Population structure analysis of Burkholderia cepacia genomovar III: varying degrees of genetic recombination characterize major clonal complexes

Microbiology (2002), 149, 77–88

DOI 10.1099/mic.0.25850-0

Population structure analysis of Burkholderia cepacia genomovar III: varying degrees of genetic recombination characterize major clonal complexes Tom Coenye3 and John J. LiPuma Department of Pediatrics and Communicable Diseases, University of Michigan, Ann Arbor, MI-48109, USA

Correspondence Tom Coenye [email protected]

Received 4 July 2002 Revised

28 August 2002

Accepted 17 September 2002

Infection with bacterial species belonging to the Burkholderia cepacia complex contribute significantly to morbidity and mortality in persons with cystic fibrosis (CF). The majority of isolates recovered from CF patients belong to B. cepacia genomovar III and several distinct ‘epidemic’ strains have been described. This study examined the population structure of B. cepacia genomovar III by using multilocus restriction typing, indexing allelic variation at five chromosomal genes by restriction analysis of PCR-amplified genes. A collection of 375 isolates, recovered from CF and non-CF patients and natural environments in North America, Europe and Australia, was examined. Among these isolates 144 different restriction types were found. Overall, the population is at linkage disequilibrium, indicating that it has a clonal structure. The majority (86?7 %) of restriction types grouped into three major clonal complexes, comprising the epidemic ET12, PHDC and Midwest clonal lineages. The analysis indicates that these complexes are geographically widespread and demonstrate varying degrees of genetic recombination. These differences in population structure among major clonal complexes within the same species are likely related to differences in evolutionary history and ecology. The observation that genetic recombination is frequent within some B. cepacia genomovar III populations has important implications for the biotechnological use of B. cepacia complex species.

INTRODUCTION Indexing of allelic variation in sets of chromosomal loci provides a basis for estimating overall levels of genotypic variation in populations of microorganisms (Musser, 1996). Differences in the ratio of genetic change introduced by recombinational events relative to de novo mutation lead to a spectrum of population structures, from the extremes of strictly clonal (no recombination has occurred in the evolutionary history of the species, e.g. Salmonella enterica) to non-clonal or panmictic (recombinational events are frequent enough to randomize alleles in a population and prevent the emergence of stable clones, e.g. Helicobacter pylori). However, most bacterial populations occupy a middle ground where recombination is highly significant in the evolution of the population, but is not sufficiently frequent to prevent the emergence of epidemic clonal lineages (e.g. Neisseria meningitidis). Since bacterial populations are complex, single species may not always be adequately described by one model; mixtures of clonal 3Present Address: Laboratorium voor Microbiologie, Universiteit Gent, K.L. Ledeganckstraat 35, B-9000 Gent, Belgium. Abbreviations: CF, cystic fibrosis; DLV, double-locus variant; RT, restriction type; SLV, single-locus variant.

0002-5850 G 2002 SGM

Printed in Great Britain

and non-clonal elements within a given species are likely related to differences in ecology and epidemiology (Musser, 1996; Maiden et al., 1998; Spratt & Maiden, 1999; Feil & Spratt, 2001; Day et al., 2001). Insights into the structure of bacterial populations have proved valuable in enhancing our understanding of the biology of various human pathogenic species, including N. meningitidis, Staphylococcus aureus, Streptococcus pyogenes and Mycobacterium tuberculosis (Musser, 1996; Maiden et al., 1998; Spratt & Maiden, 1999; Feil & Spratt, 2001; Day et al., 2001). Bacteria belonging to the Burkholderia cepacia complex [a group of nine phylogenetically closely related genomic species or genomovars (Coenye et al., 2001; Vandamme et al., 2002)] are important opportunistic human pathogens in persons with cystic fibrosis (CF) or chronic granulomatous disease (Speert et al., 1994; Govan & Deretic, 1996; LiPuma, 1998; Johnston, 2001). Infections in immunocompetent persons occur less frequently, but nosocomial outbreaks and pseudo-epidemics have been reported (LiPuma, 1998). Recent studies demonstrate a markedly disproportionate representation of B. cepacia complex species among isolates recovered from CF patients, with the majority of isolates belonging to B. cepacia genomovar III (LiPuma et al., 2001; Speert et al., 2002). Infection with specific strains of this 77

T. Coenye and J. J. LiPuma

species may be associated with increased rates of morbidity and mortality (Aris et al., 2001; DeSoyza et al., 2001; Ledson et al., 2002). Recent work has also identified B. cepacia genomovar III in the natural environment as a commensal of certain soil types and the rhizosphere of various plants (Balandreau et al., 2001; Fiore et al., 2001). Considerable evidence indicates that B. cepacia complex strains can spread between CF patients. This can occur via simultaneous hospital admission or social contact outside the hospital, and strict infection control measures are recommended to reduce the risk of patient-to-patient transmission (Govan & Deretic, 1996; LiPuma, 1998). Nevertheless, several ‘epidemic’ strains have been described as infecting multiple patients. The ET12 clone, responsible for infecting many CF patients in Canada and the UK, is believed to have spread as a result of contact among infected patients at CF summer camps (Jonhson et al., 1994; Sun et al., 1995; Clode et al., 2000; Ledson et al., 2002). Isolates belonging to this clone are characterized by distinctive cblAencoded cable pili (Sun et al., 1995) and esmR [also called the B. cepacia epidemic strain marker (BCESM)], a 1?4 kb putative ORF with homology to negative transcriptional regulators (Mahenthiralingam et al., 1997). Infection with the ET12 clone has been associated with high rates of mortality (Ledson et al., 2002) and the so-called ‘cepacia syndrome’, which is characterized by sepsis and progressive respiratory failure (Isles et al., 1984). Another epidemic clone, referred to as PHDC (Chen et al., 2001), is responsible for nearly all B. cepacia complex infections in patients attending CF treatment centres in the mid-Atlantic region of the USA. PHDC isolates described to date contain neither cblA nor esmR (Chen et al., 2001). Recently, isolates belonging to this clone have been recovered from agricultural soil samples suggesting that acquisition from the natural environment plays a significant role in the epidemiology of B. cepacia complex infection in CF (LiPuma et al., 2002). A third epidemic B. cepacia genomovar III clone, here referred to as the Midwest clone, is responsible for infecting numerous patients attending CF treatment centres in the midwestern region of the USA (LiPuma et al., 1988; Kumar et al., 1997). The relationships between these epidemic clones, between epidemic and seemingly unique strains, and between clinical and environmental isolates are not clear. A better appreciation of these relationships and the population structure of this species in general would provide critical insight into its epidemiology, evolution and ecology as a human pathogen. To date, the methods that have proved most useful in such studies have been multilocus enzyme electrophoresis (MLEE) and multilocus sequence typing (MLST), in which variation at several genomic loci is indexed through the electrophoretic mobility of the corresponding enzymes or by direct nucleotide sequencing, respectively (Musser, 1996; Maiden et al., 1998). Previous studies documented the utility of MLEE to assess the molecular epidemiology of B. cepacia (McArthur et al., 1988; Johnson et al., 1994; Wise 78

