INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Studies on the conjugal transfer (tra) systems of several plasmids in gram-negative bacteria, including IncF, IncP, IncW, IncQ, IncI, and IncX plasmids, have demonstrated that these systems have functional similarities and extensive sequence similarities at the DNA and protein levels (reviewed in references 32 and 67). This family of DNA transfer systems also includes the virulence (vir) regulon carried by Ti (tumor-inducing) plasmids of Agrobacterium spp., which are responsible for the transfer of tumorigenic DNA to higher plants (75).

Conjugal DNA processing in these plasmids requires plasmid-encoded proteins that interact with a small DNA sequence known as the origin of transfer (oriT) to introduce a strand-specific cleavage at a unique site referred to as the nic site (25, 42, 44, 58, 65). Following cleavage, one protein remains covalently bound to the 5' end of the cleaved strand, and this strand is unwound in the 5'-to-3' direction (45). According to a widely held model, complementary-strand synthesis is initiated from the free 3' end of the cleaved strand in a manner resembling rolling-circle replication (14). In this model, a second cleavage is introduced into the reconstituted oriT after unwinding, and the enzyme ligates the 3' and 5' ends of the unwound DNA (46). The mechanism by which the processed strand is transferred into the recipient cell is as yet unknown, but transfer and unwinding are believed to occur simultaneously (32).

In the IncF, IncW, IncP, IncI, and IncQ conjugative systems and the Agrobacterium vir system, cleavage at oriT requires the action of two proteins, the smaller of which binds to the oriT in a plasmid-specific manner (15, 39, 43, 44). It is believed that binding of the smaller protein is required for recognition of the nic site by the second, larger protein, which cleaves and reseals the oriT. Such enzymes are often designated relaxases. Three conserved sequence motifs have been identified within these relaxase domains, suggesting that they may all have a common ancestry (2, 47). In the IncF and IncW proteins, the C-terminal portions of these proteins contain a helicase activity that is believed to unwind the cleaved strand (37, 65), while the IncP, IncI, and IncQ relaxases are thought to lack this helicase activity. Unwinding of plasmid DNA in the latter plasmids is believed to be carried out by a host-encoded helicase (32, 67).

The conjugal transfer systems of two IncN plasmids, pKM101 (51, 52, 70-73) and pCU1 (26, 48, 49, 56), have also been described. The pilus-encoding region of pKM101 was previously described at the sequence level and shown to contain 10 genes that are required both for conjugative transfer of pKM101 and for sensitivity to several donor-specific bacteriophages that bind to the plasmid's conjugal pilus (8, 51). Earlier complementation experiments indicated that the remaining portion of the tra region contained four complementation groups that are required for conjugation but not required for sensitivity to these phages (70). These complementation groups were therefore thought to direct the processing of plasmid DNA during conjugation. These genes are flanked by the oriT and the fip gene. fip is not required for conjugation but inhibits the fertility of coresident IncP plasmids (72).

Plasmid pCU1 shows striking conservation of restriction sites present in pKM101 over the entire tra region, and the nucleotide sequence of the oriT region is identical to that of pKM101 (9, 48). These nic sites also resemble that of the IncW plasmid R388 (49). The transfer systems of pCU1 and pKM101 are quite similar to that of the IncW plasmid R388 in other respects, including heterologous complementation of tra functions and the pattern of bacteriophage sensitivities imparted by their pili (4, 33, 35).

