OUP user menu

Improved flagellin genotyping in the Burkholderia cepacia complex

Craig Winstanley
DOI: http://dx.doi.org/10.1016/S0378-1097(03)00721-3 9-14 First published online: 1 December 2003


Oligonucleotide primers designed to N- and C-terminal sequences of Burkholderia cepacia complex fliC genes were used to amplify and sequence fliC genes from a strain of Burkholderia vietnamiensis and three isolates of Burkholderia multivorans with large fliC genes. Alignments incorporating the new sequences enabled the design of polymerase chain reaction (PCR) primers for extension of a published PCR/restriction fragment length polymorphism typing method, to include isolates that previously failed to yield fliC amplicons. Most B. vietnamiensis isolates and hitherto non-typable Burkholderia cenocepacia isolates contained much smaller fliC genes than previously reported. Although B. multivorans strains with larger fliC genes clustered together, relationships between strains based on fliC sequences did not generally correlate with genomovar status.

  • Burkholderia cepacia
  • Flagellin
  • fliC

1 Introduction

What is now known as the Burkholderia cepacia complex was subdivided by DNA–DNA hybridisation, whole-cell protein pattern similarity and phenotypic markers into a number of genomic species or genomovars. The B. cepacia complex currently comprises at least nine recognised species, all of which have been isolated from patients with cystic fibrosis (CF) [1].

Members of the B. cepacia complex are also associated with other mainly nosocomial opportunistic infections [24]. Although isolates of Burkholderia cenocepacia (formerly genomovar III) and Burkholderia multivorans predominate amongst clinical isolates, there can often be significant numbers of isolates falling into other genospecies. For example, in a recent study of mainly non-CF clinical isolates from Brazil, 10% of the isolates were identified as Burkholderia vietnamiensis [5].

The bacterial flagellin gene (fliC) is a highly variable biomarker and has been targeted in a number of bacteria as an indicator of genetic diversity [6]. Previous studies have indicated that the majority of isolates of B. cepacia could be classified into two types on the basis of flagellin gene and protein size [7]. Type I and type II flagellins were reported as approximately 55 kDa and 45 kDa respectively. Type I and type II fliC amplicon sizes of 1.4 kb and 1.0 kb respectively were obtained following polymerase chain reaction (PCR) amplification with the oligonucleotide primer pair BC4/BCR12. In addition, type I flagella were wider in diameter than type II. One strain (E197) yielded a larger amplicon (1.6 kb) [7]. In contrast, a much earlier study had reported smaller flagellins (31 kDa) in some strains [8]. Considerable heterogeneity amongst the fliC genes of the B. cepacia complex has been demonstrated [7,9]. Although some gene sequence data were obtained [7], inter-strain variation was observed mainly by using PCR/restriction fragment length polymorphism (RFLP) analysis of B. cepacia fliC amplicons. In these studies, the genomovar designations were known for only a few of the strains. More recently, fliC PCR/RFLP genotyping was applied to a panel of strains representative of the B. cepacia complex [10]. The genotyping of characterised strains confirmed the heterogeneity of fliC, indicated a predominance of type II flagellins, and suggested that type I flagellins were found only in true B. cepacia (genomovar I) and B. multivorans. However, the oligonucleotide primers BC4/BCR12 did not amplify the fliC genes of B. vietnamiensis [10]. These primers are also ineffective when applied to Burkholderia dolosa (formerly genomovar VI) or Burkholderia ambifaria (unpublished observation). In a study of isolates from Brazil, fliC PCR amplification was unsuccessful for some isolates from genomovars that have not previously led to such failures, including B. cepacia and B. cenocepacia [5]. fliC RFLP typing has been included in a multilocus restriction typing scheme for analysis of B. cenocepacia [11,12]. The inclusion of fliC in such schemes is currently limited by the inability to obtain amplicons from some members of the B. cepacia complex. The aim of this study was to improve the fliC genotyping method to include strains and genospecies from which fliC amplicons had not been obtained previously.

