Diverse pathogenicity of Burkholderia cepacia complex strains in the Caenorhabditis elegans host model


  • Silvia T. Cardona,

    Corresponding author
    1. Departments of Microbiology and Immunlogy, Infectious Diseases Research Group, Siebens-Drake Research Institute, Dental Sciences Building, Room 3014, The University of Western Ontario, London, Ont., Canada N6A 5C1
      *Corresponding author. Tel.: +1 519 661 3433; fax: +1 519 661 3499., E-mail address: scardona@uwo.ca
    Search for more papers by this author
  • Julia Wopperer,

    1. Department of Microbiology, University of Zürich, Zürich, Switzerland
    Search for more papers by this author
  • Leo Eberl,

    1. Department of Microbiology, University of Zürich, Zürich, Switzerland
    Search for more papers by this author
  • Miguel A. Valvano

    1. Departments of Microbiology and Immunlogy, Infectious Diseases Research Group, Siebens-Drake Research Institute, Dental Sciences Building, Room 3014, The University of Western Ontario, London, Ont., Canada N6A 5C1
    2. Departments of Microbiology and Medicine, Infectious Diseases Research Group, Siebens-Drake Research Institute, The University of Western Ontario, London, Ont., Canada N6A 5C1
    Search for more papers by this author

  • Edited by C. Winstanley

*Corresponding author. Tel.: +1 519 661 3433; fax: +1 519 661 3499., E-mail address: scardona@uwo.ca


A fast screening method was developed to assess the pathogenicity of a diverse collection of environmental and clinical Burkholderia cepacia complex isolates in the nematode Caenorhabditis elegans. The method was validated by comparison with the standard slow-killing assay. We observed that the pathogenicity of B. cepacia complex isolates in C. elegans was strain-dependent but species-independent. The wide range of observed pathogenic phenotypes agrees with the high degree of phenotypic variation among species of the B. cepacia complex and suggests that the taxonomic classification of a given strain within the complex cannot predict pathogenicity.


The genus Burkholderia comprises a diverse group of Gram negative microorganisms that thrive in different ecological niches including soil, water, the rhizosphere, and humans. In particular, the Burkholderia cepacia complex consists of ten closely related species or genomovars that can be isolated from both environmental and clinical sources[1]. Usually non-pathogenic for healthy individuals, B. cepacia complex isolates cause a variety of infections in immunocompromised patients, and in patients with chronic granulomatous disease (CGD) and cystic fibrosis (CF)[1]. Two species of the B. cepacia complex, B. multivorans (formerly genomovar II) and B. cenocepacia (formerly genomovar III), account for the majority of isolates from CF patients in North America [2,3] and Europe[4]. B. cepacia complex isolates differ in utilization of carbon sources, susceptibility to antibiotics [5–7], and the presence of a pathogenicity island[8]. Also, individual differences among strains from the same species have been detected in amoeba[9], murine[10] and plant infection models[11].

The nematode Caenorhabditis elegans, lacking adaptive immunity, is a useful host model for studying innate immune responses to bacterial pathogens [12,13]. This model is genetically tractable from the perspectives of both host and pathogen, and thus, serves to investigate evolutionary conserved mechanisms of microbial pathogenesis and innate immunity [14–17]. A previous study employed a small number of the B. cepacia complex strains and demonstrated that they can cause infection in C. elegans[18], and the infection-like process was characterized using the B. cenocepacia strain H111[19]. However, a systematic analysis of the pathogenic diversity within B. cepacia complex species has not been performed in this model. In this study, we have developed a rapid screening method to characterize in C. elegans the pathogenic phenotypes of a representative number of environmental and clinical B. cepacia complex isolates. We observed a wide range of pathogenic phenotypes and demonstrate that the pathogenicity of B. cepacia complex isolates in C. elegans is strain-dependent but species-independent.

2Materials and methods

2.1Bacterial and nematode strains

Caenorhabditis elegans Bristol N2 and DH26 strains were obtained from the Caenorhabditis Genetics Center, University of Minnesota, Minneapolis. A collection of B. cenocepacia K56–2 transposon mutants with survival defects in the rat lung model of infection[20] was used to develop the 48-well plate mortality assay. A subset of these mutants is listed in Table 1. Strains belonging to the B. cepacia complex are listed in Table 2. All species were represented by at least three isolates with the exception of one single strain of B. ubonensis (genomovar X). Most of the strains from the B. cepacia complex experimental strain panel [21,22] were also included in the study. When available, information on the geographic and biological sources of the isolates was included in Table 2. Most isolates were obtained from North America and Europe. A few strains were obtained from Argentina, Australia, Senegal and Vietnam.

