ABSTRACT
Rhizobium leguminosarum strain VF39SM contains two plasmids that have previously been shown to be self-transmissible by conjugation. One of these plasmids, pRleVF39b, is shown in this study to carry a set of plasmid transfer genes that differs significantly from conjugation systems previously studied in the rhizobia but is similar to an uncharacterized set of genes found in R. leguminosarum bv. trifolii strain WSM2304. The entire sequence of the transfer region on pRleVF39b was determined as part of a genome sequencing project, and the roles of the various genes were examined by mutagenesis. The transfer region contains a complete set of mating pair formation (Mpf) genes, a traG gene, and a relaxase gene, traA, all of which appear to be necessary for plasmid transfer. Experimental evidence suggested the presence of two putative origins of transfer within the gene cluster. A regulatory gene, trbR, was identified in the region between traA and traG and was mutated. TrbR was shown to function as a repressor of both trb gene expression and plasmid transfer.
INTRODUCTION
Rhizobia are agriculturally important bacteria that are capable of forming nitrogen-fixing nodules on the roots of legumes, such as peas, lentils, and beans. In fast-growing rhizobia, such as the genera Rhizobium and Ensifer (Sinorhizobium), the genes required for establishment of this symbiosis (nod, nif, and fix genes) are usually located on one of the large plasmids, known as pSyms. In addition, other large rhizobial plasmids carry genes beneficial for bacterial fitness and competitiveness in the rhizosphere, such as genes encoding bacteriocin production, lipopolysaccharide production, exopolysaccharide production, utilization of different carbon sources, and synthesis of specific vitamins (1–9).
Although plasmid conjugation has not been shown to enhance bacterial competitiveness directly, it is one of the most important methods for bacteria to acquire genetic information and adapt to changing environmental conditions. There is considerable evidence for rhizobial plasmid transfer both under laboratory conditions (10–13), and in natural environments (14–16). In addition, genome sequences of many rhizobial strains and rhizobial plasmids have revealed potential conjugation genes encoding DNA transfer and replication (Dtr) components and mating pair formation (Mpf) components, as well as putative origins of transfer (oriT).
Two types of rhizobial conjugation systems have been characterized, including the quorum sensing (QS)-regulated conjugation system (type I) and the RctA-repressed conjugation system (type II). In a recent review, we demonstrated that these two conjugation systems are phylogenetically separate, consistent with their characterized transfer regulation (8). In addition, we proposed the presence of the type III conjugation system on plasmids pRL10JI, pRL11JI, and pRL12JI of Rhizobium leguminosarum bv. viciae 3841 (17), and pRleVF39d, pRleVF39e, pRleVF39f of R. leguminosarum bv. viciae VF39SM (18) based on sequence information (8). None of the type III plasmids were self-transmissible due to the lack of an Mpf component. In a recent study, the relaxase on pSmeLPU88b of Ensifer meliloti LPU88 has been characterized (19). Phylogenetic analyses showed that pSmeLPU88b relaxase together with relaxases from pSmed03 of Ensifer medicae WSM419, pSmeGR4a of E. meliloti GR4, pSmeSM11c from E. meliloti SM11, and pRleVF39b from R. leguminosarum bv. viciae VF39SM form a distinct phylogenetic cluster and hence were named type IV rhizobial relaxase.
R. leguminosarum bv. viciae VF39SM contains six large, single-copy plasmids, named pRleVF39a to pRleVF39f from the smallest to the largest plasmid. The two smallest plasmids, pRleVF39a and pRleVF39b, were demonstrated to be transmissible or at least mobilizable from R. leguminosarum VF39SM to Agrobacterium tumefaciens UBAPF2 (20), at frequencies of about 10−5 and 10−7 transconjugants per recipient, respectively (21). However, since at least two conjugation systems are present in the donor background of R. leguminosarum bv. viciae VF39SM, conclusions could not be made as to whether these two plasmids were both self-transmissible. PCR and hybridization using the transfer genes from pRL7JI and pRL8JI of the available R. leguminosarum 3841 genome sequence revealed that pRleVF39a carried a conjugation system similar to pRL8JI of 3841, whereas pRleVF39b does not carry any conjugation system similar to the ones in 3841 (8).
In this study, characterization of the conjugation system on pRleVF39b was carried out. We determined the genetic requirements that are responsible for the transfer of pRleVF39b, and identified a repressor gene, trbR, on pRleVF39b in the vicinity of the transfer genes that controls the transcription of the Mpf component genes.
MATERIALS AND METHODS
Bacterial strains, plasmids, and growth conditions.The bacteria and plasmids used in this study are listed in Table 1. R. leguminosarum and A. tumefaciens strains were grown in tryptone-yeast extract (TY) medium (22) or in Robertsen's minimal medium (RMM) (23) at 30°C. A. tumefaciens was grown on Penassay medium (Difco antibiotic medium 3) when selection against Rhizobium was required. Escherichia coli strains were grown at 37°C in LB medium (24). When required, the following antibiotic concentrations were used: for Rhizobium, streptomycin, 600 μg ml−1, tetracycline, 5 μg ml−1, kanamycin, 25 μg ml−1, neomycin, 100 μg ml−1, gentamicin, 30 μg ml−1, and spectinomycin, 300 μg ml−l; for A. tumefaciens, rifampin, 100 μg ml−l, erythromycin, 150 μg ml−l, tetracycline, 5 μg ml−1, kanamycin, 25 μg ml−1, neomycin, 100 μg ml−1, gentamicin, 50 μg ml−1, and spectinomycin 300 ml−l; and for E. coli, tetracycline, 10 μg ml−1, kanamycin, 50 μg ml−1, gentamicin, 15 μg ml−1, erythromycin, 150 μg ml−1, spectinomycin, 100 μg ml−l, and ampicillin, 100 μg ml−1.
Bacterial strains and plasmids used in this study
Construction of new VF39SM genomic libraries.An R. leguminosarum bv. viciae VF39SM genomic cosmid library was constructed essentially as previously described (25). Briefly, the total DNA of VF39SM was partially digested by Sau3A under conditions optimized so that DNA fragments were in the 40-kb range and ligated into either the broad-host-range cosmid vector pRK7813 (26) or the nonreplicating cosmid vector pJQ210SK (27), followed by in vitro packaging using the Novagen Phage Maker packaging kit. The reaction mixtures were used to transfect the mobilizer strain E. coli S17-1 (28). A total of 1,152 pRK7813 cosmids were individually selected from transfection plates and stored in 12 96-well plates. The remaining cells were scraped off from each transfection plate and stored as 12 pools of cosmids. A total of 1,200 pJQ210SK cosmids were individually selected from the transfection plates and stored as 12 pools of 100 cosmids.