et al., 1995, 1996), but these were performed before the recognition that multiple species constitute the B. cepacia complex and most studies were limited to environmental isolates only. We recently developed multilocus restriction typing (MLRT) as an alternate method for studying the global epidemiology of B. cepacia complex infection in CF (Coenye & LiPuma, 2002). With MLRT, variation at chromosomal loci is indexed by restriction analysis of PCR-amplified genes. Using this method we previously identified several subgroups within B. cepacia genomovar III that seemed to be characterized by differences in rates of recombination. The aim of the present study was to expand these observations and examine in greater detail the population structure of B. cepacia genomovar III. We investigated 375 isolates recovered from various ecological niches, including CF and non-CF patients, and the natural environment in North America, Europe and Australia between 1966 and 2001.

METHODS Sampling strategy. Three hundred and seventy-five B. cepacia genomovar III isolates were included in this study. These were obtained from clinical specimens [both CF (303 isolates) and nonCF (11 isolates)] from the USA (from 28 different states), Canada (from the provinces British Columbia, Alberta, Manitoba, Ontario, Quebec, Nova Scotia and Newfoundland) and Europe (the UK and Belgium). All isolates were from different patients and were isolated between 1966 and 2001. We also included 60 isolates recovered from the natural environment (including soil and plants) in the USA, Canada and Australia (the majority of environmental isolates were previously described by Balandreau et al., 2001 and LiPuma et al., 2002). Isolates were selected [based on known epidemiology (i.e. geography) and genotyping data obtained in previous studies] to include sufficient numbers of the major epidemic lineages (ET12, PHDC and Midwest) previously identified and to include isolates that did not belong to any of the major groups (Kumar et al., 1997; Chen et al., 2001; LiPuma et al., 2001, 2002; Coenye & LiPuma, 2002). Growth conditions and identification of isolates. Strains were

grown aerobically on Mueller–Hinton Broth (Becton Dickinson) supplemented with 1?8 % (w/v) agar and incubated at 32 ˚C. All isolates were identified by recA RFLP fingerprinting as described previously (Mahenthiralingam et al., 2000). For isolates in which species identification remained equivocal following recA RFLP fingerprinting (i.e. if they belonged to RFLP types not previously described), complete recA nucleotide sequences were obtained to confirm that they belonged to B. cepacia genomovar III (Mahenhiralingam et al., 2000; LiPuma et al., 2001). In addition, several isolates were confirmed to belong to B. cepacia genomovar III by extensive phenotypic characterization (Henry et al., 2001), SDS-PAGE of whole-cell proteins and/or DNA–DNA hybridizations (Vandamme et al., 1997) as described previously. Multilocus restriction typing. DNA preparation, PCR amplifica-

tion of nearly complete sequences of the recA, gyrB, fliC, cepIR and dsbA genes, enzymic restriction digests and separation of the resulting restriction fragments were performed as described previously (Coenye & LiPuma, 2002). Data analysis. Gel images were digitized using a GelDoc 2000 (Bio-Rad) and stored as TIFF files. Digitized images were converted

Microbiology 149

Population structure of B. cepacia genomovar III and normalized with Molecular Analyst Fingerprinting Plus software (Bio-Rad). Following normalization we defined a set of bands to each normalized densitometric curve. Similarity between the patterns was calculated using the Dice coefficient as implemented in the Molecular Analyst Fingerprinting Plus software (both tolerance and optimization parameters were set to 1?0 %). Restriction patterns for each locus were considered to be the same if they had the same number of bands and if they demonstrated at least 92 % similarity following cluster analysis using the above-defined parameters. Different restriction patterns for each locus were considered to represent different alleles and each allele was assigned an arbitrary integer. The allelic profile for each isolate (consisting of the allele designation for each locus) was entered into a Microsoft Excel database; each unique allelic profile was assigned an arbitrary integer [restriction type (RT)]. The START software package (Jolley et al., 2001) was used to group the allelic profiles using the BURST algorithm and convert the dataset to a distance matrix in NEXUS format. Distance matrices were used for split decomposition analysis (Bandelt & Dress, 1992; Dopazo et al., 1993) using SplitsTree 3.1 (Huson, 1998). Where necessary, higher resolution of the splits graph was obtained by progressively pruning resolved branches. The standardized index of association (sIA), mean genetic diversity (Hmean) and genetic diversity at each locus were calculated using the LIAN 3.1 software package (Haubold & Hudson, 2000). Difference from linkage equilibrium was statistically tested by a previously described parametric method (Haubold et al., 1998) as implemented in LIAN 3.1 (Haubold & Hudson, 2000). Detection of genes encoding putative transmissibility factors. Isolates were analysed by dot-blot hybridizations and/or

