|Home | About | Journals | Submit | Contact Us | Français|
Despite considerable interest in studying Burkholderia cepacia complex in the environment, we still do not have efficient methods to detect, isolate, and screen large numbers of B. cepacia isolates. To better describe the ecology and diversity of B. cepacia complex, a colony hybridization assay was developed to detect specifically all species of the complex based on polymorphism of the variable V3 region of the 16S rRNA sequence. The sensitivity of the assay was dramatically enhanced by using a probe consisting of three repeats of a B. cepacia complex-specific probe, each separated by a phosphoramidite spacer. In addition, a duplex PCR targeting B. cepacia complex-specific recA and 16S rRNA sequences was developed to enable a fast and reliable diagnostic assay for members of the complex. When applied to maize rhizosphere samples, colony hybridization results were in good agreement with those of most-probable-number duplex PCR, both indicating a >100-fold fluctuation of abundance between individual plants. Using restriction analysis of recA for a total of 285 confirmed isolates of the B. cepacia complex, up to seven B. cepacia complex species were identified; however, their diversity and abundance were not evenly distributed among individual plants, and several allelic variants were commonly found from the same rhizosphere sample. These results indicate that not only complex communities of B. cepacia complex species and closely related strains of the same species may coexist at high population levels but also species composition and abundance may dramatically vary between individual plants.
In the genus Burkholderia of the β subdivision of the proteobacteria, the Burkholderia cepacia complex consists of at least nine closely related Burkholderia species, namely, B. cepacia (formerly genomovar I), B. multivorans (II), B. cenocepacia (III), B. stabilis (IV), B. vietnamiensis (V), B. dolosa (VI), B. ambifaria (VII), B. anthina (VIII), and B. pyrrocinia (IX) (10, 11, 22, 49-53). They share moderate levels of DNA-DNA hybridization (30 to 50%) but high sequence similarity at the level of 16S rRNA (98 to 99%) and recA (94 to 95%) genes. Moreover, they all possess an unusually complex and plastic genome that consists of two to four replicons (chromosomes), which harbor several insertion sequences that may increase the frequency of genetic mutations and recombinations (7, 30). This may favor their adaptation to different environments, and it is not surprising that strains of the complex are found in habitats as diverse as soil, plant, and animal surfaces, rhizosphere, and water (9, 42).
This remarkable versatility is further exemplified by the wide array of properties they possess. Although first described as plant pathogens causing sour skin of onion (6), strains of the B. cepacia complex have also been reported to be effective biological control agents against soilborne (5, 25), foliar (28), and postharvest (27) plant pathogens, to promote plant growth (16, 48), and to decontaminate polluted soil or groundwater (29, 39). Despite the useful properties those organisms may offer to bioremediation or as an alternative to pesticides in agriculture, commercial application has been hampered by the risk B. cepacia complex may pose to human health. Indeed, in the last decade, strains of the complex have been found associated with nosocomial infection of immunocompromised patients (26) and chronic or rapid decline in pulmonary function in patients with cystic fibrosis (31).
To date, no clear distinction has been established between clinical and environmental isolates on the basis of phenotypic or genotypic criteria (1, 3, 34), and all B. cepacia complex species have been isolated from sputum cultures of cystic fibrosis patients (9). Moreover, the emergence of new strains each year in cystic fibrosis patients suggests the ongoing acquisition of strains from the environment (8, 32, 34).
Although B. cepacia complex strains deposited in culture collections have been sampled from various habitats, information concerning major reservoirs of B. cepacia complex in the environment are still lacking. Indeed, there are few studies concerning the ecology of the B. cepacia complex in the environment, and they deal with a few strains isolated sporadically or were done when the description of the complex was limited to one species or a few species (9, 42). Recently, several studies have examined the rhizospheres of trees (43) and crop plants (1, 3, 14, 19) as potential reservoirs for B. cepacia complex. Although most of the latter studies dealt with 50 to 100 isolates, a more thorough description of B. cepacia complex populations in the environment is needed, despite the laborious work of isolation and screening that this entails. The main issue, however, is not B. cepacia complex identification because molecular identification of B. cepacia complex species is made possible using specific primers targeting recA (37) and 16S rRNA (2, 33) sequences. Furthermore, recA restriction analysis has been shown to be a convenient tool to distinguish each of the species in the complex (37). The greater difficulty is in the way those isolates are arbitrarily selected on dilution plates prior to identification. Indeed, Miller et al. (41) reported the presence of B. cepacia complex in more than 82% of 91 urban soils as detected by culture-independent methods, while B. cepacia complex colonies could be isolated in only 15% of all soils investigated. This underlines the need for better screening techniques to recover B. cepacia complex even when present in a sample at low population levels.