DNA manipulation, sequencing, and constructions.Standard techniques were used for DNA manipulations (24). Plasmids were isolated using EZ-10 spin column plasmid DNA Minipreps kit (Bio Basic, Inc.). Total DNA was isolated using a mi-Bacterial genomic DNA isolation kit (Metabion, Inc.). Primers were synthesized by Sigma-Genosys (Sigma-Aldrich, Canada). PCRs were set up using either Taq PCR master mix kit (Qiagen) or high-fidelity Phusion polymerase (New England BioLabs) based on the experimental requirements and used according to the manufacturer's instructions. PCRs were carried out using a MultiGene II thermal cycler (Labnet International). Restriction endonucleases were purchased from Invitrogen or Fermentas and were used according to the manufacturer's specifications. DNA fragments were isolated from agarose gels using QIAquick gel extraction kit (Qiagen). Southern hybridizations were performed using the digoxigenin (DIG) labeling system supplied by Roche Applied Science. Sanger sequencing was performed by Quintara Biosciences (Berkley, CA).
The construction of UBAPF2(ΔpRleVF39b::Tn5-M6) was achieved by swapping the Km resistance gene in UBAPF2(pRleVF39b::Tn5mob) (21) with Sp resistance gene using pJQ173 as previously described (34). The resultant ΔpRleVF39b::Tn5-M6 was found to be smaller than pRleVF39b::Tn5mob in size, implying the occurrence of a deletion. The deletion that occurred in pRleVF39b has often been observed and is very likely related to the deletion of the transposon-like restriction modification system on pRleVF39b, which allows the host DNA to be resistant to PstI digestion (29). This was confirmed in the case of ΔpRleVF39b::Tn5-M6, whose DNA was sensitive to PstI digestion. PCR of the repB, traA, and trbE genes of pRleVF39b was carried out to confirm that ΔpRleVF39b::Tn5-M6 carried these pRleVF39b genes. The transfer behavior of ΔpRleVF39b::Tn5-M6 was also found not to be different from that of pRleVF39b::Tn5-M8 (Table 2). Therefore, UBAPF2(ΔpRleVF39b::Tn5-M6) was used in subsequent gene replacement experiments involving the pCosJB11::EZTn5s and pJQ200SK::trbE::Kmr.
Self-transfer of ΔpRleVF39b::Tn5-M6 and its derivatives from R. leguminosarum bv. viciae VF39SM or A. tumefaciens UBAPF2 to the recipient UBAPF2recA
In vitro transposition was carried out on cosmid pCosJB11 using the EZ-Tn5 <Kan-2> insertion kit (Epicentre Biotechnologies), followed by transformation of E. coli S17-1 competent cells. The resulting pCosJB11::EZ-Tn5 clones were selected on LB with Tc and Km, and the EZ-Tn5 insertion sites were determined by sequencing using EZ-Tn5 internal primers. Cosmids pCosJB11::EZ-Tn5 with insertions within trbE, trbI, traG, orf16, trbR, orf19, and traA were individually selected and mobilized into VF39SM and UBAPF2(ΔpRleVF39b::Tn5-M6). Plasmid pPH1JI (30), which is incompatible with pCosJB11, was introduced into each VF39SM or UBAPF2(ΔpRleVF39b::Tn5-M6) carrying pCos138::EZ-Tn5, forcing the gene replacement via homologous recombination, as previously described (31). Correct gene replacement was confirmed by Southern hybridization analysis or PCR. Plasmid pPH1JI was cured by the introduction of pRK415::sacB, which is incompatible with pPH1JI. The loss of pPH1JI was confirmed by Eckhardt gel analysis and loss of the Gmr. To cure the pRK415::sacB, individual colonies were streaked on solid medium containing 5% sucrose, which induces the expression of the lethal gene sacB.
An independent trbE mutant was constructed using the suicide vector pJQ200SK-based method (27). Briefly, a 1.3-kb EcoRI fragment containing trbE internal fragment was subcloned from pCosJB11 into pBluescript SK+. The Kmr cassette from pBSL99 (32) was inserted as a BamHI fragment within the BglII site in the trbE fragment. The trbE::Km fragment was then subcloned as an ApaI/BamHI fragment from pBluescript to pJQ200SK. The resulting pJQ200SK::trbE::Km was used in gene replacement for trbE in the wild-type VF39SM and UBAPF2(ΔpRleVF39b::Tn5-M6) as previously described (27). Mutants with correct gene replacement were confirmed by Southern hybridization.
To make recA mutants of VF39SM derivatives, a 1.9-kb EcoRI fragment containing the recA gene of R. leguminosarum was excised from pMS102 (33), from which a DNA probe was made and hybridized to a VF39SM genomic cosmid library (34). Cosmid pCos74 was identified to contain the recA gene. A 3.2-kb SstI/XhoI fragment was subcloned from pCos74 to pBluescript. A 2.1-kb HindIII fragment carrying the Kmr cassette from pHPΩ45Km (35) was inserted into the unique HindIII site within the recA gene. The whole recA::Km fragment was subcloned as an SstI/XhoI fragment into pJQ200SK.
To make recA mutants of plasmid-free Agrobacterium derivatives, the recA::Ery fragment was subcloned as a BclI fragment from pJM54 (36) into the BamHI site of pJQ200SK (27), resulting in pJQ200SK::recA(Agro)::Ery, which was used in gene replacement experiments.
To isolate the replication gene-containing cosmids, the 12 pJQ210SK cosmid pools were conjugated en masse into A. tumefaciens UBAPF2, such that any transconjugants would harbor a cosmid with a full set of replication and partitioning genes. Eight cosmids were found to replicate and were isolated from UBAPF2. Two of these cosmids (pCosRep4 and pCosRep7) were identified by Southern hybridization to contain the repABC region derived from pRleVF39b. The minimal replicon, pJQ200SK::repABCb was constructed from pCosRep4 using restriction sites determined from the VF39SM genome sequencing project (9). An ∼15-kb AgeI fragment with the repABC region of pRleVF39b was subcloned from pCosRep4 into the XmaI site of pJQ200SK. The resulting pJQ200SK subclone was then BamHI digested and religated to result in pJQ200SK::repABCb (∼7 kb).