PCR for the presence of cblA and esmR as described previously (Mahenthiralingam et al., 1997; Chen et al., 2001; LiPuma et al., 2001).

used the BURST algorithm (Day et al., 2001; Feil & Spratt, 2001). Using this algorithm, clonal complexes were defined as groups in which each isolate is identical to at least one other isolate at three or more of the five loci. In addition, within each major clonal complex we identified the putative ancestral genotype as the RT differing from the highest number of other RTs within the complex at only one locus. We then identified so-called ‘single-locus variants’ (SLVs) (differing from the ancestral genotype at one locus), ‘double-locus variants’ (DLVs) (differing from the ancestral genotype at two loci) and satellite RTs (defined as a member of the clonal complex but not an ancestral genotype or an SLV or DLV of the ancestral genotype). Secondary ancestral genotypes are SLVs of the ancestral genotype that in turn define more SLVs that have not been previously defined. Most of the isolates (n5354, 94?4 %) grouped into one of 11 clonal complexes, while 21 isolates belonged to 19 RTs that differ from all other genotypes at more than two loci (singleton RTs) (Table 2). The majority of isolates (n5325, 86?7 %) clustered in three major clonal complexes designated RT-6-complex, RT-46-complex and RT-88-complex (clonal complexes are named after their putative ancestral genotype). The relationships between RTs belonging to these three major complexes, as determined by using the BURST algorithm and split decomposition analysis, are shown in Fig. 2. Measures of association

RTs and clonal complexes

We calculated the sIA for all isolates and for all RTs; the values obtained (0?3295 and 0?1752, respectively) were both significantly different from 0 (P,0?0001) (Table 3), suggesting a clonal population structure. However, when we calculated the sIA for the three major clonal complexes separately, we found significant differences in population structure (Table 3, Fig. 2). The RT-6-complex shows evidence of an epidemic population structure: sIA.0 for the whole dataset but the evidence of association disappears when each RT is treated as a single individual entry. The population structure of the RT-88-complex shows evidence of clonality at all levels (sIA.0), while the opposite is true for the RT-46-complex.

A total of 144 RTs were found among the 375 B. cepacia genomovar III isolates examined (Table 2). To group RTs into clusters of closely related RTs (clonal complexes) we

Distribution of genes encoding putative transmissibility factors

RESULTS Diversity of recA, gyrB, fliC, cepIR and dsbA genes The chromosomal loci studied ranged in length from 1043 bp (recA) to 2160 bp (cepIR) and the number of different alleles present per locus ranged from 13 (recA) to 72 (dsbA) (Table 1). The frequency with which each allele occurred in the sample population is shown in Fig. 1; the genetic diversity (H) at each locus is shown in Table 1.

Table 1. Genetic diversity at five chromosomal loci Locus

Fragment size (bp)

No. alleles

Genetic diversity (H)

recA gyrB fliC cepIR dsbA

1043 ¡1200 1000–1200 2160 ¡1800

13 24 30 30 72

0?6806 0?6259 0?6896 0?5560 0?9155

http://mic.sgmjournals.org

The distribution of cblA and esmR is shown in Table 2. Positive results for cblA were only obtained with isolates belonging to the RT-6-complex. Specifically, most were representatives of the ET12 clone (comprising RT-5 and RT17); however, low-proficiency amplification of cblA was observed in several representatives of other RTs belonging to the RT-6-complex. Isolates that were positive for esmR were found in the majority of RTs belonging to either the RT-6complex or the RT-46-complex (77?3 % and 62?5 %, respectively). However, only 27?9 % of the RTs belonging to the RT-88-complex contained isolates that were positive for esmR. 79

T. Coenye and J. J. LiPuma

Fig. 1. Allele frequencies in the B. cepacia genomovar III sample population. For each of the five loci indicated, the number of times that each allele occurs in the isolate collection is shown. The frequencies are shown (from left to right) in order of most to least frequent.

DISCUSSION Population structure of B. cepacia genomovar III The sIA values for all isolates and for all RTs were significantly different from 0. This suggests that the overall population has a clonal structure, that recombination within the total population has been relatively rare and that the population is at linkage disequilibrium (Maynard Smith et al., 1993). Subpopulations within B. cepacia genomovar III Our data clearly show different subpopulations among the B. cepacia genomovar III isolates studied. The RT-6-complex 80

(which contains RT-5 and RT-17, the RTs comprising representatives of the trans-Atlantic ET12 clone) was characterized by an sIA significantly different from 0, indicative of linkage disequilibrium (Table 3). However, when the analysis was repeated with each RT treated as an individual, the evidence for linkage disequilibrium disappeared, indicating an epidemic population structure. This provides strong evidence that this subpopulation is recombining in the long term but that a few RTs have recently become abundant and widespread (Maynard Smith et al., 1993). Using the BURST algorithm, RT-6 was designated as the ancestral genotype; RT-21, RT-31 and RT-41 were designated as secondary ancestral genotypes (Fig. 2). The most frequent RT is RT-17, which contained Microbiology 149

Population structure of B. cepacia genomovar III

Table 2. Properties of B. cepacia genomovar III clonal complexes RT RT-6-complex 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 RT46-complex 46 47 48 49 50 51 52

No. strains

Source

Geographic location*

Year(s) of isolation

cblA3

esmR3

2 1 1 1 1 6 1 1 1 1 1 2 1 1 1 1 28 1 1 1 4 1 3 1 2 1 1 2 1 3 1 1 1 1 1 1 1 1 1 1 1 1 1 1

CF, non-CF CF CF CF CF CF CF CF CF CF CF CF CF CF CF CF CF non-CF non-CF CF CF CF CF CF CF CF CF CF CF CF CF CF CF non-CF CF CF CF CF CF CF CF CF non-CF CF

