrF/pCY23	6.8 × 10-1	Complementation with ltrF
pM2036ΔltrF/pCY16	< 10-8	Complementation with pcfF
pM1014	< 10-8	Transfer-deficient pRS01 derivative
pM1014/pCY20	2.5 × 10-2	Complementation with ltrB
pM1014/pCY19	< 10-8	Complementation with pcfG

^aTransfer frequencies of pCF10 and its derivatives were tested in E. faecalis. Exponential donor and recipient cells were mixed at a 1:10 ratio and mated for 1 h at 37°C. After dilutions, cells were plated on selective agar plates to select for donor or transconjugants. Plasmids transfer frequencies are expressed as the number of transconjugants per donor cell (T/D) and represents the average of at least two independent experiments.

^bTransfer frequencies of pRS01 derivatives were tested in L. lactis. Exponential donor and recipient cells were mixed at a 1:1 ratio and mated on GM17 agar plates for 20 h. Plasmids transfer frequencies are expressed as the number of transconjugants per donor cell (T/D) and represents the average of at least two independent experiments.

Based on the finding that pcfF is essential for conjugation, we generated an ltrF in-frame deletion mutant [the mutation was generated in a derivative of pRS01, pM2036, which is a co-integrant between pRS01 and pTRK28 that exhibits high frequency transfer, and contains a useful selective marker (Mills et al., 1994; 1996)]. Similar to the pCF10ΔpcfF mutant, no plasmid transfer was observed in the pM2036ΔltrF mutant, indicating that ltrF is also required for conjugation (Table 2).

Binding of PcfG and LtrB to single-stranded oriT DNA sequences

Current knowledge of relaxase function and enzymatic activities is mostly derived from work with plasmids in Gram-negative bacteria (Zechner et al., 2000). To date, only four relaxases from Gram-positive plasmids have been purified and characterized in some detail: TraA of the enterococcal plasmid pIP501 (Kurenbach et al., 2002; Kopec et al., 2005), MobM of the streptococcal plasmid pMV158 (Guzman and Espinosa, 1997; Grohmann et al., 1999; de Antonio et al., 2004), and MobA proteins of the staphylococcal plasmids pC221 and pC223 (Caryl et al., 2004). PcfG and LtrB have been assigned as relaxases based on protein sequence analysis (Mills et al., 1996; Staddon et al., 2004). To initiate biochemical characterization, pcfG and ltrB were cloned into the expression vector pET-28b(+) and expressed as C-terminal His₆-tagged proteins in Escheirchia coli. When overexpressed, both LtrB and PcfG were found in the inclusion body fractions (data not shown). Therefore, they were purified using metal ion chelating chromatography under denaturing conditions and refolded in vitro.

Electrophoretic mobility shift assays (EMSA) were used to analyse PcfG— and LtrB—DNA interactions. The oligo-nucleotides used in the assays are shown in Table 1. pCF10-S contains the sense-strand sequence of the minimal pCF10 oriT, and pCF10-AS contains the corresponding antisense-strand sequence. Previous work showed that the intergenic region between ltrD and ltrE contains a functional pRS01 oriT (Mills et al., 1998). Inside this region, we identified a segment of DNA that is highly homologous to the minimal pCF10 oriT (Staddon et al., 2006) (Fig. 1B; this sequence is referred as pRS01 oriT in the rest of the paper). Two probes were designed: pRS01-S contains the sense-strand sequence of this segment of DNA, and pRS01-AS contains the corresponding antisense-strand sequence. The oligonucleotides were digoxigenin (DIG)-labelled at the 3′-ends and incubated with purified LtrB or PcfG.

Table 1.

Bacterial strains, plasmids and oligonucleotides used in this study.

	Relevant featuresa	Reference(s)
Strains:
E. coli
DH5α	F- Φ80dlacZ ΔM15 Δ(lacZYA-argF) U169 deoR recA1 endA1 hsdR17 (rK- mK+)  phoA supE44 λ- thi-1 gyrA96 relA1	Gibco-BRL
BL21 (DE3)	F- ompT rb- mb- DE3	Novagen
EC1000	E. coli cloning host, provides RepA in trans	Leenhouts et al. (1996)
E. faecalis
OG1RF	RifR, FusR	Dunny et al. (1978)
CK104	OG1RFΔupp2	Kristich et al. (2005)
OG1ES	ErmR, StrR	Staddon et al. (2006)
OG1SSp	StrR, SpcR	Dunny et al. (1978)
L. lactis
LM2301RF	RifR, FusR	Mills et al. (1996)
MMS372	StrR, TetR, Rec-	Anderson and McKay (1983)
MMS370	StrR, NeoR, Rec-	Anderson and McKay (1983)
MG1363	Plasmid-free derivative of L. lactis ssp. lactis 712	Polzin and McKay (1992)
Plasmids:
pCF10	Pheromone-inducible conjugative plasmid	Dunny et al. (1981)
pCF10::Ll.ltrBΔORF-Kan	pcfG insertional mutant of pCF10	Staddon et al. (2006)
pM2036	Co-integrant of pRS01 and pTRK28, transfer proficient, ErmR	Mills et al. (1994)
pM1014	Co-integrant of pRS01 and pTRK28, transfer deficient, ErmR	Mills et al. (1994)
pDL278	Shuttle vector, SpcR	Zhou et al. (2000)
pMSP3535	Nisin-inducible expression vector, ErmR	Bryan et al. (2000)
pCJK21	Nisin-inducible expression vector, ErmR, SpcR	Kristich et al. (2005)
pORI280	Requires RepA in trans for replication, ErmR	Leenhouts et al. (1996)
pCK47	Carries oriTpCF10, lacZ, and P-pheS* cassette	Kristich et al. (2007)
pET-28 b(+)	E. coli protein expression vector	Novagen
pET-28 b(+)-HltrB	Expresses C-terminal His6-tagged LtrB	This study
pET-28 b(+)-HltrF	Expresses N-terminal His6-tagged LtrF	This study
pET-28 b(+)-HpcfG	Expresses C-terminal His6-tagged PcfG	This study
pET-28 b(+)-HpcfF	Expresses N-terminal His6-tagged PcfF	This study
pCY10	ΔpcfF (Δ31–327 bp) allele in pCJK47	This study
pCY12	ΔpcfD (Δ37–2127 bp) allele in pCJK47	This study
pCY14	ΔpcfE (Δ25–243 bp) allele in pCJK47	This study
pCY11	ΔpcfF (Δ31–327 bp) allele with P-pheS* cassette in pORI280	This study
pCY13	ΔpcfD (Δ37–2127 bp) allele with P-pheS* cassette in pORI280	This study
pCY15	ΔpcfE (Δ25–243 bp) allele with P-pheS* cassette in pORI280	This study
pCF10ΔpcfF	pCF10 with an in-frame deletion in pcfF	This study
pCF10ΔpcfD	pCF10 with an in-frame deletion in pcfD	This study
pCF10ΔpcfE	pCF10 with an in-frame deletion in pcfE	This study
pCY21	Spectinomycin gene cloned into BgIII/XmnI sites of pORI280	This study
pCY22	ΔltrF (Δ37–327 bp) allele in pCY21	This study
pM2036ΔltrF	pM2036 with an in-frame deletion in ltrF	This study
pDL278p23	L. lactis promoter p23 cloned into pDL278	This study
pCY16	pcfF cloned into BamHI/SphI sites of pDL278p23	This study
pCY17	pcfE cloned into BamHI/SphI sites of pDL278p23	This study
pCY18	pcfD cloned into BamHI/PstI sites of pMSP3535	This study
pCY23	ltrF cloned into BamHI/SphI sites of pDL278p23	This study
pCY19	pcfG cloned into Spe/SphI sites of pCJK21	This study
pCY20	ltrB cloned into Spe/SphI sites of pCJK21	This study
pORIT_16	pCF10 oriT in pDL278	Staddon et al. (2006)
pDL278-pRS01oriT	The 442 bp intergenic region between ltrD and ltrE cloned into pDL278	This study

