Distinguishing species of the Burkholderia cepacia complex and Burkholderia gladioli by automated ribotyping

Abstract

Several species belonging to the genus Burkholderia are clinically relevant, opportunistic pathogens that inhabit major environmental reservoirs. Consequently, the availability of means for adequate identification and epidemiological characterization of individual environmental or clinical isolates is mandatory. In the present communication we describe the use of the Riboprinter microbial characterization system (Qualicon, Warwick, United Kingdom) for automated ribotyping of 104 strains of Burkholderia species from diverse sources, including several publicly accessible collections. The main outcome of this analysis was that all strains were typeable and that strains of Burkholderia gladioli and of each species of the B. cepacia complex, including B. multivorans, B. stabilis, and B. vietnamiensis, were effectively discriminated. Furthermore, different ribotypes were discerned within each species. Ribotyping results were in general agreement with strain classification based on restriction fragment analysis of 16S ribosomal amplicons, but the resolution of ribotyping was much higher. This enabled automated molecular typing below the species level. Cluster analysis of the patterns obtained by ribotyping (riboprints) showed that within B. gladioli, B. multivorans, and B. cepacia genomovar VI, the different riboprints identified always clustered together. Riboprints of B. cepacia genomovars I and III, B. stabilis, and B. vietnamiensis did not show distinct clustering but rather exhibited the formation of loose assemblages within which several smaller, genomovar-specific clusters were delineated. Therefore, ribotyping proved useful for genomovar identification. Analysis of serial isolates from individual patients demonstrated that infection with a single ribotype had occurred, despite minor genetic differences that were detected by pulsed-field gel electrophoresis of DNA macrorestriction fragments. The automated approach allows very rapid and reliable identification and epidemiological characterization of strains and generates an easily manageable database suited for expansion with information on additional bacterial isolates.

FIG. 1

Comparative analysis of the riboprints obtained, using either EcoRI or PvuII, for a subset of Burkholderia spp. strains. The two triplets of strains undistinguished using PvuII (labeled with stars and diamonds, respectively) were each separated into two groups by using EcoRI. Within each triplet, EcoRI distinction was in agreement with the geographic origin (Table 1). For cluster analysis, the UPGMA method was used, based on the matrix of Pearson correlation (see Materials and Methods). In the dendrogram scale, correlation levels were converted to percent similarity levels.

FIG. 2

Overview of the PvuII riboprints obtained for Burkholderia spp. and R. pickettii strains, showing correspondence with 16S rRNA gene PCR-RFLP information gathered previously (41). Strain codes correspond to those listed in Table 1. Serial samples belonging to individual Dutch or German patients are indicated by lettering (patients A to G) and figures (patients 1 to 17), respectively. Only one isolate per patient was included. Note that not all strains for all patients are included. On the right are the species or genomovar names, as deduced from the ribotyping analysis and via characterization by whole-cell protein profiling of reference strains of each ribogroup (see Table 1). For cluster analysis, the UPGMA method was used, based on the matrix of Pearson correlation (see Materials and Methods). In the dendrogram scale, correlation levels were converted to percent similarity levels.

FIG. 3

Comparison of the PFGE fingerprints obtained for serial isolates of B. cepacia from individual CF patients from the Hannover region (Germany). From left to right, the lanes contain SpeI-digested DNA derived from strains isolated from patients 1, 2, 3, 4, 9, 18, and 10. The strain isolated from the latter patient is a B. gladioli isolate and is included for comparative reasons. The H-prefixed numbers identify the individual isolates, whose characteristics are listed in Table 1. Arrows identify banding pattern polymorphism among clusters of strains derived from the same patient.