WI WI OH MA UK MA, TX, OH OH OH MA PA OH OH NY NY MI MA NY, Canada, UK CO MD PA AL, MA, OH, PA AL OH OH OH OH CT AL WA MI, PA, WA OH OH MI NC MI MA TX MI FL AK AK TX MA KY

1998-2001 2000 2001 1997 1993 1997–2000 2000 2000 1999 1999 2000 2001 1999 2001 1999 1998 1992–1997 2000 1999 1997 1999–2000 2000 1999–2000 2000 1999 1997 2000 2000–2001 1999 1999–2001 1998 2001 1999 Before 1970 1998 1997 2001 2000 1999 2000 2000 2001 1998 2000

– (+) (+) (+) + (+) – – – (+) (+) (+) – (+) (+) (+) + (+) – – – – – – – – – – – – – (+) – – – – – – – – – – (+) –

+ + – + + + – – + + + + + – + – + + – + + + + – + + + + + + + + + – + + + + + + + + – –

56

CF, non-CF

1984–2001

–

+

1 1 1 4 1 1

CF CF CF CF CF CF

MI, OH, NC, NE, PA, WA, WI OH TX OH OH OH OH

2000 1998 1998 1998–2001 1998 2001

– – – –

+ + + + + +

http://mic.sgmjournals.org

81

T. Coenye and J. J. LiPuma

Table 2. cont. RT 53 54 55 56 57 58 59 60 61 RT-88-complex 63 64 65 66 67 68 69 74 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 114 115 116 117 118 119 120 121 123 124

82

No. strains

Source

Geographic location*

Year(s) of isolation

cblA3

esmR3

3 1 2 2 1 1 1 1 1

CF CF CF CF CF CF CF CF CF

MI, NC MI MI MI MI MI OK MI MI

1998–1999 2000 2000 2001 2000 1998 2000 2000 2000

–

+ – – – + + – – –

1 1 1 1 1 1 1 1 29 49 3 2 3 6 1 1 1 2 1 7 1 1 1 1 5 1 1 1 1 4 2 6 2 1 2 1 1 1 3 8 1 1 1

CF Soil soil CF CF CF CF CF CF, soil CF, soil CF CF CF CF CF CF CF CF, soil Soil Soil CF CF CF CF CF CF CF CF CF CF, non-CF CF, soil CF, soil CF CF CF CF non-CF Soil Soil Soil Soil CF CF

PA NY NY CT FL CO DC PA DC, MI, PA DC, NY, PA, OH PA, MA PA PA PA PA PA PA KY, NY NY NY MD GA DE DC PA PA PA PA PA DE, MO, NE, PA PA, NY PA, NY CA, NY AZ MI, PA MS USA NY NY NY NY TX Belgium

1998 2000 2000 1997 2000 2000 2001 2000 1981–2001 1989–2001 1997–1999 1998–1999 1997–2000 1997–1998 2000 1997 1997 1998–2000 2000 2000 2000 1998 2000 1997 1998–2000 2001 2000 1997 1998 1999–2001 1997–2000 1997–2001 2000–2001 2000 2001 2000 Before 1966 2000 2000 1999–2000 2000 2001 1993

–

–

–

+

– – – – – –

– + – – – – – – – + – + – – – – – – – – – + + + – + + + – – + + – – –

–

– – –

– – –

– – – – – – –

Microbiology 149

Population structure of B. cepacia genomovar III

Table 2. cont. RT

No. strains

Source

Geographic location*

Year(s) of isolation

1 1 1 1

CF CF non-CF CF

MA MA MA Belgium

2001 1999 2000 1993

4

Australia

1995

–

–

4

Endophytes of lupin and wheat Soil

Australia

1995

–

–

1 1

CF CF

KS DC

1998 1997

– –

– –

1 1 1 1

CF CF CF CF

PA PA CA AZ

1999 2000 1999 1998

– –

– –

–

1 1 1

Soil CF CF

Ontario MA PA

2000 2001 2000

–

–

1 1

CF CF

OH IN

1998 1999

–

– –

2 1

Soil non-CF

PA MO

2000 2000

1 1 1

Soil CF CF

PA Belgium PA

2000 1994 1999

1 1 1 2 1 1 1 1 1 1 1 1 1 1 1 2 1 1 1

Soil CF CF Soil Soil Soil CF CF CF Soil CF CF CF CF CF Soil CF CF non-CF

NY PA NC Australia NY NY FL WI OH NY KS FL OH MI TN Ontario GA PA MI

2000 1999 1998 1995 2000 1999 2001 2001 1998 1999 1997 2000 2001 1998 2000 2000 2000 1998 2000

RT-71-complex 70 71 72 73 RT-77-complex 77 78 RT-81-complex 81 82 RT-84-complex 84 85 86 87 RT-126-complex 126 127 128 RT-132-complex 132 133 RT-138-complex 138 139 RT-141-complex 141 142 143 Singleton RTs 45 62 75 76 79 80 83 122 125 129 130 131 134 135 136 137 140 144 145

cblA3

–

–

esmR3

– – + –

– – –

–

–

– –

+ –

–

– –

– + –

–

–

*If the location is a US state, the two-letter state code is given. 3+, at least one isolate within the RT is positive; –, all isolates tested within the RT are negative; (+), all isolates tested in the RT are weakly positive [low-proficiency amplification compared to the positive control (HI2844, a representative of the ET12 lineage)]. http://mic.sgmjournals.org

83

T. Coenye and J. J. LiPuma

28 isolates belonging to the ET12 clone (recovered from CF patients in Canada, the UK and the USA). The ET12 clone was first described in 1984 (Isles et al., 1984) and our

84

data suggest that its recent worldwide dissemination contributes significantly to the epidemic population structure of the RT-6-complex.

