ABSTRACT
Pulsed-field gel electrophoresis and 16S rRNA hybridization experiments showed that multichromosome genome structures and very large genome sizes (6.46 to 8.73 Mb) are prevalent in novel plant-associated Burkholderia species. 15N2 isotope dilution assays revealed unambiguous diazotrophy in these novel species. nifH gene sequence analysis, often used to determine phylogenetic relatedness among diazotrophs, showed tight clusters of Burkholderia species, which were clearly distinct from those of other diazotrophs.
Multireplicon genomes occur in members of the alpha, beta, and gamma subdivisions of Proteobacteria, but most species in each subdivision contain only a single chromosome (4). Genome structure studies of Betaproteobacteria of the Burkholderia genus have been directed mainly to species included in the Burkholderia cepacia complex, opportunistic human pathogens having large genomes (6 to 9 Mb) consisting usually of two or three chromosomes (17). The more than 40 bacterial species presently classified as Burkholderia are ubiquitous in nature (10). Until recently, among Burkholderia species, only B. vietnamiensis, a member of the B. cepacia complex (10) that exists in the rhizospheres of several plants and inside of maize (11) and sugarcane (15), was recognized to participate in N2 fixation (13). Recently, B. kururiensis (33) was identified as a diazotroph (11), and two N2-fixing legume-nodulating strains (18) were formally classified as B. phymatum and B. tuberum (31). Novel diazotrophic Burkholderia species found among rhizospheric and endophytic isolates associated with important crop plants, including B. unamae (5), B. xenovorans (14), B. tropica (26), and B. silvatlantica (23), were recently described (5, 6, 14, 23, 26, 27). More recently, B. mimosarum and B. nodosa were formally described as legume-nodulating species (8, 9). All of the presumptively diazotrophic Burkholderia species, excluding B. vietnamiensis, comprise a group of closely related species that are phylogenetically distant from the B. cepacia complex (5, 6, 14, 23, 26), but their genomic organizations, genome sizes, and nifH gene sequences are unknown.
In this study, the aim was to determine if the novel plant-associated, presumptively N2-fixing Burkholderia species contain large multireplicon genomes like those in B. cepacia complex species and to evaluate unambiguously their abilities to fix nitrogen using 15N2 incorporation assays. In addition, the genomic localization patterns of recA and nif genes were resolved; nifH gene sequences were determined for many of the strains examined and used for phylogenetic analyses.
The strains of Burkholderia species analyzed are shown in Table 1. Growth conditions were as described by Perin et al. (23). For pulsed-field gel electrophoresis (PFGE), genomic DNA was prepared (19) and electrophoresis was performed with a CHEF DRIII system (Bio-Rad) using 1× Tris-acetate-EDTA buffer at 13.5°C. Replicons were separated using two sets of conditions: an 800-s pulse, an angle of 106°, and a voltage of 110 V for 72 h in 0.8% agarose for the separation of replicons of 0.9 to 3.5 Mb and an 1,800-s pulse, an angle of 106°, and a voltage of 110 V for 76 h in 0.7% agarose for replicons larger than 3.5 Mb. Replicons separated by PFGE were transferred (28) onto Hybond-N membranes (Amersham), 32P-labeled probes were prepared using the Rediprime II random prime labeling system (Amersham), and the membranes were hybridized with a PCR product amplified with primers specific for each gene as described below. The 16S rRNA genes of most Burkholderia strains were localized by hybridizing membranes with a PCR product amplified with primers fD1 and rD1 (32) as described previously (11). 