Burkholderia latens sp. nov., Burkholderia diffusa sp. nov., Burkholderia arboris sp. nov., Burkholderia seminalis sp. nov. and Burkholderia metallica sp. nov., novel species within the Burkholderia cepacia complex

The Burkholderia cepacia complex (Bcc) is a diverse group of bacteria (Coenye et al., 2001c; Vandamme et al., 1997) which are ubiquitous in nature (Coenye & Vandamme, 2003). Some Bcc strains are recognized for their beneficial properties as plant-growth promoters, bioremediators and biocontrol strains (Parke & Gurian-Sherman, 2001), whereas others are notorious as pathogens causing infections in plants and immunocompromised individuals such as cystic fibrosis patients (Coenye & Vandamme, 2003; Mahenthiralingam et al., 2005).

The Bcc currently comprises nine species, which share a high degree of 16S rRNA gene (98–100 %) and recA (94–95 %) sequence similarity and moderate levels of DNA–DNA hybridization (30–50 %) (Coenye et al., 2001c); they are B. cepacia (Vandamme et al., 1997), B. multivorans (Vandamme et al., 1997), B. cenocepacia (Vandamme et al., 1997, 2003), B. stabilis (Vandamme et al., 1997, 2000), B. vietnamiensis (Gillis et al., 1995; Vandamme et al., 1997), B. dolosa (Vermis et al., 2004), B. ambifaria (Coenye et al., 2001b), B. anthina (Vandamme et al., 2002) and B. pyrrocinia (Vandamme et al., 2002). Identification of these Bcc species is a complicated challenge, requiring a combination of multiple molecular diagnostic procedures, and is mostly restricted to reference centres (Coenye et al., 2001c). Studies have revealed that the 16S rRNA gene, which is widely used for bacterial systematics, is limited in its ability to differentiate the Bcc species. In contrast, analysis of recA sequence variation correlated well with species status (Mahenthiralingam et al., 2000b, 2002). Analysis of species-specific polymorphisms in the recA gene has now been superseded by a single, robust multilocus sequence typing (MLST) scheme for the Bcc that provides both species and strain identification (Baldwin et al., 2005).

An efficient working algorithm for the identification of Bcc strains starts with a recA-based Bcc-specific PCR test (Mahenthiralingam et al., 2000b), followed by restriction fragment length polymorphism (RFLP) analysis of the amplicon using restriction enzymes HaeIII and MnlI, which reveal sequence variation within the complex. Currently, more than 70 Bcc HaeIII RFLP types have been found (P. Vandamme, unpublished data). Several species include multiple RFLP types (Coenye et al., 2001c; Mahenthiralingam et al., 2000b), and overlap of certain RFLP types is observed in different species (Mahenthiralingam et al., 2000a; McDowell et al., 2001; Vandamme et al., 2002). In particular, isolates showing RFLP types H, J and W should be further investigated using the second restriction enzyme, MnlI, for unambiguous species identification. At present, the large majority of these HaeIII RFLP types can be assigned to one of the established nine Bcc species. To avoid identification problems due to inaccurate or overlapping RFLP types, strains with novel RFLP types should be subjected to recA gene nucleotide sequence analysis to enable phylogenetic prediction of species status (Mahenthiralingam et al., 2002). This working strategy has revealed several new groups of isolates that could not be assigned to one of the established species. The aim of the present study was to determine the taxonomic status of five such clusters using a polyphasic approach. We also present data that demonstrate that Burkholderia ubonensis (Yabuuchi et al., 2000) should be considered a member of the Bcc based on its genetic relatedness with the other Bcc species.

Bacterial strains and growth conditions.
The strains studied represented five recA lineages that are referred to as clusters 1 to 5 (Table 1). Clusters 1 and 5 contained six and five isolates, respectively, which were all recovered from sputum samples of cystic fibrosis patients. Clusters 2, 3 and 4 comprised clinical and environmental isolates and contained 14, 17 and 21 isolates, respectively. All strains were routinely cultured on trypticase soy agar (TSA) and incubated at 28 °C unless indicated otherwise. Pure cultures were maintained on Microbank vials at –80 °C.