2 Materials and methods

2.1 Bacterial strains

The bacterial strains used in this study are listed in Table 1. Strain E242 was identified as B. cepacia genomovar I and strains E197 and E243 were identified as B. multivorans using recA-based PCR tests [13]. The isolates from Brazil [5] were also assigned to genomovars using the same PCR tests. In keeping with its designation as genomovar I, strain BC43 lacks type III secretion system genes [5]. B. multivorans strains G2, K1 and W1, isolated from CF patients, were kindly provided by Dr Jane Turton, Health Protection Agency, UK.

View this table:
Table 1

Strains used in this study

StrainSpecies (genomovar)DetailsfliC amplicon size (kb)fliC GenBank accession numberSource or reference
E242B. cepacia (I)CF isolate1.4AF011371[7]
BC43B. cepacia (I)clinical (non-CF) isolate, Brazil0.7AY331141[5]
E243B. multivorans (II)botanical strain1.0AF011370[7]
E197B. multivorans (II)CF isolate1.6AF011372[7]
G2B. multivorans (II)CF isolate1.9AY331136J. Turton, Health Protection Agency
K1B. multivorans (II)CF isolate1.9AY331135J. Turton, Health Protection Agency
W1B. multivorans (II)CF isolate1.4AY331137J. Turton, Health Protection Agency
BC6B. cenocepacia (IIIA)clinical (non-CF) isolate, Brazil0.7 (a)-[5]
BC10B. cenocepacia (IIIA)clinical (non-CF) isolate, Brazil0.7 (b)AY331140[5]
BC37B. cenocepacia (IIIA)clinical (non-CF) isolate, Brazil0.7 (a)-[5]
BC38B. cenocepacia (IIIA)clinical (non-CF) isolate, Brazil0.7 (c)-[5]
BV1B. vietnamiensis (V)clinical (non-CF) isolate, Brazil0.7 (d)-[5]
BV2B. vietnamiensis (V)CF isolate, Brazil0.7 (d)-[5]
BV3B. vietnamiensis (V)clinical (non-CF) isolate, Brazil0.7 (d)-[5]
BV20B. vietnamiensis (V)clinical (non-CF) isolate, Brazil0.7 (d)-[5]
PC259 (F655)B. vietnamiensis (V)CF isolate1.4AY331138[16]
LMG16232 (F764)B. vietnamiensis (V)CF isolate0.7 (d)-[16]
FC441 (F767)B. vietnamiensis (V)CGD isolate0.7 (e)-[16]
LMG10929 (F656)B. vietnamiensis (V)rice isolate0.7 (e)AY331139[16]
BCC232B. dolosa (VI)unknown--E. Mahenthiralingam
BCC207B. ambifaria (VII)unknown--E. Mahenthiralingam
  • Where known, fliC RFLP groupings for 0.7-kb amplicons are indicated in parentheses using letter designations a-e.

  • Chronic granulomatous disease.

2.2 PCR amplification and fliC RFLP typing

One or more colonies of bacteria were taken from growth on nutrient agar plates and resuspended in 50 µl of sterile distilled water. The suspensions were incubated for 5 min at 95°C to obtain a lysed bacterial suspension. Following centrifugation to remove cell debris, 1 µl of crude DNA preparation was used immediately in PCR amplification, carried out in 50-µl volumes containing 2.5 units of Taq DNA polymerase (Helena Biosciences), 1×TaqMaster (Helena Biosciences), 200 nM each primer (Sigma-Genosys), 1×Taq buffer, 2.5 mM MgCl2 and 100 µM nucleotides (dATP, dCTP, dGTP, dTTP) for 30 cycles consisting of 95°C (1 min), annealing temperature (1 min) and 72°C (2 min). The annealing temperatures used for oligonucleotide primer sets fliCF (5′-CTCGGAATYAAYAGCAACA-3′)/fliCR (5′-TTAYTGCAGGAGCTTCAG-3′) and BVF (5′-TCGGGCAAGCGCAYCAACA-3′)/fliCR were 55°C and 58°C respectively.

Amplified product samples (5 µl) were digested with the restriction enzyme BsuRI (HaeIII) or a combination of AluI and MspI, using the conditions recommended by the supplier (Helena Biosciences). These digests were subjected to electrophoresis on 3% (w/v) MetaPhor agarose gels (Flowgen) alongside 1-kb Plus size marker (Life Technologies).