Table 1.  Comparison between the 48-well plate mortality assay and the slow-killing assay using selected B. cenocepacia K56–2 transposon mutants
StrainKnown or putative function of the mutated genea48-Well plate mortality assay of C. elegans Bristol N2 strainSlow-killing assay of C. elegans DH26
  Appearance at day 2b% of live worms at day 2Total number of worms at day 5PSc% of live worms at day 2Median survival (days)dP-valuee
  1. NS, non-significant.

  2. aAs described by Hunt et al.[20].

  3. bThe appearance of worms was scored as sick when impaired locomotion or a distended intestine was found.

  4. cPathogenicity score 0, no signs of disease; 1, 2 and 3, one two or three symptoms of disease, respectively (see Section 2).

  5. dObtained from Kaplan–Meier survival plots (see Fig. 1).

  6. eP values calculated from pair wise comparisons (log rank test) by each transposon mutant strain versus K56–2 wild-type strain.

4A7paaE, ferredoxin reductaseNormal100100–500097Undefined<0.0001
3A3Cation efflux pumpNormal10050–1000713<0.0001
10F1Hypothetical proteinNormal9050–1000733<0.0001
16H8ugpB, glycerol 3-P binding periplasmic proteinNormal7250–1000713<0.0001
28D9Translation initiator inhibitor tdcF and yjgFSick6650–1001963<0.0001
36B4cpxA, capsular polysaccharide export ATP-binding proteinSick10050–1001693<0.0001
38E2hemK, methyltransferaseSick10050–1001833<0.0001
33H3wbiI, epimerase/dehydrataseSick5020–502603<0.0001
34A1Transcriptional regulatorSick5010–2022720.5059 NS
28D8UTP-glucose-1-phosphateSick40032020.2162 NS
20D2d-lactate-dehydrogenase/oxidoreductaseSick47033420.3695 NS
K56–2Parental strainSick2203242
OP50E. coli feeding strainNormal10050–1000100Undefined<0.0001
Table 2.  Pathogenicity of B. cepacia complex strains in C. elegans Bristol N2
SpeciesGVaStrainCommentsAppearance at day 2% of live worms at day 2Number of worms at day 5PSb
  1. Comparable results were obtained in at least two independent experiments.

  2. aGenomovar.

  3. bPathogenicity score as described in Section 2.

  4. cB. cepacia complex experimental strain panel.

