OUP user menu

Burkholderia anthina sp. nov. and Burkholderia pyrrocinia, two additional Burkholderia cepacia complex bacteria, may confound results of new molecular diagnostic tools

Peter Vandamme, Deborah Henry, Tom Coenye, Sazini Nzula, Marc Vancanneyt, John J. LiPuma, David P. Speert, John R.W. Govan, Eshwar Mahenthiralingam
DOI: http://dx.doi.org/10.1111/j.1574-695X.2002.tb00584.x 143-149 First published online: 1 June 2002


Nineteen Burkholderia cepacia-like isolates of human and environmental origin could not be assigned to one of the seven currently established genomovars using recently developed molecular diagnostic tools for B. cepacia complex bacteria. Various genotypic and phenotypic characteristics were examined. The results of this polyphasic study allowed classification of the 19 isolates as an eighth B. cepacia complex genomovar (Burkholderia anthina sp. nov.) and to design tools for its identification in the diagnostic laboratory. In addition, new and published data for Burkholderia pyrrocinia indicated that this soil bacterium is also a member of the B. cepacia complex. This highlights another potential source for diagnostic problems with B. cepacia-like bacteria.

  • Cystic fibrosis
  • Identification
  • Burkholderia cepacia complex
  • Burkholderia pyrrocinia
  • Burkholderia anthina

1 Introduction

During the past decade, analysis of the microbiological diversity of Burkholderia cepacia-like organisms by using a polyphasic taxonomic approach has clarified problems associated with the identification of these bacteria. It demonstrated the presence of several novel organisms such as Pandoraea species, which can be misidentified as B. cepacia[1,2]. Even more troublesome, it highlighted the unsurpassed taxonomic complexity of this organism that is rightfully considered as a species complex [2,3]. The B. cepacia complex comprises at least seven genomic species (referred to as genomovars): B. cepacia (genomovar I), Burkholderia multivorans (formerly B. cepacia genomovar II), B. cepacia genomovar III, Burkholderia stabilis (formerly B. cepacia genomovar IV), Burkholderia vietnamiensis (also known as B. cepacia genomovar V), B. cepacia genomovar VI, and Burkholderia ambifaria (B. cepacia genomovar VII) [36]. These genomovars share a high degree of 16S rDNA (98–100%) and recA (94–95%) sequence similarity, and moderate levels of DNA–DNA hybridisation (30–60%) [27]. Isolates from cystic fibrosis patients from different geographical regions and from novel environmental niches are being examined to continue to explore the biodiversity of this organism. The results from these studies rapidly challenge newly developed molecular identification approaches [711] by the discovery of novel ‘atypical’ isolates that do not fit into the current classification system.

The current study describes a group of 19 atypical B. cepacia-like organisms that could not be allocated to one of the current seven members of this species complex. The name Burkholderia anthina sp. nov. is proposed to accommodate this eighth genomovar within the B. cepacia complex and several approaches for its identification are described. In addition, analysis of the Burkholderia pyrrocinia type strain indicated that this soil bacterium represents a ninth B. cepacia complex genomovar.

2 Materials and methods

2.1 Bacterial strains and growth conditions

All B. anthina isolates examined are listed in Table 1. Reference strains of other Burkholderia species and genomovars were described previously [36]. All strains were grown aerobically on trypticase soy agar (BBL) and incubated at 28°C unless otherwise indicated.

View this table:
Table 1

List of B. anthina strains studied

Strain numberOther strain designationsSource
LMG 16670J2552, BPC51rhizosphere of Carludoucas palmata (UK)
LMG 20980W92B, J2863, CCUG 46047rhizosphere soil (Nashville, USA)
LMG 20982C1658hospital environment (Manchester, UK)
LMG 20983C1765sputum, cystic fibrosis patient (Blackpool, UK)
R-131J2553, BPC22Sanseviera leaf (Edinburgh, UK)
R-4183W92rhizosphere soil (Nashville, USA)
R-4190W89Brhizosphere soil, vine plant (Edinburgh, UK)
R-9942FC0976environmental strain
R-11752J2927Petunia rhizosphere (Dundee, UK)
R-11753J2928Petunia rhizosphere (Dundee, UK)
R-11755J2950Begonia rhizosphere (Dundee, UK)
R-11757J2944Begonia rhizosphere (Dundee, UK)
R-11758J2951Dracaena rhizosphere (Dundee, UK)
R-11759J2949Begonia rhizosphere (Dundee, UK)
R-11761J2941Petunia rhizosphere (Dundee, UK)
R-11763J2946Begonia rhizosphere (Dundee, UK)
R-11766J2945Begonia rhizosphere (Dundee, UK)
R-13393C1660hospital environment (Manchester, UK)
R-16022AU1293sputum, cystic fibrosis patient (USA)
  • LMG, BCCM/LMG Bacteria Collection, Laboratorium voor Microbiologie Gent, Universiteit Gent, Ghent, Belgium.