Table 1. B. cepacia complex isolates studied Culture collections: ATCC, American Type Culture Collection, Manassas, VA, USA; CCUG, Culture Collection, University of Göteborg, Department of Clinical Bacteriology, Göteborg, Sweden; DMST, Department of Medical Sciences Thailand, Nonthaburi, Thailand; LMG, BCCM/LMG Bacteria Collection, Laboratorium voor Microbiologie, Universiteit Gent, Gent, Belgium; NCTC, National Collection of Type Cultures, Health Protection Agency, London, UK; R, Research Collection, Laboratorium voor Microbiologie, Universiteit Gent. Other strain numbers are personal designations of the depositor. CF, Cystic fibrosis; ND, not determined.

DNA preparation.
For PCR and RFLP experiments, DNA was prepared by alkaline lysis as described before (Storms et al., 2004). For DNA–DNA hybridization experiments and the determination of the DNA base composition, DNA was prepared as described by Pitcher et al. (1989) or by Marmur (1961). The latter procedure was followed when the former did not yield sufficient pure DNA.

Bcc-specific recA gene PCR.
The Bcc recA gene (1040 bp) was amplified using primers BCR1 and BCR2, as described previously (Mahenthiralingam et al., 2000b).

recA RFLP and recA gene sequence analysis.
Amplified recA fragments of all isolates were subjected to HaeIII-based RFLP analysis (Mahenthiralingam et al., 2000b). When the recA RFLP pattern was type H, J or W, MnlI RFLP analysis was also performed. Electrophoretic separation of the restriction fragments was performed by PAGE, as described previously (Vanlaere et al., 2005). Restriction patterns were analysed using the Bionumerics 4.5 software package (Applied Maths) and compared to those of Bcc reference strains. recA sequence analysis was performed as described previously (Mahenthiralingam et al., 2000b). If not available, nucleotide sequences of the recA gene from representative strains were determined and deposited in the EMBL public database. All other sequences were available from public databases. Multiple alignment was performed by using the CLUSTAL_X program (Thompson et al., 1997). The aligned sequences were analysed phylogenetically using the Bionumerics 4.5 software. Distances were calculated using the Jukes & Cantor algorithm. Phylogenetic trees based on the neighbour-joining method were constructed with bootstrap values of 1000 replications.

16S rRNA gene sequence analysis.
16S rRNA gene sequence analysis was performed as described previously (Coenye et al., 2001a). 16S rRNA gene sequences were determined and deposited in the EMBL public database.

Determination of the DNA base composition.
DNA was degraded enzymically into nucleosides as described previously (Mesbah & Whitman, 1989). The nucleoside mixture obtained was then separated using a Waters Breeze HPLC system and XBridge Shield RP18 column thermostatted at 37 °C. The solvent was 0.02 M NH₄H₂PO₄ (pH 4.0) with 1.5 % (v/v) acetonitrile. Non-methylated lambda phage (Sigma) and Escherichia coli LMG 2093 DNA were used as calibration reference and control, respectively.

DNA–DNA hybridization experiments.
Hybridization reactions were performed with photobiotin-labelled probes in microplate wells as described before (Ezaki et al., 1989; Goris et al., 1998), using an HTS7000 Bio Assay Reader (Perkin-Elmer) for the fluorescence measurements. The hybridization temperature was 50 °C.

BOX-PCR fingerprinting.
BOX-PCR profiles were obtained as described previously (Coenye et al., 2002). The BOX profiles were analysed using the Molecular Analyst-Fingerprinting (MA-F) software. A dendrogram was derived from the UPGMA linkage of correlation coefficients between the BOX-PCR fingerprint patterns from Bcc isolates. Data for reference strains were available from previous studies (Chen et al., 2001; Coenye et al., 2002, 2004; LiPuma et al., 2002) or were unpublished (J. J. LiPuma).