The low isolate yield may be attributed to the fact that semiselective media, e.g., trypan blue tetracycline medium (TB-T) (24) or Pseudomonas cepacia azelaic acid tryptamine medium (PCAT) (4), were developed when the taxonomic description of the B. cepacia complex was not yet fully established. Although unavoidable biases exist when recovering B. cepacia complex on selective or semiselective media (46), it is the only way to obtain isolates in pure cultures for further phenotypic and genotypic characterizations and to better describe their ecology and diversity.
The objectives of this study were to develop and assess the efficiencies of molecular assays to determine the population density and diversity of culturable organisms of the B. cepacia complex from environmental samples. The methods include colony hybridization (CH) and duplex PCR combined with most-probable-number PCR (MPN-PCR). The methods were first developed and tested using a representative collection of reference B. cepacia complex and non-B. cepacia complex species and further validated and compared with each other by thoroughly characterizing B. cepacia complex populations from a heterogeneous habitat, i.e., the maize rhizosphere.
All reference B. cepacia complex and non-B. cepacia complex strains were selected from the collection of the Burkholderia cepacia Research Laboratory and Repository (at the University of Michigan, Ann Arbor). Twenty-four strains representing the nine B. cepacia complex species were used to validate the molecular techniques. In the following list, a strain name followed by a T and/or P superscript indicates that the strain is a species type strain and/or a strain included in previously published experimental strain panels (12, 36), respectively. Their designations in the American Type Culture Collection (ATCC) and the Laboratorium voor Microbiologie (LMG) (Ghent University, Ghent, Belgium) are provided when available. The reference B. cepacia complex and non-B. cepacia complex strains were B. cepacia genomovar I PC783PT (ATCC 25416, LMG 1222), HI2140P, and HI2284; B. multivorans HI229P (ATCC 17616, LMG 17588), HI2132P, and HI2240P; B. cenocepacia J2315TP (ATCC BAA-245, LMG 16656), HI3240P, and HI3248P; B. stabilis HI2210TP (ATCC BAA-67, LMG 14294) and AU0244; B. vietnamiensis PC259P (LMG 18835, CEP40, PC259), AU1344, and HI2278P; B. dolosa AU0645TP (LMG 18943, R-5670), AU0158, and HI2238; B. ambifaria AMMDTP (ATCC BAA-244, LMG 19182), AU0212 (LMG 19466), and HI2474P; B. anthina AU1293P (LMG 21821) and HI2725P; and B. pyrrocinia BC011P (LMG 21823) and HI2710P. Negative controls consisted of six strains that were not B. cepacia complex strains but were closely related to them, i.e., Burkholderia gladioli AU3923, AU5002, AU5085, and HI2870; Burkholderia xenovorans LB400T (LMG 21463) (23); and Ralstonia pickettii PKO1 (21). Reference strains and rhizosphere isolates were routinely grown at 29°C on plate count agar (Difco Laboratories, Detroit, Mich.) and stored at −80°C in Luria-Bertani broth medium (Difco) containing 30% glycerol.
B. cepacia complex-specific primers targeting 16S rRNA and recA (Table (Table1)1) were synthesized by Integrated DNA Technologies (Coralville, Iowa). Using Primer3 software (44), a new forward primer (i.e., Burkf) was designed to amplify specifically 16S rRNA sequences of B. cepacia complex based on a CLUSTAL W (47) sequence alignment of B. cepacia complex and closely related genera and species.
