pSymA-Dependent Mobilization of the Sinorhizobium meliloti pSymB Megaplasmid

Bacterial species belonging to the family Rhizobiaceae usually carry large plasmids encoding proteins with diverse functions that determine their lifestyles in the soil and rhizosphere. In many rhizobia, the genes needed to establish nitrogen-fixing symbiotic associations with leguminous plants are located in the so-called symbiotic plasmids (pSyms). Conjugative transfer of these plasmids has probably been of great importance for the evolution of the rhizobia. Conjugative plasmid transfer among rhizobia has frequently been reported; however, most rhizobial plasmids transfer at low or undetectable levels under laboratory conditions, probably due to the existence of tight regulation systems. In studies of these conjugal transfer regulatory mechanisms, two rhizobial plasmid classes have been reported: quorum sensing (QS)- and rctA-regulated plasmids (reviewed in reference 2). The first class includes diverse plasmids from various species in which plasmid transfer occurs in response to the population density, similar to the Ti plasmid of Agrobacterium tumefaciens. In contrast, transfer of the rctA-regulated plasmids, such as Rhizobium etli pRetCFN42d and pSymA of Sinorhizobium meliloti, is repressed by the transcriptional regulator RctA (10, 18).

S. meliloti 1021 carries two megaplasmids: pSymA (1,354 kb) contains most of the genes needed for nodulation and symbiotic nitrogen fixation, whereas pSymB (1,683 kb) harbors exopolysaccharide biosynthetic genes, which are also required for the establishment of symbiosis. However, both plasmids also encode functions for the utilization of various substrates as nitrogen or carbon sources and are therefore important for bacterial fitness in the rhizosphere (1, 3, 9). Furthermore, pSymB has been referred to as a second chromosome since it carries genes encoding proteins with essential housekeeping functions (5), despite carrying genes encoding proteins with typical plasmid replication functions (20). Regarding conjugal transfer, pSymA carries an oriT and traA1, traCDG, and virB1-virB11 operons encoding all necessary functions for self-transmissibility. Although indirect evidence for potential transfer had been reported (12), under common laboratory conditions, pSymA transfer is undetectable because of the repression of conjugation genes by a transcriptional repressor encoded by the rctA gene. Thus, pSymA transfer is only observed under conditions that relieve this repression, as in rctA mutants (7, 10). In contrast, pSymB conjugal transfer has never been demonstrated. The only conjugation-related functions identified in the genome sequence of this megaplasmid are a traA2 gene encoding a hypothetical relaxase that is homologous to traA1 from pSymA and a hypothetical oriT that is homologous to the pSymA oriT sequence.

Here, we report that pSymB is a mobilizable plasmid that relies on pSymA functions for conjugal transfer.

We have previously shown that the conjugal transfer of megaplasmid pSymA from S. meliloti 1021 is repressed by the transcriptional regulator RctA under laboratory conditions (10). In the absence of RctA, the repression of tra and virB gene transcription is relieved, allowing plasmid conjugation to a suitable recipient (7, 19). In contrast to pSymA, the only conjugation-related functions identified in pSymB are the traA2 gene and an adjacent putative oriT, resembling a typical organization in mobilizable plasmids (15). A number of additional indirect lines of evidence indicated that pSymB could be a mobilizable plasmid. TraA2 displays 97% sequence identity to TraA1, whereas the respective oriT regions (300-bp sequences around the hypothetical nic sites) of the S. meliloti symbiotic plasmids pSymA and pSymB display 95% sequence identity. Furthermore, the 300-bp fragment spanning the pSymB oriT could be mobilized from a rctA mutant of R. etli (11).

Before testing pSymB mobilization, we examined whether traA2 expression could also be subjected to repression by RctA. A traA2::gusA fusion was constructed by cloning an EcoRI/HindIII fragment from plasmid pJBSB (11), containing the traA2 promoter region, into vector pFUS1 (14), yielding pFustraA2::gus. This construction was introduced into strains 1021 and 1021RctA⁻ by conjugation, using the Escherichia coli mobilizing strain S17-1, and β-glucuronidase activity was assayed as previously described (10). The expression of the traA2::gusA fusion was 2.6-fold higher in an rctA mutant than in the wild-type strain (35.7 versus 13.5 β-glucuronidase units), showing that, similar to the expression of the traA1 and virB genes from pSymA, pSymB traA2 expression is also repressed by RctA.