Microbiology 149

Population structure of B. cepacia genomovar III

Isolates belonging to the RT-46-complex were mostly isolated in the Midwestern region of the USA (87?2 % of all isolates were from Michigan or Ohio), but were also recovered from single CF patients in North Carolina, Washington and Pennsylvania. The sIA within this clonal complex was not significantly different from 0 (Table 3), indicating that this subpopulation is at linkage equilibrium (Maynard Smith et al., 1993). The putative ancestral genotype (RT-46) also was the most frequent, containing 71?8 % of all isolates. In contrast to the other major clonal complexes, the population structure of the RT-88-complex showed linkage disequilibrium at all levels (sIA is significantly different from 0, even when RTs are treated as individual entries) (Table 3). This clonal complex comprised several RTs that contained both isolates recovered from CF patients and isolates recovered from soil. We have previously shown that these isolates belong to the PHDC lineage (Chen et al., 2001; LiPuma et al., 2002). The data from the present study showed that this lineage is not restricted to the mid-Atlantic region of the USA but is much more widespread (Table 2). The underlying reasons for the observed differences in population structure among the major clones within B. cepacia genomovar III are at present unclear. The frequency of recombinational events is determined by the availability and efficiency of genetic exchange mechanisms (including transformation, transduction and conjugation) and by ecological factors (different strains must be present within the same niche) (Feil & Spratt, 2001). The distribution of the various exchange mechanisms among different B. cepacia genomovar III lineages is largely unknown, although most members of the B. cepacia complex (including representatives of the ET12 clone) are sensitive to transducing bacteriophages (Nzula et al., 2000). It has also been reported that the genome of B. cepacia complex organisms is rich in insertion sequences that may promote genomic rearrangements and recruit foreign genes (Lessie et al., 1996; Tyler et al., 1996). Our data indicate that isolates belonging to different RTs can be recovered from the same soil samples (data not shown). In contrast, data regarding co-colonization of CF patients with multiple genotypes are limited. The observation that the RT-88-complex is clonal suggests that it has a reduced ability to exchange genetic material with other lineages, that it is spatially isolated from other lineages, or both. Elucidation of the distribution of genetic exchange

mechanisms and the extent of mixed infections would allow a better appreciation of how these factors might contribute to the observed differences in population structure detected within B. cepacia genomovar III. Distribution of genes encoding putative transmissibility factors cblA and esmR are markers that have been associated with B. cepacia complex strains for which there is evidence of interpatient spread (Sun et al., 1995; Mahenthiralingam et al., 1997). However, the exact role and function of these putative transmissibility factors and their distribution within the B. cepacia complex population have yet to be determined. cblA has been found primarily in the ET12 clone and a limited number of other B. cepacia genomovar III strains (Sun et al., 1995; Clode et al., 2000; Speert et al., 2002). Recently, however, several B. cepacia genomovar I strains possessing a variant of this gene were described (LiPuma et al., 2001). In this study we found that the presence of cblA was limited to members of the RT-6complex, most notably representatives of the ET12 clone (RT-5 and RT-17). Isolates belonging to several other RTs within the RT-6-complex (mostly SLVs of RT-17 and/or RT-5) reacted with the primers used for the PCR amplification of cblA but the resulting PCR product was very weak. There is considerable sequence variation in cblA of B. cepacia complex isolates belonging to different species or lineages (Sun et al., 1995; Richardson et al., 2001) and sequence variation in the primer-binding regions of cblA might explain the low-proficiency PCR amplification observed in this study. esmR has previously been found in isolates belonging to the ET12 clone as well as other epidemic lineages and in several non-epidemic isolates (including isolates belonging to Burkholderia multivorans and B. cepacia genomovar I) (Mahenthiralingam et al., 1997; Clode et al., 2000; LiPuma et al., 2001; Speert et al., 2002) but not in the PHDC clone (Chen et al., 2001). Our data show that esmR is widespread both in the RT-6 and RT-46-complexes, but is much less frequent in the RT-88-complex or any of the smaller clonal complexes (Table 2). In addition, esmR appears to be unstable, since there are several RTs that contain both isolates in which this marker is present and isolates in which it is absent (data not shown). Instability of the genomic region containing esmR was already suggested by Mahenthiralingam et al. (1997).

Fig. 2. (on facing page) Schematic representation of the major clonal complexes. (a) RT-6-complex, (b) RT-46-complex, (c) RT-88-complex. Each number represents a restriction type (RT). Data are presented as split graphs and BURST diagrams. Split graphs were progressively pruned (if necessary) to improve resolution (indicated by rectangles and arrows). In the BURST diagrams, the central circle of each clonal complex contains the ancestral clone of each complex, single-locus variants (SLVs) lie within the next concentric circle (solid line) and double-locus variants (DLVs) lie within the outer circle (dotted line). Singlelocus and double-locus differences between RTs are denoted by solid and dotted lines, respectively. Secondary ancestral genotypes are SLVs of the ancestral genotype that in turn define more SLVs that have not been previously defined. Secondary ancestral genotypes and associated SLVs and DLVs were treated as primary ancestral clones and were assigned in order according to the number of SLVs they define. In cases where an isolate defines a secondary ancestral genotype but does not represent an SLV of another ancestral genotype it is shown as a separate circle. http://mic.sgmjournals.org

85

T. Coenye and J. J. LiPuma

Table 3. Measures of association between loci in B. cepacia genomovar III

All RT-6-complex RT-46-complex RT-88-complex

Sample size

Hmean*

VD3

Ve 4

sIA1

Lpara||

Ppara"

Linkage disequilibrium?