Each DNA preparation consisted of 1-μl loopful of a colony grown overnight that was lysed in 4 μl of lysis solution (50 mM KCl, 0.1% Tween 20, 10 mM Tris-HCl [pH 8.3]) at 99°C for 10 min. Sixteen microliters of PCR cocktail was subsequently added to the cell lysate. The PCR cocktail consisted of 1× PCR buffer (50 mM KCl, 10 mM Tris-HCl [pH 9.0], 0.1% Triton X-100, 1.5 mM MgCl2; Promega Corp., Madison, Wis.), bovine serum albumin (0.5 g liter−1; Sigma, St. Louis, Mo.), 5% dimethyl sulfoxide (Sigma), 200 μM concentrations of the four deoxynucloside triphosphates (Promega), a 0.50 μM concentration of each primer targeting 16S rRNA (i.e., Burkf and CeMuVi-16-2), a 0.40 μM concentration of each primer targeting recA (i.e., BCR1 and BCR2), and 1.5 U of Taq DNA polymerase (Promega). After an initial denaturation step (3 min at 94°C), 30 PCR cycles (1 cycle consisting of 30 s at 94°C, 30 s at 55°C, and 1 min at 72°C) and a final extension step (10 min at 72°C) were used. PCR products were electrophoresed on 1.5% agarose gels in 1× Tris-acetate-EDTA (TAE) and stained with ethidium bromide by standard protocols (45). A 100-bp ladder (Invitrogen Corp., Carlsbad, Calif.) was used as molecular size markers.
PCR cocktail composition was as described above, except that only primers BCR1 and BCR2 (0.40 μM each) were used. After PCR amplification, 5 μl of amplified product was used for restriction analysis with 2 U of either HaeIII or MnlI (New England Biolabs, Inc., Beverly, Mass.) in a total volume of 15 μl. After 3-h digestion at 37°C, restriction fragments were separated by electrophoresis on 2.5% agarose gels by standard procedures (45). A 100-bp ladder and a 50-bp ladder (Invitrogen) were used as molecular size markers. Restriction analysis was generally done once and was repeated if restriction patterns could not be assigned to any of the reference patterns, i.e., those obtained by recA restriction fragment length polymorphism (RFLP) with reference B. cepacia complex strains of the nine B. cepacia complex species.
To determine 16S rRNA sequences, PCR products were purified with the Qiaquick PCR purification kit (QIAGEN, Inc., Valencia, Calif.), and both strands were sequenced using universal eubacterial primers. Alternatively, when specific PCR products could not be obtained, PCR amplicons were first cloned with TOPO TA cloning kit for sequencing (Invitrogen) and then sequenced using Taq dye terminator chemistry and an ABI cycle sequencer (Genomic Technology Support Facility, Michigan State University). Both strands of recA sequences were sequenced using PCR primers BCR1, BCR2, BCR3, and BCR4 (37).
Sequences determined in this study were used to search for similarities among known sequences using the basic local alignment tool (BLAST) at the National Center for Biotechnology Information (Bethesda, Md.), and SEQUENCE MATCH of the Ribosomal Database Project. Multiple sequences yielding the highest similarity scores were aligned using CLUSTAL W, and sites presenting gaps or unknown bases were excluded from analysis.
Evolutionary distances were calculated according to the two-parameter Kimura model, and phylogenies were reconstructed on the basis of the neighbor-joining method of the Phylogeny Inference Package (PHYLIP) (17). Input order was randomized 5 to 10 times in all phylogenetic reconstructions. Nodal support of the inferred trees was assessed by 1,000 bootstrap replicates. DNApars in the PHYLIP package was also used to perform ordinary parsimony analyses using exhaustive search, and nodal support was assessed with 100 bootstrap replicates to confirm the phylogenetic positioning of the clades.