The experiments presented below characterize the conjugal DNA-processing regions of the IncN plasmids pCU1 and pKM101, showing them to be essentially identical at the DNA sequence level. Although this region was previously thought to contain four tra complementation groups, only three tra genes were identified in the sequence presented in this study. The three genes are transcribed on the same strand, probably from a promoter that lies near or within the oriT. The products of these tra genes rather strongly resemble their counterparts in the IncW plasmid R388 and more weakly resemble Tra proteins of other plasmids. Directly downstream from these genes is the fip gene, whose product is sufficient for the fertility inhibition of coresident IncP plasmids and which may be expressed as part of this putative tra operon. We also present the sequence and functional analysis of the leading region of these plasmids, which is located on the opposite side of oriT. This region contains three genes whose products appear to be required to prevent plasmid instability that can arise as a consequence of interplasmidic homologous recombination.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Construction of recombinant plasmids. The bacterial strains, plasmids, and bacteriophages used in this study are listed in Table 1. Standard cloning techniques were employed (57), using buffers and reaction conditions recommended by the enzyme suppliers. pSP34 was constructed by cloning the RK2 oriT as a BamHI fragment from pNH-Kan/oriT (22) into the BamHI site of pCU1 derivative pCU5714 (48). pET-traK was constructed by cloning the traK coding region as a BstYI-HpaI fragment from pSP34 into the BamHI-HincII sites of pET23d. pET-traJ was created by using PCR amplification to create a DNA fragment containing traJ flanked by BamHI and EcoRI sites and cloning this fragment into the BamHI-EcoRI gap of pET23d. pET-traI was constructed by cloning the BglII-HindIII fragment from pSP34 into the BamHI-HindIII sites of pET23d. Each of the resulting junctions between the T7 promoter and the tra gene was checked by DNA sequencing. Plasmid pMIM101 was created by ligating a 3.5-kb BglII fragment containing the traI and fip genes of pKM101 into the BamHI site of pTZ18R such that these genes are transcribed from the Plac promoter of the vector. pSW345 was created by digesting pKM101 155::Tn5 (30) with SmaI and HindIII and inserting the fragment that contains fip, traKJI, stbABC, and orfD into the SmaI-HindIII gap of plasmid pUC12Cm.

Transposon mutagenesis. pSP34 was mutagenized with transposon TnphoA (38) by incubating 1 ml of log-phase CC118(pSP34) with 1 ml of a ::TnphoA lysate (ca. 10⁹ phage) overnight at 30°C with gentle shaking. The culture was centrifuged and resuspended in 1 ml of Luria-Bertani (LB) medium, and 0.2 ml of this suspension was spread onto LB plates containing kanamycin (50 µg/ml) and chloramphenicol (60 µg/ml). A 0.5-ml portion of a kanamycin solution (10 mg/ml) was added to the center of each plate and allowed to be absorbed into the agar. The plates were incubated overnight at 37°C. The high level of kanamycin caused a zone of clearing containing a small number of isolated colonies. These colonies were selected for further analysis and found to contain TnphoA derivatives of pSP34. pET-traI was mutagenized with TnphoA by transforming it into Escherichia coli PN1, which carries TnphoA on the chromosome of strain CC118. The transformation mixture was incubated in LB medium overnight at 37°C and then plated on LB plates containing chloramphenicol and kanamycin. After overnight incubation at 37°C, the confluent colonies were recovered from the plate and plasmid DNA was extracted. A portion of this DNA was introduced into DH5 by electroporation and plated on LB plates containing chloramphenicol and kanamycin. Transformants were found by restriction digestion to contain TnphoA derivatives of pET-traI. Plasmids pMIM101 and pSW345 were mutagenized with transposon MudII1734 by published procedures (7).

DNA sequencing. pKM101-derived DNA was sequenced by using derivatives of pMIM101 and pSW345 containing insertions of MudII1734 as template DNA. These plasmids were purified by using SpinBind columns (FMC Bioproducts) and sequenced by using a 373A Stretch DNA sequencer (ABI) and primers that hybridize to the left or right end of MudII1734. Sequencing reactions were carried out on both DNA strands with Taq DNA polymerase and DyeDeoxy Terminator Sequencing kits (ABI) and deoxynucleoside triphosphate substrates. Custom-made primers that hybridize to traI or fip DNA were used as needed to complete the sequence.