MLST analysis.
MLST analysis was performed as described by Baldwin et al. (2005). A phylogenetic tree of concatenated sequences (2773 bp) [including fragments of seven genes: atpD (443 bp), gltB (400 bp), gyrB (454 bp), recA (393 bp), lepA (397 bp), phaC (385 bp) and trpB (301 bp)] of each isolate was constructed based on the neighbour-joining method using MEGA software package version 3 (Baldwin et al., 2005; Kumar et al., 2004). The significance of branching within the trees was evaluated by bootstrap analysis of 1000 computer-generated trees. This publication made use of the Bcc MLST website (http://pubmlst.org/bcc/) developed by Keith Jolley and sited at the University of Oxford (Jolley et al., 2004).

Biochemical characterization.
A biochemical characterization was performed as described previously (Henry et al., 1997, 2001).

B. ubonensis is member of the Bcc
As mentioned above, Bcc species typically share a high degree of 16S rRNA gene (98–100 %) and recA (94–95 %) sequence similarity and moderate levels of DNA–DNA hybridization (30–50 %) (Coenye et al., 2001c). B. ubonensis shares at least 98 % of its 16S rRNA gene sequence and 94 % of its recA sequence with the other members of the Bcc. The recA gene was amplified using the Bcc-specific recA PCR (Mahenthiralingam et al., 2000b). DNA–DNA hybridization values obtained between B. ubonensis LMG 20358^T and the other established members of the Bcc ranged from 40 to 56 % (Supplementary Table S1, available in IJSEM Online), which indicate a close relationship with Bcc species, since values between strains of Bcc species and other Burkholderia species are typically below 30 % (Coenye et al., 2001c). By MLST analysis, B. ubonensis shares at least 93 % similarity with other Bcc species. These results demonstrate that B. ubonensis constitutes a tenth species within the Bcc and corroborate the initial phylogenetic positioning of its recA sequence within the complex (Payne et al., 2005). Although B. ubonensis is very similar on the molecular level, it differs biochemically in two major tests; B. ubonensis is positive for arginine dihydrolase and negative for β-galactosidase activity (Table 2). At present, only a single B. ubonensis strain, LMG 20358^T, has been reported. In the course of our biodiversity analyses of Bcc and similar bacteria, we identified a second strain, LMG 24263, as B. ubonensis. The latter strain was isolated from a nosocomial infection in Thailand (S. Dejsirilert, personal communication).

Table 2. Biochemical characteristics useful for the differentiation of established Bcc species and novel Bcc clusters Species: 1, B. latens sp. nov. (cluster 1; 6 strains); 2, B. diffusa sp. nov. (cluster 2; 6); 3, B. arboris sp. nov. (cluster 3; 13); 4, B. seminalis sp. nov. (cluster 4; 13); 5, B. metallica sp. nov. (cluster 5; 3); 6, B. cepacia (22 strains); 7, B. multivorans (108); 8, B. cenocepacia (137); 9, B. stabilis (27); 10, B. vietnamiensis (35); 11, B. dolosa (12); 12, B. ambifaria (17); 13, B. anthina (18); 14, B. pyrrocinia (5); 15, B. ubonensis (2). All strains grow on BCSA, assimilate D-glucose, D-gluconate, L-malate, D-mannose and citrate, acidify D-glucose and exhibit oxidase activity. L-Tryptophanase and urease activity is not observed for all strains. For Bcc species: +, >90 % of all isolates positive; V, 10–90 % positive; –, <10 % of strains positive. For the five novel Bcc clusters, the number of positive reactions is indicated for strain-dependent reactions. The reaction of the type strain is given in parentheses for variable reactions. β, β-Haemolysis; W, weak. Results for B. cepacia, B. multivorans, B. cenocepacia, B. stabilis, B. vietnamiensis, B. dolosa and B. ambifaria are taken from Henry et al. (2001). Data for three additional strains of B. dolosa and for B. anthina, B. pyrrocinia and B. ubonensis are unpublished data from D. Henry and D. Speert.