B. cepaciaIATCC 25416Onion, U.S.A., BccespcSick6750–1001
  CEP0509CF, Australia, BccespNormal95>5000
  FC124CGD, CanadaSick9450–1001
  MC353Environmental (onion)Sick803
  MC76Environmental (onion)Sick1303
  MC81Environmental (onion)Sick1720–503
B. multivoransII249–2Laboratory, U.S.A., BccespNormal10050–1000
  ATCC 17616Soil, U.S.A., BccespNormal100100–5000
  C0514CF, sputum, CanadaNormal100100–5000
  C3430CF, sputum, CanadaNormal93100–5000
  C4297CF, sputum, CanadaSick8850–1001
  C5274CF, sputum, CanadaNormal10050–1000
  C5393CF, Canada, BccespNormal73100–5000
  C5568CF, sputum, CanadaSick6650–1001
  CEP0108CGD, Lung, U.S.A.Normal100100–5000
  CEP484CF, U.S.A.Normal100100–5000
  FC0147CGD, CanadaNormal100100–5000
  FC0442CGD, blood, U.S.A.Normal100100–5000
  LMG 16660CF, U.K., Bccesp, Glasgow epidemic strainNormal100100–5000
  LMG 16665brain abscess, U.K.Normal100100–5000
  CEP1016CF, U.K.SickND02
  CEP1017CF, U.K.Normal80100–5000
  CEP1019CF, IrelandSick8820–502
B. cenocepaciaIIIBC7CF, Canada, BccespNormal8550–1000
  C1484CF, CanadaSick2903
  C3865CF, CanadaSick9202
  C4455CF, CanadaSick100>5001
  C5424CF, Canada, BccespNormal98100–5000
  CEP024CF, U.S.A., BccespNormal8450–1000
  CEP0931CGD, endotracheal tube, U.S.A.Sick5502
  CEP1067CGD, blood, CanadaSick2803
  CEP511CF, Australia, BccespNormal90100–5000
  CEP054CF, U.S.A.Normal10050–1000
  CP 706-JCF, U.S.A.Sick6050–1001
  F28368-82CF, CanadaSick3120–503
  F38192-89CF, CanadaSick4403
  J2315CF, U.K., BccespNormal9720–501
  K56–2CF, Canada, BccespSick2203
  L10CF, CanadaSick570–102
  PC 527-ICF, U.S.A.Sick3003
  PC 701-JCF, U.S.A.Sick780–102
B. stabilisIVC6061CF, sputum, CanadaNormal5750–1000
  CEP0559CF, CanadaNormal7850–1000
  FC0473CF, BelgiumNormal10050–1000
  LMG 14086respirator, U.K., BccespSick4510–203
  LMG 14294CF, Belgium, BccespNormal80100–5000
  LMG 18870CF, Canada, BccespNormal7450–1000
  LMG 18888Human blood, Belgium, BccespSick460–103
B. vietnamiensisVC2822CF, sputum, CanadaNormal81100–5000
  FC0369Rhizosphere, VietnamNormal10050–1000
  FC0441CGD, CanadaSick46>5002
  CCUG 31370CF, Sweden. BccespNormal100100–5000
  LMG18835CF, U.S.A., BccespSick100100–5001
B. dolosaVICEP021CF, U.S.A.Normal96100–5000
  CFLGCF, ArgentinaNormal9250–1000
  L6CF, CanadaSick8110–202
  LMG 18943CF, U.S.A.Sick6450–1001
  LMG 21443Alysicarpus glumaceus, root nodule, Senegal, BccespSick1000–102
  LMG 21820CF, U.K., BccespNormal8720–501
B. ambifariaVIILMG 17828Corn roots, U.S.A., BccespSick5702
  LMG 19182pea rhizosphere, U.S.A., BccespSick830–202
  LMG 19467CF, Australia, BccespSick403
B. anthinaVIIILMG 16670Carludovica palmata, rhizosphere. U.K., BccespSick3603
  LMG 20980Soil rhizosphere, U.S.A., BccespSick10010–202
  LMG 20983CF, sputum, U.K., BccespSick7502
  LMG 21821CF, U.S.A., BccespSick910–102
B. pyrrociniaIXLMG 14191Soil, Fujisawa Pharm. Co. Bccesp, Patent strainSick490–103
  LMG 21822Cornfield soil, U.S.A., BccespSick2803
  LMG 21823Water, U.S.A., BccespSick1703
  LMG 21824CF, U.K., BccespSick4003
B. ubonensisXLMG 20358Surface soil, Thailand. BccespSick1003

2.2Forty-eight-well plate mortality assay

Forty-eight-well plates containing 600 μl of NG agar[19] were inoculated with 20 μl of overnight bacterial cultures, incubated at 37 °C for 3 h, and then again overnight at room temperature to allow the formation of a bacterial lawn. Five to ten hypochlorite-synchronized L4 larvae of C. elegans strain Bristol N2 were deposited onto each well and incubated at 20 °C. The percentage of live worms and their morphological appearance was registered after two days. The total number of nematodes including the parental worms (if still alive) and the progeny nematodes (if any) was scored after five days. The non-pathogenic Escherichia coli OP50 strain was used as a negative control. From preliminary experiments, comparing infections with E. coli OP50 and B. cenocepacia K56-2, we established that a given strain of the B. cepacia complex was pathogenic for C. elegans if one of the following criteria was met: (i) a sick appearance at day 2, which included reduced locomotive capacity and the presence of distended intestine; (ii) percentage of live worms at day 2 leqslant R: less-than-or-eq, slant 50%; and (iii) total number of worms at day 5 leqslant R: less-than-or-eq, slant 50. The presence of any one, two or three of these criteria was scored as 1, 2, and 3, respectively, differentiating mild from severe infections (Tables 1 and 2). Any given strain was considered pathogenic when at least one criterion was observed (pathogenicity score 1, 2 or 3). Conversely, a strain was considered non-pathogenic when no symptoms of disease were observed during the course of the infection experiment (pathogenicity score 0).