  • See reference [25].

2.2 DNA extraction

DNA used for DNA–DNA hybridisation, AFLP fingerprinting, and determination of the DNA base composition was prepared as described by Pitcher et al. [12]. DNA used for PCR amplification of the 16S rRNA and the recA gene, and for the recA-based PCR tests was prepared by heating one to two colonies (picked from an overnight grown plate) at 95°C for 15 min in 20 µl lysis buffer containing 0.25% (v/v) sodium dodecyl sulphate (SDS) and 0.05 M NaOH. Following cell lysis, 180 µl distilled water was added to the lysis buffer and the DNA solutions were stored at −20°C.

2.3 16S rDNA sequencing

The nearly complete sequences of the 16S rRNA gene of strain R-4183 was amplified by PCR using conserved primers (5′-AGAGTTTGATCCTGGCTGAG-3′ and 5′-AAGGAGGTGATCCAGCCGCA-3′). The PCR products were purified using a QIAquick PCR purification kit (Qiagen GmbH) according to the manufacturer's instructions. Sequence analysis was performed using an Applied Biosystems 310 DNA Sequencer and the protocols of the manufacturer (Perkin-Elmer) using the ABI Prism™ dye terminator cycle sequencing ready reaction kit. The sequencing primers are those given by Coenye et al. [1]. Sequence assembly was performed using the programme AutoAssembler™ (Perkin-Elmer). Phylogenetic analysis was performed using the BioNumerics software package (Applied Maths), based on the neighbour-joining method [13]. Approximately 1460 bases were used and unknown bases were excluded from the calculations.

2.4 recA gene sequence and restriction fragment length polymorphism (RFLP) analysis

recA gene sequence and RFLP analyses were performed as described before [7]. The sequences were aligned and compared with recA sequences determined from B. cepacia complex strains analysed in previous studies [4,6,7] using the BioNumerics software package (Applied Maths).

2.5 16S rRNA- and recA-based PCR assays

Assays for the identification of members of the B. cepacia complex based on the rRNA operon and the recA gene were performed as described previously [7,9].

A B. anthina-specific recA-based PCR test was developed based on recA sequences determined during this and previous studies. PCR assays were performed in 25 µl standard reaction mixtures [7] using 20 pmol of primers, BCRG81, 5′-TACGGTCCGGAATCGTCG-3′, and BCRG82, 5′-CGCACCGACGCATAGAAT-3′ (see Section 3 for design criteria). Amplification was carried out using a Techne Touchgene Thermal Cycler (Jencons-PLS, Leighton Buzzard, UK). Thirty amplification cycles were completed, each consisting of 30 s of denaturation at 94°C, 45 s at an annealing temperature of 61°C and 60 s of extension at 72°C. A final elongation of 7 min at 72°C completed the cycle. One third of the PCR mixture was then analysed by gel electrophoresis for the presence of a 473 bp product.

2.6 AFLP (amplified fragment length polymorphism) fingerprinting

The preparation of template DNA (using restriction enzymes Apa I and Taq I), amplification (using primers A00 and T00 in the preselective PCR and primers B07 [labelled with the fluorescent dye 6-FAM] and T11 in the selective PCR), separation of the fragments using an ABI Prism 377 automated DNA sequencer and numerical analysis were performed as described previously [14]. Data for the reference strains of B. cepacia genomovars were generated during previous studies [5,6,14].

2.7 Determination of the DNA base composition

DNA was enzymatically degraded into nucleosides as described by Mesbah et al. [15]. The obtained nucleoside mixture was then separated by high-performance liquid chromatography using a Waters SymmetryShield C8 column thermostated at 37°C. The solvent was 0.02 M NH4H2PO4 (pH 4.0) with 1.5% acetonitrile. Non-methylated lambda phage DNA (Sigma) was used as the calibration reference.