recA RFLP analysis and recA gene sequencing
Among the isolates studied, five different HaeIII recA RFLP patterns were found, designated W (cluster 1 strains), AP (cluster 2 strains), H (cluster 2, 3 and 4 strains), J (cluster 3 strains) and S (cluster 5 strains). Patterns H and J have also been observed for B. cenocepacia strains (Mahenthiralingam et al., 2000b), pattern J for B. stabilis (Mahenthiralingam et al., 2000b) and pattern W for B. ubonensis (Supplementary Fig. S1). Therefore, these isolates were further examined by MnlI recA RFLP. Cluster 1 and cluster 4 isolates were each characterized by a unique MnlI restriction profile. Cluster 2 and some of the cluster 3 isolates shared the same distinct MnlI restriction profile; the remaining cluster 3 isolates again had a unique MnlI restriction profile. For all isolates, the HaeIII and MnlI recA RFLP types are listed in Table 1. Computer-generated gel images illustrating the different HaeIII and MnlI restriction profiles among the strains studied are shown in Supplementary Fig. S1.

Strains representing each RFLP type were chosen for recA sequence analysis. All sequences obtained were more than 94 % similar to those of other members of the Bcc. As mentioned above, phylogenetic analysis of these sequences revealed five recA clusters, referred to as clusters 1 to 5. A recA sequence-based phylogenetic tree, which illustrates the five recA lineages supported by high bootstrap values (>98 %), is shown in Fig. 1.

(41K): Fig. 1. Phylogenetic tree derived from analysis of recA gene sequences of Bcc bacteria and five novel Bcc clusters. Different B. cenocepacia recA lineages are designated IIIA, IIIB, IIIC and IIID (Vandamme et al., 2003). The tree was constructed using the neighbour-joining method based on the Jukes & Cantor parameter. Accession numbers are given in parentheses. Bootstrap values (>70 %) are shown for 1000 replicates. B. xenovorans LB400T was used as the outgroup. Bar, 1 % sequence dissimilarity.

16S rRNA gene sequencing and phylogenetic analysis
The 16S rRNA gene sequence of one representative strain per cluster was determined. Comparison of these 16S rRNA gene sequences with those of other Bcc strains revealed similarity levels above 98 % (data not shown). These values were similar to those obtained between strains of other Bcc species. Similarity levels of less than 97.5 % were calculated towards representatives of other Burkholderia species (data not shown).

DNA–DNA hybridization and determination of the DNA base composition
DNA–DNA hybridization experiments were performed between representative strains of the five clusters and established members of the Bcc including B. ubonensis. For each cluster, two strains were analysed. For clusters 2 and 3, strains with different HaeIII recA restriction profiles were selected. An overview of the DNA–DNA hybridization values and G+C contents is given in Supplementary Table S1. The values obtained among isolates within each cluster were high (>80 %), including clusters 2 and 3. Moderate to high DNA–DNA hybridization values, in the range of 39 to 67 %, were obtained between strains representing the five clusters and established species of the Bcc.

All strains investigated had G+C contents between 66 and 68 mol%. These DNA–DNA hybridization values and G+C contents are typical for Bcc species (Coenye et al., 2001c) and demonstrate that these five recA lineages can be considered to represent five novel species within the Bcc.

BOX-PCR fingerprinting
Strain typing methods such as repetitive element PCR using BOX A1R primers are used for typing Bcc strains (Coenye et al., 2002). Nevertheless, numerical analysis of BOX-PCR fingerprints often yields species-specific clusters of strains (Goris et al., 2002; Rademaker et al., 2000). Representative strains of the five clusters and members of the established Bcc species were examined by BOX-PCR fingerprinting. Supplementary Fig. S2 shows the result of a numerical analysis of the BOX profiles. Strains of four clusters grouped together, in congruence with results obtained by other methods in the present study, confirming their genetic homogeneity. Cluster 3 isolates, however, were divided in two clades, revealing more substantial diversity.