To test pSymB mobilization from an rctA pSymA mutant, pSymB in strains 1021RctA⁻ and 10OTSS (a 1021 derivative carrying a Sm/Sp resistance cassette in the otsA gene in pSymA; 10) was tagged by transduction of the gentamicin resistance gene inserted into pSymB of the pSymA-cured strain A818::Gm (13), a derivative of SmA818 carrying a Tn5-M8S insertion into pSymB (9, 13). Transduction was performed as previously described (4), thus obtaining strains 1021RctAHB and 10OTSSHB. Conjugative transfer of pSymA and pSymB was tested in matings with the plasmidless strain A. tumefaciens GMI9023 as the recipient (16). As expected, no plasmid transfer from strain 10OTSSHB was detected. However, conjugal transfer of both pSymA- and pSymB-associated markers from the rctA strain 1021RctAHB was readily observed (Table 1). The level of pSymA transfer was nearly 1 log higher than that of pSymB, whereas simultaneous transfer of both megaplasmids occurred at very low frequencies (Table 1). To verify that transfer of the intact plasmids had occurred, the genomic DNAs of 8 randomly chosen transconjugants of each class were used for PCR amplification of several plasmid-specific markers (SMa1077 and SMa2402 from pSymA and SMb20094, SMb21314, and SMb20946 from pSymB). The location of these genes within each of the plasmids is shown in Fig. S1 in the supplemental material. The oligonucleotides used for PCR amplifications are described in Table 2. As expected, all pSymA transconjugants were positive for the two pSymA markers and negative for the pSymB ones, whereas all pSymB transconjugants amplified the three pSymB-specific gene fragments and were negative for the pSymA markers (data not shown). Among the transconjugants selected for simultaneous transfer of both pSymA and pSymB plasmids, 6 (75%) were positive for all pSymA and pSymB markers, indicating the presence of both intact plasmids. However, 2 of the 8 transconjugants (25%) amplified one marker from each plasmid, SMa1077 and SMb20094, but were negative for the other markers (see Fig. S2 in the supplemental material), which suggested the occurrence of rearrangements leading to the formation of hybrid plasmids in a minority fraction of this type of transconjugant.

Since pSymB carries no mating pore formation genes, it was reasonable to think that its transfer was dependent on pSymA-encoded conjugation functions like the virB-encoded type IV secretion system (T4SS). To test this hypothesis, we constructed an rctA virB double mutant derivative of 1021. A SMa1322-virB11 deletion was generated in vitro after the amplification (see Table 2 for primers used) and cloning into vector pK18mobsacB of an upstream 960-bp EcoRI/HindIII fragment and a downstream 1,130-bp HindIII/BamHI fragment flanking the pHP45ΩKm cassette (HindIII), yielding plasmid pK18-ΔvirB (8). This construction was introduced by conjugation into S. meliloti 1021, and allele replacement events were selected as described previously (17), yielding strain 1021RctAΔvirB. Later, the gentamicin resistance gene inserted into pSymB from strain A818::Gm was transduced into strain 1021RctAΔvirB to give strain 1021RctAΔvirBHB, which was used as the donor in matings with A. tumefaciens GMI9023. As expected, no conjugal transfer of pSymA from the rctA virB double mutant could be detected. Likewise, no pSymB mobilization was observed (Table 1), indicating that conjugal transfer of pSymB is totally dependent on the pSymA T4SS encoded by the virB operon. In summary, the results indicated that, similar to the transcription of the pSymA conjugal transfer genes, transcription of the traA2 gene in pSymB is also repressed by rctA, suggesting coregulation of conjugal transfer genes in both S. meliloti 1021 megaplasmids. Furthermore, we demonstrated for the first time conjugal transfer of S. meliloti pSymB, as a mobilization event that is dependent on the pSymA virB-encoded T4SS.

