Molecular Analysis of Burkholderia cepacia Complex Isolates from a Portuguese Cystic Fibrosis Center: a 7-Year Study

Bacterial strains and culture.The 122 putative B. cepacia complex bacteria examined in this study were isolated from the respiratory secretions of 23 patients with CF who attended the Santa Maria Hospital CF Center in Lisbon from January 1995 to June o2002. This CF center is attended by 85% of the CF population living in the Lisbon area, in the south of Portugal, and in the Madeira and Azores islands. Both adult and pediatric patients receive care at this CF treatment center (Tables 1 and 2). Sputum samples are obtained whenever patients are periodically examined to monitor their clinical status. The systematic study was only launched in 1997, but the isolates obtained before and kept by the hospital were also included. The 1996 isolates could not be examined because they were lost due to a prolonged storage period. All the serial isolates obtained from a same patient, during the surveillance period, went through the same molecular analysis. The identification of isolates as members of the B. cepacia complex was carried out by a triphasic analysis: (i) growth on the selective B. cepacia solid medium (Selectatab; Mast Diagnostics, Merseyside, United Kingdom); (ii) positive identification using the commercial system API 20NE (Biomerieux, Marcy L'Etoile, France); (iii) molecular identification by PCR assays based on the rRNA operon (3). Among the 122 isolates tested, only the 113 isolates tested positive for B. cepacia complex are listed in Table 1. Strain J2315 (representative of the highly epidemic genomovar III-A ET12 lineage) (13), kindly provided by J. R. W. Govan, was used as a reference strain.

Bacterial cultures were maintained on Pseudomonas Isolation Agar plates (Difco, Detroit, Mich.), and liquid cultures were grown overnight in Luria-Bertani broth (Sigma), with orbital agitation, at 30°C.

Ribotyping of B. cepacia isolates.Isolation of total DNA from B. cepacia isolates was carried out using the miniprep procedure described by Ausubel et al. (2). DNA restriction with EcoRI (Gibco BRL), separation of DNA fragments using agarose gel electrophoresis, DNA blotting, and hybridization with [γ-³²P]ATP-labeled 16S and 23S rRNA from Escherichia coli were carried out using standard protocols (25). Radioactive labeling of 16S and 23S rRNA was carried out using the 5′-DNA Terminus Labeling system from Amersham following the manufacturer's instructions. Lambda phage DNA restricted with HindIII was used as the molecular mass standard. The sizes of the hybridization restriction fragments were estimated with DNA Simdex software from GenetX. Isolates producing ribopatterns that did not match others within the strain collection were considered to be unique.

Genomovar status identification.Preliminary identification of genomovar status was performed by a PCR-based procedure for the identification of sequence motifs within the 16S and 23S ribosomal DNA, which uses three sets of primer pairs with different specificities, excluding single species in a stepwise manner (16, 35).

Genomovar status was determined by amplification of the entire recA gene using primers specific for bacteria belonging to the B. cepacia complex (19) and by performing analysis of the recA gene restriction fragments length polymorphisms (RFLPs) with the endonuclease HaeIII (19). Genomovar assignment was confirmed by PCR of the recA gene with genomovar-specific primers (19). Subdivision of genomovar III isolates into distinct recA-based phylogenetic subgroups was performed as described previously (19). For isolates whose species identification remained equivocal, complete recA gene sequences were obtained and analyzed as described elsewhere (19); sequences were aligned and compared with recA sequences from B. cepacia complex strains analyzed in previous studies, using CLUSTAL W (29) and the data analysis software in the Molecular Biology package (DAMBE; http://web.hku.hk/∼xxia/software/software.htm).