2.8 DNA–DNA hybridisation

DNA–DNA hybridisations were performed with photobiotin-labelled probes in microplate wells as described by Ezaki et al. [16], using a HTS7000 Bio Assay Reader (Perkin-Elmer) for the fluorescence measurements. The hybridisation temperature was 50°C.

2.9 Polyacrylamide gel electrophoresis (PAGE) of whole-cell proteins

After an incubation period of 48 h, whole-cell protein extracts were prepared and SDS–PAGE was performed as described before [17]. The densitometric analysis, normalisation and interpolation of the protein profiles, and numerical analysis were performed using the GelCompar software package version 4.2 (Applied Maths). Similarity levels between the patterns were calculated using the Pearson product moment correlation coefficient and are expressed as percentage similarity for convenience. Data for the reference strains of B. cepacia genomovars were generated during previous studies [36].

2.10 Fatty acid methyl ester analysis

After an incubation period of 24 h at 35°C, a loopful of well-grown cells was harvested and fatty acid methyl esters were prepared, separated and identified using the Microbial Identification System (Microbial ID) as described before [18].

2.11 Biochemical characterisation

Tests were performed as described previously [19]. Briefly, pure cultures were stored at −70°C in Mueller Hinton broth (Difco) with 8% dimethylsulfoxide. Frozen isolates were subcultured to Columbia agar containing 5% sheep blood (PML Microbiologicals) before testing. The primary identification system used was the API Rapid NE system (Biomerieux Vitek Inc.), supplemented with glucose, maltose, lactose, xylose, sucrose and adonitol oxidation-fermentation (OF) sugars, and an adaptation of Moeller lysine and ornithine decarboxylases (Difco). All incubations were done at 35°C except where otherwise mentioned. Growth on trypticase soy agar (Becton Dickinson) at 35°C and 42°C was observed for appearance and pigment production.

3 Results

3.1 recA RFLP analysis

The recA RFLP patterns of the B. anthina isolates differed from those of the recognised members of the B. cepacia complex and from that of the B. pyrrocinia type strain. Two different patterns, designated T and AS could be distinguished among the 19 isolates examined (Fig. 1). The B. pyrrocinia type strain had a unique recA RFLP pattern, designated M, and occupied a distinct position in the dendrogram (Fig. 1).

Figure 1

Numerical analysis and computer-generated reproductions of Hae III –recA gene restriction profiles of B. cepacia complex bacteria. Data for non-B. anthina strains were taken from previous studies [57]. Roman numerals indicate genomovar numbers. RFLP types D and E represent B. cepacia (genomovar I), RFLP types C, F and O represent B. multivorans, RFLP types G, H, I and J' represent B. cepacia genomovar III, RFLP type J represents B. stabilis, RFLP types A and B represent B. vietnamiensis, RFLP type Q represents B. cepacia genomovar VI, RFLP types AE, N and L represent B. ambifaria, and RFLP types AS and T represent B. anthina (types J and J' are identical but represent different genomovars [7]).

3.2 PCR assays

In the 16S rRNA-based PCR assays, DNA fragments of the expected size were amplified for all B. anthina isolates investigated with the primer pair RHG-F and RHG-R (specific for the genera Burkholderia, Pandoraea and Ralstonia) and with the primer pair PC-SSF and PC-SSR (specific for B. cepacia genomovars I and III, B. stabilis and B. ambifaria). Using primer pairs BC-GII and BC-R, and BC-GV and BC-R (specific for B. multivorans, and for B. multivorans and B. vietnamiensis, respectively), no amplification product was obtained (data not shown).

None of the B. anthina isolates yielded PCR products with the recA-based PCR assays described for the other B. cepacia complex genomovars (data not shown).