16S rRNA gene probes (ca. 1.5 kb) were PCR products from B. unamae MTl-641T. PFGE profiles of representative Burkholderia species are shown in Fig. 1, and replicon numbers and sizes for all species analyzed are summarized in Table 1. The genomes of three strains each of B. unamae and B. silvatlantica contained four replicons, with average sizes ± standard deviations of 3.31 ± 0.06 Mb for the largest replicon and 1.11 ± 0.14 Mb for the smallest. The three strains of B. tropica each contained five replicons, and the replicons ranged in size from 3.24 ± 0.06 Mb for the largest to 0.53 ± 0.08 for the smallest. According to sequence data, B. xenovorans strains LB400T and CAC-124 (7) have genomes containing three and two replicons, respectively, with total genome sizes larger than (LB400T) or similar to (CAC-124) those of the B. unamae, B. tropica, and B. silvatlantica strains described above. Single strains of B. vietnamiensis and B. sacchari each displayed three replicons, while two strains of B. kururiensis each contained only two. The similarity in replicon size and number between B. kururiensis strains KP23T and M130 strengthens our contention that the not yet officially described species “B. brasilensis” strain M130 mentioned in the literature (3) is instead a strain of the species B. kururiensis (6). Total genome sizes (6.46 to 6.90 Mb) for B. vietnamiensis, B. kururiensis, and B. sacchari were somewhat smaller than those for the Burkholderia species discussed above, and the percentages of the total genome size corresponding to the different replicons were essentially identical for the strains having the same numbers of replicons. For sequenced Burkholderia genomes that contain megaplasmids, these replicons accounted for less than 5% of the total genome size. In all of the sequenced Burkholderia genomes, the smallest chromosome accounts for a minimum of 10% of the genome. For the genomes analyzed in this study (Table 1), only the smallest replicon encountered in the B. tropica strains accounted for less than 10% of the total genome size. Based on size, all of the replicons resolved in this study, except for the smallest from B. tropica, were probably chromosomes. The presence of 16S RNA genes is indicative of a replicon's being a chromosome rather than a megaplasmid (4). Southern hybridizations with 16S rRNA as a probe showed that the gene was present on all of the replicons from each species except for replicons 3 and 5 from B. tropica (Fig. 2A and B) and replicon 3 from B. xenovorans LB400T (7).
PFGE analysis of undigested whole-genome DNA. (A) Results obtained using an 1,800-s pulse for 76 h in 0.7% agarose. Lanes: 1, Hansenula wingei (DNA marker); 2, B. sacchari LMG 19450T; and 3, B. vietnamiensis MMi-302. (B) Results obtained using an 800-s for 72 h in 0.8% agarose. Lanes: 1, B. unamae MTl-641T; 2, B. tropica Ppe8T; 3, H. wingei (DNA marker); 4, B. silvatlantica SRMrh-20T; 5, B. vietnamiensis MMi-302; and 6, B. sacchari LMG 19450T.
Autoradiograms of a Southern blot of undigested whole-genome DNA hybridized with a 16S rRNA gene probe (replicons resolved by PFGE with an 800-s pulse for 72 h in 0.8% agarose) (A and B) and a recA gene probe (replicons resolved by PFGE with an 1,800-s pulse for 76 h in 0.7% agarose) (C). (A) Lanes: 1, B. unamae MTl-641T; 2, B. silvatlantica SRMrh-20T; and 3, B. tropica Ppe8T. (B) Lanes: 1, B. vietnamiensis MMi-302; 2, B. xenovorans CAC-124; 3, B. unamae MCo-762; and 4, B. silvatlantica SRCL-318. (C) Lanes: 1, B. xenovorans CAC-124; 2, B. kururiensis KP23T; 3, B. vietnamiensis MMi-302; 4, B. tropica Ppe8T; 5, B. sacchari LMG 19450T; 6, Schizosaccharomyces pombe (DNA marker); 7, B. unamae MTl-641T; 8, B. kururiensis M130; and 9, B. silvatlantica SRMrh-20T.
Results from analyses of strains of Burkholderia species to determine genome structures; locations of 16S rRNA, recA, and nifH genes; and nitrogen-fixing abilitiesa
Sequence analysis of the recA housekeeping gene, present in only one copy in the sequenced genomes, provides an important method for distinguishing between B. cepacia complex species (17). The recA gene (ca. 869 bp) of B. tropica Ppe8T was amplified using the BUR1/BUR2 primers (21), and the PCR product was used as a probe to localize the recA genes of the Burkholderia strains. Consistent with genome sequence data for different Burkholderia species (www.ncbi.nlm.nih.gov/genomes/lproks.cgi), in which recA is always carried on a chromosome (usually the largest one), we found recA present only on the largest replicon of each species analyzed in this study (Table 1; Fig. 2C). On this basis, recA genes may also be useful for differentiating between the novel Burkholderia species analyzed in this work, in addition to 16S rRNA sequences and other features used in their taxonomic identification.