Detection of transmissibility markers.The detection of the two putative transmissibility markers cblA and BCESM on the genome of the B. cepacia isolates was carried out by PCR, according to standard techniques. The 1.4-kb BCESM fragment was amplified with primers BCESM1 (5′-CCACGGACGTGACTAACA-3′) and BCESM2 (5′-CGTCCATCCGAACACGAT-3′) (18); the 664-bp cblA gene was amplified using primers PILI1 (5′-CCAAAGGACTAACCCA-3′) and PILI2 (5′-ACGCGATGTCCATCACA-3′) (24). Amplification reactions were carried out by an initial step of denaturation, for 5 min at 95°C, followed by 30 cycles of 1 min at 95°C for denaturation, 1 min at primer annealing temperature, and 2 min at 72°C for polymerization. Primer annealing was carried out at 63°C for detection of BCESM or at 55°C, for detection of cblA. Isolate J2315 (13) was used as the positive control in both amplification experiments. The amplification mixtures contained, in a total volume of 50 μl, 200 ng of DNA template, a 0.4 mM concentration of each deoxynucleoside triphosphate, 2.5 mM MgCl₂, a 4 μM concentration of each primer, 2.5 U of Taq polymerase (Biotaq Taq Polymerase; Bioline), and 5 μl of reaction buffer (10×; supplied by the polymerase manufacturer).

Nucleotide sequence accession data.The recA gene sequences for isolate IST431 (RFLP type AG) and isolates IST438 and IST445 (both RFLP type AN) have been submitted to GenBank under the following accession numbers, respectively: AF456062, AF456063, and AF456123.

Incidence and prevalence rates of infection with B. cepacia complex bacteria.From January 1995 to June 2002, 122 putative B. cepacia complex isolates were obtained from the respiratory secretions of 23 CF patients receiving care at the Santa Maria Hospital in Lisbon. The bacteria were isolated at the hospital, during routine bacteriological analysis, onto B. cepacia selective agar and then were subjected to phenotypic analysis, using a commercial identification system. After genotypic analysis based on the ribosomal DNA and recA genes, performed outside the hospital by the partner laboratories in this study, nine of the isolates examined were found to be from outside the complex.

The molecular analysis reported in this work was focused on the 113 isolates (Table 1) confirmed to belong to the B. cepacia complex. Within the 7-year surveying period, the incidence and prevalence of infection with B. cepacia complex bacteria, ranging from 1.2 to 7.8% and from 7.1 to 13.3%, respectively, could only be accurately determined from January 1998 to June 2002 (Table 2). The age of the infected CF patients at the time of first isolation of B. cepacia complex bacteria varied between 2 months and 19 years, with an estimated mean age of 7.7 years. Interestingly, although the CF population examined includes an identical percentage of males and females, during this 7-year study, the B. cepacia complex-infected patients were predominantly (70%) of the female gender. Patients were considered to be persistently infected if at least three positive cultures for B. cepacia complex bacteria persisted for a 6-month period; according to this criterion, more than one-half of the 23 patients infected with B. cepacia complex were persistently infected during the period of surveillance. Eight of the CF patients under study (34.8%) died in the course of the study; five of these patients were infected with B. cepacia complex bacteria at time of death: three of them developed the cepacia syndrome, while two patients also harbored Pseudomonas aeruginosa strains prior to death.

Ribotyping analysis.The genomic relatedness of the 113 B. cepacia complex isolates from 23 CF patients was characterized by ribotyping with EcoRI (Fig. 1). A significant percentage of the isolates under study were serial isolates collected from 14 patients, 12 of whom were persistently infected. Sixteen distinct ribopatterns, schematically presented in Fig. 1 and numbered from 1 to 16, were generated, suggesting a wide diversity in the sources of infection. More than one-half of the infected CF patients were found to harbor a unique strain, as would be expected in a CF treatment center with an isolation policy. Strains giving rise to ribopatterns 1 and 2 were the most frequent among the Portuguese CF subpopulation examined, being isolated from four and five patients, respectively. Two small clusters of possible nosocomial infection during 1995 were identified involving the isolates which gave rise to ribopatterns 2 (patients A, E, and F) and 3 (patients A and D), also indistinguishable by RFLP-pulsed-field gel electrophoresis analysis, as reported before (22). Strains exhibiting ribopatterns 2 or 3 were not isolated again until 2001 and 2002, when they were obtained from patients W, V, and X, who were hospitalized but never simultaneously. Strains with ribopatterns 1, 7, 9, 11, or 14 were also isolated from two or more patients, but based on the existing clinical records, the occurrence of any spreading event involving these patients inside the hospital is unlikely. Although social contact outside the hospital is strongly discouraged for these patients, it cannot be excluded.