The full-length recA gene sequences from strains LMG 16670, LMG 20980, LMG 20982, LMG 20983, R-4190 and R-11761 (EMBL accession numbers AF456060, AF456059, AF456034, AF456051, AF456058 and AF456061, respectively) were aligned against published recA sequences for each B. cepacia complex genomovar [4,6,7] and 48 novel recA sequences representative of distinct recA gene RFLP types from other B. cepacia complex isolates (E. Mahenthiralingam, unpublished data). The nucleotide at position 210, G, in the recA gene was found to be conserved in all B. anthina sequences and absent in other B. cepacia complex recA genes, except for B. ambifaria and two sequences from strains of currently undetermined genomovar. Base position 648, A, was completely specific to B. anthina recA genes. Two 18 nucleotide primers, BCRG81 (position 193–210 in the recA gene) and BCRG82 (the reverse complement of position 665–648 in the recA gene), were designed to detect the latter bases and specifically amplify a 473 bp PCR product from B. anthina strains. All B. anthina strains tested positive with these primers (data not shown). DNA from strains representative of 51 recA gene RFLP types, including all current genomovars of the B. cepacia complex and several novel recA RFLP types, were negative in the B. anthina PCR (data not shown).

3.3 PAGE of whole-cell proteins

Duplicate protein extracts were prepared to check the reproducibility of the growth conditions and the preparation of the extracts. The correlation level between duplicate protein patterns was more than 93% (data not shown). The overall whole-cell protein profiles of B. anthina isolates resembled those of B. cepacia genomovars I and III and B. ambifaria isolates, but were clearly different from those of other B. cepacia complex species and other Burkholderia reference species present in the database (data not shown). Numerical analysis of the profiles did not allow differentiation of B. anthina, B. cepacia genomovars I and III, and B. ambifaria isolates (data not shown).

3.4 AFLP fingerprinting

The reproducibility was checked by preparing PCR products in duplicate and was always higher than 92% (data not shown). Following numerical analysis, all of the B. anthina isolates examined formed a single cluster delineated above a similarity level of 40%, which was clearly different from the clusters comprising other Burkholderia species or B. cepacia complex genomovars (data not shown).

3.5 Phylogenetic analysis

The 16S rRNA gene sequence of strain R-4183 (EMBL accession number AJ420880) was compared with the available 16S rRNA gene sequences of representatives of the beta-Proteobacteria. Strain R-4183 clustered among members of the B. cepacia complex, with similarity levels ranging from 98.7% to 99.6%. Similarity levels towards other Burkholderia species were below 97.7%. Similarity levels towards representatives of other taxa belonging to the beta-Proteobacteria were below 95.1%.

3.6 DNA–DNA hybridisation and G+C content analysis

The DNA–DNA hybridisation results between strains LMG 16670, R-4183, and LMG 20980 revealed binding values above 71% (Table 2). The DNA–DNA hybridisation values between the latter strains and representatives of (i) B. cepacia genomovar I (53–61%), (ii) B. multivorans (34–45%), (iii) B. cepacia genomovar III (51%), (iv) B. stabilis (53–63%), (v) B. vietnamiensis (42–59%), (vi) B. cepacia genomovar VI (40–55%), (vii) B. ambifaria (61%), and (viii) B. pyrrocinia (31%) were intermediate or low (Table 2).

View this table:
Table 2

DNA–DNA binding values and mol% G+C of all strains examined

DNA binding values with B. anthina strain:
LMG 16670R-4183LMG 20980T
B. anthina
LMG 16670100
LMG 2098071100
B. cepacia genomovar I
LMG 1222T5561
LMG 140875359
B. multivorans
LMG 13010T4534
B. cepacia genomovar III
LMG 1665951
B. stabilis
LMG 14294T5363
B. vietnamiensis
LMG 10929T4259
B. cepacia genomovar VI
LMG 189414055
B. ambifaria
LMG 19182 T61
B. pyrrocinia
LMG 14191T31

The G+C contents for B. anthina strains LMG 16670, R-4183, and LMG 20980 were 66.6, 66.4, and 66.5 mol%, respectively.