Maize rhizosphere samples were obtained in mid-July, 2003, from the Michigan State University, W. K. Kellogg Biological Station Long-Term Ecological Research (KBS) site (Hickory Maizeers, Mich.) (http://lter.kbs.msu.edu). The maize field was managed under maize monoculture (one annual crop with no cover crop), with conventional tillage, and without application of fertilizers, herbicides, or insecticides. Entire root systems were sampled at 5, 10, 20, 120, 150, and 160 m from the first sample, and they were assigned sample number 2, 3, 4, 5, 6, and 7, respectively (Table (Table2).2). For each sample, 1 g of root system with closely adhering soil (rhizospheric soil) was transferred to a sterile tube containing 10 ml of 0.85% NaCl solution, and the tube was shaken at 350 rpm at room temperature for 30 min (Fig. (Fig.1A).1A). The bacterial suspension was diluted 10-fold in 0.85% NaCl, and 100 μl of each dilution was plated in triplicate on B. cepacia complex semiselective medium trypan blue tetracycline medium (TB-T) (24), which contains (per liter) 2 g of glucose, 1 g of l-asparagine, 1 g of NaHCO3, 0.5 g of KH2PO4, 0.1 g of MgSO4 · 7H2O, 20 g of agar, 5 mg of crystal violet, and 50 mg of trypan blue. The pH was adjusted to 5.5, and 20 mg of tetracycline and 100 mg of cycloheximide were added after autoclaving. TB-T plates were incubated at 29°C for 4 days prior to counting the number of CFU. TB-T was chosen over other B. cepacia complex semiselective media, because all known B. cepacia complex species can grow on it, it yields more B. cepacia complex isolates than other media (41; our unpublished results), and it was formulated for environmental samples (24).
A CLUSTAL W alignment of rRNA sequences of B. cepacia complex species and closely related organisms was inspected manually to identify putative B. cepacia complex-specific probes. Candidate probes were evaluated in silico using the Probe Match tool of the Ribosomal Database Project (38). A universal probe (for members of the domain Bacteria) served as a positive control for DNA fixation on the membranes. It consisted of a 510-bp fragment (produced with primers RHG-F and RHG-R) (33) that hybridized with all bacterial DNA spotted on the membrane. This probe was washed off the nylon membrane prior to hybridization with B. cepacia complex-specific probes.
For dot blot hybridization, full-length PCR-amplified 16S rRNA was spotted onto Hybond-N+ membranes (Amersham, Piscataway, N.J.), and fixed by UV cross-linking (Stratalinker; Stratagene, La Jolla, Calif.). For CH, colonies were grown at 29°C either overnight on PCAT or for 4 days on TB-T dilution plates. Prior to applying colonies to membranes, it was generally necessary to incubate the colonies overnight at 4°C to increase colony adhesiveness. For TB-T dilution plates, larger plates (Omnitray; Nalge Nunc International, Rochester, N.Y.) were used instead of round petri dishes to maximize distance between colonies. DNA from colonies adhering to the membranes was denatured and neutralized following the manufacturer's recommendations (Amersham) and fixed by UV cross-linking.
Oligoprobes were synthesized by Integrated DNA Technologies. They were diluted to a concentration of 1.25 μM and labeled according to the instructions for the Alkphos direct labeling kit (Amersham) for 30 min for fragment probes and for at least 2 h for oligoprobes. Prehybridization, hybridization, and posthybridization stringency washes were done at 40°C using the manufacturer's recommended buffers and protocols (Amersham). After signal generation in the presence of CDP-Star chemiluminescent substrate (Amersham), light emission was detected by autoradiography using film sensitive to blue light (Hyperfilm ECL; Amersham).
Prior to hybridizing the same membranes with a new probe, probes were removed by dipping membranes into boiling 0.5% (wt/vol) sodium dodecyl sulfate and allowing them to cool down to room temperature with gentle agitation. The absence of signal was checked by autoradiography before further hybridization with other probes.
A dilution endpoint PCR (DE-PCR) was performed by the method of McSpadden Gardener et al. (40) with the following modifications. Ten microliters of the initial rhizospheric wash suspension was diluted 10-fold six times in a 96-well microtiter plate with liquid TB-T as the diluent (Fig. (Fig.1B).1B). Four replicate samples were prepared, which enabled the simultaneous analysis of four rhizospheric samples in each 96-well plate. Plates were incubated at 29°C with gentle shaking (150 rpm) for 4 days. Five microliters from each well was transferred to 95 μl of lysis buffer using a 12-channel pipetman, and cell lysis was performed as described above. Two microliters of the DNA lysate was used in 10-μl screening duplex PCR with the PCR cocktail composition and cycling conditions described above. Population levels were estimated on the basis of the last dilution in which a PCR product could be observed after electrophoresis on standard agarose gels. Based on the same PCR results, estimates of B. cepacia complex population levels were also calculated using the MPN method and its corresponding 95% confidence intervals (20). DE-PCR and MPN-PCR differ only in the way PCR results are used to calculate final population levels.