MLST analysis
Recently, Baldwin et al. (2005) reported a single MLST scheme that allowed differentiation of all established Bcc species. Clusters 1 to 5, including four, eight, fifteen, seven and one representative strains, respectively, were analysed by MLST. Phylogenetic analysis of concatenated sequences demonstrated that strains of the five clusters occupied distinct positions in the MLST tree supported by 100 % bootstrap values, except for cluster 3, which showed a bootstrap value of 84 % (Fig. 2). In clusters 1 to 5, one, eight, eleven, five and one different sequence types (STs), respectively, have been found. None of these clusters contained alleles that were found in STs of other Bcc species. For clusters 2, 3 and 4, which show at least five STs, an index of association (Ia) value was calculated to examine whether or not those populations were clonal (Baldwin et al., 2005; Haubold & Hudson, 2000). Cluster 2 had an Ia of 0.3683 (P<0.05) and cluster 3 had a Ia of 0.2933 (P<0.05), whereas cluster 4 had a Ia of 0.0185 (P>0.05). As the Ia of cluster 4 was not significantly greater than zero, it implies that cluster 4 is a freely recombining population, whereas the other two clusters have Ia values significantly greater than zero and are therefore more clonal populations, though further investigations by MLST are necessary to obtain larger numbers of STs in order to study their population biology. Nevertheless, the present study demonstrates the usefulness of MLST for strain typing and species identification in bacterial taxonomy. Comparison of our results with data from other studies showed that strains belonging to clusters 1 to 3 have been reported in previous studies under the designations Bcc 1, 2 and 3, respectively (Baldwin et al., 2005).

(28K): Fig. 2. Phylogenetic tree of concatenated nucleotide sequences from seven loci (atpD, gltD, gyrB, recA, lepA, phaC and trpB) of Bcc bacteria and novel Bcc clusters using the neighbour-joining method (Jukes & Cantor). Different B. cenocepacia recA lineages are designated IIIA, IIIB, IIIC and IIID (Vandamme et al., 2003). Bootstrap values are shown for 1000 replicates. B. pseudomallei ATCC 23343T was used as the outgroup. Bar, 1 % sequence dissimilarity.

Biochemical characterization
In general, species of the Bcc are phenotypically nearly indistinguishable, making their biochemical differentiation difficult, even with an extended panel of biochemical tests (Henry et al., 2001). Biochemical characteristics were determined for clusters 1 to 5, including six, six, thirteen, thirteen and three strains, respectively. A comparison of the biochemical characteristics of the five novel species with those of established members of the Bcc is shown in Table 2. Additional characteristics are given in the species descriptions below.

In conclusion, the aim of the present study was to determine the taxonomic position of five recA gene clusters of Bcc isolates using a polyphasic taxonomic approach. The 16S rRNA and recA gene similarity levels, the MLST data and the intermediate DNA–DNA binding values demonstrate that these five clusters represent five novel species within the Bcc. Biochemical identification of these species is difficult, as is the case for most Bcc species (Henry et al., 2001; LiPuma et al., 2007; Vandamme et al., 2007). However, identification of these novel species can be accomplished through recA gene sequence analysis, MLST and BOX-PCR profiling and by recA RFLP analysis. For diagnostic laboratories, recA gene sequence analysis offers the best combination of accuracy and simplicity. Based on these results, we propose to classify clusters 1 to 5 as Burkholderia latens sp. nov., Burkholderia diffusa sp. nov., Burkholderia arboris sp. nov., Burkholderia seminalis sp. nov. and Burkholderia metallica sp. nov., respectively.

Description of Burkholderia latens sp. nov.
Burkholderia latens (la'tens. L. fem. part. pres. latens concealed, hidden, because the taxonomic status of these strains remained unclear for a long time).