RNA methods and transcript analysis.Total RNA was extracted using the Ambion RiboPure-Bacteria kit (Applied Biosystems). cDNA synthesis was accomplished by reverse transcription using the QuantiTect reverse transcription kit (Qiagen) according to manufacturer's instructions using specific primers instead of the kit-supplied primer mixture (for primer sequences, see Table S1 in the supplemental material). The synthesized cDNA was used as a template in the subsequent PCRs for targeting the upstream genes in the putative operon.
Bacterial conjugation experiments.Bacterial cultures were grown overnight in TY broth at 30°C. Equal volumes (500 μl) of the donor and recipient cultures were mixed and centrifuged in a microcentrifuge at 13,000 rpm for 2 min. The pellet was washed with TY or RMM, and then resuspended in 50 or 100 μl of the selected medium (TY or RMM) and spotted on the surface of an agar plate (TY or RMM). The mating plate was then incubated at 30°C or at room temperature (20 to 22°C) for 24 or 48 h. The cells were collected from the mating spot and resuspended in 1 ml of autoclaved deionized H2O. Different dilutions were plated on the appropriate selection medium, and the plates were incubated at 30°C until colonies appeared.
To determine the transfer frequency, the resuspended mating mixture was serially diluted to 10−7. Appropriate dilutions were plated out on media selecting for either the recipient or the transconjugant. The colony numbers for recipients and transconjugants, as well as donors in some experiments, were counted, and the transfer frequency was then calculated as transconjugants per recipient. Some data based on frequencies per donor are also provided in some cases.
Plasmid curing by incompatibility.R. leguminosarum VF39SMrecA(−pRleVF39b) was constructed by curing pRleVF39b from VF39SM through incompatibility. Briefly, construct pJQ200SK::repABCb, which contains the incompatibility determinants of pRleVF39b, was introduced into VF39SMrecA::Km through conjugation. Transconjugants with pJQ200SK::repABCb were selected and subjected to Eckhardt gel analysis to confirm the loss of pRleVF39b. The construct pJQ200SK::repABCb was eliminated using the sacB-based system by plating on medium containing 5% sucrose (27). The resulting colonies were analyzed for the loss of the replication genes and transfer genes of pRleVF39b by PCR.
Construction of gene fusions and β-glucuronidase assasy.To construct gene fusions, the putative promoter region of the traA gene and the trb operon were PCR amplified. The traA promoter region was amplified from the primer set containing the forward (F) primer (gagaaagcttAACGCCACGAGCAACTGT) and reverse (R) primer (gagagaattcCTTGACGACGACCGCATTG). The capitalized sequence corresponds to sequence on pRleVF39b. The lowercase sequences are adapter sequences containing restriction sites for HindIII and EcoRI (underlined). The amplified fragment was digested with HindIII and EcoRI, and subcloned as a 653-bp fragment in pFUS1 (37). For the trb operon promoter, primers used were as follows: F primer, TCTGCTGGTCAACGAACGGATG; and R primer, GCATTATAGGCAGCGGCGACAA. The amplified fragment was digested with internal sites XhoI and SphI and subcloned as a 469-bp fragment in pFUS1. Both fusion constructs were confirmed by DNA sequencing.
All transcriptional fusion constructs were mobilized into Rhizobium by conjugation. The wild-type and mutant strains carrying the fusion constructs were grown in TY. The β-glucuronidase assay was performed as previously described (34).
Bioinformatic analyses.BLAST (Basic Logic Alignment Search Tool) (38) searches were used to identify sequences homologous to a particular gene (nucleotide BLAST, blastX) or protein (protein BLAST) within the NCBI database. Sequence alignment was performed using ClustalX 2.0 (39). The phylogenetic tree of a particular protein from different sources was generated from the ClustalX sequence alignment by the neighbor-joining method with a bootstrap value of 1,000.
Nucleotide sequence accession number.The sequence and annotation of the novel conjugation system examined in this study has been submitted to GenBank under accession no. HQ711365.
RESULTS
Transfer of pRleVF39b in different genetic backgrounds.In this study, we examined the self-transfer of pRleVF39b in the otherwise plasmid-free UBAPF2, using another A. tumefaciens or Rhizobium as recipients. To be able to detect the transfer, recipients with different markers were constructed, including UBAPF2recA, which carried an erythromycin resistance marker allowing for selection against the donor UBAPF2, and VF39SMrecA(−pRleVF39b), which is cured of pRleVF39b, and hence would not inhibit pRleVF39b transfer by either entry exclusion or by plasmid incompatibility.
Bacterial conjugations were carried out on different media, including TY agar, RMM agar with different sole carbon sources, including mannitol, adonitol, erythritol, rhamnose, glycerol, and pea seed exudates, and on TY agar using Phytagel as gelling agent, in order to ascertain whether mating conditions affected transfer frequencies. Bacterial matings were also carried out at 30°C and at room temperature (20 to 22°C). It was revealed that the different media and different temperatures tested did not seem to have any significant influence on the conjugative transfer of pRleVF39b (data not shown).
When UBAPF2 was used as the donor, ΔpRleVF39b::Tn5-M6 self-transferred at frequencies of (6.5 ± 1.1) × 10−5 transconjugants per recipient and (1.1 ± 0.3) × 10−4 transconjugants per donor from UBAPF2 to UBAPF2recA and (8.0 ± 1.6) × 10−7 transconjugants per recipient from UBAPF2 to R. leguminosarum VF39SMrecA(−pRleVF39b). The resulting transconjugants from this mating were confirmed to carry pRleVF39b by Eckhardt gel analysis. When R. leguminosarum VF39SM was used as the donor, ΔpRleVF39b::Tn5-M6 transferred to UBAPF2 at a frequency of (4.3 ± 0.7) × 10−6 transconjugants per recipient and (7.5 ± 2.00) × 10−4 transconjugants per donor cell plated.