Oligonucleotides	Sequence (5′–3′)	Application
pCF10-S	AAATTCGCAACATGCTAGCATGTTGCTCCGCTTGCAAAAAGAAAGCCG	ssDNA EMSA assay
pCF10-AS	GATCCGGCTTTCTTTTTGCAAGCGGAGCAACATGCTAGCATGTTGCG	ssDNA EMSA assay
pRS01-S	TCCATTTCTTAAATTCCGTAAGATGCTATCATCTTACTATGCTTGCAAAAGGT  CAAGGAA	ssDNA EMSA assay
pRS01-AS	TTCCTTGACCTTTTGCAAGCATAGTAAGATGATAGCATCTTACGGAATTTAA  GAAATGGA	ssDNA EMSA assay
pCF10-S1	GGGTCATTCAAAATATCGCAACATGCTAGCATGTTGCTCCGCTTGCAAAAA  GAAAGCCTA	nic site determination
pCF10-AS1	TAGGCTTTCTTTTTGCAAGCGGAGCAACATGCTAGCATGTTGCGATATTTTG  AATGACCC	nic site determination
pCF10-L	GCAACATGCTAGCATGTTGCTCCGCTTG	nic site determination
pRS01-L	GTAAGATGCTATCATCTTACTATGCTTG	nic site determination
pCF10-IR	GCAACATGCTAGCATGTTGC	dsDNA EMSA assay
pRS01-IR5	GTAAGATGCTATCATCTTAC	dsDNA EMSA assay
pRS01-IR3	GTAAGATGATAGCATCTTAC	dsDNA EMSA assay
Tn5252-IR5	GATATTGTGGACACAATATC	dsDNA EMSA assay
Tn5252-IR3	GATATTGTGTCCACAATATC	dsDNA EMSA assay
pC221-S	CGAAGTGGCTAGAATATACGACGCTTGGCAA	dsDNA EMSA assay
pC221-AS	TTGCCAAGCGTCGTATATTCTAGCCACTTCG	dsDNA EMSA assay
R2-XhoI	GCTCGAGGTAAATGGATAAGGAAAGAAA	dsDNA EMSA assay
B-1270	GGGATCCGGATTATGTTAATATTGTCT	dsDNA EMSA assay
B-1200	GGGATCCCCCCTTTGGGAGGAAGAGCC	dsDNA EMSA assay
pCF10oriTF3	GGTAAGTCGAAACGTCAAT	dsDNA EMSA assay
pCF10oriTR1	CTCCTTAGTTTCGACAATTG	dsDNA EMSA assay

^aRif^R, rifampicin resistant; Fus^R, fusidic acid resistant; Erm^R, erythromycin resistant; Str^R, streptomycin resistant; Spc^R, spectinomycin resistant; Tet^R, tetracycline resistant; Neo^R, neomycin resistant.

We first tested the binding capacities of PcfG and LtrB to their cognate oriT ssDNA. PcfG specifically bound to the pCF10-S probe, and the protein—DNA complex was abolished in the presence of a 10-fold cold pCF10-S oligonucleotide (Fig. 2A, lanes 3 and 4). No binding was observed with the pCF10-AS probe even at higher concentrations of PcfG (Fig. 2A, lane 5). Similarly, LtrB also only specifically bound to the pRS01-S probe (Fig. 2B). At higher protein concentrations, two protein—DNA complexes were observed with both PcfG and LtrB (data not shown).

Fig. 2.

Binding specificity of PcfG and LtrB to single-stranded oriT DNA sequences. PcfG or LtrB was incubated with 3′-end DIG-labelled oligonucleotides (16 fmol) at room temperature for 15 min. The reactions were analysed on 8% native polyacrylamide gels. The components in each lane are indicated at the top of the gel. Sequences of the oligonucleotides are listed in Table 1. PcfG and LtrB concentrations: 150 nM when using pCF10-S and pRS01-S probes; 750 nM when using pCF10-AS and pRS01-AS probes. pCF10 oriT ssDNA: S*, 3′-end DIG-labelled pCF10-S; AS*, 3′-end DIG-labelled pCF10-AS; S, 10-fold molar excess of cold pCF10-S. pRS01 oriT ssDNA: S*, 3′-end DIG-labelled pRS01-S; AS*, 3′-end DIG-labelled pRS01-AS; S, 10-fold molar excess of cold pRS01-S.

A. PcfG bound to pCF10 and pRS01 sense-strand oriT ssDNA.

B. LtrB bound to only pRS01 sense-strand oriT ssDNA.

We then tested whether PcfG and LtrB could recognize non-cognate oriT ssDNA. LtrB did not bind to either pCF10-S, or pCF10-AS oligonucleotides (Fig. 2B, lanes 8 and 9). However, PcfG could bind to the pRS01-S DNA, whereas no interaction between PcfG and pRS01-AS was detected (Fig. 2A, lanes 9 and 11). The PcfG-pRS01-S complex could be totally abolished by a 10-fold molar excess of pCF10-S DNA (Fig. 2A, lane 12). On the contrary, the PcfG—pCF10-S complex was only partially competed out by a 10-fold molar excess of pRS01-S DNA (Fig. 2A, lane 6), indicating that PcfG has a higher affinity for its cognate oriT DNA.

Determination of pCF10 nic site

Although the functional pCF10 oriT has been mapped, the nic site (relaxase cleavage site) remained to be determined. In addition to its high similarity with pRS01 oriT, there are sequence and secondary structural identities between pCF10 oriT and those of staphylococcal plasmids pC221 and pC223 (Fig. 1B) (Staddon et al., 2006). A stretch of six nucleotides, located to the right of an inverted repeat, is highly conserved among these oriT (Staddon et al., 2006). The experimentally mapped pC221 and pC223 nic sites are within this conserved sequence (GCTTG/C; ‘/’ represents the cleavage site) (Smith and Thomas, 2004). Furthermore, PcfG, LtrB and MobA proteins of pC221 and pC223 are all classified as IncP-type family relaxases (Caryl et al., 2004; Staddon et al., 2006). Given the fact that members of the same relaxase family often target the same nic sites, we proposed that PcfG and LtrB share the same cleavage sites as that of MobA. To test this, we performed ssDNA cleavage and strand-transfer assays using purified proteins. These assays are based on the ability of relaxase to cleave and recombine two nic sites on oligonucleotides of different lengths to create hybrid oligonucleotides that can be resolved by gel electrophoresis (Lanka and Wilkins, 1995).