SDS-polyacrylamide gel electrophoresis (SDS-PAGE). Derivatives of E. coli BL21(DE3) containing a derivative of pET23d containing individual tra genes were cultured to an optical density at 590 nm of 0.4 in M9 medium supplemented with ampicillin (30 µg/ml), thiamine (40 µg/ml), and all 20 amino acids (40 µg/ml) except methionine and cysteine. IPTG (isopropyl--D-thiogalactopyranoside) was added to some cultures to a final concentration of 0.4 mM, and incubation was continued at 37°C for 40 min, at which time rifampin (200-µg/ml final concentration) was added. After a further 20 min at 37°C, 10 µCi of [³⁵S]methionine was added to the culture, and incubation was continued at 37°C for 20 min. One milliliter of each culture was centrifuged, resuspended in 50 µl of protein loading buffer, and boiled for 5 min. The lysate was centrifuged for 5 min in a microcentrifuge, and 5 µl of the supernatant was loaded onto a sodium dodecyl sulfate (SDS)-12.5% discontinuous polyacrylamide gel (20). The gel was stained with Coomassie blue and autoradiographed with BioMax X-ray films (Dupont).

Genetic complementation of tra mutations. Derivatives of strain JC2926 containing a derivative of pKM101 and a derivative of pSW345 were cultured in LB medium supplemented with kanamycin (50 µg/ml) and chloramphenicol (50 µg/ml) to an optical density at 600 nm of approximately 0.5. A single culture of the conjugal recipient (MM294) was cultured to a similar optical density and concentrated 50-fold by centrifugation. Fifty microliters of each donor culture was combined with 50 µl of the concentrated suspension of MM294 and spotted onto Millipore filters that had been placed on prewarmed LB agar medium. These plates were returned to a 37°C incubator for exactly 1 h, at which time the cells were resuspended in 1% NaCl. These cells were serially diluted and plated onto defined medium (AB salts and buffer) containing chloramphenicol (50 µg/ml) to select for transfer of pSW345 derivatives into MM294. Donor cells were enumerated by plating on LB agar supplemented with chloramphenicol (50 µg/ml), kanamycin (50 µg/ml), and streptomycin (250 µg/ml). Transfer efficiency was calculated as the number of recovered transconjugants per recovered donor per hour.

Plasmid cointegration assays. Filter matings between recipient strain HB101rif and donor strain S17-1 containing pSP27 and a derivative of pSP34 were conducted as previously described (48). pSP34 contains the oriT site of plasmid RK2 and is therefore efficiently mobilized by the tra system of S17-1. pSM34 and pSP27 also contain the oriT of pCU1, and the assay measures the efficiency of cointegration between these sites. Transconjugants were selected on LB plates containing rifampcin (100 µg/ml) and either chloramphenicol (60 µg/ml) to select for transfer of pSP34 or tetracycline (20 µg/ml) to select for cointegrative transfer of pSP34 and pSP27.

Plasmid curing assays. Strains were cultured from frozen permanent stocks into LB medium containing ampicillin (50 µg/ml) to minimize the accumulation to plasmid-free cells. When these cultures had reached saturation, they were serially diluted 10⁴-fold, and 0.1 ml of the resulting cell suspension was used to inoculate a 10-ml culture of LB medium. Cultures were incubated to the stationary phase (approximately 20 generations) in 125-ml Erlenmeyer flasks with vigorous aeration. They were then serially diluted 10⁶-fold in 100-fold increments. A 0.1-ml portion of the 10⁴-fold dilution was used to inoculate a fresh 10-ml LB culture, while 0.1 ml of the 10⁶ dilution was plated to determine the fraction of plasmid-free colony-forming units. The procedure was repeated two additional times (60 generations total).

Nucleotide sequence accession numbers. The sequences reported here have been deposited in the GenBank DNA sequence database (accession no. U43676, AF000361, and AF109305).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