A novel transfer region on pRleVF39b of R. leguminosarum VF39SM.The R. leguminosarum bv. viciae VF39SM genome sequencing project revealed the presence of three types of conjugation systems: one is similar to the QS-regulated conjugation system, which has been confirmed to be on pRleVF39a, as determined by PCR and Southern hybridization (8), one is similar to the sequence on pRLG203 in R. leguminosarum bv. trifolii WSM2304, which is the focus of this work, and the third type, present in multicopies, is similar to the Dtr systems from pRL10JI, pRL11JI, and pRL12JI, the three largest plasmids of R. leguminosarum 3841 Presumably, these copies of Dtr systems are from pRleVF39d, -e, and -f, based on previous hybridization results (8).
After preliminary assembly, two contigs were found to be homologous to the putative tra locus on pRLG203, with a small gap between these contigs predicted to be less than 1 kb. PCR was used to close the gap between these two contigs. The sequence for the putative novel conjugation system was annotated by the web-based Xbase bacterial genome annotation server (40), which integrates Glimmer (41), tRNAScan-SE (42), and RNAmmer (43). Minor editing was done based on sequence alignment with the tra region on pRLG203 made by ClustalX version 2.0 (39) and tBlastX (38). This novel conjugation system was located on plasmid pRleVF39b, as determined by Southern hybridization between a trbE internal PCR probe and the pRleVF39b band on an Eckhardt gel showing plasmid profiles of R. leguminosarum VF39SM (data not shown).
The genetic organization of this conjugation system is shown in Fig. 1A. The entire transfer region contains 12 trb genes, encoding a putative Mpf component, a traG gene encoding the coupling protein, and the relaxase gene traA located about 9 kb downstream of the traG. BlastX analyses of each putative gene revealed that other than the conjugation system on pRLG203, nothing else in the NCBI databases (including both the nonredundant protein and metagenomic protein database) was highly similar (>50% identity at the protein level) to the pRleVF39b-borne conjugation system. A more comprehensive comparison of each tra gene between pRleVF39b and pRLG203 is shown in Table S2 in the supplemental material.
Genetic structure of the conjugation system on pRleVF39b and alignment with other type IV rhizobial plasmid conjugation systems. (A) The transfer genes (in gray) on pRleVF39b contain a trb operon, a coupling protein gene, traG, and the relaxase gene traA. (B) Alignment of the Mpf component genes of the type IV rhizobial conjugation systems, from pRLG203 of R. leguminosarum bv. viciae WSM2304, pSmed03 of E. medicae WSM419, and pAtS4a of A. vitis S4 and plasmid 1 of Chelativarons sp. strain BNC1, as well as A. tumefaciens pTiC58 (type I) and Rhizobium etli CFN42 plasmid pRetCFN42d (type II). (C) Alignment of the relaxase gene of the type IV conjugation systems.
Recently, the relaxase of pRleVF39b has been shown to belong to the MOBP0 type relaxase group (19). However, the pRleVF39b conjugation system differs from the MOBP0 conjugation systems in several ways. First, the relaxase phylogenetic tree clearly shows that the pRleVF39b relaxase is on a separate branch from the rest of the type IV relaxases, MobZ, proposed by Giusti et al. (19). This is consistent with our phylogenetic analyses of rhizobial relaxases (Fig. 2A). In addition, the phylogenies of the TraG (Fig. 2B) and TrbE/VirB4 (Fig. 2C) genes associated with rhizobal conjugative plasmids displayed almost identical topologies to that of the relaxase phylogenetic tree (Fig. 2A), with pRleVF39b on a different branch from the rest of the type IV systems. Second, all of the proposed type IV Dtrs are organized similarly, with mobC and mobZ (relaxase gene) transcribed divergently from a parA-like gene (Fig. 1C), with oriT located between mobC and parA. The pRleVF39b-borne conjugation system, however, does not have this genetic organization. No putative mobC gene is found in the vicinity of traA or in the unfinished VF39SM genome. A parA-like gene (orf26), transcribed in the same orientation as traA, is present two open reading frames (ORFs) upstream of the traA gene. It has to be noted that the conjugation system on pRLG203 also has the same synteny as the pRleVF39b over the transfer region. Due to these differences, we grouped the pRleVF39b- and pRLG203-borne Dtr components as type IVA, whereas type IVB includes the rest of the MOBP0 Dtrs. The putative Mpf components associated with all type IV Dtrs are found to have the same synteny, with putative trb/virB/avhB genes and a coupling protein gene traG/virD4 immediately downstream (Fig. 1B).
