The conjugative plasmid of a bean-nodulating Sinorhizobium fredii strain is assembled from sequences of two Rhizobium plasmids and the chromosome of a Sinorhizobiumstrain

Plasmid pSfr64b is required for symbiosis but pSfr64a is dispensable

Strain GR64 contains two plasmids: pSfr64a (183 kb) and pSfr64b (~400 kb) (Figure 1A, Table 1). A band corresponding to a megaplasmid (~1300 kb), has been visualized [13], but is not always clearly apparent in the gels. Plasmid pSfr64b was identified as the symbiotic plasmid [13], because it hybridizes with the nifH gene. Nodulation assays confirmed that the genetic information in pSfr64b is necessary and sufficient to establish symbiosis. Table 2 shows that all derivatives carrying pSfr64b, were able to form nodules (GR64, CFN2001-1, GMI9023/pSfr64b), and that the construct lacking pSfr64b (GR64-4) was unable to nodulate beans. Consistent with previous findings [14, 15], the number of nodules was decreased in an Agrobacterium genomic background. On the other hand, lack of pSfr64a had no effect on the symbiotic process (GR64-2), and its presence in Agrobacterium did not confer nodulation capacity to the receptor, indicating that pSfr64a encodes none of the essential symbiotic genes.

Plasmid pSfr64a shares sequences with the R. etli pSym, pRet42a, and with the chromosome of Sinorhizobium frediiNGR234

We sequenced plasmid pSfr64a (GenBank accession number: CP002245). The main features of this plasmid are shown in Figure 2 and Additional File 1. Plasmid pSfr64a is 183 612 bp long. The genetic organization of this plasmid clearly reveals its chimeric nature, since 38 (23%) of the 166 ORFs encoded in the plasmid presented highest similarity to sequences of the chromosome of Sinorhizobium fredii NGR234, while 87 (52%) were most similar to ORFs encoded in R. etli CFN42 plasmids pRet42a (36 ORFs, 22%) and pRet42d (51 ORFs, 31%).

Among the ORFs shared between pSfr64a and pRet42a, the self-transmissible plasmid of CFN42, most are related to conjugative transfer (20 ORFs), only two were ascribed to macromolecular metabolism. Interestingly, both are related to DNA metabolism, one was classified as a putative nuclease, and the other as a probable DNA methylase. In Figure 3, it can be appreciated that the genomic region shared between pRet42a and pSfr64a is markedly colinear. Colinearity is disrupted by the absence of an homolog to the regulatory gene cinR of pRet42a, and the presence of pSfr64a ORFs 147 and 148, which encode hypothetical proteins. The correspondence between pSfr64a and pRetCFN42 ORFs is presented in Additional File 1. Figure 2 shows that the segment of pSfr64a shared with pRet42a has a high GC content, compared to the rest of the plasmid. This feature is also present in the similar pRet42a sequence.

The ORFs similar to the pSym of CFN42 (pRet42d) include the repABC genes (Figure 2, Table 3). This is congruent with our finding that pSfr64a and pRet42d are incompatible (data not shown). The pSfr64a-pRet42d-shared ORFs are mainly involved in small molecule metabolism (26 ORFs), and carbohydrate transport (13 ORFs). It is noteworthy that, in spite of the fact that pRet42d carries genes engaged in symbiotic functions, none of these are present in pSfr64a.

Within the region similar to pRet42d (ORFs 46 to 110), the colinearity is restricted to small segments, some of them in inverse orientation. (Figure 3, Additional file 1). The repABC genes (pSfr64a ORFs 164 to 166) were adjacent to the transfer region, separated from the other pRet42d genes. It has been amply documented that plasmid pRet42d is subject to frequent genomic rearrangements, due to the presence of reiterations and a high density of insertion sequences [16–20]. R. etli ORFs encoding transposon-related proteins located near to the sites where colinearity is disrupted are indicated in Figure 2 (purple arrows) and Additional File 1. For example, pSfr64a ORFs 122 to 146 are colinear with pRet42a ORFs 139 to 162. The adjacent ORF on pRet42a (ORF 138) encodes a transposon-related protein. It is possible that these sequences are related to the generation of rearrangements, causing the interruptions in colinearity. ORFs 114, 115, 116, 117, 118 and 121 show homology to ORFs encoded in another Rhizobium etli strain; IE4771 [21]. The genome of this strain has been sequenced, but not assembled, so we cannot assign them to the pSym, although that is their most probable localization.

The majority of the ORFs shared between pSfr64a and the chromosome of NGR234 are related to small molecule metabolism (15 ORFs), and to the transport of small molecules (11 ORFs). As shown in Figure 3 and Additional File 1 this region is also highly colinear with the corresponding genes on the chromosome of NGR234. Data presented in this section suggest that pSfr64a was assembled during evolution as a chimeric structure, harboring segments from two separate R. etli plasmids and the chromosome of a Sinorhizobium strain, such as NGR234.

