Use of Suppression-Subtractive Hybridization To Identify Genes in the Burkholderia cepacia Complex That Are Unique to Burkholderia cenocepacia

Abstract

We have previously shown differences in virulence between species of the Burkholderia cepacia complex using the alfalfa infection model and the rat agar bead chronic infection model. Burkholderia cenocepacia strains were more virulent in these two infection models than Burkholderia multivorans and Burkholderia stabilis strains. In order to identify genes that may account for the increased virulence of B. cenocepacia, suppression-subtractive hybridization was performed between B. cenocepacia K56-2 and B. multivorans C5393 and between B. cenocepacia K56-2 and B. stabilis LMG14294. Genes identified included DNA modification/phage-related/insertion sequences and genes involved in cell membrane/surface structures, resistance, transport, metabolism, regulation, secretion systems, as well as genes of unknown function. Several of these genes were present in the ET12 lineage of B. cenocepacia but not in other members of the B. cepacia complex. Virulence studies in a chronic lung infection model determined that the hypothetical YfjI protein, which is unique to the ET12 clone, contributes to lung pathology. Other genes specific to B. cenocepacia and/or the ET12 lineage were shown to play a role in biofilm formation and swarming or swimming motility.

Cystic fibrosis (CF) is the most common life-threatening autosomal recessive disease in the Caucasian population, affecting more than 50,000 individuals worldwide (25) with an incidence of 1 in 3,200 live births (51). Despite a better understanding of CF and improved therapies and life expectancy over the last decades, 90% of the morbidity and mortality involving this genetic disorder is caused by progressive loss of lung function (12, 25, 37, 51). Respiratory failure in CF lungs is related to severe host inflammatory responses caused by chronic infection leading to progressive airway obstruction, long-term tissue damage, and early death (15, 37, 45). The most common pathogens affecting CF lungs are Pseudomonas aeruginosa, Staphylococcus aureus, Haemophilus influenzae, and the Burkholderia cepacia complex (Bcc) (53). Despite the fact that P. aeruginosa remains the most prevalent CF pathogen and that P. aeruginosa infection is the most common cause of mortality in CF patients, several studies report that clinical outcomes of CF patients infected with Bcc are often worse than those infected with P. aeruginosa (32). Bcc infection can result in “cepacia syndrome,” which is characterized by a rapid pulmonary decline with necrotizing pneumonia, fever, and occasional septicemia (58).

Bcc is a group of closely related bacterial species recognized to be important opportunistic pathogens in immunocompromised patients, including those with CF and chronic granulomatous disease (10, 24). To date, there are nine Bcc species that have been isolated from sputum of CF patients (8, 9, 35). Burkholderia cenocepacia (formerly genomovars III) and Burkholderia multivorans (formerly genomovars II) are the two species most commonly recovered from CF sputum, but their proportions vary geographically. In Canada, 80% of the Bcc CF isolates are B. cenocepacia compared to 9% B. multivorans (62); however, in the United States, B. multivorans is recovered from 38% of CF patients colonized with Bcc compared to 50% for B. cenocepacia (35). Highly transmissible B. cenocepacia clones, such as the ET12, Midwest, and PHDC (Philadelphia-DC) lineages, have been identified in outbreaks in Europe and North America (6, 39).

Strains belonging to the B. cenocepacia ET12 lineage contain both the B. cepacia epidemic strain marker (BCESM) (39) and the cable pili gene (cblA) (63). ET12 strains were shown to be responsible for outbreaks in Canada and the United Kingdom and were linked to patient-to-patient transmission (62, 63). The PHDC clone has also been implicated in outbreaks, yet it lacks these two putative virulence markers (6, 8), demonstrating that the BCESM and cable pili are not necessary for transmissibility. The BCESM has recently been shown to be part of a genomic island, the B. cenocepacia island (cci), which is involved in pathogenicity and metabolism (2).