Phylogenetic analysis of relaxases, coupling proteins, and TrbE/VirB4 proteins in rhizobial plasmid conjugation systems. Trees were generated based on protein sequence alignment generated by ClustalX, using the bootstrap neighbor-joining method with bootstraps set at 1,000. Only bootstrap values lower than 90% are displayed. The scale bar represents the estimated frequency of the amino acid substitutions per sequence position. (A) A phylogenetic tree of relaxases shows four different types of rhizobial relaxase. Type I contains relaxase from QS-regulated conjugation system, type II contains relaxase from the RctA-repressed conjugation systems, type III contains uncharacterized relaxases from plasmids that do not have associated Mpf components, and type IV contains MOBP0-type relaxases. Type IVA and type IVB systems are clearly on separate branches. The accession numbers of the relaxases used in building this tree are as follows: NP_396650 (pTiC58), YP_002540050 (pTiS4), YP_002542670 (pAtS4c), YP_002551269 (pAtK84b), YP_002539500 (pAtS4e), YP_771015 (pRL8JI), YP_001965642 (pSmeSM11b), NP443828 (pNGR234a), ABR62390 (pSmed01), ABR64161 (pSmed02), YP_471748 (pRetCFN42a), YP_770819 (pRL7JI), NP_066693 (pRi1724), YP_001961052 (pRi2659), YP_002978881 (pR132503), YP_002551445 (pAtS4b), NP_659868 (pRetCFN42d), YP_001984448 (pRetCIAT652b), YP_001985502 (pRetCIAT652c), AAK65163 (pSymA), NP_4437206 (pSymB), AEH83757 (pSmeSM11d), YP_002823261 (pNGR234b), NP_3960469 (pAtC58), YP_086778 (pAgK84), YP_003329392 (pSmeSM11a), YP_770499 (pRL10JI), YP_771309 (pRL11JI), YP_765073 (pRL12JI), YP_002978744 (pR132504), YP_002984810 (pR132502), YP_002973152 (pR132501), ACY71086 (pRleVF39b), YP_002283997 (pRLG203), AEU04532 (pSmeLPU886), AEH81303 (pSmeSM11c), YP_001314900 (pSmed03), YP_002546478 (pAtS4a), ABG61150 (p1_BNC1_Meso_4166), AAA26445 (RSF1010), NP_040369 (ColE1), and YP_001687708 (RP4). (B) Phylogenetic tree of TrbE/VirB4 as a representation of the Mpf components of rhizobial conjugation systems. The accession numbers of the TrbE/VirB4 proteins used in this tree are as follows: YP_771030 (pRL8JI), YP_001965662 (pSmeSM11b), YP_443809 (pNGR234a), NP_066708 (pRi1724), YP_001961066 (pRi2659), YP_001314081 (pSmed02), YP_001965661 (pSmeSM11b_TrbEb), YP_002540109 (pTiS4), NP_396555 (pTiC58), YP_002551329 (pAtK84b), YP_002542722 (pAtS4c), YP_002978853 (pR13205), NP_659891 (pRetCFN42d), YP_001984424 (pRetCIAT652b), YP_001985528 (pRetCIAT652c), NP_435962 (pSymA), AEH82016 (pSmeSM11c_pC0943), NP_396095 (pAtC58), YP_002823234 (pNGR234b), YP_005724130 (pSmeSM11c_pC0248), YP_001314917 (pSmed03), YP_002546488 (pATS4a), ABG61161(p1_BNC1, Meso_4177), YP_002283974 (pRLG203), and ACY71077 (pRleVF39b). (C) Phylogenetic tree of the coupling protein TraG. The accession numbers of the TraG coupling proteins used in this tree are as follows: YP_002551442 (pATS4b), YP_0025440047 (pTiS4), YP_002539497 (pAtS4e), YP_002542667 (pAtS4c), NP_396647 (pTiC58), YP_002551266 (pAtK84b), YP_001965639 (pSmeSM11b), YP_771012 (pRL8JI), YP_002978884 (pR132503), YP_001961049 (pRi2659), NP_066690 (pRI1724), YP_001314097 (pSmed02), NP_443831 (pNGR234a), YP_471745 (pRetCFN42a), NP_396043 (pAtC58), YP_002823258 (pNGR234b), NP_435748 (pSymA), YP_001985505 (pRetCIAT652c), YP_001984445 (pRetCIAT652b), NP_659871 (pRetCFN42d), YP_665800 (p1_BNC1, Meso_4169), YP_002546479 (pAtS4a), YP_001314908 (pSmed03), YP_005724121 (pSmeSM11c, pC0239), ACI59591 (pRLG203), ACY71085 (pRleVF39b), YP_002973150 (pR132501), YP_002979545 (pR132505), YP_76071 (pRL12JI), YP_770497 (pRL10JI), YP_002984812 (pR132502), YP_771307 (pRL11JI), YP_086775 (pAgK84), and YP_003329338 (pSmeSM11a).
Isolation of the oriT-containing cosmids.Since the sequence of the pRleVF39b conjugation system was available, we carried out PCR using trbE internal primers to identify cosmids from a new genomic library. Four cosmids, pCosDB6, pCosGE7, pCosJB11, and pCosID2, were identified to give amplification signals. Some fragments from these cosmids were isolated and sequenced. Two fragments from pCosJB11 were found to contain sequence homologous to trbE and traA, respectively. Presumably, the sequence would contain the region between trbE and traA; therefore, it was selected for the subsequent in vitro transposon mutagenesis. The isolated pCosJB11 was confirmed to be mobilizable from VF39SMrecA to UBAPF2recA. Hence, it should contain the oriT.
The novel conjugation system is responsible for the self-transfer of pRleVF39b.To determine whether the conjugation system led to the self-transfer of pRleVF39b, mutagenesis studies were carried out in the otherwise plasmid-free A. tumefaciens UBAPF2 carrying pRleVF39b. However, the available UBAPF2(pRleVF39b::Tn5mob) carried the same antibiotic marker (Kmr) as the EZTn5<Kan-2>. For easy selection during gene replacement, a spectinomycin cassette from Tn5-M6 was swapped with the Kmr gene within Tn5mob on pRleVF39b, using pJQ173 as previously described (44). Eckhardt gel analyses of the resulting UBAPF2(pRleVF39b::Tn5-M6) colonies revealed that spontaneous deletions occur in the plasmid (see Fig. S1 in the supplemental material); hence, it was named UBAPF2(ΔpRleVF39b::Tn5-M6).
After gene replacement, conjugation experiments were carried out to examine the effect of each mutation. UBAPF2(ΔpRleVF39b::Tn5-M6) and mutants carrying EZTn5s within trbE, trbI, traG, orf16, trbR, orf19, and traA were used as the donors, whereas UBAPF2recAEry was used as the recipient (Table 2). It was revealed that disruption of either trbE, trbI, traG, or traA completely abolished the transfer of ΔpRleVF39b::Tn5-M6 (Table 2), whereas mutations in orf16 or orf19 did not affect the transfer of ΔpRleVF39b::Tn5-M6 (data not shown). Since the EZ::Tn5 insertions generate polar mutations, traA and traG are thus absolutely required for the transfer of pRleVF39b. Interestingly, mutation within the trbR gene allowed for almost a 100-fold increase in the transfer frequency of ΔpRleVF39b::Tn5-M6 from UBAPF2 to UBAPF2recA and a striking ca. 1,000-fold increase from R. leguminosarum to UBAPF2recA (Table 2), implying a negative regulatory role of TrbR.
Genetic structure of the novel plasmid conjugation system.To reveal whether traG and its upstream trb genes were within the same operon structure, transcript analyses were performed. Total RNA was isolated from a 2-day-old mating spot with UBAPF2(pRleVF39b::Tn5mob) and UBAPF2recA. cDNA was then synthesized using the traG reverse primer (see Table S1 in the supplemental material). The synthesized cDNA was used as a template for PCRs targeting internal fragments within traG and different upstream trb genes, including trbI, trbF, trbL, trbC, and trbN (see Table S1). Amplification signals were detected for traG, trbI, trbF, trbL, and trbC (Fig. 3A). No amplification signals were detected when the total RNA was used as the template, confirming that no genomic DNA contamination was present in the RNA preparation. These results indicated that traG is within the trb operon. Similar transcript analysis was carried out for traA and its upstream genes, revealing that traA is monocistronic (Fig. 3B) (see Table S1 for the primers used for this study).