• Open in new tab
• Download powerpoint

FIG. 1.

Schematic representation of the ribopatterns of B. cepacia complex clinical isolates examined in this study. Ribopatterns are indicated by numbers across the bottom of the figure, and the molecular sizes of the EcoRI hybridization bands are arranged in a linear scale at left.

Remarkably, two sisters living together (patients N and O), were persistently infected with different strains, as indicated by the different ribopatterns produced by the isolates obtained over 2.5 years. Only in 2002 was the strain with ribotype 12, which persistently infected patient N, isolated from her sister (patient O), strongly suggesting recent transmission. However, since we have only studied one of the colonies grown on B. cepacia selective medium obtained from the sputum of each patient, at a specific date of isolation, the sisters may have been infected with isolates with common genotypes that were not detected. During this 7-year study, persistent colonization with isolates of single ribotypes was the most common situation. Less than 20% of the patients harbored two or more different strains, and among them, only patient A was transiently infected. The five patients whose deaths were linked to B. cepacia infection were infected with single unrelated strains.

Prevalence of B. cepacia complex species and genomovars.The identification of the isolates recovered from Portuguese CF patients at the species or genomovar level was based on the polymorphisms of the recA gene with HaeIII, using specific primers (19). Nine distinct HaeIII-derived RFLP patterns were obtained (Table 1). The RFLP types F (B. multivorans); G, H, I, and AN (all corresponding to genomovar III); and J (B. stabilis) (Table 1) were previously described (19). RFLP type K was identified by Vermis et al. (34) as genomovar I. However, subsequent results revealed a considerable heterogeneity among type K strains, necessitating further analysis of the homogeneity of this taxon by means of DNA-DNA hybridization experiments (P. Vandamme and E. Mahenthiralingam, unpublished observations). In Table 1, we tentatively classified type K isolates as genomovar I. Two new recA RFLP profiles were obtained and designated AG and H2 (Table 1). Strains corresponding to these new RFLP types failed to react with genomovar-specific primers. Subsequent recA sequence analysis indicated that RFLP type AG corresponds to B. cepacia genomovar I by phylogenetic analysis (data not shown). The taxonomic status of the strain possessing RFLP type H2 was not known, but its reactivity with recA primers BCR1 and BCR2 (19) confirmed that it is a member of the B. cepacia complex.

The 112 B. cepacia complex isolates examined were found to belong to genomovars I and III and to the species B. multivorans and B. stabilis (Table 1). The genomovar status of isolate IST426 could not be determined because this isolate was lost. In agreement with previous reports (1, 17, 27), B. cenocepacia (genomovar III) was the most prevalent species or genomovar in the Portuguese subpopulation examined, being isolated from 52.2% of patients infected with the B. cepacia complex (Table 3). These isolates were separated into subgroups III-A (46.3%) and III-B (53.7%) with no significant differences. B. multivorans, the second most predominant CF pathogen among the three CF populations examined so far (1, 17, 27), was recovered from two patients only (9%). Contrasting with previous studies, a significant percentage of the Portuguese CF subpopulation examined was infected with B. cepacia genomovar I (36%; eight patients) and B. stabilis (18%; four patients) (Table 3).

Among the few patients infected with more than one strain, one patient harbored two different strains, both from B. cenocepacia or genomovar III (recA lineages III-A and III-B); two patients harbored strains of B. cepacia genomovars I and III; and one patient harbored strains belonging to the species B. cenocepacia, B. multivorans, and B. stabilis (Table 1).