Virulence factors known to play a significant role in the pathogenesis of B. cenocepacia ET12 strains include two quorum-sensing systems, CepIR (61) and CciIR (2), iron acquisition via siderophore production (60, 66), a protease (14), a type III secretion system (65), and a type IV secretion system (TFSS) (19). Invasion and survival in epithelial cells (4, 7, 56), macrophages (42), amoebae (41), and acanthamoeba (33) have also been demonstrated. Analysis of a signature-tagged mutagenesis (STM) library in a chronic pulmonary infection model identified several B. cenocepacia genes directly involved in host survival, including genes involved in cellular metabolism, regulation, DNA replication and repair, cell surface proteins, and polysaccharide production (31).

Recently, we have shown differences in virulence between species of the Bcc using alfalfa and rat agar bead infection models (3). For most strains tested there was a correlation in virulence between the two models, and B. cenocepacia was one of the most virulent species in these two infection models (3). This observation is consistent with the clinical profile since CF patients infected with B. cenocepacia frequently experience worse outcomes than patients infected with other Bcc species (32, 40). Burkholderia stabilis (formerly genomovars IV) and B. multivorans were generally avirulent in the alfalfa infection model. In the rat agar bead chronic infection model, lung pathology changes were significantly less for B. multivorans strains than B. cenocepacia, and B. stabilis strains did not persist in the lung as well as other species (3).

The objectives of this study were to identify genes that may account for the differences in virulence between B. cenocepacia, B. multivorans, and B. stabilis in CF patients, plants (alfalfa), and animals (rats). Suppression-subtractive hybridization (SSH), which has previously been used to identify virulent determinants between closely related Burkholderia species (17, 52), was performed independently between B. cenocepacia K56-2 and B. multivorans C5393 and B. stabilis LMG 14294, respectively. Select unique B. cenocepacia K56-2 genes were evaluated for their distribution within the Bcc, their implication in virulence using a chronic lung infection model, as well as their role in biofilm formation and swarming and swimming motility.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions.

The bacterial strains and plasmids used in this study are described in Table 1. Cultures were grown in Luria broth (LB) (Invitrogen, Burlington, Ontario, Canada) or on 1.5% LB agar plates supplemented with antibiotics when required and incubated at 37°C for 24 h. When appropriate, antibiotics were used at the following concentrations: 50 μg/ml of kanamycin (Km) and 1,200 μg/ml of trimethoprim (Tp) for Escherichia coli and 100 μg/ml of Tp for B. cenocepacia. Plasmids were purified from overnight cultures in LB medium supplemented with appropriate antibiotic by using a QIAprep Spin Miniprep kit (QIAGEN, Mississauga, Ontario, Canada). For animal studies, overnight cultures were grown in TSB-DC medium (46). For swarming and swimming motility assays, cultures were grown in nutrient broth (NB) (Difco) supplemented with 0.5% glucose. All chemicals were purchased from Sigma-Aldrich Canada Ltd. (Oakville, Ontario, Canada).