Figure 3A shows a schematic representation of the assays. pCF10-S1 (60-mer ssDNA) contains the minimal pCF10 oriT and flanking sequence (15 bp upstream and 5 bp downstream). pCF10-L contains the 28 bp sequence upstream of the putative nic 3′-terminus. The 60-mer was DIG-labelled at the 3′-end and incubated with purified PcfG. After a 30 min incubation, a 10-fold molar excess of the 28-mer was added and incubated for another 30 min. The reaction mixture was treated with proteinase K and separated on 16% denaturing polyacrylamide gels. As seen in Fig. 3B, PcfG cleaved the labelled pCF10-S1, resulting a 15-mer that was covalently linked to a peptide resulted from proteinase K digestion (Fig. 3B, lane 3). Adding an excess amount of unlabelled pCF10-L to the cleavage reaction produced a new 43-mer recombinant product (Fig. 3B, lane 5), indicating that PcfG is capable of transferring the 15-mer to a preformed nic 3′-end terminus. No cleavage or recombinant products were generated with the pCF10-AS1 ssDNA (Fig. 3B, lanes 4 and 6). These results, in combination with the ssDNA binding data described above, demonstrated that PcfG is the relaxase of pCF10, and the nic site is located to 5′-(32843)GCTTG/CAAA(32853) on the sense strand. We were unable to determine the pRS01 nic site utilizing these assays, possibly because the recombinant LtrB protein was not fully active. However, mutating the putative nic GC nucleotides to TA, TT, GG or TC, all abolished pRS01 oriT function. Plasmids containing the mutated oriT were all mobilized at frequencies similar to that of the vector control (10^-6 transconjugants/donor), while a wild-type oriT-containing plasmid was mobilized at a frequency of 10^-1 transconjugants/donor, suggesting that these two nucleotides are important for pRS01 transfer.

Fig. 3.

Site-specific ssDNA cleavage and stand-transfer catalysed by PcfG.

A. Schematic representation of the cleavage and strand-transfer assays. The DIG molecules are indicated with stars.

B. PcfG (755 nM) was incubated with 3′-end DIG-labelled ssDNA and the reactions were carried out as described in the Experimental procedures. Lanes 1–6: pCF10 ssDNA. S1*, 3′-end DIG-labelled pCF10-S1 (60-mer); AS1*, 3′-end DIG-labelled pCF10-AS1 (60-mer); L, pCF10-L (28 mer). Lanes 7 and 8: pRS01 ssDNA. S*, 3′-end DIG-labelled pRS01-S (60-mer); L, pRS01-L (28-mer).

PcfF and LtrF specifically bind to pCF10 and pRS01 double-stranded oriT DNA sequences

In-frame deletions in pcfF or ltrF totally abolished transfer, suggesting their products are essential components of the transfer machinery. Both pcfF and ltrF encode small basic proteins with isoelectric point of 8.9 (PcfF) and 9.2 (LtrF), implying they might bind to DNA. Therefore, these genes were cloned into pET-28b(+) and expressed as N-terminal His₆-tagged proteins. Soluble recombinant proteins were purified from E. coli and their DNA binding abilities were tested in EMSA assays. LtrF and PcfF were capable of binding to double-stranded DNA (dsDNA) fragments containing the intergenic regions between ltrD-ltrE and pcfE-pcfF respectively (data not shown). These intergenic regions contain several inverted repeats, which are likely to be protein binding sites. Therefore, additional probes containing segments of the intergenic regions were designed to determine the minimal binding sequence. Figure 4A shows the schematic diagrams of the probes designed. The pCF10-20 and pRS01-20 probes contain only the inverted repeats within the oriT of pCF10 and pRS01. Specific protein—DNA interactions were detected between LtrF and probes pRS01-180, -110, -60 and -20 (Fig. 4B, and data not shown), and between PcfF and probes pCF10-186, -43, and -20 (Fig. 4C, and data not shown). Even at the highest protein concentrations tested, only one protein—DNA complex was detected (Fig. 4B, lane 4; Fig. 4C, lane 4).

Fig. 4.

Interactions of PcfF and LtrF with double-stranded oriT DNA sequences. dsDNA probes pRS01-20 (8 fmol) and pCF10-20 (8 fmol) were incubated with purified LtrF or PcfF for 15 min at room temperature. The reactions were analysed on 8% native polyacrylamide gels. Lanes 1 and 11, DNA probes only. Protein concentration in each lane is as follows. For LtrF (B): lane 2, 170 nM; lane 3, 340 nM; lane 4, 680 nM; lanes 5–10, and 12–15, 680 nM. For PcfF (C): lane 2, 78 nM; lane 3, 156 nM; lane 4, 312 nM; lanes 5–10, and 12–15, 312 nM. In the competition assays (lanes 5–10, and 12–15), cold specific or non-specific dsDNA was added to the preformed F protein and oriT complexes. Unlabelled cold competitor in each lane is indicated at the top of the gel. The concentrations of cold pCF10-20 and pRS01-20 competitors (molar excess) are as following: lanes 5 and 8, 10-fold; lanes 6 and 9, 100-fold; lanes 7 and 10, 1000-fold. The concentration of cold Tn5252 oriT (putative) competitor (molar excess) is as following: lane 13, 100-fold; lane 14, 1000-fold; lane 15, 10 000-fold.

A. Schematic representations of the pcfE-pcfF and ltrD-ltrE intergenic regions. Inverted repeats are indicated as pairs of arrows. The linear segments below the map represent probes used in the EMSA assays.

B. LtrF specifically bound to pRS01 and pCF10 oriT dsDNA.

C. PcfF specifically bound to pCF10 and pRS01 oriT dsDNA.

LtrF and PcfF were also assayed against their noncognate oriT sequences. LtrF specifically bound to the pCF10-20 probe (Fig. 4B, bottom figure). Similarly, PcfF could recognize the pRS01-20 probe (Fig. 4C, bottom figure). Competitive EMSA assays suggest that LtrF had a higher affinity for its cognate pRS01 oriT sequence (Fig. 4B, lanes 5–7 compared with lanes 8–10), while PcfF bound both oriT sequences with similar affinities (Fig. 4C, lanes 5–7 compared with lanes 8–10). The specificity of these interactions was further examined by testing the interactions between LtrF/PcfF and two nonspecific dsDNA molecules: a 20 bp sequence containing the inverted repeat from the putative Tn5252 oriT and a 31 bp sequence surrounding the pC221 nic site (Fig. 1B). No binding was observed even using a protein/DNA ratio of 7500 (data not shown). In addition, a 10 000-fold molar excess of cold Tn5252 oriT DNA competitor did not disrupt LtrF—pRS01-20 (Fig. 4B, lane 15) and PcfF—pCF10-20 (Fig. 4C, lane 15) complexes. In summary, LtrF and PcfF specifically recognize the inverted repeats within the oriT of pRS01 and pCF10. The two proteins are unable to distinguish between their cognate, and the highly related non-cognate sequences.

Interactions between relaxases and the PcfF/LtrF—oriT DNA complexes

While PcfF and LtrF specifically bind to oriT sequences and might be involved in DNA processing, their precise function is unknown. It was interesting that PcfF and LtrF bound to the inverted repeats adjacent to the nic sites (Fig. 4) and had no observable ssDNA binding activities (data not shown). In contrast, PcfG and LtrB interacted with ssDNA (Fig. 2). These observations led us to develop a super-shift EMSA assay to detect stable relaxase—PcfF/LtrF—oriT complexes and possible interactions between these proteins at the oriT regions of dsDNA probes.