The nifH genes of Burkholderia strains were PCR amplified with primers IGK (24) and NDR-1 (30) by using mixtures and conditions similar to those described previously (30). nif genes were localized by hybridizing membranes with nifH gene probes (1:1 mixtures of PCR products) from B. unamae MTl-641T and B. xenovorans LB400T. For analysis, total DNA was digested with EcoRI, electrophoresed, blotted, and hybridized under the conditions described previously (11). The conditions for the cloning and sequence analysis of nifH genes were the same as those described previously (30). A new primer pair to amplify the second complementary strand of the nif genes was designed; the forward primer nifH-IF (5′-GGCSATGTACGCGGCSAAC-3′) and the reverse primer nifH-IR (5′-GTTSGCCGCGTACATSGCC-3′) corresponded to B. unamae MTl-641T at positions 442 and 460, respectively. PCR products (1.2 kb) amplified with primers IGK and NDR-1 were used as templates in PCR assays using the primer sets NDR-1/nifH-IF and IGK/nifH-IR under the PCR conditions described above. The products amplified with primers IGK/nifH-IR (450 bp) and NDR-1/nifH-IF (750 bp) were cloned into pCR2.1 (Invitrogen) and sequenced. For the phylogenetic analysis, NifH protein sequences were retrieved from the NCBI database. Deduced amino acid sequences were aligned using Clustal W (29), and neighbor-joining phylogenies were constructed with MEGA 3.1 (16) by using pairwise deletion of gaps and missing data. For all of the B. unamae and B. tropica strains, nifH hybridized only with the smallest replicon, while for all of the B. silvatlantica strains, it hybridized only with the largest (Table 1; Fig. 3A). For B. xenovorans strains, nifH hybridized only with the second largest replicon. Southern blots of total EcoRI-digested DNA hybridized with either a nifH or recA probe showed a single hybridization band for all of the Burkholderia species analyzed, with the exception of the B. xenovorans strains, which contain an EcoRI site in their nifH gene sequences (Fig. 3B), and B. vietnamiensis MMi-302, which apparently contains an EcoRI site in its recA sequence (Fig. 3C), as occurs in the recA gene on chromosome 1 of B. vietnamiensis G4 (GenBank accession no. NC_009256). These results confirm that nifH and recA genes are present in only one copy in these genomes.
(A) Autoradiogram of a Southern blot of undigested whole-genome DNA (replicons resolved by PFGE with an 800-s pulse for 72 h in 0.8% agarose) hybridized with a nifH gene probe. Lanes: 1, B. kururiensis KP23T; 2, B. sacchari LMG 19450T; 3, B. silvatlantica SRMrh-20T; 4, B. unamae MTl-641T; 5, B. tropica Ppe8T; 6, B. silvatlantica SRCL-318; 7, B. tropica MTo-672; and 8, B. xenovorans CAC-124. (B and C) Autoradiograms of a Southern blot of the total EcoRI DNA fingerprints hybridized with a nifH gene probe (B) and with a recA gene probe (C). (B) Lanes: 1, DNA marker 1 kb Plus; 2, B. sacchari IPT-101T; 3 and 4, B. unamae SCCu-23 and MTl-641T; 5, 6, and 7, B. tropica Ppe8T, MOc-725, and MTo-672; 8 and 9, B. silvatlantica SRMrh-20T and SRCL-318; 10, B. vietnamiensis MMi-302; 11 and 12, B. kururiensis KP23T and M130; and 13 and 14, B. xenovorans CAC-124 and LB400T. (C) Lanes: 1, B. sacchari LMG 19450T; 2 and 3, B. unamae SCCu-23 and MTl-641T; 4, 5, and 6, B. tropica Ppe8T, MOc-725, and MTo-672; 7 and 8, B. silvatlantica SRMrh-20T and SRCL-318; 9, B. vietnamiensis MMi-302; 10 and 11, B. kururiensis KP23T and M130; 12 and 13, B. xenovorans CAC-124 and LB400T; and 14, DNA marker 1 kb Plus.