TABLE 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Relevant characteristicsa	Reference or source
E. coli strains
DH5α	F− φ80d lacZΔM15Δ(lacZYA-argF)U169 recA1 endA1 hsdR17(rK− mK+) phoA supE44 λ−thi-1 gyrA96 relA1	Invitrogen
HB101	supE44 hsdS20(rB mB) recA13 ara-14 proA2 lacY1 galK2 rpsL20 xyl-5 mtl-1	57
TOP10	F−mcrA Δ(mrr-hsdRMS mcrBC) φ80lacΔM15 ΔlacX74 deoR recA1 araD139 Δ(ara-leu)7697 galU galK rpsL(Smr) endA1 nupG	Invitrogen
Bcc strains
B. cepacia
ATCC 25416T	Onion isolate (United States)	38
ATCC 17759	Soil isolate (Trinidad)	38
CEP509	CF sputum isolate (Australia)	38
LMG 17997	UTI isolate (Sweden)	38
B. multivorans
C5393	CF sputum isolate (Canada)	38
LMG 13010T	CF sputum isolate (Belgium)	38
C1576	CF sputum isolate (United Kingdom)	38
CF-A1-1	CF sputum isolate (United Kingdom)	38
JTC	CGD isolate (United States)	38
C1962	Clinical isolate (United Kingdom)	38
ATCC 17616	Soil isolate (United States)	38
249-2	Laboratory isolate (United States)	38
B. cenocepacia
J2315	CF sputum isolate (United Kingdom); ET12 lineage, cblA+, recA subgroup A	38
BC7	CF sputum isolate (Canada); ET12 lineage, cblA+, recA subgroup A	38
K56-2	CF sputum isolate (Canada); ET12 lineage, cblA+, recA subgroup A	38
C5424	CF sputum isolate (Canada); ET12 lineage, cblA+, recA subgroup A	38
C6433	CF sputum isolate (Canada)	38
C1394	CF sputum isolate (United Kingdom)	38
PC184	CF sputum isolate (United States); Midwest lineage	38
CEP511	CF sputum isolate (Australia)	38
J415	CF sputum isolate (United Kingdom)	38
ATCC 17765	UTI isolate (United Kingdom)	38
Pc715j	CF sputum isolate (Canada)	14
SB002	K56-2 derivative; ΔyfjI::tp, Tpr	This study
SB003	K56-2 derivative; ΔVgr-related protein::tp, Tpr	This study
SB004	K56-2 derivative; Δphospholipase-like gene::tp, Tpr	This study
SB006	K56-2 derivative; Δautotransporter adhesin::tp, Tpr	This study
SB007	K56-2 derivative; ΔcpaC::tp, Tpr	This study
10F1	K56-2 derivative; pTnMod insertion into a hypothetical protein, Tpr	31
17D8	K56-2 derivative; pTnMod insertion into fkbH, Tpr	31
32D2	K56-2 derivative; pTnMod insertion into wbiG, Tpr	31,47
34D8	K56-2 derivative; pTnMod insertion into rmlD, Tpr	31,47
36B4	K56-2 derivative; pTnMod insertion into cpxA, Tpr	31
SRC2	K56-2 derivative; ΔvirB1-virB2::tp, Tpr	This study
B. stabilis
LMG 14294	CF sputum isolate (Belgium)	38
C7322	CF sputum isolate (Canada)	38
LMG 14086	Respiratory isolate (United Kingdom)	38
LMG 18888	Clinical isolate (Belgium)	38
B. vietnamiensis
PC259	CF sputum isolate (United States)	38
LMG 16232	CF sputum isolate (Sweden)	38
FC441	CGD isolate (Canada)	38
LMG 10929T	Rice isolate (Vietnam)	38
B. dolosa
LO6	CF sputum isolate
AU0645	CF sputum isolate (United States)	11
LMG 19468T		D. P. Speert
B. ambifaria
ATCC 53266	Soil isolate (United States)	11
AMMDT	Soil isolate (United States)	11
CEP0996	CF sputum isolate (Australia)	11
B. anthina
W92T	Soil isolate (United States)	11
C1765	CF sputum isolate (United Kingdom)	11
J2552	Rhizosphere isolate (United Kingdom)	11
LMG 20982		D. P. Speert
B. pyrrocinia
ATCC 15958T	Soil isolate (Japan)	11
ATCC 39277	Soil isolate (United States)	11
Plasmids
pBluescript SK+	Cloning vector; Apr	Stratagene
pCR 2.1-TOPO	Cloning vector for PCR products; Apr, Kmr	Invitrogen
pRK2013	Mobilizing vector, ColE1 Tra (RK2)+Kmr	22
p34E-Tp	Source of trimethoprim cassette; Tpr	18
pEX18Tc	Suicide vector; sacB Tcr	28

^aAll PCR products were amplified from B. cenocepacia K56-2 genomic DNA. CF, cystic fibrosis; UTI, urinary tract infection; CGD, chronic granulomatous disease; Ap, ampicillin; Km, kanamycin; Tc, tetracycline; Tp, trimethroprim.

Construction of suppression-subtractive hybridization libraries.

