Molecular Comparison of Isolates of Burkholderia multivorans from Patients with Cystic Fibrosis in the United Kingdom

Abstract

Burkholderia multivorans strains from 47 cystic fibrosis (CF) patients in 28 hospitals were compared by pulsed-field gel electrophoresis (PFGE) and flagellin (fliC) PCR-restriction fragment length polymorphism (PCR-RFLP) analysis. A considerable degree of genetic variation was evident, with each patient harboring a strain with a unique PFGE profile. Four sizes of fliC amplicons were produced, and these amplicons gave 13 RFLP types with restriction enzyme MspI. B. multivorans did not appear to spread between patients, suggesting that most CF patients acquire the organism from the natural environment.

Bacteria of the Burkholderia cepacia complex are recognized as important opportunistic pathogens among patients with cystic fibrosis (CF). Colonization of the lungs of CF patients with these organisms is of great concern, since it is associated with a decrease in long-term survival; in addition, a minority of patients may develop “cepacia syndrome,” which leads to an acute clinical decline that is frequently fatal (4). The B. cepacia complex comprises a number of genomically distinct species (1) (presently, nine have been described), of which B. cepacia genomovar III, now called B. cenocepacia (17), and B. multivorans (genomovar II) are the most commonly cultured from the sputa of CF patients (3, 7, 9, 10, 15). Certain strains of B. cenocepacia, for example, strains of the electropherotype 12 (ET12) clonal lineage, responsible for infecting many patients with CF in the United Kingdom and Canada, are associated with high transmissibility between patients, and CF patients colonized with these strains have a poorer prognosis and higher mortality than CF patients who have not been infected (10) and are normally segregated from other CF patients. While much attention has understandably been focused on this genomovar, much less information is available on B. multivorans, and much of this information is conflicting.

In contrast to B. cenocepacia, B. multivorans does not appear to spread from patient to patient, and mortality associated with this organism was minimal in comparison with that associated with B. cenocepacia, according to studies of patients with CF in Canada (9, 15). However, a sibling pair did transiently share the same strain. Similarly, there was no evidence of nosocomial spread of B. multivorans among the Danish CF population (9, 13). Nevertheless, both an outbreak among children with CF in a hospital in Glasgow, Scotland, which was associated with some deaths (18), and a further outbreak in Cardiff, Wales, involving four adult patients (11) were attributed to B. multivorans (9). Segonds et al. (14) described two transmissible strains of B. multivorans (identified by restriction fragment length polymorphism [RFLP] ribotyping) in multiple CF centers, and one of these strains was associated with fatal septicemia. In a more recent paper, Heath et al. (5) compared isolates of B. multivorans from 26 patients treated at a lung transplantation center in North America and found that isolates from 4 of these patients were identical on the basis of pulsed-field gel electrophoresis (PFGE) results. Those authors were able to identify overlapping periods of hospitalization when transmission between these patients could have occurred. However, transmissibility markers associated with epidemic strains of B. cenocepacia (the cblA gene, associated with the ET12 clone [16], which encodes the protein for cable pilus production, and the B. cepacia epidemic strain marker) have not been found in isolates of B. multivorans (3, 7). With such conflicting reports, more information would clearly be helpful in determining appropriate strategies for the clinical management of patients infected with B. multivorans.

In our experience, B. multivorans is strongly represented among Burkholderia species isolated from CF patients. During the first 11 months of 2002, our laboratory received Burkholderia isolates (B. cepacia complex organisms and other Burkholderia species) from 111 CF patients, of which 43 (39%) were infected with B. multivorans and 32 (29%) were infected with B. cenocepacia. Genomovars were differentiated by using recA-targeted PCR (8). It has been suggested that B. multivorans infection occurs predominantly in children (9). Our data do not support this suggestion. The ages of half of the above-mentioned patients at the time of sampling are known; the mean age was 22.1 years (range, 8.3 to 45.4 years). The ages of these patients were not markedly different from the ages of patients infected with other members of the B. cepacia complex or from those of patients infected with B. gladioli.

