ABSTRACT
A polyphasic taxonomic study involving DNA-DNA hybridization, whole-cell protein electrophoresis, and 16S ribosomal DNA sequence analysis revealed that a group of Burkholderia cepacia-like organisms isolated from the rhizosphere or tissues of maize, wheat, and lupine belong to B. cepacia genomovar III, a genomic species associated with “cepacia syndrome” in cystic fibrosis patients. The present study also revealed considerable protein electrophoretic heterogeneity within this species and demonstrated that the B. cepacia complex consists of two independent phylogenetic lineages.
In a survey of nonnative plant rhizosphere bacteria conducted in La Côte Saint André(France) with maize and in Kapunda (South Australia, Australia) with wheat, high levels of two groups of Burkholderia strains were found. The first group was characterized by using a polyphasic approach and formed a new taxon, Burkholderia graminis(15). Strains of the second group (designated phenon B) were found to be closely related to the Burkholderia cepaciacomplex; large numbers of these strains were present on roots, and more recently, new isolates were also obtained from inside the tissues of wheat and lupine in Kapunda (Table 1). Here, characterization of this taxonomic group was revisited by including reference strains of the B. cepacia complex in DNA-DNA hybridization, whole-cell protein electrophoretic, and 16S ribosomal DNA (rDNA) sequence analyses.
Strains used in this study
Total DNA-DNA hybridization analyses were performed by using two methods, one involving tritiated reference DNAs (Table2) and one involving photobiotin-labeled probes (Table 3). In a preliminary study, the two methods showed good correlation. For instance, the levels of hybridization of strain AUS 27 DNA with DNA of strain LMG 12614 were 65% when tritiated DNA was used and 63% when photobiotin-labeled DNA was used.
Levels of total DNA hybridization with radioactively labeled DNAs of eight reference strains
DNA-DNA hybridization of environmental strains with genomovar III representatives
In the first experiments we used tritiated reference DNAs of eight isolates, including two rhizosphere isolates (AUS 27 and C3B1M), one recent cystic fibrosis isolate (1–36), and reference strains ofB. cepacia genomovar I (ATCC 25416T),Burkholderia vietnamiensis (LMG 10929T),Burkholderia multivorans (1–45), Burkholderia pyrrocinia (ATCC 15958T), and B. graminis(ATCC 700544T). DNAs from 15 of our rhizosphere isolates, reference strains belonging to B. cepacia genomovar I (ATCC 25416T and LMG 6964) and genomovar III (LMG 12614, LMG 16661, and LMG 6988), and B. pyrrocinia (ATCC 15958T), and three recent cystic fibrosis isolates (strains 751, 1–36, and 1–47) were hybridized with these radioactively labeled DNAs. When hybridized with labeled DNA of strain AUS 27, all rhizosphere isolates except m35b showed levels of DNA-DNA hybridization greater than 65% and differences in melting temperatures (ΔTm values) less than 5°C, indicating that they belong to the same genomic species (12). When they were hybridized with labeled DNA of strain C3B1M, slightly lower values (as low as 61%) were obtained, indicating a certain degree of genomic heterogeneity in this species. Strain m35b showed significant but low levels of hybridization (40 to 48%) with all reference strains and thus does not belong to any of the genomovars examined. The possibility that this strain could belong to Burkholderia stabilis was not eliminated and will be tested further. B. cepaciagenomovar III reference strains exhibited levels of hybridization of 58 to 76% with labeled DNA of strain AUS 27, indicating that the rhizosphere isolates belong to B. cepacia genomovar III. Reference strains of the other B. cepacia genomovars and ofB. pyrrocinia exhibited levels of DNA-DNA hybridization between 39 and 60%, values which are in complete agreement with values reported previously (13). The levels of hybridization with DNA of the B. graminis type strain were much lower (13 to 15%). The three recent cystic fibrosis isolates (strains 1–36, 751, and 1–47) showed levels of hybridization between 63 and 76% with AUS 27 DNA with ΔTm values less than 5°C. These data show unambiguously that these three isolates also belong to the same genomic species as AUS 27 (i.e., B. cepaciagenomovar III).
A second group of DNA-DNA hybridization experiments (Table 3) was performed in order to substantiate the relationships among B. cepacia genomovar III strains. In addition, a representative endophytic isolate was included. Values between 63 and 82% were obtained, which confirmed that all of these isolates belong to a single genomic species.
Whole-cell protein extracts were prepared from 48-h cultures of all of the B. cepacia genomovar III strains and several additional endophytic isolates. Data for the reference strains were obtained from previous studies (3a, 13, 14). Sodium dodecyl sulfate-polyacrylamide gel electrophoresis analyses were performed as described previously (13). Protein profiles were analyzed by using the GelCompar software package (version 4.2; Applied Maths, Kortrijk, Belgium). Levels of similarity between the patterns were computed by using the Pearson product moment correlation coefficient and were expressed as percentages of similarity for convenience. Considerable heterogeneity was apparent, and the strains grouped into two main protein electrophoretic clusters comprising the endophytic isolates (cluster 1) and all of the Australian rhizosphere isolates except isolate AUS 27 (cluster 2), three small clusters comprising two isolates each (clusters 3 to 5), and several isolates with distinct positions in the dendrogram (Fig.1). Cluster 3 comprises two reference strains (LMG 12614 and LMG 12615) and represents cluster xii described previously (13). Cluster 4 also comprises two reference strains (LMG 13053 and LMG 16661) and corresponds to cluster xi described previously (13). Finally, cluster 5 comprises two French rhizosphere isolates (C3B1M and m32). In spite of this protein electrophoretic heterogeneity, DNA-DNA hybridization data (Tables 2 and 3) demonstrated that all of the isolates shown in Fig. 1represent a single genomic species; most of the isolates belong to clusters 2 through 5, and one strain, strain W11.7, is a cluster 1 reference isolate. The protein electrophoretic homogeneity of the other cluster 1 isolates indicates that they are members of the genomic species as well.