Genomic DNA was isolated using a Wizard genomic DNA purification kit (Promega, Madison, Wisconsin). Suppression-subtractive hybridization was carried out using a PCR-Select bacterial genome subtraction kit (Clontech, Palo Alto, CA) as recommended by the supplier except that two hybridization temperatures of 73°C and 85°C were used in separate experiments due to the high G+C content of B. cenocepacia. Two suppression-subtractive hybridization libraries were constructed using B. cenocepacia K56-2 genomic DNA as the tester in both libraries and B. multivorans C5393 and B. stabilis LMG 14294 genomic DNA as the driver in their respective library. PCR products generated in each library were cloned into pCR2.1-TOPO (Invitrogen) using a TOPO TA cloning kit (Invitrogen) and transformed into TOP10 cells (Invitrogen). Resulting transformants were individually grown overnight in LB medium with kanamycin (50 μg/ml) at 37°C in 96-well microtiter plates and stored in 15% glycerol at −70°C.

Screening of B. cenocepacia K56-2-specific clones in each library.

Subtracted libraries were screened for tester-specific fragments using a modified dot blot hybridization previously described (50, 72). Plasmid DNA from individual clones was isolated using a QIAprep Spin Miniprep kit (QIAGEN). Approximately 200 ng of plasmid DNA was spotted in duplicate onto wet GeneScreen Plus hybridization transfer membranes (Perkinelmer Life Sciences, Boston, MA) using a BIO-DOT SF blotting apparatus (Bio-Rad Laboratories, Richmond, CA) in a 96-well format. Denatured DNA was hybridized with approximately 1 μg of RsaI-digested genomic DNA from tester and driver DNA separately, previously end-labeled with [α-³²P]dCTP (GE Healthcare).

Distribution of B. cenocepacia K56-2-specific genes within the B. cepacia complex.

Genomic DNA from 43 strains belonging to different species of the Bcc (11, 38) was isolated using Chelex 100 as previously described (68). Dot blot hybridization was carried out as described above except that DNA was spotted onto nitrocellulose membranes and probes used were obtained by PCR amplification from clones with primers M13F and M13R (Invitrogen), purified by QIAquick PCR Purification kit (QIAGEN), and labeled with [α-³²P]dCTP (GE Healthcare). For PCR analysis, primers were designed to amplify internal regions specific to genes unique to B. cenocepacia K56-2.

DNA manipulations.

Molecular biology techniques were performed as generally described (57). Restriction enzymes and T4 DNA polymerase were purchased from Invitrogen. T4 DNA ligase was purchased from New England Biolabs (Mississauga, Ontario, Canada). Oligonucleotide primers were purchased from University of Calgary Core DNA and Protein Services. DNA fragments used in cloning procedures were purified with a QIAquick gel extraction kit (QIAGEN), and recombinant plasmids were electroporated into E. coli by using a Gene Pulser (Bio-Rad Laboratories) according to the manufacturer's recommendations.

DNA sequencing and analysis.

DNA sequencing reactions were performed by Macrogen Inc. (Seoul, Korea) with an ABI 3730XL automatic DNA sequencer. Fragments identified by SSH were analyzed by using the unpublished annotation from the B. cenocepacia J2315 sequencing project at the Sanger Institute (http://www.sanger.ac.uk/projects/B_cenocepacia/) using Artemis software (55). Each open reading frame (ORF) predicted by the Sanger Institute was used to identify homologous sequences with the BLASTX (http://www.ncbi.nlm.nih.gov/BLAST/) program.

Construction of different B. cenocepacia K56-2 mutants by allelic exchange.