2.3 DNA sequence analysis

Nucleotide sequences were obtained from PCR amplicons by Lark Technologies, using the same oligonucleotide primers employed in the PCR amplification and internal primers. Sequences were edited and aligned using the GCG sequence analysis software package (Genetics Computer Group, University of Wisconsin, Madison, WI, USA). BLASTN and BLASTX searches were conducted using the site http://www.ebi.ac.uk. fliC amplicon sequences have been deposited in GenBank under the accession numbers shown in Table 1.

Sequence alignments were carried out using the ClustalW program (http://www.ebi.ac.uk). Alignments were edited and evolutionary relationships were determined using the genetic distance-based neighbour-joining algorithms of the Data Analysis in Molecular Biology software (DAMBE; http://web.hku.hk∼xxia.software.htm). Sequence input order was randomised and 100 data sets were examined by bootstrapping resampling statistics. Phylogenetic trees were drawn using TreeExplorer software (http://evolgen.biol.metro-u.ac.jp/TE/TE_man.html). The fliC sequence of Burkholderia pseudomallei E503 has been reported previously [14]. B. cenocepacia strain J2315 and Burkholderia fungorum fliC sequences were retrieved from the genome sequence databases http://www.sanger.ac.uk/Projects/B_cepacia/ and http://genome.ornl.gov/microbial/bcep/ respectively.

2.4 Electron microscopy

Carbon-reinforced, formvar-coated copper specimen support grids (400 square mesh) were coated with a drop of bacterial suspension fixed in 1% glutaraldehyde. The grid was air-dried then washed three times in a drop of distilled water to remove crystalline salts. It was then negatively stained using 1% EM grade potassium phosphotungstate, pH 7.0 (Agar Scientific, Stansted, UK). Grids were examined with a Philips 301 electron microscope.

3 Results

3.1 PCR amplification of fliC genes

On the basis of alignments between the fliC sequences of strains E242, E243, E197 and J2315 the oligonucleotide primers fliCF (forward) and fliCR (reverse) were designed for amplification of virtually the whole fliC gene from B. cepacia complex strains. The forward primer incorporates all but the start ATG codon of the fliC gene and the reverse primer is designed to incorporate the stop TAA codon. The fliCF/fliCR primer pair was tested on various members of the B. cepacia complex, including two B. multivorans strains with abnormally large fliC genes (G2 and K1), a B. multivorans strain of fliC type I (W1) and several strains of B. vietnamiensis. Of the latter, only strain PC259 (F655) yielded a PCR amplicon of approximately 1.4 kb, equivalent to a type I fliC gene. Nucleotide sequences obtained from the fliC amplicons of strains G2, K1, W1 and PC259 were included in further alignments with previously published B. cepacia complex fliC sequences. These alignments were used to design an alternative forward primer, BVF, commencing 96 bp downstream of the start of the fliC gene. The primer set BVF/fliCR was used to amplify fliC gene amplicons of approximately 0.7 kb from each of the B. vietnamiensis strains listed in Table 1 (Fig. 1). Thus, strain PC259 was revealed as having an atypically large fliC gene compared to the other B. vietnamiensis strains. An amplicon from a representative of the smaller B. vietnamiensis fliC genes (strain LMG10929 (F656)) was sequenced.

Figure 1

fliC amplicon size variations in the B. cepacia complex. The gel contains amplicons derived from B. cepacia strains E242 (lane 1) and BC43 (lane 2), and B. multivorans strains E243 (lane 3), E197 (lane 4) and G2 (lane 5), produced using the BVF/fliCR primer set. M: 1-kb Plus size marker (Life Technologies).

The BVF/fliCR primer set was further tested on a number of B. cenocepacia isolates from Brazil that had previously failed to yield fliC amplicons [5]. Atypical, smaller PCR products of approximately 0.7 kb were obtained from each of these strains, namely BC6, BC10, BC37 and BC38. The fliC amplicon obtained from strain BC10 was sequenced as a representative of this group. The 16S rRNA gene has also been amplified and sequenced from strain BC10. The best match was against sequences obtained from B. cenocepacia (genomovar III) (for example, 969/970 bp identity with the sequence GenBank AF311970), thus confirming the genomovar status of BC10. In addition, a B. cepacia (genomovar I) strain, BC43, that was PCR-negative with fliC primer set BC4/BCR12, yielded an amplicon of approximately 0.7 kb with primer set BVF/fliCR. This fliC amplicon was also sequenced. The fliC genes of strains BC10 and BC43 shared 98.8% identity over the 686 bp of common nucleotide sequence obtained. Fig. 1 demonstrates the extent of fliC amplicon size variation in the B. cepacia complex, including variations within the same genomovar.