375 isolates 144 RTs 87 isolates 44 RTs 78 isolates 16 RTs 160 isolates 43 RTs

0?6935¡0?0604 0?8264¡0?0421 0?3888¡0?1468 0?5036¡0?1597 0?1055¡0?0814 0?2817¡0?1764 0?2968¡0?1250 0?5121¡0?1289

2?2943 1?1596 1?0252 0?6542 0?3379 0?2604 1?1818 1?3641

0?9898 0?6819 0?7571 0?7397 0?3294 0?3892 0?7313 0?9168

0?3295 0?1752 0?0855 20?0289 20?0011 20?0827 0?1540 0?1220

1?0290 0?7145 0?8559 0?8608 0?3932 0?5240 0?8156 1?0915

,0?0001 ,0?0001 ,0?0001 1 1 1 ,0?0001 ,0?0001

Yes Yes Yes No No No Yes Yes

*Hmean, mean genetic diversity. 3VD, observed variance of pairwise differences. 4Ve, expected variance in case of linkage equilibrium. 1sIA, standardized index of association. ||Lpara, calculated 95 % critical value for VD. "Ppara, calculated significance.

Origin of ‘epidemic’ strains B. cepacia complex bacteria first emerged as pathogens in CF in the late 1970s. Although interpatient spread of B. cepacia complex has been demonstrated in a number of studies, it seems likely that initial introduction into the CF population occurred as a result of acquisition from the natural environment. Our finding that the RT-88-complex is characterized by the occurrence of several RTs that contain isolates recovered both from clinical samples as well as environmental samples [mostly agricultural soils planted with onions for several years (LiPuma et al., 2002)] supports this hypothesis. Following initial acquisition of an organism closely resembling RT-88 and/or RT-89, this strain and variants thereof may have then spread through the CF population by interpatient transfer (Chen et al., 2001). The fact that other RTs are also common to CF patients and soil samples suggests that there may have been multiple acquisitions from the environment. Our observation that environmental B. cepacia genomovar III strains can be found in several clonal complexes but not in the RT-6- or RT-46-complex, raises questions about the origin of these two major clonal complexes. Indeed, a recent study looking for B. cepacia genomovar III in the environment failed to recover isolates belonging to either of these two complexes, despite the fact that samples were taken in the same geographic region as where the majority of the RT-6- and RT-46-complex infected patients reside (our unpublished data). A possible explanation for this is that the sites surveyed were significantly different from the agricultural soils surveyed in other studies (Balandreau et al., 2001; LiPuma et al., 2002) and may not be a suitable ecological niche for B. cepacia genomovar III. Nevertheless, initial acquisition of RT-6-complex organisms from the environment probably occurred with an isolate resembling RT-6; subsequent mutations and recombination then gave rise to 86

the ET12 clone which seems to have spread epidemically through the CF population. The fact that cblA is predominantly found in RT-6 and SLVs of RT-6 suggests an important role for this gene in the rapid spread of this group of organisms. Similarly, the initial acquisition of RT46-complex organisms from the environment most likely occurred with an isolate resembling RT-46. The finding of singleton RTs, and the fact that several minor clonal complexes contain isolates recovered both from the environment and from clinical samples (Table 2) also support the hypothesis of frequent environmental acquisition of B. cepacia genomovar III. Implications for biotechnological use of B. cepacia complex organisms Besides being human pathogens, members of the B. cepacia complex have also attracted considerable attention as biocontrol and bioremediation agents. However, the potential hazard of these strains to the CF community is unclear (Holmes et al., 1998; Parke & Gurian-Sherman, 2001). Our data show that the possibility of recombination exists in at least two major clonal complexes of B. cepacia genomovar III and this potential for gene transfer may have important implications for the biotechnological use of B. cepacia complex strains (Parke & Gurian-Sherman, 2001). Whether or not this potential for gene transfer can be extended to other members of the B. cepacia complex remains to be determined, but it is worth noting that interspecies recombinational events have previously been described in other genera, including Neisseria, Salmonella and Streptococcus (Dowson et al., 1990; Bowler et al., 1994; Lan & Reeves, 1996). Implications for infection control To reduce the risk of B. cepacia complex spreading from Microbiology 149

Population structure of B. cepacia genomovar III

patient to patient, stringent infection control measures (including segregation of colonized patients) are recommended (Govan & Deretic, 1996; LiPuma, 1998). Our data and results from previous work (Govan et al., 2000; Balandreau et al., 2001; LiPuma et al., 2002) suggest that environmental acquisition of ‘epidemic’ B. cepacia genomovar III strains may be more common than previously assumed. In cases where good infection control practices do not drastically reduce or eliminate new B. cepacia complex infections, environmental acquistion of this organism should be considered. Recently it was suggested that infection control strategies should be based on the presence or absence of several markers, including cblA and esmR (Clode et al., 2000), but this seems ill-advised since some transmissible strains do not contain either of these markers (Chen et al., 2001). Our data confirm that the presence of cblA and/or esmR are of limited value to predict virulence or transmissibility, since (i) they are not found in all epidemic lineages, (ii) they are not stable within a given lineage and (iii) variation in the sequence of these putative transmissibility markers may complicate their accurate detection.

ACKNOWLEDGEMENTS The authors gratefully acknowledge the cooperation of participating CF centres and microbiology laboratories for the submission of isolates. This work was supported by a grant from the Cystic Fibrosis Foundation (to J. J. L.). T. C. is supported by the Caroll Haas Research Fund in Cystic Fibrosis.

Coenye, T., Vandamme, P., Govan, J. R. W. & LiPuma, J. J. (2001).