To assess their genetic relatedness, 55 isolates of B. multivorans obtained between 2001 and early 2003 from 47 patients in 28 hospitals across the United Kingdom were compared by PFGE. They were also typed by flagellin gene PCR-RFLP analysis. PFGE was carried out as outlined by Kaufmann (6) by using a two-step lysis procedure in which agarose-embedded cells were lysed in a solution containing 6 mM Tris-HCl, 100 mM EDTA, 1 M NaCl, 0.5% (wt/vol) Brij 58, 0.2% (wt/vol) sodium deoxycholate, 0.5% (wt/vol) N-lauroyl sarcosine, and 1 mM MgCl₂ (pH 7.5) containing 0.5 mg of lysozyme ml⁻¹ at 37°C overnight, followed by incubation in 0.5 M EDTA-1% (wt/vol) N-lauroyl sarcosine (pH 9.5) containing 60 μg of proteinase K ml⁻¹ at 56°C overnight. Following extensive washing, the agarose-embedded DNA was digested with XbaI and the macrorestriction fragments were separated by PFGE on a CHEF DRII apparatus (Bio-Rad Laboratories, Hemel Hempstead, United Kingdom) in a 1.2% agarose gel in 0.5× Tris-borate-EDTA buffer at 12°C. A linear ramp of 5 to 35 s was used, and gels were run for 30 h at 6 V cm⁻¹. Following staining with ethidium bromide, gels were photographed and stored as TIFF files. Gel images were analyzed with BioNumerics software (Applied Maths, Kortrijk, Belgium), and the percentage of relatedness was calculated by use of the Dice coefficient. The unweighted pair group method with arithmetic averages was used for clustering to produce a dendrogram with a band position tolerance of 0.6%. Flagellin typing was performed essentially as described by Coenye and LiPuma (2), except that a Taq PCR core kit (QIAGEN, Crawley, United Kingdom) was used. Reaction mixtures (25 μl) contained 1× PCR buffer, 1× Q solution, a 250 μM concentration of each deoxynucleoside triphosphate, 1.5 mM MgCl₂, 20 pmol each of primers BC4 and BCR12, 1 U of Taq DNA polymerase, and 3 μl of extracted DNA. The final total concentration of MgCl₂, including that in the PCR buffer, was 3 mM. For DNA extraction, three to five colonies were suspended in 100 μl of tissue culture water, heated in a Dri-block apparatus at 100°C for 5 min, and centrifuged at 10,000 × g for 5 min. Three microliters of the supernatant was used in the PCR. Following amplification, 8 μl of the PCR product was digested with 5 U of MspI in a total volume of 20 μl at 37°C for 4 h. PCR products (5 μl) and restriction fragments (10 μl) were analyzed by agarose gel electrophoresis in 1.2 and 1.5% gels, respectively. A 123-bp ladder (Invitrogen, Paisley, United Kingdom) was used as a size marker. All of the isolates in the study were PCR positive with the B. multivorans species-specific primers BCRBM1 and BCRBM2, described by Mahenthiralingam et al. (8). To ensure that all included isolates were indeed B. multivorans, they were tested by this method both on receipt and at the end of the study. To protect patient and hospital identities, isolates were labeled with a letter(s) representing a hospital (A through Z, AA, and BB) and a number representing a patient at that hospital (1 through 6). When more than one isolate was obtained from a patient, these isolates were in addition given small Roman numerals in the order in which they were received.

The PFGE results (Fig. 1) showed a high degree of genetic diversity among isolates from different patients. The gel images, as well as the dendrogram, are included in Fig. 1 to allow comparison by eye. This diversity was in marked contrast to the constancy of the DNA profiles of isolates of B. cenocepacia of recA subgroup III-A (data not shown), many of which are representative of the ET12 lineage prevalent in the United Kingdom (12). The only isolates that clustered at greater than 80% similarity were those from the same patient; the remaining isolates, all from different patients, gave unique fingerprints. Some isolates from different patients (isolates A1 and K2 and isolates D3 and D4i) did cluster together at relatively high percentages of similarity (between 70 and 80%). Closer inspection revealed at least seven different band differences between isolates A1 and K2; isolates D3 and D4i, which were from patients from the same hospital, did show similarity, although there were at least five band differences between them. These isolates were unusual among the set in having some very large bands (approximately 510 kb). However, when these two isolates were compared by using SpeI (data not shown), eight band differences were seen. The faint bands of high molecular weight in the XbaI PFGE profiles of some isolates (e.g., D3 and the D4 isolates) persisted even when digests were carried out three times in succession, indicating that they were not due to partial digestion; these bands were not scored in the analysis. Of interest is that some band differences were seen between isolates from the same patients (C2, D4, and Y1 isolates), suggesting a relatively high rate of genetic change in this organism.

FIG. 1.

Dendrogram of XbaI PFGE profiles of B. multivorans isolates, showing percentages of similarity of banding patterns. Isolates were labeled by a letter(s) representing the hospital, a number representing the patient, and, where multiple isolates from the same patient were received, small Roman numerals. The flagellin type of each isolate is also given (last column).