3.2 Restriction digestion of fliC amplicons

Amplicons from those B. vietnamiensis and B. cenocepacia isolates with smaller fliC genes were digested using BsuRI and a combination of AluI and MspI. Some discrimination was achieved between both B. vietnamiensis and B. cenocepacia isolates (Table 1). B. cenocepacia strain BC10 produced a different BsuRI RFLP pattern from strains BC6, BC37 and BC38 (Fig. 2). Digestion with AluI and MspI produced slightly different groupings, with BC10 matching BC38 and BC6 matching BC37 (data not shown). B. vietnamiensis strains BV1, BV2, BV3, BV20 and LMG16232 all shared the same fliC BsuRI RFLP pattern, whereas strains FC441 and LMG10929 shared the same alternative fliC BsuRI RFLP pattern (Fig. 2). The same groupings were obtained using digestion with a combination of AluI and MspI (data not shown).

Figure 2

BsuRI digestion of fliC amplicons. The gel shows RFLP patterns derived from the fliC amplicons of B. cenocepacia strains BC10 (lane 1), BC38 (lane 2), BC6 (lane 3) and BC37 (lane 4), and B. vietnamiensis strains BV1 (lane 6) and FC441 (lane 7). M: 1-kb Plus size marker.

3.3 Relationships based on FliC sequences

Predicted proteins derived from fliC sequences of the B. cepacia complex were aligned together and with sequences from B. pseudomallei, B. fungorum and Escherichia coli. The alignments confirmed the domainal structure of flagellin genes and proteins, whereby N- and C-terminal regions are conserved, with the central region subject to considerable variation in both size and sequence. Alignments using 123 amino acid residues from the N-terminal conserved region (commencing at an alanine residue, position 40 on the FliC protein sequence, and ending just prior to the point at which gaps appeared in the alignment) were used to construct a phylogenetic tree (Fig. 3). With the exception of a cluster of B. multivorans strains, relationships based on FliC did not correlate well with genomovar designation. Trees based on whole FliC sequences produced similar relationships (data not shown).

Figure 3

Phylogenetic tree based on N-terminal flagellin protein sequences.

4 Discussion

A fliC gene of a size indicative of type I was found in only one of the B. vietnamiensis isolates. All others were of the smaller fliC gene size also found amongst previously non-typable isolates of B. cepacia and B. cenocepacia. The existence of this group supports the earlier observations of smaller flagellins (31 kDa) in some strains of the B. cepacia complex [8]. This group was not identified in subsequent studies, which included the representative panel of strains [7,9,10]. However, it is now clear that the smaller fliC genes were present amongst B. vietnamiensis isolates in the representative panel, but were not detected because these isolates proved PCR-negative with the primers used previously.

It has been demonstrated previously that there are flagellar diameter variations in the B. cepacia complex [7]. This variation can occur between or within genomovars. Fig. 4 demonstrates the increased diameter of B. multivorans K1 flagella compared with B. multivorans E243 flagella. As reported previously [7], the greater diameter correlates with greater flagellin protein size.

Figure 4

Negatively stained electron microscopy of flagella from B. multivorans strains E243 (left) and K1 (right). Magnification 150 000.