Samples (1 g) were shaken in liquid TB-T medium 4 days at 29°C (Fig. (Fig.1C).1C). Total DNA was extracted and purified using the UltraClean soil DNA isolation kit (Mo Bio Laboratories, Inc., Carlsbad, Calif.). Duplex PCR was performed as described above using 4 μl of undiluted or 10-fold-diluted soil extracts for a total volume of 20 μl.
B. cepacia complex population levels were determined at least twice, and representative results of one experiment are presented. Restriction analysis to determine B. cepacia complex species was done for one of those experiments. The numbers of CFU expressed per gram of fresh roots were log10 transformed (35) prior to statistical analysis. Analysis of variance, regression analysis followed by Bonferroni's two-tailed significance test, and nonparametric tests were done using SYSTAT (version 8.0; SPSS, Inc., Chicago, Ill.), and a P value of <0.05 was considered statistically significant.
16S rRNA and recA sequences have been deposited in the GenBank and EMBL databases under accession numbers AY769902 to AY769913 for 16S rRNA sequences and AY769895 to AY769901 for recA sequences.
In order to determine quickly and reliably whether an isolate belongs to the B. cepacia complex, a duplex PCR assay was designed to specifically differentiate B. cepacia complex from closely related bacteria. Existing primers and a new forward primer targeting 16S rRNA sequences (Table (Table1)1) were combined to enable the simultaneous amplification of recA (1,040 bp) and 16S rRNA (320 bp) in only B. cepacia complex species (Fig. (Fig.2).2). All reference B. cepacia complex strains used in this study produced the double band, whereas closely related genera and species did not. The Burkf primer was designed to provide more specificity to the 16S rRNA gene-based PCR assay than the combination of eubacterial primer Eub-16-1 with the B. cepacia complex-specific CeMuVi-16-2 (2, 41). Initially, the two primer pairs were mixed at equal concentrations, but this sometimes produced a weaker signal for the 16S rRNA gene band than for the recA band. Comparable band intensities were obtained by using 25% more of the 16S rRNA gene primers than of the recA primers. As a result, a reliable and efficient method to identify all B. cepacia complex species was obtained, eliminating closely related isolates that could produce weak, unspecific fragments (e.g., Fig. Fig.2,2, lanes 10 to 12). This was particularly valuable for high-throughput PCR screening and rapid identification of environmental isolates displaying the characteristic double band (data not shown).
We identified the variable V3 region of the 16S rRNA sequence (positions 430 to 457 in B. multivorans LMG 13010T; accession number Y18703) as the best candidate to differentiate B. cepacia complex species from closely related species and genera, such as B. gladioli, Ralstonia spp., Chryseobacterium meningosepticum, Stenotrophomonas maltophilia, Comamonas acidovorans, and Pseudomonas spp. These bacteria are often found in mixed cultures with B. cepacia complex isolates from either environmental or clinical samples, often have similar growth requirements as B. cepacia complex species, and thus, may often be misidentified as B. cepacia complex species (9). In the B. cepacia complex, sequence polymorphism in the V3 region consisted of two or three mismatches distantly located from each other, whereas more than seven mismatches (i.e., with B. gladioli) were found for non-B. cepacia complex species. Generally, about 10 mismatches were obtained in the genus Burkholderia, with the exception of Burkholderia andropogonis, which displayed 15 mismatches. Other closely related genera listed above displayed either 15 to 20 mismatches or complete deletions (e.g., Chryseobacterium meningosepticum with gaps at seven sites [data not shown]).
Several probes ranging from 30 to 150 bases long, spanning the V3 region of the 16S rRNA, were tested for specificity using PCR-amplified DNA from B. cepacia complex and closely related organisms. Non-B. cepacia complex strains were chosen on the basis of the high similarities of their rRNA gene sequences, since they would be more difficult to differentiate from the B. cepacia complex, and consisted of B. gladioli, B. xenovorans, and Ralstonia spp.