Cells are Gram-negative, aerobic, non-sporulating rods. Colonies are moist. The isolates can show a mucoid and a non-mucoid morphotype. Known strains grow on MacConkey agar. Strains grow on B. cepacia-selective agar (BCSA), forming acids. Growth is observed at 30, 37 and 42 °C; some isolates produce a diffusing, melanin-like, brown pigment at 37 and 42 °C on Luria–Bertani agar. Pigmented strains or haemolysis have not been detected. Known strains assimilate glucose, L-arabinose, D-mannose, D-mannitol, N-acetylglucosamine, D-gluconate, caprate, adipate, L-malate, citrate and phenylacetate, but not maltose. Acidification of glucose, maltose, lactose, xylose, sucrose and adonitol is observed. Nitrate reduction is absent. Oxidase, β-galactosidase and lysine decarboxylase activities are present, but not ornithine decarboxylase, tryptophanase, arginine dihydrolase, aesculin hydrolase or urease. Gelatin liquefaction is strain-dependent. The G+C content is 67 mol%. Strains have been isolated from sputum samples of cystic fibrosis patients.

The type strain is FIRENZE 3^T (=LMG 24064^T =CCUG 54555^T) and was recovered from sputum of a cystic fibrosis patient in Italy in 2000. Phenotypic and biochemical characteristics are as described above for the species.

Description of Burkholderia diffusa sp. nov.
Burkholderia diffusa (dif.fu'sa. L. fem. adj. diffusa widespread).

Cells are Gram-negative, aerobic, non-sporulating rods. Colonies are moist or smooth. Known strains grow on MacConkey agar. Strains can grow on BCSA, turning the medium either alkaline or acidic. Growth is observed at 30 and 37 °C, but is strain-dependent at 42 °C. So far, pigmented strains have not been detected and haemolysis has not been observed. Known strains assimilate glucose, L-arabinose, D-mannose, D-mannitol, N-acetylglucosamine, D-gluconate, caprate, adipate, L-malate, citrate and phenylacetate, but not maltose. Acidification of glucose, maltose, lactose, xylose and sucrose is observed, but acidification of adonitol is strain-dependent. Nitrate is reduced. Oxidase, β-galactosidase and lysine decarboxylase activities are present, but not ornithine decarboxylase, tryptophanase, arginine dihydrolase, aesculin hydrolase or urease. Gelatin liquefaction is strain-dependent. The G+C content is 67 mol%. Strains have been isolated from sputum samples of cystic fibrosis patients, blood samples from non-cystic fibrosis patients and from soil and water.

The type strain is AU1075^T (=LMG 24065^T =CCUG 54558^T) and was recovered from sputum of a cystic fibrosis patient in the USA in 1999. Phenotypic and biochemical characteristics are as described above for the species.

Description of Burkholderia arboris sp. nov.
Burkholderia arboris (ar.bo'ris. L. gen. n. arboris of the forest, referring to the source of isolation of the type strain, the Morris Arboretum in Philadelphia).

Cells are Gram-negative, aerobic, non-sporulating rods. Colonies are moist. Known strains grow on MacConkey agar. Strains can grow on BCSA and turn the medium either alkaline or acidic. Growth is observed at 30 and 37 °C; only a few strains are able to grow at 42 °C. Strains R-13059 and R-20536 produce a purple pigment after several days. Six of thirteen strains known show β-haemolysis, a characteristic not commonly observed among Bcc species. Known strains assimilate glucose, D-mannose, D-gluconate, adipate, L-malate and citrate. The assimilation of L-arabinose, D-mannitol, N-acetylglucosamine, maltose, caprate and phenylacetate is strain-dependent. Acidification of glucose, sucrose, maltose, lactose and adonitol is observed, but not of xylose. Nitrate reduction is strain-dependent. Oxidase and β-galactosidase activities are present, but not tryptophanase, arginine dihydrolase or urease activity. Gelatin liquefaction and activities of aesculin hydrolase and lysine and ornithine decarboxylases are strain-dependent. The G+C content is 67 mol%. Strains have been recovered from sputum of cystic fibrosis patients and from environmental sources.