The majority of the available nifH gene sequences of diazotrophic Burkholderia species were obtained in this study. nifH gene phylogenies are often used to determine relatedness among diazotrophic bacteria (25). All of the Burkholderia species analyzed (Table 1) formed a tight cluster with other plant-associated Burkholderia species and Cupriavidus taiwanensis LMG 19424T (Fig. 4). This cluster was well separated from others containing symbiotic or free-living diazotrophs from the alpha, gamma, and delta subdivisions of the Proteobacteria, as well as members of the Archaea and Firmicutes. However, it is noteworthy that the legume-nodulating Burkholderia species, except B. tuberum, formed a single cluster that included Cupriavidus taiwanensis, a diazotroph nodulating Mimosa species that is closely related to the genus Burkholderia and whose nifH gene may be derived from a Burkholderia ancestor (1).
Phylogenetic inference for nitrogen-fixing Burkholderia species and other diazotrophic bacteria, based on NifH amino acid sequences. The analysis included 205 sites. A bootstrapping analysis was used to test the statistical reliability of tree branch points; 1,000 bootstrap samplings were performed. The NCBI GenBank accession numbers for the nifH gene sequences are indicated in parentheses.
Previous analyses of the nitrogen-fixing abilities of Burkholderia species have been based on acetylene reduction activity and/or the presence of nifH genes (5, 11, 13, 14, 23, 26). Since relating acetylene reduction activity to the actual quantity of nitrogen fixed is prone to error (12) and the presence of nifH, encoding nitrogenase reductase (a key enzyme in the process of nitrogen fixation), only implies the ability of an organism to fix nitrogen, in this study 15N2 isotopic dilution experiments were developed to unequivocally confirm diazotrophy. For these experiments, cells grown in BAz liquid medium (11), lacking azelaic acid and containing succinic acid (5 g/liter) and yeast extract (0.5 g/liter), were harvested by centrifugation and washed three times with sterile deionized water. Inocula were standardized using the Bradford method. Each suspension was combined with 15N2 (Isotec; 99% 15N2 atom excess) and sterile air in a Vacutainer tube to give an isotopic dilution equivalent to a 74.25% 15N2 atom excess. The cultures were incubated for 13 or 21 days at 28°C. The percentage of 15N atom excess was determined by emission spectrometry (2) using an atomic absorption spectrophotometer (model NOI-6E; Fischer Analysen Instrumente, Leipzig, Germany). This is the first time that diazotrophy in Burkholderia has been demonstrated using 15N2, quantitatively confirming the N2-fixing abilities of all of the Burkholderia strains analyzed (Table 1), as well as that of the legume-nodulating B. tuberum STM678T (data not shown). No assay detected nitrogen fixation in B. sacchari, which lacks nifH.
We have shown here that novel N2-fixing, plant-associated Burkholderia species, as well as the closely related nondiazotrophic species B. sacchari, possess two to five large replicons, usually chromosomes. This finding extends the observation that large (usually >7.0-Mb) multichromosomal genomes are a characteristic feature of the genus Burkholderia. Although the selective advantage conferred by multireplicons is not clear, it has been proposed previously that in B. cepacia complex and Vibrio species the presence of multichromosomal genomes is essential and advantageous for the fitness of the species in particular environments (17, 20). The large genomes and chromosomal multiplicity in the novel diazotrophic Burkholderia species analyzed in this study may explain their great abilities to colonize the rhizospheres and endophytic environments of a wide range of host plants (5, 6, 13, 22, 23, 26), which along with their metabolic versatility in using volatile compounds (6, 33), as well as their ability to fix nitrogen, produce siderophores, and solubilize insoluble phosphates (6), are probably significant for the survival and success of Burkholderia species in different environments.
Nucleotide sequence accession numbers.
The nifH gene sequences determined in this study were deposited in GenBank under accession numbers EF158799 through EF158811.
ACKNOWLEDGMENTS
This research was partially funded by grant DGAPA-UNAM IN229005.
We thank Jaime Mora (CCG-UNAM) for access to the CHEF DRIII system and J. A. Vera-Núñez for assistance with 15N2 analyses. We acknowledge Catherine Boivin-Masson (INRA-CNRS, France), who kindly provided B. tuberum strain STM678T, and José M. Igual (CSIC, Spain) for supplying B. ferrariae strain FeGI01T.
FOOTNOTES
- Received 22 January 2008.
- Accepted 13 May 2008.
↵▿ Published ahead of print on 23 May 2008.
- American Society for Microbiology