2.3Slow-killing assay

Slow-killing assays were performed using the C. elegans strain DH26[23]. Six-well plates containing 5 ml of NG agar[19] were inoculated with 50 μl of the overnight cultures adjusted to an OD600 of 1.7 and incubated 24 h at 37 °C to allow the formation of a bacterial lawn. Twenty to forty hypochlorite-synchronized L4 larvae of C. elegans strain DH26 were inoculated to each plate and incubated at 25 °C. Each assay was performed in triplicate. Plates were scored for live and dead worms every 24 h. For each killing assay nematode survival was calculated by the Kaplan–Maier method, and survival differences were tested for significance by the use of the log rank test (GraphPad Prism, version 4.0).

3Results and discussion

3.1Validation of the 48-well plate mortality assay

To validate the 48-well plate mortality assay in the C. elegans model as a rapid method for screening large numbers of strains, we exploited our collection of transposon mutants in B. cenocepacia strain K56-2, which display reduced survival rates in the rat agar bead model of lung infection[20]. We reasoned that these mutants would elicit a wide range of pathogenic phenotypes in the C. elegans model, thus, facilitating the assessment of the pathogenicity criteria. These parameters were based on the percent survival and appearance of worms at two days post-infection, and the total number of parental and progeny nematodes after five days, as described in Section 2. As controls, we performed infections with the non-pathogenic E. coli strain OP50 (data not shown) and the parental B. cenocepacia K56–2 whose pathogenic phenotypes were assigned scores of 0 and 3, respectively. From the 75 mutants screened, we found 31 (41%) with a pathogenicity score of 3, comparable to that of the parental K56–2 strain. This suggested that the mutated genes in these strains had no effect in attenuating the infection in C. elegans. In contrast, 44 mutants displayed various levels of attenuation, including 33 mutants with pathogenicity score 2, five mutants with pathogenicity score 1, and 6 mutants with pathogenicity score 0.

To compare the 48-well plate mortality assay with the more established slow-killing assay, we selected a subset of the transposon mutants with different pathogenicity scores (Table 1), which were employed in killing assays using the C. elegans DH26 strain[23]. This nematode strain has a temperature sensitive mutation in the spermatogenesis fer-15 gene. As worms are sterile at 25 °C, it is possible to count the original worms for longer periods of time after infection without the interference of progeny worms. The rate of killing by B. cenocepacia K56–2 on C. elegans DH26 at 25 °C and on C. elegans Bristol N2 at 20 °C was indistinguishable (data not shown). The B. cenocepacia K56–2 mutants examined in C. elegans DH26 also exhibited different degrees of pathogenicity (Fig. 1) that generally correlated well with the results of the 48-well plate mortality assay (Table 1). The transposon mutants with pathogenicity scores 0 and 1 in the 48-well killing assay were highly attenuated with respect to the K56–2 wild-type strain when tested in the slow-killing assay (P-value < 0.0001). While the mutant 4A7 did not show any significant nematode killing activity over a 5-day period, C. elegans infected with mutants 3A3, 10F1, 16H8, 28D9, 36B4, and 38E2 displayed a median survival of three days (Table 1). In contrast, mutants 28D8 and 20D2 (pathogenicity score 3) were as pathogenic as K56–2 (Table 1). The mutant 33H3 (pathogenicity score 2) showed an attenuated phenotype (P-value < 0.0001 and median survival of three days) in the slow-killing assay. However, the percent of live worms at day 2 was lower than the rest of the transposon mutants that also showed an attenuated phenotype and had pathogenicity scores 0 and 1. The only exception to this correlation was the mutant 34A1, which showed a pathogenicity score 2 in the 48-well mortality assay but was not significantly different from K56–2 in the slow-killing assay. This disparity could be due to an overestimation in the number of counted worms at day 2 in the 48-well plate mortality assay. Therefore, we conclude from these observations that the 48-well plate mortality assay correlates with the slow-killing assay. Further studies are underway in our laboratory to characterize in detail the function of the mutated genes in relation to infection in C. elegans.

Figure 1.

Kaplan–Meier survival plots of selected transposon mutants. The killing ability of wild-type B. cenocepacia K56–2 strain (n= 76) was compared with that of the STM-mutants 4A7 (n= 109), 3A3 (n= 77), 10F1 (n= 71), 16H8 (n= 72), 28D9 (n= 87), 36B4 (n= 45), 38E2 (n= 35), 33H3 (n= 109), 34A1 (n= 110), 28D8 (n= 71) and 20D2 (n= 93) in slow-killing assay experiments using C. elegans DH26 strain. n, Number of worms at day 0. Dashed lines with squares, B. cenocepacia K56–2 strain; solid lines with triangles, any given transposon mutant of K56–2 strain (indicated in the title of each graphic).