The conjugal DNA-processing regions of pKM101 and pCU1. The DNA sequence of an 8.7-kb region between the SmaI-1 site and BglII-3 site of plasmid pKM101 (30) revealed eight open reading frames (ORFs) that are likely to encode proteins (Fig. 1). Four of these ORFs (designated traK, traJ, traI, and fip) are transcribed from right to left, while the remaining four ORFs (stbA, stbB, stbC, and orfD) are transcribed from left to right. traK and stbA are separated by a 513-nucleotide intergenic region that contains the oriT. The intergenic regions between the traK, traJ, traI, and fip genes contain 1, 10, and 1 nucleotides, respectively, suggesting that these genes are probably transcribed as an operon from a promoter that lies within or near oriT. The region containing traK to fip was sequenced independently in the laboratory of one of the authors (R.W.), who obtained an identical sequence (GenBank AF0000361). We also sequenced the traK, traJ, and traI genes of pCU1 (3.4 kb in all), and found that these sequences were identical to those of pKM101 at all but three bases in traI. One of the DNA sequence differences altered TraI residue 206 from alanine to threonine. The two remaining DNA sequence differences did not alter the protein sequence.

View larger version (25K): [in this window] [in a new window]   FIG. 1.   Genetic map of the conjugal DNA-processing regions of pKM101 and pCU1 and the stb region of pKM101. Open triangles, positions and orientations of TnphoA derivatives of pSP34; filled triangles, insertions of MudII1734 in plasmid pSW345; circles, Tn5 insertion derivatives of pKM101 (70). pGW277 and pGW276 are deletion derivatives of pKM101 (30); solid lines show DNA retained in these plasmids. pSP27 is a derivative of pMK2004 containing oriT of pCU1. Plasmids pET-traK, pET-traJ, and pET-traI are derivatives of pET23d that overexpress the corresponding tra products. Short vertical lines indicate the scale, measured in kilobases.

Visualization of the TraK, TraJ, and TraI proteins. The traK, traJ, and traI genes of pCU1 were each cloned individually into the expression vector pET23d (Fig. 2). These constructs were introduced into E. coli BL21(DE3) to visualize the corresponding Tra proteins by SDS-PAGE. Plasmid pET-traK contains 43 bp upstream of the putative ATG start codon (Fig. 2) and hence should produce a native TraK protein (with 139 amino acids and a molecular mass of 15.3 kDa). pET-traK also contains 113 codons of traJ translationally fused to 33 codons of the pET23d vector (146 codons in all, encoding a peptide of 16.6 kDa). This plasmid directed the synthesis of three detectable proteins (Fig. 3, lane 4). The two most strongly expressed proteins (14.3 and 18 kDa) probably correspond to the native TraK protein and the TraJ fragment, respectively. The fainter protein (16 kDa) could be a truncated version of the TraJ fragment initiated from an internal start codon 20 codons downstream of the ATG used to initiate the 18-kDa TraJ peptide (Fig. 2). The detection of the TraJ peptides strongly suggests that traK and traJ are translationally coupled and therefore are expressed from a single promoter.

View larger version (14K): [in this window] [in a new window]   FIG. 2.   5' ends of pCU1 tra genes cloned into the pET23d expression vector. Nucleotide sequences are shown in the upper lines, with corresponding amino acid sequences below. pCU1 nucleotide sequences are shown in lowercase letters, while pET nucleotide sequences are shown in uppercase letters. The amino acid sequences derived from pET23d are shown in roman type, while the amino acid sequences derived from pCU1 genes are italic. Restriction sites, ribosome binding sites (RBS), and putative ATG start codons are underlined.

View larger version (70K): [in this window] [in a new window]   FIG. 3.   SDS-PAGE of pCU1 Tra proteins. Positions of molecular mass standards (in kilodaltons) are indicated at the left. Lanes 1 and 2, pET23d; lanes 3 and 4, pET-traK; lanes 5 and 6, pET-traJ; lanes 7 and 8, pET-traI. The cultures used to make the cell extracts in lanes 2, 4, 6, and 8 were treated with 0.4 mM IPTG prior to addition of rifampin. Proteins corresponding to predicted Tra proteins are indicated at the right.