Plasmid pSfr64a is transmissible and required for transfer of pSfr64b

Genomic background determines functionality of conjugative plasmids

In order to assess the specificity of pSym transfer induction, we constructed derivatives containing diverse plasmid combinations, in either R. etli or S. fredii genomic backgrounds, as described in Materials and Methods, and determined the transfer frequency of the self-transmissible and symbiotic plasmids (Table 4). Analysis of a derivative containing the R. etli self-transmissible plasmid pRet42a in S. fredii background (GR64-3) showed a dramatic decrease in the transfer ability of the plasmid as well as no transfer of the GR64 pSym. These results suggest that the genome of GR64 contains an inhibitor of pRet42a transfer. The decrease in pRet42a transfer could mask its function as helper for pSym transfer induction. The fact that pRet42a transfer is also decreased in a derivative lacking the pSym of GR64 (GR64-5), points to a chromosomal location of the putative inhibitor locus. Similarly, S. fredii pSfr64a was unable to perform conjugative transfer or induce transfer of pSfr64b in R. etli genomic background (CFN2001-3). Only R. etli pRet42a was still able to induce pSfr64b transfer in the R. etli background (CFN2001-2).

The pSym of GR64 differs from the typical R. etlipSym

To further analyze the bean-nodulating S. fredii strain GR64, we performed a phylogenetic analysis with chromosomal genes (recA, rpoB), and with the plasmid-encoded genes nifH and repB. The results (Figure 4) show that, based on the phylogeny of the chromosomal genes, GR64 clusters within the fredii clade, while nifH and repB genes group strain GR64 with other bean-nodulating Sinorhizobium strains isolated from the South of Spain (Granada and Sevilla) [22, 23] and from the North of Africa (Tunisia) [24] (Figure 4C). The data obtained indicate that GR64 has a S. fredii chromosome but carries a pSym that allows nodulation of Phaseolus. However, this plasmid differs from typical R. etli pSyms in its replication genes, allowing it to coexist with plasmid pSfr64a, which does share its replication genes with the R. etli pSym. Another feature that differentiates this pSym is the presence of a single copy of the nifH gene.

Bacterial strains and plasmids

The bacterial strains and plasmids used in this work are described in Table 1. R. etli strains were grown at 30°C on PY medium [33]. Escherichia coli and Agrobacterium tumefaciens strains were grown on Luria-Bertani (LB) medium [34] at 37°C and 30°C respectively. When required, antibiotics were added at the following concentrations (in μg ml^-1): nalidixic acid (Nal) 20, spectinomycin (Sp) 75, kanamycin (Km) 15, neomycin (Nm) 60, rifampicin (Rif) 100, streptomycin (Str) 50, tetracycline (Tc) 2 for Rhizobium and 10 for E. coli, gentamicin (Gm), and erythromycin (Ery) 100.

Genetic manipulations

Conjugation experiments were performed on PY plates at 30°C, using overnight cultures grown to stationary phase. Donors and recipients were mixed in a 1:2 ratio and incubated overnight. The mixtures were collected and suspended in 1 ml of 10 mM MgSO₄-0.01% Tween 40 (vol/vol). Serial dilutions were plated on suitable selective media. The transfer frequency was expressed as the number of transconjugants per donor.

A derivative of GR64 carrying a Tn5mob-labeled pSym was constructed by mating GR64 with strain S-17/pSUP5011 and selecting for resistance to Nal and Nm. Tagged plasmids were mobilized to A. tumefaciens GMI9023 [35] in triparental crosses, using pRK2013 [36] as helper, and selecting for Rif^R Nm^R transconjugants. Transconjugants carrying the tagged pSym (pSfr64b) were identified using Eckhardt type gels.

To determine the presence of transmissible plasmids, we randomly labeled strain GR64 with Tn5-GDYN, by mating it with E. coli S17/Tn5-GDYN [17] and selecting Nal^R Sp^R transconjugants. The labeled transconjugants were used as donors in conjugations with A. tumefaciens strain GMI9023. As the transposon integrates randomly into the chromosome or plasmids present in a strain, its integration into a transmissible plasmid confers a selective marker to the plasmid. Plasmids present in the selected transconjugants were visualized with Eckhardt gels.

The Tn5-GDYN element contains the sacR-sacB genes, which confer sucrose sensitivity in several gram-negative bacteria, so that selection of sucrose-resistant colonies allows the isolation of plasmid-less derivatives [17]. Plasmid-curing was carried out by plating overnight cultures of the transposon-labeled strains on PY plates containing 12.5% sucrose. Sucrose-resistant colonies were selected and verified as Sp^S. Plasmid profiles of such colonies were analyzed in Eckhardt type gels.

