ABSTRACT
We present a PCR procedure for identification of Burkholderia cepacia, Burkholderia multivorans, andBurkholderia vietnamiensis. 16S and 23S ribosomal DNAs (rDNAs) of B. multivorans and B. vietnamiensiswere sequenced and aligned with published sequences for definition of species-specific 18-mer oligonucleotide primers. Specific antisense 16S rDNA primers (for B. cepacia, 5′-AGC ACT CCC RCC TCT CAG-3′; for B. multivorans, 5′-AGC ACT CCC GAA TCT CTT-3′) and 23S rDNA primers (for B. vietnamiensis, 5′-TCC TAC CAT GCG TGC AA-3′) were paired with a general sense primer of 16S rDNAs (5′-AGR GTT YGA TYM TGG CTC AG-3′) or with a sense primer of 23S rDNA (5′-CCT TTG GGT CAT CCT GGA-3′). PCR with these primers under optimized conditions is appropriate to specifically and rapidly identify B. multivorans, B. vietnamiensis, and B. cepacia (genomovars I, III, and IV are not discriminated). In comparison with the polyphasic taxonomic analyses presently necessary for species and genomovar identification within the B. cepacia complex, our procedure is more rapid and easier to perform and may contribute to clarifying the clinical significance of individual members of the complex in cystic fibrosis.
The spectrum of pathogens isolated from respiratory specimens of cystic fibrosis (CF) patients has changed within the last 60 years (1). The prevalence ofPseudomonas aeruginosa has increased, in contrast with that of Staphylococcus aureus (18). Furthermore, new species have emerged, e.g., Burkholderia(Pseudomonas) cepacia, which was first detected in CF patients in about 1971 (20). The prevalence ofB. cepacia was 7.1% among 283 patients at CF centers in Munich, Germany, between 1987 and 1997 (4) and 5.6% among 180 patients of the CF centers at the University Clinic in Essen, Germany, between 1993 and 1998 (8). Govan et al. (15) reported a prevalence of 7% in Scotland in 1992. An average prevalence of 3.6% was reported for patients from 113 Cystic Fibrosis Foundation-accredited centers in the United States by S. FitzSimmons (11a).
Since 1984 (20), it has been known that CF patients who acquire B. cepacia may follow one of three different clinical courses: persistent colonization without additional deterioration of lung function, significantly accelerated reduction of lung function, or—in about 20% of the patients—rapid deterioration with signs and symptoms of acute necrotic pneumonia with or without septicemia and fatal outcome within several weeks or months (“cepacia syndrome” [20]). The factors which determine this dissociation may be either pathogen- or patient-related, or both.
Provided that the incoming Burkholderia strain strongly determines the further course of CF lung infection, the strains isolated from patients who develop the “cepacia syndrome” may be different from strains of CF patients without rapid clinical decline. So far, all Burkholderia isolates collected from CF patients were uniformly identified by conventional methods to be B. cepacia and did not appear to be heterogeneous. Thus, it was not possible to correlate the heterogeneity in the clinical course with a specific Burkholderia subgroup. This is, however, of major importance for the individual patient and for the community of persons with CF.
In fact, Vandamme et al. (31) reidentified 128 isolates conventionally characterized as B. cepacia by DNA-DNA and DNA-ribosomal DNA (rDNA) hybridization, assimilation of organic carbon sources, whole-cell protein pattern, and fatty acid profile. They subclassified the strains into five species, namely B. vietnamiensis (genomovar V), B. multivorans (genomovar II), and the B. cepacia genomovars I, III, and IV (these have no names, as their identification by phenotypic markers is inconclusive so far). Strains of B. vietnamiensis(14), B. multivorans (8), and theB. cepacia genomovars I, III, and IV (31) have been detected in respiratory specimens of CF patients. This subclassification of strains previously identified as B. cepacia stimulated the advancement of the hypothesis that the nature of the incoming Burkholderia strain determines or strongly influences the clinical course of CF lung disease. It has been difficult to evaluate this hypothesis because appropriate procedures for reliable and rapid identification of the Burkholderiaspecies and genomovars have not been available. There are, however, reports suggesting that B. cepacia genomovar III strains were predominantly associated with outbreaks in North America and Europe (24a, 33). This underlines the importance of accurate genomovar identification.