Agarose gel electrophoresis pictures of the PCR products from the transcript analyses of the trb operon (A) and the traA (B). The total RNA was isolated from a 2-day-old mating spot containing UBAPF2(pRleVF39b::Tn5mob) and UBAPF2recA. The cDNA was synthesized by reverse transcription from the total RNA using the traG reverse primer and traA reverse primer for the trb operon and the traA transcript analyses, respectively. For the trb operon transcript analysis, PCRs were carried out targeting internal fragments of traG, trbI, trbF, trbL, and trbC, whereas for the traA transcript analysis, PCRs were carried out targeting internal fragment of traA, orf28, orf27, orf26, and orf25. The expected sizes (bp) of the PCR products are listed in the parentheses. The positive amplification bands with the expected size are indicated by white arrows. PCRs using total RNA (without reverse transcriptase [−RT]) as a template were included as a negative control (i.e., there was no genomic DNA contamination in the RNA preparation). In the traA transcript analysis, genomic DNA was also used as a positive control, showing that the PCR conditions were feasible.
Isolation of clones containing the origin of transfer (oriT).The conjugation system on pRleVF39b does not closely resemble any studied rhizobial conjugation systems in terms of similarity of predicted proteins or gene arrangement, particularly the genes encoding the Dtr components. Therefore, predictions about the oriT location could not easily be made. To identify the oriT, we took an in vivo experimental approach based on the definition of oriT, which is able to convert a nontransmissible vector into a mobilizable plasmid.
It was confirmed that pCosJB11 was mobilizable from VF39SMrecA to UBAPF2recA. Since oriT often is located in the noncoding region in the vicinity of transfer genes, we predicted its location should lie between traG and traA. Fragments within the region between traA and traG were subcloned into the broad-host-range vector pBBR1MCS-5 (45), resulting in pBBR-HD constructs (Table 1). To simplify the subcloning process, we took advantage of the pCos138::EZTn5 mutants for which we had identified the insertion sites, including EZTn5s inserted with the traG, trbR, and traA genes. Subcloning of the fragments containing the EZTn5 and flanking region would allow for easy selection of the kanamycin-resistant fragments. A graphic representation of the fragments subcloned in pBBR1MCS-5 is shown in Fig. 4. Each pBBR-HD construct was then mobilized into VF39SMrecA. Bacterial matings between VF39SMrecA carrying the pBBR1MCS-5 and pBBR-HD constructs and UBAPF2recA were carried out, and the mobilization frequency of each pBBR1MCS5 with a possible oriT-containing fragment was calculated (Fig. 4). It was found that pBBR1MCS5, pBBR-HD1, pBBR-HD2, and pBBR-HD6 were not mobilizable. Interestingly, two nonoverlapping regions were found to be mobilizable by VF39SMrecA. One is contained by pBBR-HD3 (positions 14637 to 20460 in GenBank accession no. HQ711365) in the vicinity of the trbR gene, and the other is covered by fragments pBBR-HD4 and pBBR-HD5 (positions 23718 to 25531 in GenBank accession no. HQ711365) in the vicinity of the traA gene. Sequence comparison of previously identified nic sites from rhizobial and agrobacterial plasmids and a search within the two nonoverlapping regions identified putative oriT nic sites (nic1 positions 19280 to 19299 and nic2 positions 24349 to 24332) within each of the two regions. The sequence alignment of these two putative nic sites with other rhizobial and agrobacterial nic sites is shown in Fig. S2 in the supplemental material. A core 7-bp TTGCGCC sequence with the nic site (as indicated by the inverted triangle in Fig. S2) was conserved among these oriTs.
Identification of the origin of transfer (oriT) through in vivo mobilization experiments. Mobilization of pBBR-HD and pBBR-nic constructs from VF39SMrecA into UBAPF2recA. Each fragment cloned in pBBR1MCS-5 is represented as a black line, and its mobilization frequency (per recipient) from VF39SMrecA into UBAPF2recA is shown in parentheses. n.d., not detected. The gray vertical dotted lines indicate the possible regions containing oriT.
To confirm the functionality of both proposed nic sites, PCR primers (see Table S1 in the supplemental material) were designed to amplify the flanking regions of the two proposed nic sites with approximate lengths of 200 bp, 1 kb, and 2 kb, designated nic1-1 (∼200 bp), nic1-2 (∼1 kb), nic1-3 (∼2 kb), nic2-1 (∼100 bp), nic2-2 (∼1 kb), and nic2-3 (∼2 kb). The amplified fragments were then subcloned into pBBR1MCS5 (24) to test their ability to convert nontransmissible pBBR1MCS5 into a mobilizable plasmid. Only pBBR1MCS5::nic2-3 was found to be mobilized by VF39SMrecA into UBAPF2recA, at a frequency of 6 × 10−7 per recipient (Fig. 4). The mobilization was found to be pRleVF39b-dependent since pBBR1MCS5::nic2-3 was not mobilizable when VF39SMrecA(−pRleVF39b) was the donor.
Transcriptional regulation of pRleVF39b transfer genes.To examine the transcriptional regulation of the transfer genes, we constructed transcriptional fusions in pFUS1, using the putative promoters of the trb operon and the traA gene to drive the expression of the reporter gene gusA. The resulting constructs, pFUS1::PtraA and pFUS1::Ptrb, were mobilized into different genetic backgrounds, and the activities of the reporter gene product, β-glucuronidase, were measured as an index of the promoter activity (Table 3).
Expression of pFUS1::Ptrb and pFUS1::PtraA in Rhizobium and Agrobacterium UBAPF2 backgrounds
Since pRleVF39b transfers at higher frequencies when UBAPF2(ΔpRleVF39b::Tn5-M6) was used as the donor than when R. leguminosarum VF39SM was the donor (Table 2), we measured the promoter activities of PtraA and Ptrb in VF39SM and UBAPF2(ΔpRleVF39b::Tn5-M6). It was found that the expression of both traA and the trb operon in UBAPF2(ΔpRleVF39b::Tn5-M6) is about 14-fold higher than their respective expression levels in R. leguminosarum VF39SM (Table 3).