Putative transmissibility marker genes.The presence of the cblA and BCESM genetic markers in the 112 B. cepacia isolates was investigated by PCR. The cblA gene was detected in 13.4% of the isolates (Table 3), corresponding to 2 of the 16 strains examined (12.5%). The cblA-positive isolates were recovered from three infected patients and belong to genomovars I and III (12.5 and 0.9% of the 112 isolates examined, respectively) (Table 3). The seven strains found positive for the presence of the BCESM marker (25% of the 112 isolates examined) belong exclusively to B. cepacia genomovar III (Table 3). The esmR gene was found in one-half of the genomovar III isolates, being significantly more frequent in isolates from genomovar III subgroup III-A (35.2%) than those from subgroup III-B (16.7%) (Tables 1 and 3). Among the 113 isolates examined, no isolate was found to exhibit both cblA and esmR genes. The study of several isolates, most of them serial isolates of the same strain (with a common ribopattern), indicates that the presence of the BCESM region is not a stable trait (Table 1). The instability of the BCESM region was registered in both genomovar III subgroups (Tables 1 and 3), but no significant differences were detected in the two subgroups, as reported before (21).

Clinical outcome and epidemiology in relation to genomovar and transmissibility markers.During the 7-year surveillance period, three patients died with symptoms of the cepacia syndrome. From one of these patients (patient G), three different strains were isolated over 8 months: two of them belonging to genomovar III-B, both being BCESM-positive, and the other isolate belonging to genomovar I; close to the time of death, an isolate belonging to genomovar III-B was isolated from this patient. Another patient deceased from the cepacia syndrome (patient J) simultaneously harbored B. multivorans and genomovar III-A strains. Notably, the third patient (patient H) who developed the cepacia syndrome was persistently infected with B. stabilis. Interestingly, this same B. stabilis strain (ribopattern 1) chronically infected another patient (patient B) for 5 years and was apparently eradicated. The same B. stabilis strain was once transiently isolated from patient J, who is now deceased, and more recently was isolated from patient U, who was infected over 7 months with no apparent impact on the clinical status. The high prevalence of B. stabilis among the Portuguese CF subpopulation under study results from the presence of this same strain in 4 of the 23 CF patients under surveillance. Although this may suggest its spread among the patients, there is no evidence of patient contact within the hospital or of any other way of nosocomial infection.

Two other deceases (patients N and T) were registered during the surveillance period, following progressive deterioration during persistent infection with a unique strain of genomovar I (during 5 years) or with a BCESM-positive genomovar III-B strain (for more than 1 year). The ribopatterns exhibited by these two strains were not detected in any other isolate recovered from other patients, with one exception. The isolate IST4128 (genomovar I, cblA-positive), which chronically infected patient N over a 5-year period, has recently been recovered from her sister (patient O). Notably, although these sisters lived together and most of the isolates recovered from patient N harbor the giant cable pilus subunit gene, cross-infection with this strain was never detected previously. During this period, two other strains of genomovar III-A (BCESM positive) and III-B were isolated from patient O, whose clinical situation has currently deteriorated. During the surveying period, other patients (patients E, F, and Q), who had been transiently infected with strains of genomovar I and III-A, died, but these deceases were not related to B. cepacia complex infection.

B. multivorans and genomovar III-A strains were alternately isolated from patient J, throughout a 4-year period, indicating that during such an extended period the replacement of the B. multivorans strain with the genomovar III-A strain did not occur, as suggested by other studies (20).

Patients sharing the same strains (with a common ribopattern), persistently infected with the same strain, or exhibiting lung deterioration were observed for all the four B. cepacia complex genomovars and species represented in the collection examined. There were also cases of transient infection with antibiotic clearance with strains belonging to the four genomovars, including genomovar III (subgroups III-A, III-B, and BCESM-positive strains).

The eight patients infected with strains from genomovar I were children (ages ranging from 3 months to 12 years), while three of the four patients infected with B. stabilis were adults (>16 years old). Due to the very small CF population examined, results do not support or reject a previous observation that patients with genomovar III infection are significantly older than those infected with B. multivorans (20).