First, we tested if PcfG could super-shift PcfF—oriT_pCF10, and PcfF—oriT_pRS01 complexes. PcfG was able to completely super-shift PcfF—oriT_pCF10 into two higher molecular mass complexes (Fig. 5A, lanes 1 and 2, complex 1 and 2). At the same protein concentration, only one weak super-complex was observed between PcfG and PcfF— oriT_pRS01 (Fig. 5A, lane 7). No interactions between PcfG and dsDNA were observed (Fig. 5A, lane 3; Fig. 5B, lanes 3 and 9). Interestingly, PcfG could also super-shift LtrF— oriT_pCF10 into two weak super-complexes (Fig. 5B, lane 8), yet was unable to interact with LtrF—oriT_pRS01 complex, even at a twofold higher protein concentration (Fig. 5B, lanes 1 and 2).

Fig. 5.

Interactions between PcfG and PcfF/LtrF—oriT DNA complexes. dsDNA probes pRS01-180 (16 fmol) and pCF10-186 (16 fmol) were incubated with purified PcfG (150 nM) and LtrF (314 nM) or PcfF (158 nM) for 15 min at room temperature. The reactions were analysed on 8% native polyacrylamide gels. Protein components in each lane are shown on the top of the gel. PcfG concentration in lane 1 was 300 nM.

A. Interactions between PcfG and PcfF—oriT DNA.

B. Interactions between PcfG and LtrF—oriT DNA.

Similarly, LtrB was able to super-shift LtrF—oriT_pRS01 completely into one super-complex (Fig. 6A, lanes 4 and 5). From the size, it seems that this complex may correspond to the super-complex 2 of PcfG—PcfF—oriT_pCF10. LtrB could only weakly interact with LtrF—oriT_pCF10 (Fig. 6B, lanes 2 and 3), and was unable to super-shift PcfF—oriT_pCF10 or PcfF—oriT_pRS01 (Fig. 6A, lane 8; Fig. 6B, lane 5); these results indicate that any ssDNA region created by binding of either of the F proteins to the double-stranded pRS01 probe is not sufficient to recruit LtrB, suggesting an essential role of protein/protein interactions in formation of the tripartite complex. No interactions between LtrB and dsDNA were observed (Fig. 6A, lane 3). The data also suggest that both PcfG and LtrB are capable of interacting with their cognate (Pcf/Ltr)F—DNA complexes, and that stable relaxase—(Ltr/Pcf)F protein complex formation is enhanced or stabilized by the relaxase cognate oriT DNA.

Fig. 6.

Interactions between LtrB and PcfF/LtrF—oriT DNA complexes.

Double-stranded probe pRS01-180 (16 fmol) and pCF10-186 (16 fmol) were incubated with purified LtrB (152 nM) and LtrF (314 nM) or PcfF (158 nM) for 15 min at room temperature. The reactions were analysed on 8% native polyacrylamide gels. Protein components in each lane are shown on the top of the gel. LtrB concentrations in A lane 4, and B lane 3 were 304 nM.

A. pRS01-180 DNA probe.

B. pCF10-186 DNA probe.

In vivo cross-complementation and mobilization of oriT-containing plasmids

The above biochemical data suggest that PcfG/LtrB, and PcfF/LtrF exhibit difference in oriT DNA recognition and protein—protein interactions, which possibly contribute to plasmid specificity. We therefore performed in vivo experiments to further investigate the specificity determinants involved in pCF10 and pRS01 transfer. First, we tested the ability of pCF10 and pRS01 homologues to rescue deletion mutations in the essential conjugation genes: pcfF, pcfG, ltrF and ltrB. When expressed in trans, LtrB could not complement the pcfG insertional mutant of pCF10, nor could PcfG complement pM1014 [a derivative of pRS01 that contains a splicing deficient intron in ltrB (Mills et al., 1994; 1996)] (Table 2). LtrF was able to partially complement the pCF10ΔpcfF mutant when expressed in trans. In contrast, PcfF could not complement the pM2036ΔltrF mutant (Table 2).

We also tested whether pCF10 and pRS01 could mobilize non-cognate oriT-containing plasmids. The oriT regions of pRS01 and pCF10 were cloned into plasmid pDL278 to generate plasmids pDL278-pRS01oriT and pORIT_16 respectively. The mobilization frequencies of these plasmids by pCF10 or by pM2036 were examined in the conjugation assays. In L. lactis matings, pDL278-pRS01oriT was efficiently mobilized by pM2036, in contrast to poor mobilization of pORIT_16 and pDL278 (Table 3). In E. faecalis matings, pCF10 could efficiently mobilize pORIT_16. In addition, pDL278-pRS01oriT could also be mobilized by pCF10, at a frequency of 100-fold higher than that of the vector control pDL278 (Table 3).

Table 3.

Mobilization of pCF10 and pRS01 oriT-containing plasmids.

Plasmids in donor	Mobilization frequency (T/D)
L. lactisa
pM2036/pDL278	3.2 × 10-5
pM2036/pDL278-pRS01oriT	3.4 × 10-2
pM2036/pORIT_16c	2.4 × 10-5
E. faecalisb
pCF10/pDL278	5.4 × 10-6
pCF10/pDL278-pRS01oriT	7.7 × 10-4
pCF10/pORIT_16c	4.3 × 10-1

^aL. lactis MMS370 harbouring pCF10 or pRS01 oriT-containing plasmids, or the vector control pDL278 was used as the donor. L. lactis LM2301RF was used as the recipient. Plasmids mobilization frequencies are expressed as transconjugant per donor (T/D) ratios, normalized to the pM2036 transfer frequency in the same experiment.

^bE. faecalis OG1RF harbouring pCF10 or pRS01 oriT-containing plasmids, or the vector control pDL278 was used as the donor. E. faecalis OG1ES was used as the recipient. Plasmids mobilization frequencies are expressed as transconjugant per donor (T/D) ratios, normalized to the pCF10 transfer frequency in the same experiment.

^cpORIT_16 contains the minimal pCF10 oriT cloned into pDL278 (Staddon et al., 2006).

The data represent the average of at least two independent experiments.

Discussion

While most of E. faecalis pheromone-responsive plasmid pCF10 is unrelated to the L. lactis conjugative plasmid pRS01, the oriT regions and the putative DNA-processing proteins are highly similar (Fig. 1). Although the functional transfer origins of both plasmids were mapped previously, very little was known about their DNA-processing mechanisms. We therefore determined the dispensability of each gene in the processing region for plasmid transfer and characterized protein—DNA interactions involved in this process. We precisely determined the nic site of pCF10 and we investigated the specificity determinants of DNA processing in these two plasmids.