3.7 Cellular fatty acid analysis

The average whole-cell fatty acid profile was determined for nine B. ambifaria and 10 B. anthina isolates. Following components were present in B. ambifaria and B. anthina, respectively (in percent): 14:0 (3.6±0.3, 5.0±0.8), 16:1 ω7c (11.0±3.5, 17.1±1.1), 16:0 (26.3±2.0, 28.4±0.9), 17:0 cyclo (11.3±4.9, 4.6±2.5), 16:1 2OH (trace amounts in both species), 16:0 2OH (trace amounts in both species), 16:0 3OH (6.6±0.7, 5.8±0.3), 18:1 ω7c (25.6±6.3, 27.6±5.2), 18:0 (1.1±0.5, 1.0±0.2), 19:0 cyclo Ω8c (4.8±2.8, 1.3±0.4), 18:1 2OH (1.4±0.4, 1.0±0.4), 14:0 3OH (5.5±1.1, 5.9±0.5). In addition, B. ambifaria also contains 1.1±0.2% 12:0. In these profiles, summed feature 2 (comprising 14:0 3OH, 16:1 iso I, an unidentified fatty acid with equivalent chain length value of 10.928 Da, or 12:0 ALDE, or any combination of these fatty acids), and summed feature 3 (comprising 16:1 ω7c or 15 iso 2OH or both) are listed as 14:0 3OH and 16:1 ω7c, respectively, as these fatty acids have been reported in Burkholderia species [20].

Overall, all B. cepacia complex isolates had the same fatty acid components in their whole-cell profiles. However, B. anthina isolates had strikingly high levels of the unsaturated fatty acids 16:1 ω7c and 18:1 ω7c, and low levels of the cyclic fatty acids 17:0 cyclo and 19:0 cyclo ω8c, suggesting that their physiological condition after 24 h cultivation at 35°C is younger compared to other B. cepacia complex bacteria. This allows a clear differentiation between B. anthina and B. cepacia genomovars I and III, which are otherwise very similar. The same is true, but to a lesser extent, for B. ambifaria.

3.8 Biochemical characterisation

Biochemical characteristics were determined for 16 B. anthina isolates (isolates R-131, R-4183 and R-16022 were not included). Growth was observed at 30 and at 37°C, but not at 5°C. Growth at 42°C was strain dependent (when present, growth was poor). No pigmented strains were detected, nor was a melanin-like pigment produced on tyrosine agar. Growth on BCSA and MacConkey agar and Simmons citrate agar was observed, and there was no haemolysis on sheep blood agar.

Oxidase activity (slow reaction), and oxidation of glucose, lactose, maltose, and xylose were present in all isolates investigated. Ornithine decarboxylase and liquefaction of gelatin were absent. Hydrolysis of aesculin, nitrate reduction, β-galactosidase and lysine decarboxylase activity, and oxidation of sucrose and adonitol are strain dependent.

The API identification scores (as ‘B. cepacia’) were 1067577 (three isolates), excellent identification; 1067777 (nine isolates) very good identification; 1047577 (one isolate), very good identification; 1467777 (one isolate) very good identification; and 0047577 (two isolates) low discrimination.

Two characteristics were particularly helpful to differentiate B. anthina from most other B. cepacia complex isolates. On BCSA agar, the isolates turned the medium alkaline (pink colour production), which is most unusual for organisms capable of utilising sucrose. In addition, the colony colour and morphology would be best described as ‘creamy’ in colour and texture whereas most other B. cepacia complex isolates were typically grey, and moist or dry (depending on genomovar).

4 Discussion

In an on-going multicenter effort to improve the diagnosis of B. cepacia complex infections by studying the taxonomic relationships of large collections of isolates obtained from various ecological niches, we discovered 19 isolates with whole-cell protein profiles resembling those of B. cepacia genomovars I and III, and B. ambifaria isolates. These isolates could however not be assigned unequivocally to one of these established genomovars within the complex and reacted negative in the recA-based PCR assays for the identification of B. cepacia genomovars I and III, and B. ambifaria. Subsequent analysis by recA RFLP revealed two distinct restriction profiles (Fig. 1) that were different from those established in previous studies [6,7]. Analysis of the same isolates by means of AFLP strongly suggested that they represented a single genomic species [14]. The DNA–DNA hybridisation values among three of these isolates generated the typical high (>70%) hybridisation levels found within B. cepacia complex genomovars and the characteristic intermediate to low hybridisation levels (30–60%) towards representatives of the other B. cepacia complex genomovars [2] (Table 2). Comparative sequence analysis of the 16S rDNA gene of strain R-4183 confirmed its assignment to the B. cepacia complex. The DNA base ratio of about 66 mol% was also within the range of the B. cepacia complex. These data unambiguously indicated that the 19 isolates represented a novel B. cepacia complex genomovar for which we propose the name B. anthina. The specific epithet is derived from the Greek adjective anthinos (hence, N.L. adj. anthinus), meaning ‘from the garden’ or ‘from flowers’ (referring to the origin of the majority of the isolates in the present collection). The type strain is LMG 20980 (=CCUG 46047) which was isolated in 1997 from the rhizosphere of a house-plant in Nashville (USA).