Only probe V3-30 specifically detected all B. cepacia complex species (Fig. (Fig.3A),3A), while longer probes lacked specificity (data not shown). However, the hybridization assay using probe V3-30 was not sensitive enough to satisfactorily detect small colonies growing on dilution plates (Fig. (Fig.3B),3B), for which a weak signal was often difficult to distinguish from the low level of background hybridization signal.
Longer probes were designed by linking together several copies of the V3-30 probe using one or two phosphoramidite spacers (Table (Table1),1), because signal intensity was directly related to probe length due to probe labeling with alkaline phosphatase, but probes of increased sequence length were found to lack specificity. The sensitivities of dimer (V3x2) and trimer (V3x3) probes (Table (Table1)1) were substantially higher than that of V3-30, especially when working with colonies (Fig. (Fig.3B),3B), without losing the specificity of the monomer (data not shown). Generally, the limit of detection of the CH assay using V3x3 ranged from 0.1 to 1 ng, and the limit of detection of the assay using V3-30 or V3x2 was >1 ng (Fig. (Fig.3B;3B; our unpublished data).
The difference in the sensitivities of the V3-30 and V3x3 probes was further confirmed by a preliminary experiment in which 1,100 isolates from a maize rhizosphere sample were randomly selected from TB-T dilution plates, transferred to PCAT plates, and submitted to CH with either V3-30 or V3x3. Among the 700 isolates that were identified as B. cepacia complex strains by using V3-30, only 78% were confirmed by duplex PCR. In comparison, 99% of the 548 colonies identified as B. cepacia complex with V3x3 were duplex PCR positive. This can be explained by the weaker hybridization signals that are generated with V3-30 compared with those of V3x3; hence, when using V3-30, more doubtful colonies had to be picked on the plates to be screened by duplex PCR. In conclusion, probe V3x3 was the best probe in terms of both specificity and sensitivity and was used in subsequent studies on B. cepacia complex population densities in maize rhizospheres.
Seven individual rhizosphere samples were used to determine the B. cepacia complex population densities, species abundance, and diversity in the same field. The number of total bacteria growing on TB-T plates was high, with an average of 6.01 ± 0.83 log (CFU/g of root) (mean ± standard deviation) and with noticeable differences between maize plants (e.g., plants 4 and 7 [Table [Table2]).2]). CH revealed that B. cepacia complex population levels were generally found within 0.5 to 1 log unit below the total number of bacteria growing on TB-T plates, with an average of 5.64 ± 0.84 log (CFU/g of root), very close to the estimates previously found in Italy on maize with 6.20 ± 0.20 log (CFU/g of root) (15). As for total counts on TB-T, B. cepacia complex colonization was found to vary by several orders of magnitude between individual rhizospheres (e.g., plants 4 and 6, with a 140-fold difference). Experimental repeatability that consisted of sampling the same root system at different times ranged from 0.5 to 1.5 log units, depending on the rhizosphere sample considered (data not shown). The percentage of B. cepacia complex isolates recovered from TB-T plates ranged from 17 to 81, which led to the isolation of about 40 ± 18 isolates per rhizosphere for a total of 285 isolates (Table (Table22).
Restriction analysis of recA amplicons with two restriction enzymes (HaeIII and MnlI ) for all B. cepacia complex strains identified by CH was used to determine species abundance and richness. Samples giving patterns that were not identical to those of reference strains were sequenced further to ascertain the correct species (Table (Table22 and Fig. Fig.44).
Sequencing of the recA and 16S rRNA gene amplicons was performed for seven randomly selected isolates belonging to the main recA types and allelic variants to confirm that they were B. cepacia complex isolates and determine their phylogenetic positions. The 16S rRNA gene-based phylogeny indicated that all selected isolates were indeed B. cepacia complex isolates (Fig. (Fig.4A).4A). They clustered with B. ambifaria (KBC-1, -5, -6, and -7), B. cepacia (formerly genomovar I) (KBC-2 and -3) or close to but separately from B. anthina (genomovar VIII) (KBC-4) (Fig. (Fig.4A4A).