We mapped the nic site of pCF10 to 5′-(32843)GCTTG/ CAAA(32853) on the sense strand, which is located 8 bp downstream of the inverted repeat within the minimal oriT region (Fig. 1B). This cleavage site is inside a six-nucleotide sequence that is completely or largely conserved among the oriT regions of several plasmids, including the staphylococcal plasmids pC221 and pC223 (Staddon et al., 2006), whose nic sites are the same as that of pCF10. Further sequence comparison revealed that this six-nucleotide sequence is homologous to the IncP-type oriT, one of the five oriT core families defined previously (Zechner et al., 2000). It is likely that pRS01 utilizes the same nic site, as suggested both by the homology between the target sites and by the results of mutagenesis of this site. Correspondingly, PcfG and LtrB are grouped into the IncP relaxase family (Staddon et al., 2006). In contrast, the nic sites of the enterococcal pheromone plasmids pAD1 (experimentally mapped) (Francia and Clewell, 2002) and pAM373 are not similar to any of the five oriT groups and represent a new family (Francia and Clewell, 2002).

Genetic studies have shown that PcfG and LtrB are essential for pCF10 and pRS01 transfer (Mills et al., 1996; Staddon et al., 2006). These two relaxases are 50% identical, yet they exhibit interesting differences in substrate binding. Both proteins recognize the sense-strand oriT ssDNA (the strand that contains the nic site), but not the complementary strand. LtrB is highly specific for its cognate pRS01 oriT. In contrast, PcfG is capable of binding to both pCF10 and pRS01 oriT sequences. The inverted repeats upstream of the nic sites are necessary to form stable relaxase—DNA complexes in the EMSA assays, because deletion of either arm of the repeat abolished complex formation (data not shown). Between the right arm and the nic site, there are only two nucleotide differences between pRS01 and pCF10 oriT (Fig. 1B), suggesting that these two bases may make specific contact with the relaxases. Nucleotides in the inverted repeats also can affect binding and specificity, as it was recently demonstrated that base substitutions in the inverted repeat reduced the binding affinity of the N-terminus domain of F TraI relaxase to its oriT ssDNA (Williams and Schildbach, 2006). Although relaxases are commonly considered to be highly specific for their cognate oriT DNA sequences, our data suggest that some relaxases may be less stringent in substrate recognition. These results are similar to the observations with the N-terminal relaxase domains (TraI36) of F factor and the highly homologous plasmid R100. These two proteins are 91% identical and the oriT of the two plasmids differ only 2 bp near the nic sites (Frost et al., 1994; Harley and Schildbach, 2003). Even though both proteins are highly specific for their cognate binding sites, F TraI36 cleaves the R100 oriT sequence poorly, while R100 TraI36 can cleave the F sequence relatively well (Stern and Schildbach, 2001; Harley and Schildbach, 2003). In vivo, R100 TraI also can utilize both oriT, while F TraI is only active against its own oriT (Fekete and Frost, 2000; Stern and Schildbach, 2001). Here we did not observe detectable cleavage and ligation products by PcfG on the pRS01 oriT DNA in vitro. However, pRS01 oriT-containing plasmids could be mobilized by pCF10 in E. faecalis matings, indicating that cleavage of the pRS01 oriT sequence does occur in vivo.

Both pcfF and ltrF are small genes, whose stop codons are adjacent (pcfF) or partially overlapping (ltrF) with the relaxase genes. By in-frame deletion mutation analysis, we found that pcfF and ltrF are absolutely required for plasmid transfer. In vitro, purified PcfF and LtrF are capable of binding to dsDNA, yet no conserved DNA binding domains are evident. Pfam database search (http://www.genome.wustl.edu) classified these two proteins into the bacterial mobilization protein (MobC) family (accession number:PF05713). Members of this family contain a (iGnNiNQiA) motif in the protein carboxyl terminus. MobC proteins are small in size (less than 200 amino acids), and are found in a large number of plasmids from both Gram-negative and Gram-positive bacteria (the database already contains 118 members). Many IncP/ColE1 plasmids encode MobC proteins between oriT and the cognate relaxase gene. Sequence alignments of several selected MobC family proteins are shown in Fig. S1. Two other MobC family members that have been characterized are the MobC proteins encoded by pC221 and pC223. Both are essential for plasmid mobilization and relaxosome formation, and possess oriT DNA binding abilities (Caryl et al., 2004; Smith and Thomas, 2004; Caryl and Thomas, 2006). In the Pfam database, MobC family proteins are proposed to function as ‘molecular wedges’ for the relaxosome-induced melting of oriT DNA. However, the precise roles of PcfF and LtrF in conjugation remain to be determined.

Surprisingly, we found that PcfF and LtrF can efficiently bind to the 20 bp inverted repeats within the oriT of both pCF10 and pRS01. The interactions are sequence specific: both proteins are unable to recognize the inverted repeat of the putative Tn5252 oriT, or the right arm of the inverted repeat of the pC221 oriT. The inverted repeats of pCF10 and pRS01 are 75% identical, differing by 5 out of 20 bases (Fig. 1B). A 6 bp segment (ATGCTA) present in both inverted repeats is likely to be the protein binding site. As PcfF and LtrF only share 47% protein identity, amino acids involved in DNA recognition might be highly conserved in these two proteins. Interestingly, the two MobC proteins (73% identical) of pC221 and pC223 also are unable to distinguish between cognate and noncognate oriT DNA (Caryl et al., 2004). In contrast, the well-characterized RP4 accessory protein TraJ, which lacks the conserved (iGnNiNQiA) MobC motif, only interacts with its cognate oriT, and not with that of the closely related R751 (Furste et al., 1989).

LtrB and PcfG are able to interact with their cognate F—oriT DNA to form stable complexes in EMSA. As no direct interactions between relaxases and dsDNA templates were observed (Figs 5 and 6), PcfF/LtrF is the component that initiates formation of the nucleoprotein complex in vivo at oriT. The relaxase protein is then recruited through direct protein—protein interactions or through binding to its target ssDNA present in the complex (see Fig. S2 for an illustration of these models). These two scenarios are not exclusive, given that there is a possible overlap between the F protein and relaxase binding sites on the oriT DNA; in either case, our EMSA data suggest that the stability of the tripartite complex is enhanced by both protein/protein and protein/DNA interactions. The proposed LtrF/PcfF-mediated alteration of the DNA structure to create a ssDNA target for relaxase binding (Fig. S2A) is similar to the mechanism of action suggested for the MobC protein of R1162 (Zhang and Meyer, 1997). LtrB was not able to super-shift PcfF—oriT_pRS01 complex (Fig. 6A), indicating a local ssDNA region (assuming that PcfF does induce such a structure) is not sufficient to recruit LtrB. In addition, LtrB could super-shift LtrF— oriT_pCF10 (Fig. 6B), even though LtrB did not interact with pCF10 oriT ssDNA (Fig. 2B). These data strongly support direct protein—protein interactions between LtrB and LtrF. However, we have not been able to demonstrate these direct interactions in the absence of DNA. It seems that interactions between LtrB and LtrF are relatively weak, or transient, compared with those between LtrB and LtrF— oriT complexes based on the super-shifted mobility profiles shown in Fig. 6. Thus, protein—protein interactions between LtrB and LtrF are enhanced or stabilized by relaxase—oriT interactions. As PcfG interacts with pRS01 oriT ssDNA, the super-shift data alone are not a compelling demonstration of direct protein—protein interactions between PcfG and PcfF.