Amino acid sequence analysis of pCU1 and pKM101 DNA-processing proteins. The traK, traJ, and traI genes encode proteins that are predicted to be largely hydrophilic, suggesting that they are soluble in aqueous environments. However, TraJ contains a strongly hydrophobic region between residues 10 and 25 (Fig. 4), which is preceded by arginine residues at residues 4 and 6. This suggests that the amino terminus of TraJ may be exported from the cytoplasm by the general protein export system (41). TraJ contains a second hydrophobic region between residues 54 and 74 followed by positively charged residues at positions 77, 79, 82, 85, 87, 88, and 91. This sequence resembles a stop transfer signal, suggesting that TraJ may have a transmembrane topology, with residues 26 to 53 located in the periplasmic space and the remainder of the protein located in the cytoplasm.

View larger version (31K): [in this window] [in a new window]   FIG. 4.   Hydropathy profiles of the TraI (A), TraJ (B), and TraK (C) proteins. The algorithm of Kyte and Doolittle (28) was used.

View larger version (110K): [in this window] [in a new window]   FIG. 5.   Protein sequence similarity between the Tra proteins of pKM101 and the Trw proteins of plasmid R388. The Clustal method (23) was used, with a gap penalty of 20 and a gap length penalty of 20.

Complementation analysis. Previous complementation studies with pKM101::Tn5 insertion mutants identified five complementation groups: traK, traJ, traH, traI, and fip. Mutations in the region from traH to traK eliminated transfer, while mutations in fip abolished fertility inhibition of the IncP plasmids (70, 72). None of these genes was required for phage sensitivity, suggesting that none is required for synthesis of the conjugal pilus. In those earlier studies, complementation analysis was carried out by using transient heterozygotes that contained two Tn5 derivatives of pKM101, one introduced by transformation. Since the DNA sequence of this region indicates that there are three genes rather than four, we repeated this analysis with stable merodiploids. To do this, a region of pKM101 DNA (Fig. 1) was subcloned into pUC12Cm, creating pSW345. This plasmid was subjected to transposon mutagenesis with MudII1734 (7), and 10 derivatives having insertions in tra genes were isolated. These were introduced into derivatives of E. coli MC4100 containing tra mutants of pKM101, and the resulting merodiploids were tested for the ability to transfer kanamycin resistance to a conjugal recipient. These experiments indicate that this region contains three complementation groups rather than four (Table 2). The mutations previously designated traH alleles have therefore been renamed traI alleles.

Intracellular site-specific recombination at oriT. Intramolecular site-specific recombination between two oriTs on the same plasmid has been observed previously and used as an assay for oriT-processing activity (5, 6). In this study, an assay involving intermolecular site-specific recombination between oriTs carried on two separate plasmids was used to determine which of the tra functions are required for cleavage and religation at oriT during transfer.

A locus required for stable plasmid inheritance. Transposon insertions within a 1.5-kb region of pKM101 were found somewhat serendipitously to decrease the stable inheritance of the plasmid, causing plasmid-free bacteria to accumulate during prolonged culturing. This locus was first described as being required for sensitivity to the donor-specific bacteriophages Ike and PRD1, even though it is not required for efficient conjugation (unpublished data). It was subsequently found that strains containing stb mutants of pKM101 formed wild-type plaques on media containing kanamycin or ampicillin, while plaques were not detectable on antibiotic-free media. This was thought to be due to plasmid-free, phage-resistant bacteria overgrowing and obscuring the plaques. Additional alleles of this locus were later identified during two subsequent screens, one for deficiency in the fertility of coresident IncP plasmids (the Fip phenotype) and another for deficiency in entry exclusion (the Eex phenotype). However, when plasmid-free bacteria were eliminated (either by using antibiotics or, as described below, by altering the host genotype), mutants with transposon insertion mutations in this region were found to be Fip⁺ and Eex⁺.

View larger version (38K): [in this window] [in a new window]   FIG. 6.   Curing of pKM101 and its derivatives during prolonged culturing in the absence of antibiotic selection. (A) AB1157(pKM101) (), AB1157(pKM101 stbA655::Tn5) (), AB1157(pKM101 stbC135::Tn5) (), AB1157(pKM101 155::Tn5) (), JC2926(pKM101) (), JC2926(pKM101 stbA655::Tn5) (), and JC2926(pKM101 stbC135::Tn5) (). (B) JC7623(pGW277) () and JC7623(pGW276) (). (C) GW4212(pGW277) () and GW4212(pGW276) ().