The phylogenetic positions of the isolates in the B. cepacia complex were further confirmed by the analysis of recA sequences (Fig. (Fig.4B).4B). Noticeably, while all selected isolates clustered similarly when recA or 16S rRNA gene sequences were considered, KBC-4 was found between B. cenocepacia III-B and III-C clusters in the recA-based phylogeny, with significant nodal support (Fig. (Fig.4B)4B) for this topology. The existence of low bootstrap values between B. cepacia complex species (e.g., between clusters III, V, VI, and VII [Fig. [Fig.4B])4B]) may suggest that nucleotide substitutions at this locus could display some phylogenetic saturation (our unpublished data).
As a whole, one to seven B. cepacia complex species were recovered on each plant rhizosphere with B. ambifaria (formerly genomovar VII) found in all samples (i.e., 64% of all isolates), followed by B. cepacia (genomovar I) with 58 isolates (i.e., 20% of the total number) found on three plants (Table (Table2).2). The other species consisted of B. multivorans (genomovar II) with three isolates found on one plant, B. cenocepacia (genomovar III) with nine isolates found on three plants, B. stabilis (genomovar IV) with a unique isolate, B. dolosa (genomovar VI) found on two plants for a total of 26 isolates, and B. pyrrocinia (genomovar IX) found on two plants for a total of three isolates. To our knowledge, this is the first report of B. stabilis and B. dolosa isolated from the environment. Noteworthy, B. multivorans and B. cenocepacia, although rarely evidenced in our study, are the most frequent species isolated in patients with cystic fibrosis. Indeed, those two species were found in up to 38 and 50%, respectively, of patients with cystic fibrosis in the United States, accounting for mortality rates of up to 15 and 45%, respectively, in cystic fibrosis patients in Canada (42).
Interestingly, in the dominant species, substantial allelic polymorphism was evidenced for recA with two and four alleles for B. cepacia (genomovar I) and B. ambifaria, respectively. This allelic polymorphism was also observed for populations from the same sample, e.g., two recA variants found in genomovar I and three to four recA variants of B. ambifaria on plants 2 and 3 (Table (Table22).
If the abundance of each species is expressed in log10 so as to identify significant difference in species abundance in a given sample, it becomes clear that although up to six B. cepacia complex species may coexist on a given maize rhizosphere (e.g., plant 3 [Table [Table2]),2]), each plant is colonized by a dominant species and/or one variant type. For instance, plant 1 is dominated by B. ambifaria variant A, while other recA variants were three to six times less numerous (Table (Table2).2). The only case where abundance was similar between species in a given sample was for plant 6 with seven to nine isolates for both B. dolosa and B. ambifaria. Note that no direct relationship was found between the location of the sample and the total population density of B. cepacia complex strains, species distribution, and abundance (P > 0.05 by regression [Table [Table22]).
In summary, our results indicate that in each maize rhizosphere, a complex community of B. cepacia complex species and variants of the same species coexist at high population densities. Since recA restriction analysis is rather a coarse method to resolve intraspecific diversity, further studies using DNA fingerprinting may prove very valuable to reveal the level of genetic differentiation of B. cepacia complex species on each individual plant.
Previous studies that examined B. cepacia complex population levels in maize rhizospheres reported that B. cenocepacia and B. ambifaria were predominant, while B. cepacia and B. pyrrocinia were found less frequently (3, 13, 19). This was partly corroborated here because B. ambifaria and B. cepacia were the most dominant in our study, whereas B. cenocepacia and B. pyrrocinia were less frequently isolated (Table (Table2).2). Furthermore, species not yet encountered on maize, such as B. multivorans and B. stabilis, were also found at significant population densities at the KBS site (Table (Table2).2). Therefore, without higher capacity methods, such as proposed here, sampling bias may distort our understanding of the diversity and ecology of B. cepacia complex strains, which may underestimate species distribution and abundance, and hence could potentially miss the identification of true environmental reservoirs.
rRNA sequencing was also done for five randomly chosen isolates that were not identified as B. cepacia complex isolates by any of the methods used in this study but that were often present at high densities on TB-T dilution plates. These organisms belonged to the α, β, and γ subdivisions of the proteobacteria. In the α subdivision, one isolate (S8 [accession number AY769909]) showed 70 to 90% identity of its rRNA sequence with Rhizobium or Agrobacterium. In the β subdivision, isolates S9 (accession number AY769910) and S12 (AY769913) displayed 98% sequence similarity with Burkholderia spp. and Variovorax spp., respectively. Two isolates belonged to the γ subdivision (S10 [accession number AY769911] and S11 [AY769912]) and shared more than 97% identity of their 16S rRNA sequences with Lysobacter spp. and Xanthomonas spp., respectively. The diversity of organisms growing on semiselective media in company with B. cepacia complex strains further underlines the necessity of using an efficient and specific screening strategy to recover B. cepacia complex organisms from environmental samples.