Currently, the B. cepacia genomovars I, III, and IV,B. multivorans, and B. vietnamiensis can be discriminated by the procedures used for their description (see above). These procedures are not suitable for routine diagnostic work. Instead, a number of biochemical reactions were proposed for phenotypic differentiation (9, 14, 21, 31); however, not all strains can be reliably identified, e.g., nonsaccharolytic strains (for example, biovar C of B. cepacia genomovar III), and the evaluation of the tests may take 3 to 5 days for metabolically deficient, slow-growing strains. We therefore tried to find a procedure for specific identification of Burkholderia spp. based on the diversity within the nucleotide sequences of their 16S and 23S rDNAs. Sequence motifs characteristic for single species were integrated into oligonucleotide primers. They allow the production of PCR products specific for B. multivorans, B. vietnamiensis, and the B. cepacia genomovars I, III, and IV. Thus, PCR-based molecular procedures to discriminate theBurkholderia species relevant in CF, includingBurkholderia gladioli (6), are now available.
MATERIALS AND METHODS
Organisms.For characterization of the strains used in this study, see Table 1.
Nucleic acid preparation.Genomic DNA was prepared and purified by using the QiaAmp purification kit (Qiagen, Hilden, Germany).
PCR. (i) For sequencing of the PCR products.Custom oligonucleotide primers (see Table 2) were purchased from MWG Biotech, Ebersberg, Germany. Amplification reactions were carried out in a 50-μl final volume with 1 U of proofreading Taq polymerase provided with the Expand High Fidelity PCR System (Boehringer, Mannheim, Germany), 5 μl of the reaction buffer supplied by the manufacturer (diluted 1:10), 10 μmol of each deoxynucleotide triphosphate, and 50 pmol of each oligonucleotide primer. Approximately 50 to 100 ng of DNA was used as a template. The PCR was carried out in a GeneAmp PCR system 9600 (Applied Biosystems, Weiterstadt, Germany) under the following conditions: denaturation for 5 min at 95°C, followed by 30 amplification cycles of 30 s at 95°C, 30 s at the annealing temperature specific for the primer, and 45 s at 72°C. The samples were then incubated at 72°C for 7 additional min and cooled to 4°C. Pyrogen-free water that had been shown to be free of contaminating DNA was used throughout the study. To inactivate contaminating DNA, the PCR mixture was exposed to UV light for 15 min prior to addition of enzyme and template. To ascertain reproducibility, all species identifications by PCR were performed in duplicate, starting each time from new cultures. The amplification products were checked by agarose gel electrophoresis and purified of salt and excess primers by using the PCR purification kit (Qiagen).
(ii) For identification of unknown strains.The same procedure as described above was followed, with some exceptions: half of the volumes of all amplifications reagents was used, and normalTaq polymerase was used instead of the proofreadingTaq of the High Fidelity PCR System. For each reaction, a specific antisense primer was combined with the universal sense primer Eub-16-1, and for identification of B. vietnamiensis the primers ViMaPs-23-1 and CeVi-23-2 were used (Table 3).
Agarose gel electrophoresis.Agarose gel electrophoresis was performed as described previously (5).
Sequence determination.The amplified rDNAs were sequenced as described previously (23). The two strands of the DNA were sequenced from different PCR products. A model 373A DNA Sequencer (Applied Biosystems) was used, following the protocol of the manufacturer for dye terminator reactions.
Analysis of the sequence data.The nucleotide sequences were aligned with reference rDNA sequences provided in the noncommercial software program package ARB (beta-version 2.4). Secondary structure analysis was done according to the method of Ludwig et al. (24).
Nucleotide sequence accession numbers.The 16S and 23S rDNA sequences of B. multivorans (B. cepaciagenomovar II) LMG 13010T and the 23S rDNA sequence ofB. vietnamiensis (B. cepacia genomovar V) LMG 10929T have been deposited in the EMBL database under the accession nos. 418703, 418704, and 418705.
RESULTS
Analysis of 16S and 23S rDNA sequences.16S rDNA sequences ofB. cepacia were found in the literature (24). We sequenced 16S rDNA of reference strains (Table1) of B. multivorans andB. vietnamiensis and of the B. cepacia genomovars I and III kindly supplied to us by P. A. R. Vandamme (Gent, Belgium). Multiple alignment of the sequences of various strains demonstrated among the species substitutions of nucleotides included within the variable helix 18 or 37/38 regions. They were used for the definition of species-specific primers (Table2). No sequences appropriate for discrimination between B. multivorans and B. vietnamiensis were detectable in the 16S rDNA. However, analysis of the 23S rDNA revealed sites with nucleotides that were different from those in the other Burkholderia spp. included but were identical in the four B. vietnamiensis strains investigated. Thus, as shown for differentiation of Burkholderia malleiand Burkholderia pseudomallei (5), the 23S rDNA sequences may be useful when no suitable nucleotide substitutions are detectable in the 16S rDNA genes.