Construction of S. fredii and R etliderivatives with diverse plasmid content

We constructed various derivatives of GR64 (Table 1): GR64-1 has pSfr64a labeled with Tn5-GDYN and pSfr64b with Tn5mob. This construct allowed us to obtain a derivative cured of pSfr64a (GR64-2). The absence of pSfr64a in GR64-2 was confirmed by Southern type hybridization of plasmid profiles probed with purified pSfr64a (Figure 1B), and of total restricted DNA (data not shown). Tn5-GDYN-labeled-pRet42a from R. etli CFN42 was introduced into GR64-2 to generate GR64-3. A derivative of GR64-2 with a Tn5-GDYN inserted in pSfr64b was constructed. This strain was used to generate GR64-4, cured of both plasmids. Tn5-GDYN-labeled-pRet42a from R. etli CFN42 was introduced into GR64-4 to generate GR64-5. To construct GR64-6, Tn5-GDYN-labeled-pSfr64a was introduced into GR64-4. CFN2001 is a derivative of R. etli CFN42 that lacks the pSym (pRet42d), and the self-transmissible plasmid pRet42a [37]. This strain was used as receptor to select transconjugants carrying the Tn5mob-labeled pSym of GR64 (CFN2001-1), the Tn5mob-labeled pSym of GR64 and Tn5-GDYN-labeled pRet42a of R. etli CFN42 (CFN2001-2), and both plasmids of GR64 (CFN2001-3). Other derivatives carried either pSfr64a::Tn5-GDYN or pSfr64b::Tn5mob in Agrobacterium strain GMI9023 genomic background.

Plasmid profiles

Plasmid profiles were visualized by the Eckhardt technique [38], as modified by Hynes and McGregor [39].

Filter blot hybridization and plasmid visualization

For Southern-type hybridizations [40], Eckhardt type gels, or 1% agarose gels where restricted DNA was electrophoresed, were blotted onto nylon membranes, and hybridized under stringent conditions, as previously reported [41], by using Rapid-hyb buffer. Probes were linearized by digesting them with appropriate restriction enzymes and were labeled with [α³²P]dCTP by using a Rediprime DNA labeling system. All restriction endonucleases, [α -³²P]dCTP, hybridization buffer, and labeling systems were purchased from Amersham Pharmacia Biotech.

Nodulation assays

Overnight cultures were used to inoculate surface-sterilized Phaseolus vulgaris cv. Negro Jamapa seeds. Plants were grown in 250-ml Erlenmeyer flasks with Fahraeus agar medium [42], without added nitrogen, at 28°C. Nodulation was scored at day 15 after inoculation. Surface-sterilized nodules were crushed on PY plates, and the plasmid pattern of single colonies was checked on Eckhardt type gels.

Amplification and sequencing of recA, rpoB, and nifHgene fragments

Partial nifH, recA and rpoB fragments were amplified with the primer pairs nifH40F/nifH817R, recA41F/recA640R and rpoB454F/rpoB 1364R as previously described [43, 44]. All amplifications were performed with Taq polymerase (USB-Amersham). Amplification products were purified using Roche's PCR product purification system. Both strands were commercially sequenced by Macrogen, Korea.

Phylogenetic inference

Reference nifH, recA, rpoB and repB sequences were retrieved via BLASTP searches from a locally maintained BLAST database containing all fully sequenced Rhizobiales genomes, and via remote BLASTP searches against NCBI's non-redundant database. The query sequences for nifH, recA and rpoB used in the BLASTP searches were those obtained from the sequenced PCR amplicons from strain GR64, while that of repB was obtained from the sequence of pSfr64a. Nucleotide sequences were translated and aligned using muscle 3.7 [45]. The resulting protein multiple sequence alignments were used as masks to generate the underlying codon alignments using custom Perl scripts.

Models of nucleotide substitution were selected by the Akaike information criterion (AIC), using MODELTEST3.7 [46]. Among-site rate variation was modelled by a gamma distribution, approximated with 4 rate categories, each category being represented by its mean. Maximum likelihood (ML) trees were inferred under the AIC-selected models of nucleotide substitution for each data set using PhyML v3.0.1 [47]. The robustness of the ML topologies was evaluated using a recently developed Shimodaira-Hasegawa-like test for branches implemented in PhyML v3.0.1 [47]. For the sake of clarity, a small selection of the most relevant sequences was performed to show herein, based on the results of the phylogenetic analysis with the full set of homologous sequences.

Sequencing of plasmid pSfr64a

Plasmid pSfr64a was purified by the Hirsch method [48], and used to construct a shotgun library with inserts of approximately 1-2 kb. A total of 1970 high-quality readings were collected by using the ABI3730XL automatic DNA sequencing machine (Applied Biosystems, Foster City, CA). Gaps were filled in by performing appropriate PCR amplification. Assemblages were obtained by the PhredPhrap-Consed software [49–51]. The quality of the final assembly was less than 1 error per 100,000 bases and had an average coverage of 6.5X.

Annotation

Open reading frames were predicted by using GLIMMER 3.0 [52, 53] and annotation was carried out with the help of BLASTX [54] comparisons against the GenBank nonredundant database [55], INTERPRO [56] searches, and manual curation by using ARTEMIS [57]. To compare partial genomic sequences with the nonredundant database of GenBank, BLASTX searches were performed, and the top hits were classified with respect to organisms with which they matched.

Nucleotide sequence accesion number

Plasmid pSfr64a accession number is GenBank: CP002245. GR64 nifH, recA, and rpoB accesion numbers are respectively GenBank: JN034672, JN034673, JN034674.