In parallel to the detection by CH, a liquid culture screening assay was developed to determine B. cepacia complex population levels (Fig. (Fig.1B).1B). CH gave the best estimate, since all dilution plates are screened, and positive isolates are further confirmed as B. cepacia complex by duplex PCR, leading to an accurate counting of B. cepacia complex at each dilution level. Liquid culture PCR screening assays gave results that showed good agreement with the results of CH (Fig. (Fig.5).5). Both DE-PCR and MPN-PCR indicated a strong linear relationship (r > 0.90, P < 0.001, R2 = 0.90) with CH estimates of population levels (Fig. (Fig.5).5). DE-PCR and CH gave the same results, ranging from 4.5 to 6.5 log (CFU/g), whereas above 6.5 log (CFU/g), DE-PCR overestimated the abundance of B. cepacia complex. For MPN-PCR, good estimates of population levels were obtained over the range 5.5 to 7 log (CFU/g) with an underestimation below 5.5 log (CFU/g). As a consequence, the three methods agreed with each other in the range 5 to 5.5 log (CFU/g) to 6.5 to 7 log (CFU/g). Below 4.5 to 5 log (CFU/g of sample), liquid culture PCR screening methods gave lower estimates of B. cepacia complex abundance than those of CH. These results are supported by previous studies reporting that MPN counts generally underestimate the actual number of bacteria in a sample (18). Therefore, a quick estimation of B. cepacia complex population abundance can be obtained using liquid culture screening methods in order to decide whether isolation should be performed by randomly picking isolates or by CH in the case of low B. cepacia complex population levels.
The actual detection limit of liquid culture PCR screening is 1,000 CFU/g, given that only 10 μl of the initial 10 ml of sample is inoculated into the wells of the dilution plate (Fig. (Fig.1B).1B). For CH, the detection limit is 10-fold lower (i.e., 100 CFU/g), since it is possible to detect one B. cepacia complex colony on the semiselective medium plate corresponding to the lowest dilution of the sample suspension. Because not finding B. cepacia complex by these methods cannot definitely rule out their presence at population levels below 100 CFU/g (i.e., what is not detected by CH), incubation of the sample in liquid TB-T medium for 4 days prior to plating on TB-T plates (Fig. (Fig.1C)1C) can be used to enrich as few as 1 to 10 cells in the original sample to levels detectable by duplex PCR after DNA extraction of the whole sample (our unpublished results). If the sample gave positive results by PCR, screening of dilution plates by CH may then be successfully used to isolate B. cepacia complex. However, due to the enrichment step, a reliable estimate of the actual population level in the sample is no longer possible. This enrichment step can prove valuable in the case of bulk soil in which B. cepacia complex abundance may range from 102 to 104 CFU/g of soil (24) or even lower (our unpublished results).
In conclusion, a combination of the methods presented here offers efficiency and flexibility, depending on the specific aims of each study. For instance, enrichment and duplex PCR may be used to screen numerous samples in parallel for just the presence or absence of the B. cepacia complex, liquid culture PCR screening assay may estimate B. cepacia complex abundance, and CH provides the best means of isolating the target organisms in pure cultures. Another variation would be to perform CH on the bacterial suspension from each well of liquid culture plates (Fig. (Fig.1B).1B). Indeed, by directly spotting liquid cultures from several plates on hybridization membranes, determining B. cepacia complex population densities in tens of samples in parallel will no long be an obstacle. This will undoubtedly yield more data about the ecology and diversity of the B. cepacia complex in various habitats and environments.
We thank Melissa S. Meyers for excellent technical assistance and Andrew T. Corbin for help in sampling at the KBS site.
This work was supported in part by NSF grant no. DEB-00755564, a postdoctoral fellowship from the Swiss National Science Foundation to A.R., and a grant from the Cystic Fibrosis Foundation to J.J.L.