It has been suggested that variation in housekeeping genes may provide a better index of genotypic variation in microbial populations than PFGE, since events such as insertions, deletions, and inversions in the genome can dramatically alter PFGE profiles and obscure similarity between epidemiologically related isolates (2). The isolates discussed above were therefore also typed by PCR-RFLP analysis at the flagellin (fliC) locus. PCR-RFLP analysis at the recA and gyrB loci was also carried out for some isolates, but little variation was observed (data not shown). The results obtained (Fig. 2) revealed a high degree of variation in the fliC genes of these isolates, with 13 different RFLP types being resolved with enzyme MspI alone. Four different approximate size classes of amplicons were identified (Fig. 2a), with most isolates producing amplicons of approximately 1.0 kb (type 1). This amplicon size corresponds to the type II flagellins described by Winstanley et al. (19). Isolates from 30 of the 47 patients (64%) produced amplicons of this size. Amplicon sizes of approximately 1.3 kb (type 2), 1.4 kb (type 3), and 1.9 kb (type 4) were also seen, with type 2 being the second-most-common size class. The amplicon size types were numbered according to increasing size. The 1.4-kb amplicons presumably correspond to the type I flagellins described by Winstanley et al. (19), who suggested that this amplicon size is characteristic of B. multivorans, although it can also be found in genomovar I. Nevertheless, the present study clearly shows that the amplicon size varies among B. multivorans isolates. Within each amplicon size type, a number of RFLP patterns were evident (Fig. 2b). Despite repeated attempts, two isolates (isolates I1 and R1) failed to produce fliC amplicons. The flagellin type of each isolate is given on the dendrogram in Fig. 1. All isolates from the same patient had the same flagellin type, but most isolates from different patients from the same hospital did not share the same flagellin type. The exceptions were the two isolates from hospital T (both of type 1b) and the isolates from patients D3 and D4, which were identified as being of the unusual 2d type. Type 1a was represented twice among isolates from patients from hospitals A and X, as was type 1b among isolates from patients from hospital X, but considering that these were the most common types, this was not surprising. Flagellin typing was useful and confirmed the PFGE results in indicating a high degree of genetic variation among the isolates. The extent of variation seen within this single gene is remarkable, and had we been successful in combining flagellin typing with PCR-RFLP analysis at other loci showing similar variation, this could have been a highly discriminatory typing approach.

FIG. 2.

Flagellin PCR-RFLP analysis of B. multivorans isolates. (a) fliC amplicons produced by isolates. The four different size classes (indicated above the lane numbers) obtained are shown. Amplicons from representatives of each of the different RFLP types (except type 2e) are included. Isolates were X4 (lane 1), F1 (lane 2), N1 (lane 3), P1 (lane 4), Y1v (lane 5), C1 (lane 6), X1 (lane 7), B1 (lane 8), M1 (lane 9), A3ii (lane 10), G1 (lane 11), W1 (lane 12), O1 (lane 13), and K1 (lane 14). Molecular size markers (123-bp ladder) are shown in lanes M. (b) RFLP types identified by using MspI as the restriction enzyme. The RFLP type assigned to each pattern is given above the lane number. Isolates were X4 (lane 1), E1 (lane 2), N1 (lane 3), X3 (lane 4), A2 (lane 5), Y1v (lane 6), C1 (lane 7), B1 (lane 8), M1 (lane 9), A3ii (lane 10), G1 (lane 11), W1 (lane 12), O1 (lane 13), and K1 (lane 14). A type 2e amplicon was also found (data not shown). Molecular size markers (123-bp ladder) are shown in lanes M.

It seems clear from our results that there is not a specific clone or clones of B. multivorans circulating in the United Kingdom, as is the case for the ET12 clone of B. cenocepacia (10, 12). Unfortunately, our study did not involve large numbers of patients from single hospitals. The data that we do have on such patients indicate that each patient harbors a distinctive strain. Since completing the present study, we have received isolates from a further five patients from hospital D and compared them with the others by PFGE; all 10 patients from this hospital harbored unique strains (data not shown). Although there was some similarity between the XbaI PFGE profiles of the isolates from patients D3 and D4 and they shared the same fliC types, the SpeI PFGE profiles of these isolates were clearly different, supporting the conclusion that these two patients harbored different strains. Certainly, there was no evidence of an outbreak strain associated with the hospital. As far as we are aware, our isolates were from patients attending CF clinics rather than from hospitalized patients. It is likely that while transmission of B. multivorans between patients, such as that between siblings, as described by Mahenthiralingam et al. (9), or that between hospitalized patients, as described by Heath et al. (5), may occasionally occur, most patients acquire the organism from the natural environment rather than from one another. Note that replacement of B. multivorans by transmissible strains of B. cenocepacia (of the III-A recA subgroup) has been reported for some patients (9) and that it is therefore important that patients with B. multivorans are segregated from those with these strains.