3.2Pathogenicity of clinical and environmental B. cepacia complex strains in the C. elegans model

The pathogenicity of representative strains from the B. cepacia complex was screened with the 48-well plate mortality assay. B. cepacia (genomovar I) was represented by six strains, four environmental isolates and two clinical strains isolated from CF and CGD patients. The pathogenic phenotypes for these isolates were not homogenous, ranging from the non-pathogenic CEP0509 to the highly pathogenic strains MC353, MC76 and MC81. Fourteen of the 18 B. multivorans (genomovar II) strains were non-pathogenic. Two isolates, C4297 and C5568 caused a mild infection in C. elegans while strains CEP1016 and CEP1019 were the most pathogenic. B. cenocepacia (genomovar III) strains were highly diverse with respect to their pathogenic phenotypes. Even the strains K56-2, J2315 and BC7, which are considered to be clonal[21], exhibited different pathogenic phenotypes. Strain K56–2 had the highest degree of pathogenicity while J2315 only caused a slight decrease in the number of progeny worms. This could be attributed to the differences in O-antigen expression between strains K56–2 and J2315[24]. However, reconstitution of O-antigen production in J2315 by complementation did not render the strain more pathogenic in C. elegans (data not shown). Hence, the differences between these two clonal strains in the pathogenicity for C. elegans might be explained by differences other than O-antigen production. Five of the seven B. stabilis (genomovar IV) isolates were highly attenuated, while the degree of pathogenicity of isolates from B. vietnamiensis (genomovar V) and B. dolosa (genomovar VI) was diverse. All isolates from B. ambifaria (genomovar VII) B. anthina (genomovar VIII), B. pyrrocinia (genomovar IX), and B. ubonensis strain LMG20358 were pathogenic for C. elegans. But we cannot conclude that all the strains from these species are pathogenic given that we only had a limited number of strains available for this study.

The distribution of pathogenic and non-pathogenic strains according to their genomovar classification is summarized in Fig. 2. On average, the proportion of strains having a pathogenic phenotype was highest in B. cepacia and B. cenocepacia, while B. multivorans and B. stabilis strains were the least pathogenic (Fig. 2). Environmental isolates appeared to be more pathogenic for C. elegans than the clinical isolates (78% of the environmental isolates and 52% of the human disease isolates). However, it cannot be concluded that the clinical and environmental isolates necessarily differ in their capacity to cause disease in humans. More likely, the observed differences may reflect the adaptation of clinical strains to the lung environment with the concomitant loss of other characteristics required for colonization and infection of C. elegans.

Figure 2.

Distribution of pathogenicity of B. cepacia complex isolates in the C. elegans Bristol N2 model. The strains shown in Table 2 were grown in NG agar plates at 37 °C and the 48-well mortality assay was performed. Percent of pathogenic and non-pathogenic strains was calculated for each genomovar (see Section 2). Numbers over the bars represent the total number of isolates in each genomovar.


In this study, we demonstrate that B. cepacia complex strains show a great diversity of pathogenic phenotypes for C. elegans. This variability, which also applies to strains within the same species, could reflect either loss or acquisition of accessory genetic material, which may provide functional diversity among individual strains. Therefore, the assignment of an environmental or clinical strain as a given species or genomovar does not predict the potential risk for infection. We also identified B. cenocepacia K56–2 transposon mutants that were non-pathogenic in both nematodes and rats. Thus, at least some of the survival-associated properties of B. cenocepacia are common to both nematodes and mammalian hosts, as it has been demonstrated for Pseudomonas aeruginosa[14,25]. A detailed analysis at the molecular level of the pathogenicity of B. cepacia complex strains in C. elegans will provide additional clues to better understand the adaptation of these microbes to multiple environments.


We thank M. Köthe for technical help with the C. elegans model, D. Henry and P. Vandamme for assisting with the classification of genomovar isolates, S. Thyssen for assisting with the statistical analysis, and Karen Keith for critical reading of the manuscript. S.T.C. was supported by a Postdoctoral Fellowship Award from the Canadian Cystic Fibrosis Foundation. This study was supported by operating grants from the Canadian Institutes of Health Research and the Canadian Cystic Fibrosis Foundation. M.A.V. holds a Canada Research Chair in Infectious Diseases and Microbial Pathogenesis.