Plasmids containing pCF10 oriT were poorly mobilized by the pRS01 transfer system in L. lactis matings. However, pCF10 could mobilize pRS01—oriT containing(****) plasmids at a frequency about 100-fold greater than that of vector controls in E. faecalis matings. The mating pores of pRS01 and pCF10 are not species-specific, because plasmids containing pCF10 oriT could be efficiently mobilized from E. faecalis into L. lactis by pCF10 (Staddon et al., 2006). Thus, the specificity of the two plasmids probably resides in the DNA-processing reactions and subsequent steps, for instance, relaxosome— coupling protein interactions. Three levels of specificity can occur at the DNA-processing step: ssDNA recognition by relaxase, accessory protein binding to oriT, and protein—protein interactions within the relaxosome. LtrB determines specificity in the pRS01 system by specifically targeting only pRS01 ssDNA. In contrast, PcfG does not share the same level of specificity as LtrB and can act on both pCF10 and pRS01 oriT, although less efficiently on the latter. However, pcfG could not complement an ltrB insertion mutant, suggesting that there are additional specificity determinants. LtrF and PcfF are not able to distinguish between pRS01 and pCF10 oriT sequences. Therefore, LtrF/PcfF oriT DNA binding is not involved in specificity. However, pcfF could not complement the ltrF mutant, and ltrF could only partially complement the pcfF mutant, indicating that these two proteins participate in specificity by interacting with other conjugation components. One such interaction is between the LtrF and PcfF proteins and relaxases. LtrB can only interact with LtrF, but not with PcfF. In contrast, PcfG can be recruited into both PcfF—oriT_pCF10 and LtrF—oriTpCF10 complexes. In addition to interacting with PcfG— oriT DNA complex, LtrF may be able to interact with other components of the pCF10 transfer machinery. Neither pcfD nor pcfE is absolutely required for conjugation, but mutations in them do reduce pCF10 transfer. Their products might be components of the relaxosome, or be involved in conjugation regulation, or function at other transfer stages. The requirements for ltrD and ltrC in pRS01 conjugation need to be tested. PcfD/LtrC and PcfE/LtrD may add additional levels of specificity by oriT DNA binding, or by protein—protein interactions. The pRS01-encoded proteins LtrM and LtrE, which lack pCF10 homologues, may also play a role in this process. In conclusion, while relaxases determine specificity by oriT DNA binding, protein—protein interactions among relaxosome components also contribute to plasmid transfer specificity.

Experimental procedures

Bacterial strains, growth conditions, plasmids and oligonucleotides

Bacterial strains, plasmids and oligonucleotides are listed in Table 1 and Table S1. E. coli DH5α (Gibco-BRL), and EC1000, a strain that expresses the pWV01 RepA protein (Leenhouts et al., 1996), were used in clonings. E. coli BL21 (DE3) (Novagen) was used for protein expression. E. coli strains were cultured in Luria broth (LB) (Difco Laboratories) or Brain Heart Infusion broth (BHI) (Difco Laboratories) and grown at 37°C with shaking. L. lactis strains were cultured in M17 (Difco Laboratories) supplemented with 0.5% glucose (GM17, pH 7.0) and grown at 30°C without shaking. E. faecalis strains were cultured in Todd—Hewitt broth (THB) (Difco Laboratories) or BHI and grown at 37°C without shaking. Antibiotics were added at the following final concentrations. For E. faecalis: erythromycin, 10 μg ml^-1 for plasmids and 100 μg ml^-1 for chromosomal markers; fusidic acid, 25 μg ml^-1; rifampicin, 200 μg ml^-1; spectinomycin, 1000 μg ml^-1 for plasmids and 250 μg ml^-1 for chromosomal markers; streptomycin, 1000 μg ml^-1; tetracycline, 10 μg ml^-1. For L. lactis: rifampicin, 100 μg ml^-1; fusidic acid, 25 μg ml^-1; spectinomycin, 300 μg ml^-1; streptomycin, 600 μg ml^-1; tetracycline, 5 μg ml^-1; neomycin, 1000 μg ml^-1. For E. coli: erythromycin, 100 μg ml^-1 in BHI; spectinomycin, 50 μg ml^-1; kanamycin, 30 μg ml^-1. All antibiotics were obtained from Sigma Chemical.

Plasmid constructions

The plasmid pCY11 was used to generate an in-frame deletion mutation in pcfF and was constructed as follows. A 1160 bp PCR fragment embracing DNA upstream and the first 30 bp of pcfF (upstream fragment) was amplified with primer pair 5F-1/5F-2. A 1120 bp fragment embracing DNA downstream and the last 27 bp of pcfF was amplified with primer pair 3F-1/3F-2 (downstream fragment). The two fragments were sequentially cloned into pCJK47 to create pCY10, using primer-added XbaI/PstI (upstream fragment) and PstI/BglII (downstream fragment) restriction sites, in a way such that a PstI site was fused in-frame with the upstream and downstream sequences to form the ΔpcfF allele. To generate pCY11, the ΔpcfF allele and downstream P-pheS* cassette were digested out from pCY10 and cloned into the XbaI/BglII restriction sites of pORI280. The plasmid pCY13 was used to generate an in-frame deletion mutation in pcfD and was constructed as follows. Upstream flanking sequence and the first 39 bp of pcfD was amplified with primer pair D5F/D5R (upstream fragment, 936 bp). Downstream flanking sequence and the last 30 bp of pcfD was amplified with primer pair D3F and D3R (downstream fragment, 906 bp). The two PCR products were purified using QIAquick PCR Purification Kit (Qiagen). One-tenth of each purified product were mixed and used as templates in a second step PCR using primer pair D5F/D3R. The resulting PCR fragment (ΔpcfD allele) was cloned into pCJK47 to create pCY12, using primer-included XbaI/XmaI restriction sites. To construct pCY13, the ΔpcfD allele and downstream P-pheS* cassette were cut out from pCY12 and cloned into pORI280 using XbaI/BglII restriction sites. The pCY15 plasmid was used to generate an in-frame deletion mutation in pcfE and was constructed similar to pCY13. The upstream fragment (896 bp) and the downstream fragment (934 bp) were amplified with primer pairs E5F/E5R and E3F/E3R respectively. The second PCR step was performed using primers E5F and E3R. The resulting PCR fragment was cloned into pCJK47 to create pCY14, using primer-included XbaI/XmaI restriction sites. pCY14 contains the two flanking sequences, the first 24 and the last 27 bases of pcfE (ΔpcfE allele). To construct pCY15, the ΔpcfE allele and downstream P-pheS* cassette were cut out from pCY14 and cloned into pORI280 using XbaI/BglII restriction sites.

pCY21 was constructed by replacing the erythromycin-resistance marker in pORI280 with spectinomycin marker. The spectinomycin gene was amplified from pDL278 using primer pair Spec-Bg/Spec-N. The PCR product was digested with BglII that was included in the 5′-end primer and cloned into pORI280 digested with BglII/XmnI. The plasmid used to create an in-frame deletion in ltrF (pCY22) was constructed as follows. The upstream fragment (1120 bp) and the downstream fragment (1267 bp) were amplified from pM2036 DNA with primer pairs pORF1/pORF2 and pORF3/PORF4 respectively. The two fragments were sequentially cloned into pCY21, using primer-added BamHI/NcoI (upstream fragment) and NcoI/BglII (downstream fragment) restriction sites. The resulting plasmid pCY22 contains the two flanking sequences, the first 30 and the last 27 bases of ltrF fused in-frame with a NcoI site (ΔltrF allele).