Investigated strains
Primers selected for PCR
Specificity of PCR.Primers were combined to form pairs and were used to establish optimal conditions for specificity. Each pair of opposite primers was analyzed for the specificity in comparison with those of all other Burkholderia species or genomovars and ofRalstonia pickettii. No amplification products were detected with heterologous Burkholderia strains (Table3). Furthermore, we studied representative strains of species which grow on tryptone soya agar supplemented with 16 μg of colistin/ml. Two strains each of the following species were included: Proteus mirabilis,Morganella morganii, Serratia liquefaciens,Alcaligenes xylosoxidans, Stenotrophomonas maltophilia, Brevundimonas diminuta, and P. aeruginosa (MIC of colistin, >64 μg/ml). In addition, strains assumed to be Burkholderia spp. were sent to us for verification. Some of them were negative according to our PCR procedure, so they were subjected to conventional identification reactions. They proved to be A. xylosoxidans (n = 15), P. aeruginosa (n = 10), S. maltophilia (n = 6), M. morganii(n = 3), and Citrobacter freundii(n = 3). The results were definitive, and no amplification products obtained with any of the specific primer pairs shown in Table 3 were detectable with templates of these strains.
Specificity of primer combinations for the identification of Burkholderia spp.
Procedure for identification of Burkholderia spp.For identification of strains shown to be Burkholderia spp. by API 20NE, four separate PCRs are run. They included primers of different degrees of specificity in order to achieve stepwise exclusion of single species: Bu-Ra excludes species of genera other thanBurkholderia or Ralstonia, CeMuVi-16-2457 excludes B. gladioli andRalstonia spp., MuVi-16-21028 excludes B. cepacia, and the 23S rDNA primers specific for B. vietnamiensis exclude B. multivorans. Except forB. cepacia, which includes the three genomovars I, III, and IV, all species can be identified by positive reactions (B. vietnamiensis, B. gladioli, and R. pickettii) or exclusion (B. multivorans). For theB. cepacia genomovars I, III, and IV, no specific signatures were found within 16S or 23S rDNA sequences. Strains yielding products of PCR with the Bu-Ra primers but negative for PCR with the other primers may be either B. mallei or B. pseudomallei. They can be identified as described previously (5).
Evaluation of the primers with CF isolates.We studied 124 isolates collected from CF patients from four European countries. They had been identified by conventional phenotypic means as B. cepacia and collected for PCR. All of the strains could be classified. Of the isolates 79 (63.7%) were found to be B. multivorans, 42 (33.9%) were B. cepacia, 2 (1.6%) were B. gladioli (Hannover, Germany), and 1 (0.8%) wasB. vietnamiensis (Amsterdam, The Netherlands). B. multivorans appears to be the dominating species amongBurkholderia isolates from European CF centers investigated in this work (7). No major spread of B. multivorans strains was observed among 10 patients of one CF center. Altogether, eight different randomly amplified polymorphic DNA types were identified, and only two of them were detected in more than one patient (8).
DISCUSSION
In CF, the basic genetic defect predisposes the individual to lung infections caused by specific bacterial pathogens. The major gram-negative organisms causing chronic recurrent lung inflammation areP. aeruginosa and Burkholderia spp. Once established, these pathogens can be eradicated only infrequently and transiently. Nevertheless, the increase in life expectancy and quality of life achieved within the last decade is mainly attributed to progress in antiinfective therapy. Therefore, even transient reduction in the concentration of pathogens at the site of infection benefits the clinical status and prognosis of the patient. As long as gene therapy is not available antimicrobial treatment must be optimized, along with measures for the prevention of early infection and the transmission of pathogens among CF patients.
These measures rely on the knowledge of the microbiological status of the CF patient at all times. Exact data on the identity of the pathogens and their concentrations in respiratory secretions (mostly sputum) as well as on their susceptibility to therapeutically relevant antibiotics allow the rational design of treatment regimens and effective strategies for controlling the spread of these pathogens from infected to pathogen-free patients.
At this time, the genus Burkholderia is thought to be composed of 16 species (10, 32, 36). Most of these are primarily plant pathogens; the CF-relevant Burkholderiaspecies are restricted to B. cepacia (genomovars I, III, and IV), B. multivorans, B. gladioli, and B. vietnamiensis. Their detection in diagnostic laboratories depends initially on the use of selective media which inhibit growth ofP. aeruginosa. All Burkholderia spp. of potential relevance in CF lung disease grow on colistin-supplemented media (12, 17, 19, 34). In a quality control trial carried out by Tablan et al. (30), only 32% of the participating 115 laboratories detected B. cepacia in the test sputum. In a surveillance study including 78 centers in Germany conducted in 1997, the prevalence of Burkholderia spp. was 4.1% when a selective medium containing colistin was used but only 1.8% when selective medium was not used (our own unpublished data). In fact, only about one-fifth of the sputa were analyzed by using aBurkholderia-selective agar. On colistin-containing media, besides Burkholderia spp., a variety of other species are able to grow (e.g., Proteus spp., Providenciaspp., M. morganii, Serratia spp., A. xylosoxidans, and staphylococci).