Taxonomy and identification of the Burkholderia cepacia complex. J Clin Microbiol 39, 3427–3436. Day, N. P. J., Moore, C. E., Enright, M. C., Berendt, A. R., Maynard Smith, J., Murphy, M. F., Peacock, S. J., Spratt, B. G. & Feil, E. J. (2001). A link between virulence and ecological abundance in natural

populations of Staphylococcus aureus. Science 292, 114–116. DeSoyza, A., McDowell, A., Archer, L., Dark, J. H., Elborn, S. J., Mahenthiralingam, E., Gould, K. & Corris, P. A. (2001). Burkholderia

cepacia complex genomovars and pulmonary transplantation outcomes in patients with cystic fibrosis. Lancet 358, 1780–1781. Dopazo, J., Dress, A. & von Haeseler, A. (1993). Split decomposition: a technique to analyze viral evolution. Proc Natl Acad Sci U S A 90, 10320–10324. Dowson, C. G., Hutchison, A., Woodford, N., Johnson, A. P., George, R. C. & Spratt, B. G. (1990). Penicillin-resistant viridans streptococci

have obtained altered penicillin-binding protein genes from penicillin-resistant strains of Streptococcus pneumoniae. Proc Natl Acad Sci U S A 87, 5858–5862. Feil, E. J. & Spratt, B. G. (2001). Recombination and the population structures of bacterial pathogens. Annu Rev Microbiol 55, 561–590. Fiore, A., Laevens, S., Bevivino, A., Dalmastri, C., Tabacchioni, S., Vandamme, P. & Chiarini, L. (2001). Burkholderia cepacia complex:

distribution of genomovars among isolates from the maize rhizosphere in Italy. Environ Microbiol 3, 137–143. Govan, J. R. W. & Deretic, V. (1996). Microbial pathogenesis in cystic

fibrosis: mucoid Pseudomonas aeruginosa and Burkholderia cepacia. Microbiol Rev 60, 539–574. Govan, J. R. W., Balandreau, J. & Vandamme, P. (2000).

Burkholderia cepacia – friend and foe. ASM News 66, 124–125. Haubold, B. & Hudson, R. R. (2000). LIAN 3.0: detecting linkage disequilibrium in multilocus data. Bioinformatics 16, 847–848. Haubold, B., Travisano, M., Rainey, P. B. & Hudson, R. R. (1998).

REFERENCES

Detecting linkage disequilibrium in bacterial populations. Genetics 150, 1341–1348.

Aris, R. M., Routh, J. C., LiPuma, J. J., Heath, D. G. & Gilligan, P. H. (2001). Lung transplantation for cystic fibrosis patients with

Henry, D. A., Mahenthiralingham, E., Vandamme, P., Coenye, T. & Speert, D. P. (2001). Biochemical and molecular methods for

Burkholderia cepacia complex: survival linked to genomovar type. Am J Resp Crit Care Med 164, 2102–2106.

determining genomovar status of the Burkholderia cepacia complex. J Clin Microbiol 39, 1073–1078.

Balandreau, J., Viallard, V., Cournoyer, B., Coenye, T., Laevens, S. & Vandamme, P. (2001). Burkholderia cepacia genomovar III is a

Holmes, A., Govan, J. & Goldstein, R. (1998). Agricultural use of

common plant-associated bacterium. Appl Environ Microbiol 67, 982–985. Bandelt, H. J. & Dress, A. W. (1992). Split decomposition: a new and

useful approach to phylogenetic analysis of distance data. Mol Phylogenet Evol 1, 242–252. Bowler, L. D., Zhang, Q. Y., Riou, J. Y. & Spratt, B. G. (1994).

Interspecies recombination between the penA genes of Neisseria meningitidis and commensal Neisseria species during the emergence of penicillin resistance in N. meningitidis: natural events and laboratory simulation. J Bacteriol 176, 333–337. Chen, J. S., Witzmann, K. A., Spilker, T., Fink, R. J. & LiPuma, J. J. (2001). Endemicity and inter-city spread of Burkholderia cepacia

Burkholderia (Pseudomonas) cepacia: a threat to human health? Emerg Infect Dis 4, 221–227. Huson, D. H. (1998). SplitsTree: analyzing and visualizing evolutionary data. Bioinformatics 14, 68–73. Isles, A., Maclusky, I., Corey, M., Gold, R., Prober, C., Fleming, P. & Levison, H. (1984). Pseudomonas cepacia infection in cystic fibrosis:

an emerging problem. J Pediatr 104, 206–210. Johnson, W. M., Tyler, S. D. & Rozee, K. R. (1994). Linkage analysis

of geographic and clinical clusters in Pseudomonas cepacia infections by multilocus enzyme electrophoresis and ribotyping. J Clin Microbiol 32, 924–930. Johnston, R. B. (2001). Clinical aspects of chronic granulomatuous

genomovar III in cystic fibrosis. J Pediatr 139, 643–649.

disease. Curr Opin Hematol 8, 17–22.

Clode, F. E., Kaufmann, M. E., Malnick, H. & Pitt, T. L. (2000).

Jolley, K., Feil, E. J., Chan, M. S. & Maiden, M. C. J. (2001). Sequence

Distribution of genes encoding putative transmissibility factors among epidemic and nonepidemic strains of Burkholderia cepacia from cystic fibrosis patients in the United Kingdom. J Clin Microbiol 38, 1763–1766.

type analysis and recombinational tests (START). Bioinformatics 17, 1230–1231.

Coenye, T. & LiPuma, J. J. (2002). Multilocus restriction typing, a novel tool for studying global epidemiology of Burkholderia cepacia complex infection in cystic fibrosis. J Infect Dis 185, 1454–1462.

http://mic.sgmjournals.org

Kumar, A., Dietrich, S., Schneider, W., Jacobson, R., Pouch Downes, F., Robinson-Dunn, B. E., Honicky, R., Smith, J. & Martin, R. (1997). Genetic relatedness of Burkholderia

(Pseudomonas) cepacia isolates from five cystic fibrosis centers in Michigan. Resp Med 91, 485–492. 87

T. Coenye and J. J. LiPuma Lan, R. & Reeves, P. R. (1996). Gene transfer is a major factor in bacterial evolution. Mol Biol Evol 13, 47–55.

transducing bacteriophages. FEMS Immunol Med Microbiol 28, 307– 312.