View larger version (38K): [in this window] [in a new window]   FIG. 7.   DNA sequence of the intergenic region between traK and stbA. (A) Putative promoters, ribosome binding sites (RBS), and translation start sites for traK and stbA are underlined. The oriT sequence is shaded, while the nucleotides that flank the proposed nic site are double underlined. A dyad symmetry is indicated by inverted arrows, and repeated sequences are indicated by square brackets. (B) Alignment of the repeated sequences.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Restriction maps of pCU1 and pKM101 suggested that they contain very similar conjugal transfer regions. The data presented in this paper demonstrate that their conjugal DNA-processing regions are identical at all but 3 nucleotides over a 5,879-bp region encompassing oriT, traK, traJ, and traI. pCU1 and pKM101 also have strong sequence similarity at the 2,366-bp kik region, which has been sequenced for both plasmids in previous studies (51, 55). The variation observed over both regions represents a substitution rate of less than 0.2%. This high degree of sequence identity at both regions combined with the lack of any restriction site polymorphisms over the entire tra region strongly suggests that the tra regions of pCU1 and pKM101 are extremely similar and diverged from a common ancestor relatively recently. Furthermore, it is possible that this identity extends past the tra region. pCU1 and pKM101 have identical restriction maps except for the regions that encode antibiotic resistances. pKM101 is a deletion derivative of a larger plasmid, R46, which contains resistance determinants against ampicillin, streptomycin-spectinomycin, sulfanomides, tetracycline, and arsenate. Some of these genes are found within an integron (18). In contrast, pCU1 confers resistance to ampicillin and streptomycin-spectinomycin. It is possible that the only significant differences between pCU1, pKM101, and R46 are found in these clustered antibiotic resistance determinants. The deletion that created pKM101 appears to have been mediated by insertion sequence IS26 (17, 31). It is possible that pCU1 was also derived from an R46-like plasmid by a similar in vivo deletion.

The nucleotide sequences of the conjugal DNA-processing genes of pKM101 and pCU1 identify three genes in this region that are required for conjugation. TraI is homologous to TrwC of the IncW plasmid R388 (Fig. 5), and both proteins are also related to the TraI protein of plasmid F. Both the R388 and F plasmid proteins are known to contain an amino-terminal oriT-specific nucleolytic function and a carboxyl-terminal helicase function (36, 37, 65). The intermolecular-recombination results presented in this study strongly suggest that the IncN TraI has a comparable nucleolytic activity. The corresponding relaxases of the TraI protein of plasmid RK2 and the VirD2 protein of Agrobacterium tumefaciens have tyrosine residues (at residues 22 and 29, respectively) which are involved in covalent binding of the protein to the 5' end of the cleaved strand and are considered to be part of the catalytic centers of these enzymes (46, 47, 66). The IncN TraI protein has four tyrosine residues at positions 18, 19, 26, and 27. This arrangement is reminiscent of the two tyrosine residues separated by three amino acids in the A protein of phage X174, which is required for rolling-circle replication (21). For the X174 system, it was postulated that the two tyrosine residues alternate in cleaving within the replication origin. This situation could assist in the intermolecular recombination event observed in this study. It is conceivable that each tyrosine residue (or one from each pair) cleaves the DNA strand and binds to separate pSP27 and pSP34 oriTs, bringing them into close proximity to each other to allow the transesterification step between the free 3' OH of one plasmid and the 5' phosphate of the other, resulting in the formation of a cointegrate plasmid.