The type strain is ES0263A^T (=LMG 24066^T =CCUG 54561^T) and was isolated in the Morris Arboretum in Philadelphia in 1999. Phenotypic and biochemical characteristics are as described above for the species.

Description of Burkholderia seminalis sp. nov.
Burkholderia seminalis (se.mi.na'lis. L. fem. adj. seminalis pertaining to the seed, referring to the rice seed surface, from which several strains were isolated).

Cells are Gram-negative, aerobic, non-sporulating rods. Colonies are moist. Known strains grow on MacConkey agar. All known strains grow on BCSA, forming acids. Growth is observed at 30 and 37 °C; growth at 42 °C is strain-dependent. Most strains are yellow-pigmented. Haemolysis has not been observed. Known strains assimilate glucose, L-arabinose, D-mannose, D-mannitol, N-acetylglucosamine, D-gluconate, caprate, adipate, L-malate, citrate and phenylacetate, but not maltose. Acidification of glucose, sucrose, maltose, lactose, xylose (weakly positive) and adonitol is observed. Oxidase, β-galactosidase activities and gelatin liquefaction are present, but no tryptophanase, arginine dihydrolase or urease activity. Ornithine and lysine decarboxylase and aesculin hydrolase activities are strain-dependent. Nitrate reduction is absent. The G+C content is 67 mol%. Strains have been isolated from sputum of cystic fibrosis patients, nosocomial infections and environmental samples.

The type strain is AU0475^T (=LMG 24067^T =CCUG 54564^T) and was recovered from sputum of a cystic fibrosis patient in the USA in 1998. Phenotypic and biochemical characteristics are as described above for the species.

Description of Burkholderia metallica sp. nov.
Burkholderia metallica (me.tal'li.ca. L. fem. adj. metallica metallic, referring to the metallic shine on the colony surface of the four known isolates).

Cells are Gram-negative, non-sporulating, aerobic rods. Colonies are moist, showing a metallic shine. Known strains grow on MacConkey agar. Known strains grow on BCSA, forming acids. Growth is observed at 30, 37 and 42 °C. Strains are not haemolytic and most strains are yellow-pigmented. Known isolates assimilate glucose, L-arabinose, D-mannose, D-mannitol, N-acetylglucosamine, D-gluconate, caprate, adipate, L-malate, citrate and phenylacetate, but not maltose. Acidification of glucose, sucrose, maltose, lactose, xylose and adonitol is observed. Nitrate reduction is absent. Activity of oxidase, β-galactosidase, lysine decarboxylase, aesculin hydrolase and gelatin liquefaction is present, but not tryptophanase, arginine dihydrolase or urease activity. The G+C content is 67 mol%. Known strains have been isolated from sputum samples of cystic fibrosis patients.

The type strain is AU0553^T (=LMG 24068^T =CCUG 54567^T) and was recovered from sputum of a cystic fibrosis patient in the USA in 1998. Phenotypic and biochemical characteristics are as described above for the species.

The five type strains have been deposited in the Belgian Co-ordinated Collections of Microorganisms/Laboratorium Microbiologie Gent and the Culture Collection of the University of Göteborg, Sweden. Additional strains (designated with LMG and CCUG numbers) have also been deposited in these culture collections (Table 1).

E. V. is indebted to the Special Research Council of Ghent University (Belgium). J. J. L. is supported by the Cystic Fibrosis Foundation (USA). A. B., C. D. and E. M. are indebted to the Cystic Fibrosis Trust UK (grant PJ535) for funding their research. D. S. and D. H. are supported by the Canadian Cystic Fibrosis Foundation. P. V. is indebted to the Fund for Scientific Research Flanders (Belgium) for research grants. We are grateful to all depositors who kindly collaborated in this study.