Dendrogram derived from unweighed pair group average linkage of correlation coefficients (expressed for convenience as percentages of similarity) for the whole-cell protein patterns of theB. cepacia genomovar III strains studied.
To investigate the phyletic relatedness of the genomovar III isolates, almost complete 16S rDNA sequences of the following two endophytic isolates and two genomovar III strains were obtained: LMG 12615, LMG 12614, LS2.4 and WS11.7. These sequences and selected GenBank 16S rDNA sequences of 19 representative strains of the B. cepaciacomplex were aligned. A total of 1351 16S rDNA sites were then selected, and sites involving indels (insertions or deletions) were excluded from further analysis. Evolutionary distances (representing the percentages of transversion type differences between sequence pairs) were computed by the method of Jukes and Cantor (9). A phylogenetic tree was inferred by using the neighbor-joining method (10), and bootstrapping was performed (9). This phylogenetic analysis divided the 16S rDNA sequences of the B. cepacia complex into two major clusters: (i) a lineage containing the B. vietnamiensis(genomovar V), B. multivorans (genomovar II), and LMG 18941 (genomovar VI) DNA sequences, and (ii) a group containing the B. stabilis (genomovar IV), B. pyrrocinia, B. cepaciagenomovar I (ATCC 25416T and ATCC 17759), B. cepacia genomovar III (LMG 12614, LMG 12615, WS11.7, LS2.4, and C3B1M), and unclassified strain m35b sequences. This division was strongly supported by 98% of the bootstrap replicates and should be obtained with any other phylogenetic markers that match classical bacterial evolutionary patterns.
Public health implications.Our 14 isolates represent environmental niches ranging from the rhizosphere to the inner tissues of wheat, lupine, and maize and were obtained in France and in South Australia. These plants are cultivated all over the world, and it is likely that our isolates represent very common bacteria.
We demonstrate here that a significant proportion of these maize-, wheat-, and lupine- associated bacteria are actually members ofB. cepacia genomovar III. This B. cepaciagenomovar is particularly relevant for cystic fibrosis as most strains associated with the “cepacia syndrome” belong to it. This syndrome is characterized by a dramatic necrotizing pneumonia that results in rapid death of the patient (6, 8). Recent deadly outbreaks which occurred in many parts of the world have been attributed to strains of genomovar III, suggesting that high transmissibility could be a characteristic of B. cepaciagenomovar III in the B. cepacia complex (13). It has been suggested that the environment is a source of new isolates, but so far no clear evidence of this has been obtained. Attempts to recover B. cepacia isolates similar to clinical isolates from soils and other environmental sites have failed, most likely because of the selective agents used (antibiotics). The environmental strains studied here were isolated by using the PCAT medium (2), whose selective power is based solely on the metabolism of unusual sources of carbon and nitrogen (viz., azelaic acid and tryptamine). Thus, our Australian and French isolates represent the first collection of genuinely environmental B. cepacia genomovar III strains. This collection opens the way for comparisons of closely related strains of environmental and clinical origin, which could provide some clue about the properties acquired by hospital-adapted strains and their pathogenicity characteristics or genes. It could also help refine strategies for avoiding acquisition of new Burkholderia strains in cystic fibrosis treatment centers. Most of the soil isolates have been deposited in the BCCM/LMG Bacteria Collection (University of Ghent, Ghent, Belgium) under the accession numbers shown in Table 1.
ACKNOWLEDGMENTS
K. Ophel-Keller (Commonwealth Scientific and Industrial Research Organisation, Adelaide, South Australia, Australia) isolated the endophyte strains which were kindly sent to us by C. Lodewyckx from M. Mergeay's laboratory (SCK/CEN, Mol, Belgium). We thank G. Chabanon and C. Segonds for sending us cystic fibrosis patient isolates ofBurkholderia. F. Fontaine efficiently contributed to the 16S rDNA sequencing of several strains.
J.B. was supported by an OECD fellowship and a CNRS-CSIRO collaborative program. P.V. is indebted to the Fund for Scientific Research-Flanders (Belgium) for a position as a postdoctoral fellow. T.C. acknowledges support received from the Vlaams Instituut voor Bevordering van Wetenschappelijk-technologisch Onderzoek in de Industrie (Belgium) in the form of a bursary for advanced study. We acknowledge the financial support provided by the Cystic Fibrosis Trust (United Kingdom) (grant RS15). We acknowledge the financial support of Centre National de la Recherche Scientifique for the research performed at UMR CNRS 5557.
FOOTNOTES
- Received 30 May 2000.
- Accepted 21 November 2000.
- Copyright © 2001 American Society for Microbiology