Intermolecular recombination at the pCU1 oriT required both the traI and traK genes, suggesting that TraK may be a functional homologue of the IncF TraY protein. The TraY proteins of both the F plasmid and R100 are required for DNA cleavage both in vivo and in vitro (25, 42) and for oriT-mediated recombination (6). Alignment of the TraK protein with the F TraY protein shows little sequence similarity. On the other hand, there is limited sequence identity, primarily in their carboxyl termini, between TraK and the TrwA protein of IncW plasmid R388. TrwA is not required for intramolecular recombination between two R388 oriT sites in vivo (33) but has been shown to enhance in vitro cleavage by TrwC (39). The fact that R388 TrwA is not required for recombination at oriT while pKM101 TraK is required may be related to the sensitivities of these two recombination assays. In the IncW study, recombination was identified by the loss of an antibiotic marker situated between two copies of the oriT (33). Detection of a recombination event required that all copies of the plasmid within the cell carry the deletion. As a result, individual recombination events within the cell could be masked by the presence of plasmids that had not undergone recombination. The assay described in this study selects for plasmids which have undergone recombination, since only those pSP27 plasmids which cointegrate into the larger pSP34 plasmid are transferred.

TraJ is the only Tra protein in this region whose hydropathy profile suggests a transmembrane topology. Homologous proteins are found in virtually all conjugation systems. Members of this family of proteins have similar hydropathy profiles and in some cases have been shown to be associated with the inner membrane (43). It has been postulated that such a protein could be used to bind both the relaxosome and the mating pore and, by doing so, to bring the DNA in juxtaposition to the pore. This could explain why traJ mutations caused such a slight reduction in conduction of pSP27. Binding of pSP27 and pSP34 DNA to the membrane via TraJ prior to cleavage could enhance intermolecular recombination by bringing the two plasmids into closer proximity. Involvement of the traJ function in the site-specific recombination event has not been observed for the TraJ homologue of any other transfer system studied to date (33, 42, 46). Alternatively, traJ::Tn5 mutations could exert polar effects on expression of the downstream traI gene, although these same mutations did not appear to be polar in complementation assays.

Downstream of traI is a gene designated fip, which abolishes the conjugation of coresident IncP plasmids (72). fip appears to be part of the traKJI operon, although the significance of this coexpression is not understood. Immediately downstream from fip is the nuc gene, which is transcribed convergently to fip (53). Like fip, nuc is also the last gene in a tra operon and has no direct role in conjugation.

It was initially surprising that three tra genes were found in this region rather than four, since this region was previously thought to have four complementation groups (70). We therefore carried out additional complementation studies, using stable heterodiploid strains rather than the transient heterodiploids used previously. We identified three complementation groups, in agreement with the sequence data. Mutations previously designated as being in the traH complementation group were therefore renamed traI mutations. It is far from clear why the earlier study indicated that traI mutations fell into two complementation groups. However, the sequence of traI suggests an internal translation start site at codon 482 (AAGAAGG-N₅-ATG). This region is immediately upstream of the putative helicase domain. This would suggest that TraI could be made in two forms, a full-length form and a truncated form containing only a helicase domain. The TraI protein of F has extremely similar properties, since a protein designated TraI* is translated by using an internal initiation codon. TraI* contains the helicase domain but not the relaxase domain (65). However, if this explanation is correct, is does not explain why only three complementation groups were obtained in the present study. We therefore cannot at this time fully understand the discrepancy between the older analysis and the present one.

We have also described a locus of pKM101 that appears to play a role in preventing recombination-mediated plasmid instability. While the cause of this instability is unclear, several other plasmids have been reported to have functionally similar genes, and in some cases these genes have been shown to encode site-specific recombination systems. These include the parA gene of RP4 (12), the D protein of mini-F (29), the per gene of R46 (11), and the cre gene of bacteriophage P1 (59), among others. The ParA and Per proteins are homologous to the resolvase protein of Tn3, while the D and Cre proteins are not homologous to other known proteins, and none is similar to StbA, StbB, or StbC or pKM101. The hallmark of most of these systems is that they are needed only in recombination-proficient hosts. It is believed that two identical copies of any multicopy plasmid can undergo homologous recombination, resulting in a head-to-tail dimer, which would effectively decrease plasmid copy number and could lead to inefficient partitioning to daughter cells during cell division (61). These site-specific recombination systems are thought to convert dimeric plasmids to monomers, thereby enhancing plasmid stability. We postulate that the stb locus of pKM101 may play a similar role. If so, it is not clear why three proteins would be required, since all the site-specific recombination systems listed above require just one protein. Perhaps only stbC is required for stability, and insertion mutations in stbA or stbB prevent expression of stbC by transcriptional polarity.