pSymB carries its own relaxase gene and oriT sequences that are very similar to the pSymA oriT, which would suggest an in trans mobilization. This was also supported by the fact that the transfer of pSymB alone was much more efficient than the simultaneous transfer of both pSymA and pSymB (Table 1). However, the transfer of pSymB seemed to be about 1 log less efficient than that of pSymA. To further investigate this difference, we compared the mobilization frequencies of the cloned oriTs from both pSyms. For this, two plasmid constructs containing 300-bp fragments spanning the oriTs from pSymA and pSymB (pJBSA and pJBSB, respectively) (11) were introduced into an rctA mutant. As shown by the results in Table 3, both plasmid oriTs were transferred to a plasmidless A. tumefaciens recipient at frequencies lower than 10⁻⁶, indicating that both oriTs are functional. The presence of the intact constructs in the transconjugants was verified after restriction enzyme or PCR analyses of the plasmid DNAs isolated from individual transconjugants. However, the mobilization frequencies of both cloned oriTs were 1 to 2 logs lower than the frequency of transfer of pSymA and severalfold lower than that of pSymB. This result resembled the situation of the cloned R. etli pSym oriT, which was mobilized much less efficiently than the full plasmid, owing to a cis-acting preference of the relaxase (11). Therefore, we constructed two plasmids harboring the complete traA1-oriT and traA2-oriT regions from pSymA and pSymB, respectively. The corresponding 5.1-kb and 5.3-kb DNA fragments were PCR amplified using oligonucleotides TA1O-F, TA1O-R, TA2O-F, and TA1O-R (Table 2). The resulting DNAs were digested with endonuclease BamHI and cloned into vector pJB3Tc19, resulting in plasmids pJBSAR (traA1-oriT) and pJBSBR (traA2-oriT). The mobilization of these two constructions from an rctA mutant was as efficient as the transfer of the full pSymA plasmid (Table 3) and 2 logs higher than the mobilization of the cloned oriTs with no traA in cis. This result provided evidence that both the traA1 and the traA2 gene are functional and that the corresponding proteins might have a nick site cis-acting preference, as was previously shown for the homologous pRetCFN42d relaxase (11). Similar to the transfer of pSymA and pSymB, no transfer of the cloned oriTs (with or without accompanying relaxase genes) was detected from an rctA virB double mutant (data not shown). All together, the results suggested that plasmid pSymB carries a functional relaxase gene and an oriT that could be mobilized from a pSymA rctA mutant, further supporting the possibility that pSymB mobilization by pSymA occurs by an in trans mechanism. The data also suggested that both plasmids' oriTs could potentially be mobilized with similar efficiencies and, therefore, that other unknown factors were responsible for the 1-log-less-efficient transfer of pSymB than of pSymA.

It has recently been reported that VBPs (VirD2-binding proteins) in A. tumefaciens are involved in the recruitment of the nucleoprotein substrate complex to the energizing components of the transport apparatus and, therefore, enhance the recruitment of the T-complex to the T4SS (6). Furthermore, VBPs are also involved in the recruitment of derivatives of plasmid RSF1010 (but not of RK2) during conjugation, thereby enhancing plasmid transfer efficiency. Two VBP homologs are encoded in the genome of S. meliloti 1021 (6): SMa0967 is encoded by a pSymA gene located near the tra region, and SMb20835 is encoded in pSymB. We hypothesized that these two proteins could be involved in the different transfer efficiencies of pSymA and pSymB. We obtained real-time RT-PCR expression profiles of SMa0967 and SMb20835 genes in S. meliloti strains 1021 and 1021RctA⁻. Aliquots (1 μg) of total RNA from S. meliloti strains 1021 and 1021RctA⁻ were reverse transcribed using Superscript II reverse transcriptase (Invitrogen) and random hexamers (Roche) as primers. The sequences of the primers used are listed in Table 2. We found that the expression of SMa0967 was increased by 2.3-fold in an rctA mutant compared to its expression in the wild type; however, the expression levels of SMb20835 were similar in wild-type and RctA⁻ strains. To test whether either of these two genes could have any role in pSym transfer efficiency, we constructed a derivative of strain 1021RctAHB carrying deleted versions of both the SMa0967 and the SMb20835 gene. Mutant 1021Δ0967RctA was first generated by deletion of the SMa0967 gene in 1021RctA. For this, a 1-kb fragment and a 0.9-kb fragment (see Table 2 for primers), upstream and downstream of SMa0967, respectively, were PCR amplified and later cloned as a HindIII/XbaI fragment into suicide vector pK18mobsacB, which was then used for allele exchange in strain 1021RctA⁻. Similar PCR amplification and cloning procedures were followed (see Table 2 for primers) to generate a deletion of SMb20835 in strain 1021Δ0967RctA, giving rise to mutant 1021ΔvbpRctA, in which both VBP-encoding genes were deleted. Later, the gentamicin resistance gene inserted into pSymB from strain A818::Gm was transferred into the 1021ΔvbpRctA genome by generalized transduction using phage ΦM12, yielding strain 1021ΔvbpRctAHB. pSym transfer from this strain into A. tumefaciens GMI9023 was compared with that from strain 1021RctAHB. No differences in pSymA or pSymB conjugal transfer were observed (data not shown), indicating that neither of these two genes encoding VBP homologous proteins are involved in the transfer efficiency of the pSyms.

In conclusion, we demonstrate for the first time that pSymB of S. meliloti 1021 is a mobilizable plasmid and that its conjugal transfer depends on functions like the T4SS encoded in pSymA. The pSymA-dependent mobilization of pSymB occurs in trans, since this megaplasmid carries a functional oriT that requires a relaxase gene in cis to achieve maximal transfer efficiency. Moreover, the transcription of traA2, which encodes the pSymB relaxase, is repressed by the pSymA-encoded RctA transcriptional regulator, suggesting the need to coregulate conjugative functions in both megaplasmids.

• Copyright © 2010 American Society for Microbiology