Ledson, M. J., Gallagher, M. J., Jackson, M., Hart, C. A. & Walshaw, M. J. (2002). Outcome of Burkholderia cepacia colonisation in an

Parke, J. L. & Gurian-Sherman, D. (2001). Diversity of Burkholderia cepacia complex and implications for risk assessment of biological control strains. Annu Rev Phytopathol 39, 225–258.

adult cystic fibrosis centre. Thorax 57, 142–145. Lessie, T. G., Hendrickson, W., Manning, B. D. & Devereux, R. (1996). Genomic complexity and plasticity of Burkholderia cepacia.

FEMS Microbiol Lett 144, 117–128. LiPuma, J. J. (1998). Burkholderia cepacia: management issues and new insight. Clin Chest Med 19, 473–486.

Richardson, J., Stead, D. E. & Coutts, R. H. A. (2001). Incidence of

the cblA major subunit pilin gene amongst Burkholderia species. FEMS Microbiol Lett 196, 61–66. Speert, D. P., Bond, M., Woodmann, R. C. & Curnutte, J. T. (1994).

LiPuma, J. J., Mortensen, J. E., Dasen, S. E., Edlind, T. D., Schidlow, D. V., Burns, J. L. & Stull, T. L. (1988). Ribotype analysis of

Infection with Pseudomonas cepacia in chronic granulomatous disease: role of nonoxidative killing by neutrophils in host defense. J Infect Dis 170, 1524–1531.

Pseudomonas cepacia from cystic fibrosis treatment centers. J Pediatr 113, 859–862.

Speert, D. P., Henry, D., Vandamme, P., Corey, M. & Mahenthiralingam, E. (2002). Epidemiology of Burkholderia cepacia

LiPuma, J. J., Spilker, T., Gill, L. H., Campbell, P. W., Liu, L. & Mahenthiralingam, E. (2001). Disproportionate distribution of

complex in patients with cystic fibrosis, Canada. Emerg Infect Dis 8, 181–187.

Burkholderia cepacia complex species and transmissibility markers in cystic fibrosis. Am J Resp Crit Care Med 164, 92–96.

Spratt, B. G. & Maiden, M. J. C. (1999). Bacterial population genetics,

LiPuma, J. J., Spilker, T., Coenye, T. & Gonzalez, C. F. (2002). An

epidemic Burkholderia cepacia complex strain identified in soil. Lancet 359, 2002–2003. Mahenthiralingam, E., Simpson, D. A. & Speert, D. P. (1997).

Identification and characterization of a novel DNA marker associated with epidemic Burkholderia cepacia strains recovered from patients with cystic fibrosis. J Clin Microbiol 35, 808–816. Mahenthiralingam, E., Bischof, J., Byrne, S. K., Radomski, C., Davies, J. E., Av-Gay, Y. & Vandamme, P. (2000). DNA-based

diagnostic approaches for the identification of Burkholderia cepacia complex, Burkholderia vietnamiensis, Burkholderia multivorans, Burkholderia stabilis and Burkholderia cepacia genomovars I and III. J Clin Microbiol 38, 3165–3173. Maiden, M. C. J., Bygraves, J. A., Feil, E. & 10 other authors (1998).

Multilocus sequence typing: a portable approach to the identification of clones within populations of pathogenic microorganisms. Proc Natl Acad Sci U S A 95, 3140–3145.

evolution and epidemiology. Philos Trans R Soc Lond B Biol Sci 354, 701–710. Sun, L., Jiang, R. Z., Steinbach, S. & 7 other authors. (1995). The

emergence of a highly transmissable lineage of cbl+ Pseudomonas (Burkholderia) cepacia causing CF centre epidemics in North America and Britain. Nat Med 1, 661–666. Tyler, S. D., Rozee, K. R. & Johnson, W. M. (1996). Identification of

IS1356, a new insertion sequence, and its association with IS402 in epidemic strains of Burkholderia cepacia infecting cystic fibrosis patients. J Clin Microbiol 34, 1610–1616. Vandamme, P., Holmes, B., Vancanneyt, M. & 8 other authors (1997). Occurrence of multiple genomovars of Burkholderia cepacia

in cystic fibrosis patients and proposal of Burkholderia multivorans sp. nov. Int J Syst Bacteriol 47, 1188–1200. Vandamme, P., Henry, D., Coenye, T., Nzula, S., Vancanneyt, M., LiPuma, J. J., Speert, D. P., Govan, J. R. W. & Mahenthiralingam, E. (2002). Burkholderia anthina sp. nov. and Burkholderia pyrrocinia,

How clonal are bacteria? Proc Natl Acad Sci U S A 90, 4384–4388.

two additional Burkholderia cepacia complex bacteria, may confound test results of new molecular diagnostic tools. FEMS Immunol Med Microbiol 33, 143–149.

McArthur, J. V., Kovavic, D. A. & Smith, M. H. (1988). Genetic

Wise, M. G., Shimkets, L. J. & McArthur, J. V. (1995). Genetic

Maynard Smith, J., Smith, N. H., O’Rourke, M. & Spratt, B. G. (1993).

diversity in natural populations of a soil bacterium across a landscape gradient. Proc Natl Acad Sci U S A 85, 9621–9624.

structure of a lotic population of Burkholderia (Pseudomonas) cepacia. Appl Environ Microbiol 61, 1791–1798.

Musser, J. M. (1996). Molecular population genetic analysis of

Wise, M. G., McArthur, J. V., Wheat, C. & Shimkets, L. J. (1996).

emerged bacterial pathogens: selected insights. Emerg Infect Dis 2, 1–17.

Temporal variation in genetic diversity and structure of a lotic population of Burkholderia (Pseudomonas) cepacia. Appl Environ Microbiol 62, 1558–1562.

Nzula, S., Vandamme, P. & Govan, J. R. W. (2000). Sensitivity of

the Burkholderia cepacia complex and Pseudomonas aeruginosa to

88

Microbiology 149