PCR/RFLP analysis was tested for the ability to discriminate between isolates carrying the smaller fliC gene. Inevitably, there were fewer restriction fragments than observed in previous studies, but some discrimination between B. cenocepacia isolates and between B. vietnamiensis isolates was possible. Using recA RFLP analysis (with HaeIII, which recognises the same target sequence as BsuRI), strains BC10 and BC38 share the same pattern, whereas BC6 and BC37 share a different pattern ([5]; data not shown). The same grouping was observed when fliC amplicons were digested with a combination of AluI and MspI, but the additional use of BsuRI subdivides the groups further. B. vietnamiensis strains BV1 and BV3 also share the same recA RFLP pattern (pattern B [13]). Strain BV2 did not yield a recA amplicon, whilst strains BV20 and LMG16232 had different recA HaeIII RFLP patterns ([5]; data not shown). Thus, since strains BV1, BV2, BV3, BV20 and LMG10929 shared common fliC RFLP patterns, recA RFLP gave better discrimination for these strains. In addition, B. vietnamiensis strains FC441 and LMG10929 share the same fliC HaeIII RFLP pattern, but lie in different recA HaeIII RFLP groups [13]. Strain FC441 shares recA HaeIII pattern A with strains LMG16232 and PC259, whilst strain LMG10929 shares recA HaeIII pattern B with strains BV1 and BV3 (as defined by Mahenthiralingam et al. [13]). Thus, there was not much concordance between RFLP typing based on recA and fliC.

The relationships inferred from alignments of flagellin protein sequences were not generally in accordance with genomovar designations. For example, the smaller fliC sequences of B. cepacia BC43, B. cenocepacia BC10 and B. vietnamiensis LMG10929 aligned more closely to each other than to fliC sequences from strains within the same genomovar. Similarly, B. cenocepacia J2315 and B. multivorans E243, both fliC type II strains, were most closely related to each other. Most strikingly, B. cepacia BC43 was identical to B. cenocepacia BC10 over the region used for the alignment. Indeed, the three sequences representative of 0.7-kb amplicons clearly clustered together despite the fact that they came from strains representing different genomovars. Only the B. multivorans fliC sequences typical of type I or larger genes aligned in a manner that might be anticipated from their genomovar status. These observations were made despite the fact that the alignments used for constructing the phylogenetic tree included only a 123-amino acid residue sequence from the N-terminal conserved regions. Thus, recombination in the central, variable region could not be responsible for these discrepancies. The relationships observed were more typical of horizontal gene transfer involving most of or the entire gene. The GC contents of fliC, ranging from 59.5% for strain W1 to 63.8% for strain E243, were consistently below the genome average for B. cepacia (66.9% for strain J2315; http://www.sanger.ac.uk/Projects/B_cepacia/). This observation further supports the involvement of gene transfer events in the evolution of fliC. Although this might preclude the use of fliC as a reliable molecular marker for phylogenetic studies, it makes the gene an interesting marker for studies seeking information about gene flux amongst B. cepacia complex populations.

Some of the strains used were assigned to genomovars on the basis of recA-specific PCR tests. These tests have been subjected to thorough evaluation. In the study of Vermis et al. [15], the assays for B. multivorans and B. cenocepacia had specificities of 100%. The assay for B. cepacia (genomovar I) was less specific but all of the false positives obtained were Burkholderia pyrrocinia. Even if strain BC43 was really a B. pyrrocinia strain rather than B. cepacia, this would not alter the fact that the FliC sequences do not cluster according to genomovar status.

The design of new primers has extended the range of fliC PCR/RFLP typing in the B. cepacia complex. In particular, the method can now be applied to B. vietnamiensis and to atypical isolates of B. cepacia and B. cenocepacia. The test now needs to be further validated using a larger collection of isolates. This method can still not be considered universally applicable to the B. cepacia complex. No amplification was achieved with B. dolosa BCC232 or B. ambifaria BCC207. Thus, there is still further variation to be characterised in the fliC genes of the B. cepacia complex. It might not be feasible to design a single primer set for amplification of fliC genes from all members of the B. cepacia complex, but further studies may lead to the design of multiplex PCR tests using ‘cocktails’ of primers.


I would like to thank Brian Getty for carrying out the electron microscopy, and Jane Turton, UK Health Protection Agency, for providing strains. I would also like to acknowledge funding from the United Kingdom Cystic Fibrosis Trust.


  1. [1].
  2. [2].
  3. [3].
  4. [4].
  5. [5].
  6. [6].
  7. [7].
  8. [8].
  9. [9].
  10. [10].
  11. [11].
  12. [12].
  13. [13].
  14. [14].
  15. [15].
  16. [16].
View Abstract