In both the QS and RctA-repressed conjugation system, the transcriptional regulatory genes, such as traI/traR/traM, and rctA/rctB are found in the vicinity of the transfer genes on the same plasmid (in cis). Therefore, it was also of interest to determine whether pRleVF39b carries any regulatory genes. We measured the expression of traA and the trb operon in the absence of pRleVF39b—i.e., in VF39SM(−pRleVF39b) and UBAPF2. It was revealed that expression of Ptrb in VF39SM(−pRleVF39b) was over 200-fold higher than that in VF39SM, implying the presence of a transcriptional repressor on pRleVF39b (Table 3). In UBAPF2, the expression of Ptrb was found to be about 2-fold higher than that in UBAPF2(ΔpRleVF39b::Tn5-M6). The expression of PtraA was about 4-fold higher in the absence of pRleVF39b in R. leguminosarum VF39SM. However, in UBAPF2, PtraA was expressed only about 15% of the level seen in UBAPF2(ΔpRleVF39b::Tn5-M6). From the mutagenesis studies, it was found that the disruption of trbR led to a 1,000-fold increase and a 100-fold increase in pRleVF39b transfer in the Rhizobium and Agrobacterium donors, respectively (Table 2). BLASTX revealed that trbR encodes a 113-amino-acid helix-turn-helix xenobiotic response element (XRE) family-like transcription regulator. Therefore, it was speculated that trbR might encode a transcriptional repressor. We measured the expression of Ptrb and PtraA in VF39SMtrbR and UBAPF2(ΔpRleVF39btrbR::EZTn5). It was found that the expression levels of Ptrb in VF39SMtrbR and UBAPF2(ΔpRleVF39b::Tn5-M6trbR::EZTn5) were about 150-fold and 3-fold higher than in their respective parent strains, VF39SM and UBAPF2(ΔpRleVF39b::Tn5-M6). The expression of PtraA in VF39SMtrbR is about six times that in VF39SM. However, the expression level of PtraA in UBAPF2(ΔpRleVF39b::Tn5-M6trbR::EZTn5) is only about 5% of that in the wild-type UBAPF2(ΔpRleVF39b::Tn5-M6).
DISCUSSION
The two smallest plasmids of VF39SM were known to be transmissible or mobilizable from VF39SM to A. tumefaciens UBAPF2 (21). However, in a VF39SM background, no definite conclusions could be made as to whether these two plasmids self-transfer, due to the possibility of interference from other resident plasmids providing in trans mobilization or mediating transfer by cointegration. We reported earlier that both pRleVF39a and pRleVF39b were self-transmissible by using the otherwise plasmid-free donor UBAPF2 (8). In this study, we continued to use UBAPF2 as the donor strain to study the self-transfer of pRleVF39b, as well as to avoid any possible interference of plasmid transfer by other plasmids.
It was found that pRleVF39b transferred at higher frequencies when donated by UBAPF2 than by VF39SM, when conjugation frequencies were measured per recipient cell plated. When frequencies per donor were examined, there was no significant difference, but we attribute this to much more rapid growth of the Agrobacterium donors in the mating spots. Transcription levels of the transfer genes (traA and the trb operon) in UBAPF2 were found to be about 14-fold higher than in VF39SM, which is consistent with the transfer experiments. When UBAPF2(ΔpRleVF39b::Tn5-M6) was used as the donor strain, different recipients also affected the conjugative transfer of pRleVF39b. When UBAPF2 was used as the recipient, the transfer frequency of pRleVF39b was about 100-fold higher than when VF39SM(−pRleVF39b) was the recipient.
To our knowledge, there are very limited studies demonstrating the effect of recipients on plasmid conjugation in Gram-negative bacteria. One study showed that recipient-produced quorum-sensing molecules induce the transfer of pRL1JI in R. leguminosarum through a QS signaling relay (46). This should not be the case for our experiments, as the plasmid-free UBAPF2 lacks any QS system. Studies using the F plasmid system have identified that mutations affecting recipient cell surface structure (such as lipopolysaccharide [LPS] production and OmpA outer membrane) lead to defects in mating bridge formation in liquid matings (47). However, the LPS or OmpA mutants were found to have only minor effects on surface conjugation (48). The difference in pRleVF39b transfer when different recipients were used could be due to the differences in the cell surface structure between UBAPF2 and Rhizobium, potentially related to the presence of five other plasmids in VF39SM.
The genome sequencing of VF39SM (9) allowed the identification of transfer genes responsible for pRleVF39b transfer. Sequence analysis revealed the presence of a closely related system, which has not yet been characterized or demonstrated to be functional, on pRLG203, although we have confirmed in our lab that pRLG203 is transmissible by conjugation (H. Ding and M. F. Hynes, unpublished data). The putative transfer genes encoding a type IV secretion system on pRLG203 were annotated as virB genes, which is misleading since there is no evidence that these genes are responsible for virulence. We therefore annotated our type IV secretion system as trb genes instead.
In a recent study, the relaxase from pSmeLPU88b of Ensifer meliloti LPU88 was characterized by Giusti et al. (19). Phylogenetic analyses showed that this relaxase together with relaxases from pSmed03 of S. medicae WSM419, pAtS4a from Agrobacterium vitis S4, plasmid 1 of Chelativorans sp. strain BNC1(p1_BNC1), and pRleVF39b form one distinct branch from the previously proposed type I, II, and III relaxases (8). Therefore, they classified these as the type IV rhizobial relaxase. It was also shown that the type I, II, and III rhizobial relaxases belonged to the MOBQ relaxase family, whereas type IV relaxases are more related to the MOBP family. In this work, we showed that all type IV relaxases have their associated Mpf genes organized in an almost identical arrangement (Fig. 1B). However, it is noticeable from the relaxase phylogenetic trees, both as presented by Giusti et al. (19) and in this article, that the pRleVF39b relaxase is clearly on a different branch from the rest of the type IV relaxases. The organization of the pRleVF39b genes in the vicinity of relaxase is also different from those in other type IV systems. The phylogenies of TraG and TrbE/VirB4 (Fig. 2B and C) are also consistent with the relaxase phylogeny, with distinct type I, II, III, and IV (IVA and IVB) systems. Based on the similarity and the differences, we named the pRleVF39b-borne and the pRLG203-borne conjugation systems as type IVA systems, and classify the remainder as the type IVB system.