For identification of Burkholderia spp., the procedures used for their taxonomic description (13, 28) (see Introduction) are inappropriate in clinical laboratories. The aim of our work was to establish procedures which meet the demands of clinical practice where determination of both the identity of a pathogen and its antibiotic susceptibility is called for. For this purpose, culturing of the organisms is indispensable. Thus, our identification procedure is based on culture. The procedure may also be adapted to detectBurkholderia spp. in situ, e.g., in sputum specimens. The major advantage of the PCR-based identification is rapidity. While evaluation of the biochemical reactions proposed for phenotypic differentiation (9, 14, 21, 31) may take up to 7 days (17), the PCR procedure is performed within 5 h. The biochemical tests for reliable identification are not commercially available now and have to be prepared and controlled for quality within individual laboratories. They are inappropriate for the identification of nonsaccharolytic strains (e.g., biovar C of B. cepaciagenomovar III [31]).
Molecular identification of pathogens relies on the diversity of the nucleotide sequences of the 16S and 23S rDNA sequences with specific motifs conserved at different phylogenetic levels, e.g., genus and species. This diversity has been used for species-specific PCR procedures (22, 26, 28). Our 16S and 23S rDNA sequence data for B. vietnamiensis, B. multivorans, and theB. cepacia genomovars I, III, and IV demonstrated options for discriminatory primers. Among them, the oligonucleotides listed in Table 2 were found to be appropriate for the identification of B. multivorans, B. vietnamiensis, and the B. cepacia genomovars I, III, and IV, which could not be separated until now. The different PCRs may be run separately to discriminate simultaneously B. cepacia (genomovars I, III and IV are included), B. multivorans, or B. vietnamiensisfrom colonies identified conventionally as a Burkholderiaspecies. A mixture from different colonies (instead of a pure culture) should be tested to find out whether the specimen contains more than one species.
A molecular identification procedure for B. cepacia that uses 16S rDNA sequences was proposed by Campbell et al. (11). This work preceded the changes of taxonomy in 1997 (31) and therefore could not include efforts to differentiate the five B. cepacia complex genomovars. Analysis of the primers described by Campbell et al. demonstrated specificity for B. cepacia. No PCR products were amplified from DNA templates of B. multivorans and B. vietnamiensis (our own unpublished data), while discrimination ofB. gladioli had already been shown by the authors. Whitby et al. (35), in addition to the primers described by Campbell et al. (11), used a primer pair, G-1 and G-2, targeting the 130-bp spacer region between 16S and 23S rDNA. However, the spacer-directed primers were less specific than the 16S-rDNA-derived primers. The PCR procedures which we present facilitate rapid and specific identification of B. multivorans, B. vietnamiensis, B. gladioli (6), and B. cepacia. However, the primers proposed for identification ofB. cepacia do not discriminate among the B. cepacia genomovars I, III, and IV. Work for molecular differentiation of these genomospecies is in progress.
The basic question has been asked whether members within a genus of mostly phytopathogenic species can be primary causative pathogens in CF or whether they are just indicators of a late stage of lung disease. The pros and cons have been presented and discussed by different authors (13, 16). It is so far uncertain whether among theBurkholderia spp. in CF patients, there are species with increased virulence (25) which on their own or in combination with cofactors (e.g., preceding damage of lung tissue by other microorganisms) cause the rapid clinical deterioration of the patient. Major outbreaks in Canada and the United Kingdom were caused by the spread of particular strains which produce cblA protein (27, 29) and appear to be B. cepacia genomovar III. This may indicate that virulence might be associated with a particular type or species of pathogen and that accurate identification of the species of the B. cepacia complex is of major clinical importance. The PCR-based procedures proposed to discriminateB. cepacia, B. multivorans, B. vietnamiensis, and B. gladioli should help to clarify their prevalence, epidemiology, antibiotic susceptibility, and pathogenetic potential in CF (2, 3).
ACKNOWLEDGMENTS
We thank H. Bärmeier, Erlangen, J. Dankert, Amsterdam, L. Möller, Groningen, J. Govan, Edinburgh, N. Høiby, Copenhagen, F. Ratjen and K. Hübner, Essen, D. Speert, Vancouver, and B. Tümmler, Hanover, for providing Burkholderia isolates from their CF patients. We are most grateful to Peter Vandamme, Gent, Belgium, for stimulating discussions and verification of species identification obtained by the PCR-based procedures described.
Part of the work on this project was supported by a grant from the German Ministry of Defense and by the Christiane-Herzog-Stiftung for CF patients, Berlin, Germany.
FOOTNOTES
- Received 20 October 1998.
- Returned for modification 28 December 1998.
- Accepted 7 February 1999.
- Copyright © 1999 American Society for Microbiology