Each of these recombinases act at a particular DNA sequence, denoted the par site in RP4, the rfsF site in mini-F, the per site in R46, and the lox site in P1. Each of these sites is composed of direct or inverted repeats that provide binding sites for the resolvase proteins. In all cases except the lox-cre system, these sites are located directly upstream of the respective recombinase genes, and each recombinase serves as a transcriptional autorepressor as well as a recombinase. This provides a simple mechanism for the synthesis of sufficient amounts of protein to saturate the binding site. We hypothesize that if in fact the stb operon encodes a site-specific recombinase, the direct repeats found directly upstream of stbA could provide a cognate resolution site. If so, binding of one or more of these proteins could cause negative autoregulation.

As described above, one of these recombinase systems (per) is found on plasmid R46. Interestingly, R46 is the direct parent of pKM101, which was derived by an in vivo deletion of 15 kb of R46 DNA (17, 31). This deleted DNA includes the per gene. Therefore, if our model for stb function is accurate, it would appear that R46 has two such systems, one encoded by stb and the other encoded by per. In agreement with previous studies, we find that R46 is not detectably lost from a population of bacteria even after 60 generations (unpublished data). In constrast, pKM101 is lost at a detectable rate, which is consistent with published data that per mutants of R46 are detectably unstable (11). In all cases, instability occurred only in a recombination-proficient host.

Plasmid pKM101 (36,255 bp) has now been sequenced in its entirety. Figure 8 shows the positions and transcriptional orientations of all previously characterized genes of this plasmid as well as 10 uncharacterized ORFs. Approximately half of the DNA of this plasmid encodes proteins that play some role in conjugation. pKM101 also contains genes that direct vegetative replication, stable plasmid inheritance, inhibition of host restriction systems, and error-prone repair of damaged DNA (see the legend to Fig. 8 for more detail). A more detailed description of pKM101 and its parent R46 is in preparation and will be the subject of a future study.

View larger version (20K): [in this window] [in a new window]   FIG. 8.   Functional map (A) and physical map (B) of pKM101. Labelled arrows denote characterized genes, while unlabelled arrows denote uncharacterized ORFs. Previous maps of pKM101 started at the unique EcoRI site (31). To avoid confusion between the locations of genes in pKM101 and its parental plasmid R46, we have numbered the sequence from the first nucleotide of IS26 (which is common to both plasmids). Gene designations: bla, oxa2 -lactamase (19); IS26, insertion sequence IS26, (formerly denoted IS46); kikA, required for killing of Klebsiella strains during conjugation (24); traM to -G, required for conjugation and for sensitivity to donor-specific bacteriophages (51); traI to -K, conjugal DNA processing (this study); korA and korB, corepressors of the korB, traL, and traN promoters (40); eex, entry exclusion (52); nuc, periplasmic endonucleolytic DNase (53); fip, fertility inhibition of coresident IncP plasmids (70); stbA to -C: stable plasmid inheritance (this study); ardA and ardB, inhibition of host DNA restriction enzymes (3); ardR and ardK, regulators of ardA, ardB, ccgAI, ccgAII, ccgC, ccgD, and repA (10); ccgAI, ccgAII, ccgC, and ccgD, genes of unknown function that are regulated by ArdK and ArdR (10); mucA and mucB, error-prone DNA repair (50); mpr, unknown function, possible metalloprotease (10); repA, plasmid vegetative replication (10); uvp1, site-specific recombinase (64); int, conserved gene within integron (19). The complete DNA sequence of pKM101 was compiled by using sequence data contained in the GenBank DNA sequence database (accession no. AF000360, U09868, U43676, U72482, U00430, U00434, L09114, M81860, Y00358, and X06046).