The type IV systems all have their coupling protein gene, traG, located immediately downstream of the trb operon. Transcriptional analysis confirmed that traG is within the trb operon. This is another difference between the type IV system and the type I, II, and III systems, in which traG is often found in the vicinity of the relaxase gene, suggesting that traG expression in type I, II, and III systems might be coordinated with the expression of the relaxase. The coupling protein TraG plays an important role in connecting the Dtr components and the Mpf components. The differential coordination of their expression with the Dtr and Mpf components implies that type IV conjugation systems are regulated differently from the type I, II, and III systems.
Through mutagenesis studies, we showed that the trb genes, the coupling protein gene traG, and the relaxase gene traA are all required for the self-transfer of pRleVF39b. In the type I, II, III, and IVA rhizobial conjugation systems, the origins of transfer (oriT genes) are all located between the two divergently transcribed transcripts, traA(FBH) and traCDG. In the type IVB system, however, there is no such analogous region. In addition, sequence identity between the pRleVF39b-borne system and the rest of the conjugation systems in the NCBI database was too low to allow for meaningful in silico predictions of the oriT location. In vivo mobilization experiments have allowed for the identification of two nonoverlapping regions that could convert pBBR1MCS-5 into a mobilizable plasmid. Further investigation of these regions by PCR amplification of the regions inside these nonoverlapping clones identified that one of these oriT sites may be located upstream of traA and does not correspond to the potential oriT site identified in Fig S2 in the supplemental material. It is also possible that it lies on a short region between the end of clone nic2-2 and the pBBR-HD4 clone, although this seems less likely. We have been unable to identify any sequences resembling known oriT nicking sites in either of these regions, so further experimental work will be required to confirm them. The location of the other oriT site on the large subclone pBBRHD-3 remains unclear but also does not correlate to the predicted nicking site shown in the alignment in Fig S2. Since this clone is very large and contains a number of genes of unknown function, it remains uncertain whether it carries a genuine oriT that is recognized by the pRleVF39b conjugation system, whereas the other putative oriT, most likely upstream of traA, definitely requires the presence of pRleVF39b for mobilization.
Through mutagenesis studies, we were also able to identify a putative repressor gene, trbR, which negatively affects the transfer of pRleVF39b in both VF39SM and UBAPF2 donor backgrounds. Using plasmid-borne transcriptional fusion assays, we demonstrated that the expression of the trb operon is repressed by TrbR under experimental conditions (Table 3). However, the trbR product seems to enhance the transcription of traA in UBAPF2(ΔpRleVF39b::Tn5-M6) (Table 3). This is inconsistent with the results of transfer experiments. There might be several explanations for this. First, plasmid conjugative transfer is believed to occur only when conditions are favorable. The Dtr genes might therefore only be expressed for a short period during conjugation in response to environmental or internal physiological signals. It is possible that at the time point at which pFUS1::PtraA measurements were taken, the cells were not expressing the traA gene. Therefore, the use of reporter gene fusions to measure this transient transcription within a population of cells might not be appropriate. On the other hand, the trb operon encodes the Mpf component on the membranes of the donor cell, which could be viewed as the “prerequisite” for plasmid conjugation. When conditions are favorable, expression of Mpf component genes would allow the donor population enter a “standby mode,” ready to receive signals that would trigger the expression of the Dtr genes. Second, given the low overall activities of all of the traA fusions it is possible that traA might have a second promoter that is responsive to trbR, and that we are not seeing the whole story. Alternatively, there may be another regulator produced in the VF39SM background that is necessary for TrbR to function as a repressor at the traA promoter, and in the absence of this regulator, it actually acts as an activator. Some Xre-type proteins, such as the QseC regulator of conjugation in Mesorhizobium loti, are known to act as both activators and repressors (49).
TrbR is predicted to be a xenobiotic response element (XRE)-like family transcriptional regulator, implying that environmental substances might directly trigger the transfer of pRleVF39b mediated by TrbR, although pea seed exudates did not seem to play such a role. Since R. leguminosarum VF39SM nodulates a variety of legume plants, we would not be able to rule out the possibility that signals from other host plants might interact with TrbR.
In the NCBI database, the only conjugation system closely related (∼80% identity at the protein level) to the pRleVF39b-borne system is on pRLG203 of R. leguminosarum bv. trifolii WSM2304. Our lab has a large collection of rhizobial field isolates, including members from the genera Rhizobium and Ensifer. Hybridization of the trbE internal probe to the blots of Eckhardt gels showing the plasmid profiles of rhizobial strains were carried out under high-stringency conditions (>90% identity at the nucleotide level). A total of five Ensifer meliloti strains, 12 R. leguminosarum strains, and 13 Rhizobium gallicum strains were used in this screen. It was revealed that the trbE gene hybridized to the second or third smallest plasmid (which comigrate in the Eckhardt gel) in R. leguminosarum 309 and several plasmids of R. gallicum strains, including strains SO14B4, SO13A1, SO19B5, SO13B2, and SO07A5 (data not shown). The identification of conjugation systems similar to that encoded by pRleVF39b in various Canadian R. gallicum field isolates implies that this conjugation system is not restricted geographically, since R. leguminosarum bv. viciae VF39SM is a European isolate; R. leguminosarum bv. trifolii WSM2304 was isolated from Uruguay, whereas, all of the R. gallicum strains were field isolates from 30 sites within a 250-km region of Saskatoon, Saskatchewan, Canada.
ACKNOWLEDGMENTS
The work was supported by an NSERC Discovery grant to M.F.H. H.D.'s studies were partially supported by a scholarship from the Dean of Graduate Studies, University of Calgary.
The technical assistance by Stacy Bruce in constructing pRK415::sacB is acknowledged, as is the aid of Glen Ong in cosmid library construction. We thank S. Farrand for the gift of pJM54 and A. Schlüter for pMS102.
FOOTNOTES
- Received 17 July 2012.
- Accepted 3 November 2012.
- Accepted manuscript posted online 9 November 2012.
Supplemental material for this article may be found at http://dx.doi.org/10.1128/JB.01234-12.
- Copyright © 2013, American Society for Microbiology. All Rights Reserved.