The identification of B. anthina, and particularly its differentiation from B. cepacia genomovars I and III, using previously established PCR tests proved not possible. However, as demonstrated in the present study, recA RFLP analysis and the newly developed recA-based PCR assay provide a simple means to identify this organism. Using recently developed rRNA-based PCR assays [9], B. anthina strains could be distinguished from B. multivorans, B. vietnamiensis and B. cepacia genomovar VI, but not from the other members of the B. cepacia complex. Biochemically, these isolates were difficult to distinguish from B. cepacia genomovars I and III. Characteristics useful to separate B. anthina from B. cepacia genomovar I and III isolates are primarily the alkaline reaction (pink colour change) on BCSA agar, and the distinctive creamy colony morphology [21]. Other methods that facilitate identification of these isolates to the genomovar level include AFLP fingerprinting, DNA–DNA hybridisation (Table 2) and recA gene sequence analysis, but these techniques are primarily research tools. Analyses of the cellular fatty acid components demonstrated that the whole-cell fatty acid profiles of B. anthina isolates were similar to those of B. ambifaria isolates. However, this approach was particularly useful to separate B. anthina from B. cepacia genomovars I and III.

Most of the present B. anthina strains have been isolated from the rhizosphere soil of garden flowers and house-plants, and from sputum samples of cystic fibrosis patients. The presence of these bacteria in soil of various flowers and green plants, suggests that this can serve as a reservoir for infections. However, during the past years, several thousands of cystic fibrosis-related isolates have been stored and examined in various national reference or referral centres (see http://go.to/cepacia), and thus far only few human B. anthina isolates have been collected. One of the isolates (R-16022) was recently described by LiPuma et al. [22] as a strain with ‘indeterminate’ genomovar status. In the latter study, only one out of 606 B. cepacia complex infected cystic fibrosis patients carried B. anthina. At present, this patient has been chronically colonised for a period of two years (J.J. LiPuma, unpublished information).

An evaluation of published DNA–DNA hybridisation [3,23] and 16S rDNA sequence [23] data for B. pyrrocinia, a soil bacterium described in the 1960s [24], revealed hybridisation and similarity levels as reported between other B. cepacia complex bacteria. This was confirmed in the present study by the analysis of the recA gene sequence which again revealed values similar to those observed between B. cepacia complex bacteria. This was also substantiated by the results of the recA RFLP technique (Fig. 1), as the primers used for the initial amplification of the recA gene were chosen such that they were specific for B. cepacia complex bacteria [7]. Altogether, these data indicate that B. pyrrocinia should be considered a ninth genomovar within the B. cepacia complex and highlights additional sources of identification problems. Our unpublished observations (J.J. LiPuma, E. Mahenthiralingam, and P. Vandamme) indicated that B. pyrrocinia indeed accounts for a considerable percentage of the putative B. cepacia complex that cannot be identified to the genomovar level. A full taxonomic characterisation of these cystic fibrosis-related and environmental isolates will be published elsewhere.


The authors are indebted to the Fund for Scientific Research — Flanders (P.V.), the Canadian Cystic Fibrosis Foundation (D.P.S.), the United States Cystic Fibrosis Foundation (J.J.L.), the United Kingdom Cystic Fibrosis Trust (J.R.W.G., E.M. and P.V.) and the Carroll Haas Research Fund in Cystic Fibrosis (T.C.) for financial support. E.M. is grateful to Julie Fadden for excellent technical assistance.


  1. [1].
  2. [2].
  3. [3].
  4. [4].
  5. [5].
  6. [6].
  7. [7].
  8. [8].
  9. [9].
  10. [10].
  11. [11].
  12. [12].
  13. [13].
  14. [14].
  15. [15].
  16. [16].
  17. [17].
  18. [18].
  19. [19].
  20. [20].
  21. [21].
  22. [22].
  23. [23].
  24. [24].
  25. [25].
View Abstract