The constitutive promoter p23 was amplified from L. lactis strain MG1363 chromosome DNA with primer pair p23F/p23R. The 0.2 kb p23 promoter was digested with EcoRI/BamHI and cloned into pDL278 to create pDL278p23. Plasmid pDL278-pRS01oriT was made by cloning the 442 bp intergenic region between ltrD and ltrE (amplified using primer pair pRS01_oriTF1/pRS01_oriTR1) into the EcoRI/BamHI sites of pDL278.

The plasmids used in the complementation experiments were generated as follows. The ORFs of pcfF, pcfE and ltrF and their upstream ribosome binding sites were amplified using primer pairs PpcfF-B/PpcfF-S (product, 419 bp), PpcfE-B/PpcfE-S (product, 393 bp) and PltrF-B/PltrF-S (product, 508 bp) respectively. The PCR fragments were digested with BamHI/SphI, and cloned into pDL278p23 to generate plasmid pCY16 (used to complement the pCF10ΔpcfF mutant), pCY17 (used to complement the pCF10ΔpcfE mutant) and pCY23 (used to complement the pM2036ΔltrF mutant). The ORF of pcfD and upstream ribo-some binding site (2220 bp) was amplified using primer pair PpcfD-B/PpcfD-P and cloned into pMSP3535 BamHI/PstI sites to generate pCY18, which was used to complement the pCF10ΔpcfD mutant.

A PCR fragment containing pcfG or ltrB (including their upstream ribosome binding sites) was cloned into pCJK21 SpeI/SphI sites to create plasmid pCY19 (used to complement the pcfG insertion mutant pCF10::Ll.ltrBΔORF) and pCY20 (used to complement pM1014). The cloned fragments were amplified with primers PcfG-S/PcfG-sp. (1746 bp, for pCY19) and LtrB-S/LtrB-sp. (1768 bp, for pCY20) respectively.

The plasmids used for protein expression in E. coli were constructed as follows. ltrB and ltrF were amplified from plasmid pM2036 using primer pairs ltrB-N/ltrB-X and ltrF-N/ pBE1-B respectively. pcfG and pcfF were amplified from pCF10 with primer pairs pcfG-N/pcfG-X and pcfF-N/pcfF-B respectively. The resulting PCR products were digested with NcoI/XhoI (for ltrB and pcfG) and NdeI/BamHI (for ltrF and pcfF). The PCR primers used for amplification contained cleavage sites for these enzymes. Digested PCR products were then cloned into pET-28b(+) to generate pET-28b(+)-HltrB, pET-28b(+)-HltrF, pET-28b(+)-HpcfG and pET-28b(+)-HpcfF. In the final constructs, a C-terminal His₆-tag was fused in-frame to ltrB and pcfG, and an N-terminal His₆-tag was fused in-frame to ltrF and pcfF.

Protein purification

Purification of His-LtrB and His-PcfG: 5 ml of E. coli BL21 (DE3) carrying pET-28b(+)-HltrB or pET-28b(+)-HpcfG was grown in LB (with kanamycin 30 μg ml^-1) at 37°C till OD₆₀₀ reached 0.5. The culture was stored at 4°C overnight. Next day, 1 ml of the culture was used to inoculate 100 ml of fresh LB and grown at 37°C till OD₆₀₀ reached 0.6. The culture was then induced with IPTG at a final concentration of 0.1 mM for 3 h at 30°C. The cells were collected and the pellet was resuspended in 5 ml of CelLytic^TM B II (Sigma) and the inclusion bodies were isolated following the manufacture’s instructions. The inclusion bodies were washed with buffer 1 (0.2 M urea, 20 mM Tris HCl, pH 7.9, 5 mM imidazole, 0.5 M NaCl) twice and dissolved in buffer 2 (8 M urea, 20 mM Tris HCl, pH 7.9, 5 mM imidazole, 0.5 M NaCl). After filtering through a 0.4 μm filter, the dissolved inclusion bodies were loaded onto a His-column (1.5 ml of His.Bind Resin) (Novagen) equilibrated with buffer 2. The column was washed with 10-column volume of buffer 2, 10-column volume of wash buffer (8 M urea, 20 mM Tris HCl, pH 7.9, 20 mM imidazole, 0.5 M NaCl), and eluted with elution buffer [8 M urea, 20 mM Tris HCl, pH 7.9, imidazole (200 mM for LtrB and 100 mM for PcfG), 0.5 M NaCl]. The eluted fractions were analysed on a 7.5% SDS-PAGE gel. Fractions containing protein were pooled, diluted 10-fold in the elution buffer and refolded by using four steps of dialysis to decrease urea and imidazole concentrations. During dialysis, glycerol (final concentration 10%) was added to the sample and the dialysis buffer. Dialysis buffer 1, 4 M urea, 20 mM Tris HCl, pH 7.9, 200 mM imidazole, 0.3 M NaCl; buffer 2, 2 M urea, 20 mM Tris HCl, pH 7.9, 100 mM imidazole, 0.2 M NaCl; buffer 3, 0.5 M urea, 20 mM Tris HCl, pH 7.9, 10 mM imidazole, 0.2 M NaCl; buffer 4 (storage buffer), 20 mM Tris HCl, pH 7.9, 0.2 M NaCl, 0.1 mM EDTA, 10% glycerol (v/v). For refolding PcfG, DTT (final 0.5 mM) and glycine (final 0.5 M) were included in the dialysis buffers and the sample. Dialysis buffer was changed every 16 h at 4°C. After dialysis, His-LtrB and His-PcfG were concentrated by using iCON concentrators (Pierce) following the manufacturer’s instructions and stored at -20°C.

Purification of His-LtrF and His-PcfF: E. coli BL21 (DE3) carrying pET-28b(+)-HltrF or pET-28b(+)-HpcfF was grown and induced as described above. The cells were collected by centrifugation and the pellet was resuspended in 1× binding buffer (20 mM Tris HCl, pH 7.9, 5 mM imidazole, 0.5 M NaCl) containing 0.1 mg ml^-1 lysozyme and incubated at 37°C for 15 min. The cells were then disrupted by sonication. After centrifugation, the soluble fraction was filtered through a 0.4 μm filter and loaded onto a His-column (1.5 ml of His.Bind Resin) (Novagen) equilibrated with 1× binding buffer. The column was washed with 10-column volume of 1× binding buffer, 10-column volume of wash buffer [20 mM Tris HCl, pH 7.9, 150 mM (for His-LtrF) or 80 mM (for His-PcfF) imida-zole, 0.5 M NaCl], and eluted with elution buffer [20 mM Tris HCl, pH 7.9, 300 mM (for His-LtrF) or 150 mM (for His-PcfF) imidazole, 0.5 M NaCl]. The eluted fractions were analysed on a 15% SDS-PAGE gel. Fractions containing protein were collected, diluted fivefold with the elution buffer and dialysed against storage buffer (20 mM Tris HCl, pH 7.9, 0.2 M NaCl, 0.1 mM EDTA, 10% glycerol). After dialysis, proteins were concentrated by using iCON concentrators (Pierce) following the manufacturer’s instructions and stored at -20°C.

Recombinant protein concentrations were determined by BCA protein assay (Pierce) using bovine serum albumin as a standard.

Electrophoretic mobility shift assays

pRS01 and pCF10 oriT-containing oligonucleotides used in the ssDNA binding assays were synthesized and purified by high-performance liquid chromatography and polyacrylamide gel electrophoresis (Advanced Genetic Analysis Center and Microchemical Facility, University of Minnesota). The dsDNA molecules pRS01-180, -110 and pCF10-186 were PCR amplified with primer pairs R2-XhoI/B1270, R2-XhoI/B1200 and pCF10oriTF3/pCF10oriTR1 respectively. PCR products were purified using QIAquick PCR Purification Kit (Qiagen). pRS01-60, -20 and pCF10-43, -20 were generated by annealing the oligonucleotides pRS01-S/pRS01-AS, pRS01-IR5/pRS01-IR3, pCF10-S/pCF10-AS and pCF10-IR/ pCF10-IR respectively. The Tn5252 putative oriT and pC221 oriT dsDNA molecules were generated by annealing the oligonucleotides Tn5252-IR5/Tn5252-IR3 and pC221-S/pC221-AS. ssDNA and dsDNA molecules were labelled at the 3′-ends with DIG-11-ddUTP using the DIG Gel Shift kit (Roche) by following the manufacturer’s instructions. EMSA assays were conducted in 20 μl reactions containing the following components: DIG-labelled ssDNA or dsDNA, recombinant proteins, 1 μg of poly-[d(A-T)], 0.1 μg of poly-L-lysine, 1× reaction buffer (20 mM Tris HCl, pH 7.9, 0.1 M NaCl, 0.1 mM EDTA, 10% glycerol, 10 mM MgCl₂). The reactions were incubated at room temperature for 15 min and loaded onto to 8% polyacrylamide gels in 0.5× TBE buffer (Tris, Borate, EDTA, pH 7.9). After electrophoresis at 100 V for 1.5 h, the DNA—protein complexes and DNA probes were electrotransferred onto a nylon membrane (Roche) at 6 V for 2 h by using the GENIE electrophoretic blotter (Idea Scientific). DIG-labelled DNA fragments or oligonucleotides were visualized by an enzyme immunoassay (DIG Gel Shift Kit, Roche) following the manufacturer’s instructions. Super-shift EMSA assays were performed as the same except that PcfG or LtrB was added to samples containing LtrF/PcfF and dsDNA probes. In competitive EMSA assays, various molar excess (10-, 100-, 1000- and 10 000-fold) of cold competitors were added to the preformed probe—protein complexes.

Oligonucleotide cleavage and strand-transfer reactions

For cleavage reactions, 3′-end DIG-labelled oligonucleotides (6 fmol) were incubated with purified His-PcfG (755 nM) in the reaction buffer (20 mM Tris HCl, pH 7.9, 100 mM NaCl, 10 mM MgCl₂). After 1 h incubation at 30°C, the reactions were stopped by the addition of SDS (final 0.1%) and proteinase K (final 100 μg ml^-1) and incubated at 37°C for another 30 min. Samples were heated at 95°C for 5 min and were analysed on 16% (wt/vol) polyacrylamide gels containing 8 M urea. Electrophoresis, transfer and detection were carried out as described above. Oligonucleotide strand-transfer reactions were performed similar to the cleavage reactions, except that after a 30 min incubation at 30°C, a 10-fold (60 fmol) unlabelled pCF10-L or pRS01-L oligonucleotides were added to the samples and incubated for another 30 min. The reactions were stopped and analysed as above.

Conjugation and mobilization assays

Enterococcus faecalis donor and plasmid-free recipient cells were grown overnight in THB with appropriate antibiotics. Overnight cultures were diluted 1:10 in fresh THB and incubated at 37°C for 1 h to ensure that cultures were in the exponential phase during the matings. The cells were combined in THB with a 1:10 donor to recipient ratio and mated for 1 h at 37°C. Donor and transconjugants colony-forming units (cfu) on THB agar were quantified with relevant selective antibiotics. Transconjugant to donor ratios were calculated by dividing the number of cfu ml^-1 resistant to both a recipient chromosomal marker and a plasmid-mediated marker by the number of cfu resistant to donor antibiotic markers. When testing pCF10ΔpcfF and pCF10ΔpcfE transfer, E. faecalis CK104 was used as the host donor and OG1ES as the recipient. When testing pCF10ΔpcfD transfer, CK104 was used as the host donor and OG1SSp was used as the recipient.

GM17 agar plate matings were used to determine plasmid transfer or mobilization frequencies in L. lactis and were performed as described before (Klein et al., 2004). When testing pM2036ΔltrF transfer, L. lactis LM2301RF was used as the host donor and MMS372 was used as the recipient.

Generation of in-frame deletion mutations in pcfF, pcfD, pcfE and ltrF

Markerless genetic exchange in pCF10 was carried out as described before with several modifications (Kristich et al., 2005; 2007). Briefly, about 1 μg of plasmids pCY11, 13 and 15 DNA were introduced into E. faecalis CK104/pCF10/ pVE6007 by electrotransformation, followed by incubation at 30°C (1.5 h) and then at 37°C (3 h) to inactivate the RepA(ts) protein (supplied in trans from pVE6007) required for plasmid replication. The cells were plated on THB agar supplemented with Erm and Xgal (250 μg ml^-1) at 37°C. Blue transformants that represent integrants were restreaked on the same medium at least three times for single-colony purification. Several colonies were tested by a colony-PCR to verify that integration did not occur at the pheS locus in the E. faecalis chromosome (Kristich et al., 2007). A blue colony with the desired integration locus was then cultured in BHI without selection for approximately 20 generations at 37°C. Serial dilutions of the cell suspension were plated on MM9YEG (Kristich et al., 2007) agar supplemented with 10 mM p-Cl-Phe and Xgal (250 μg ml^-1) (Kristich et al., 2007) at 37°C. White colonies (representing cells in which integrated plasmid has been excised and lost) were restreaked on selective medium (MM9YEG agar supplemented with 10 mM p-Cl-Phe and Xgal) at least three times. Purified white colonies were analysed by a colony PCR assay (using primers that anneal to pCF10 sequences external to the cloned mutant allele sequence) to determine whether they carried the wild-type or the mutant allele (Kristich et al., 2005). The following primers were used in the colony PCR assay: pcfD31220/pcfG1 for the ΔpcfE mutant; 2795/PpcfF-S for the ΔpcfD mutant; E5F/pcfG1 for the ΔpcfF mutant.

Markerless genetic exchange in pM2036 was carried out as before with minor modifications (Leenhouts et al., 1996). About 1.5 μg of pCY22 was introduced into L. lactis LM2301RF/pM2036 by electrotransformation, followed by incubation at 30°C for 3 h. The cells were plated on GM17 agar supplemented with Spec and Xgal (250 μg ml^-1) at 30°C. About two blue colonies were obtained per electroporation. Blue transformants (integrants) were purified by restreaking on the same medium at least three times. Several purified blue colonies were then cultured in GM17 without selection for approximately 30 generations at 30°C. Serial dilutions of the cell suspension were plated on GM17 supplemented with Xgal (250 μg ml^-1) at 30°C. About 20 000 colonies were screened and white colonies appeared at a frequency of 2.5 × 10^-6 per generation. Purified white colonies were analysed by a colony PCR assay similar to that used for E. faecalis to determine whether they carried the wild-type or the mutant allele. Primer pair BE1-1/pORE3 was used in the PCR assay.