Construction of deletion mutants in B. cenocepacia K56-2 was performed as follows. Two sets of PCR primers were designed for each gene targeted for mutagenesis. Each set of primers amplified two discontinuous parts of the gene in order to create a deletion. Each primer contained a restriction site at the 5′ end in order to facilitate cloning into pEX18Tc (29). The fragment amplified with the first primer set was cloned into pEX18Tc or pBluescript for SRC2, followed by the cloning of the fragment generated by the second set of primers. A Tp resistance cassette obtained from p34E-Tp (18) was then inserted between the two fragments of the targeted gene into the restriction site generated by the primers used to clone the two fragments. This plasmid was transferred from E. coli to B. cenocepacia K56-2 by triparental mating using the mobilizing plasmid pRK2013 (22). Tp^r transconjugants were plated onto 5% sucrose to select for excision of the plasmid. Confirmation of the mutant genotype was preformed by PCR.

Biofilm assay.

Biofilm assays were performed as previously described (48) with minor modifications. Biofilms were formed on polystyrene pegs by placing a 96-peg lid (Nunc, Roskilne, Denmark) in a 96-well microtiter plate (Nunc) containing 5 μl of a 0.4 optical density at 600 nm suspension diluted in 145 μl of LB broth (Invitrogen). The plate was incubated for 24 h at 37°C on a rocking platform. The lid was removed, air dried for 10 min, stained with 200 μl of 1% crystal violet (Sigma) in a 96-well plate for 1 min, and rinsed three times in separate plates containing double-distilled water. The stained pegs were decolorized with 175 μl of 95% ethanol in a microtiter plate for 1 min. The quantity of crystal violet removed was measured using a Wallec Victor2 model 1420 multilabel counter (PerkinElmer Life Sciences) set to measure absorbance at 600 nm.

Swarming and swimming motility assays.

Motility assays were performed as previously described (34). Briefly, 1 microliter of an overnight culture was spotted in the middle of a swarm plate (NB, 0.5% glucose, 0.5% agar) or a swim plate (NB, 0.5% glucose, 0.25% agar), allowed to dry for 1 h at room temperature, and incubated for 24 h for the swarming and 12 h for swimming assays at 37°C. Diameters of swarming and swimming zones were measured.

Animal studies.

Sprague-Dawley rats (150 to 175 g) (Charles River Canada, Inc.) were tracheostomized under anesthesia and inoculated with approximately 10⁴ CFU of the appropriate strain embedded in agar beads as previously described (5). At 14 days postinfection, the lungs from four to five animals from each group were removed aseptically and homogenized (Polytron Homogenizer; Brinkman Instruments, Westbury, N.Y.) in 3 ml of phosphate-buffered saline (10 mM sodium phosphate, 150 mM NaCl, pH 7.5). The homogenates were serially diluted in phosphate-buffered saline and plated on trypticase soy agar (Difco) and B. cepacia isolation agar (27) with 100 μg/ml of trimethoprim when required. The lungs of four to five additional animals from each group were removed en bloc, fixed in 10% formalin, and examined for quantitative pathological changes. Infiltration of the lung with inflammatory cells and exudate was measured as previously described (3).

Alfalfa infection assay.

Virulence studies in the alfalfa infection model were performed as previously described (3).

Statistical analysis.

Analysis of variance (ANOVA) was performed with INSTAT software (GraphPad Software, San Diego, Calif.). A P value of <0.05 was considered statistically significant.

RESULTS

Construction of suppression-subtractive hybridization libraries and screening for B. cenocepacia K56-2-specific clones.

SSH was carried out between B. cenocepacia K56-2 and B. multivorans C5393 and B. stabilis LMG 14294 in order to identify genomic regions specific to B. cenocepacia K56-2 that could potentially be linked to its increased virulence. Genomic DNA from the tester DNA (B. cenocepacia K56-2) and the driver DNA (B. multivorans C5393 or B. stabilis LMG 14294) were digested with the restriction enzyme RsaI, ligated with two different adaptors (Clontech), and subjected to two rounds of hybridization. Hybridization temperatures of 73°C and 85°C were used in separate experiments. After two rounds of hybridization, the unhybridized tester-specific fragments were subsequently amplified twice by PCR using adaptor-specific primers (Clontech) and cloned into pCR2.1-TOPO (Invitrogen). The four subtracted libraries were screened by dot blot hybridization with both tester and driver genomic DNA to identify clones that hybridized with K56-2 but not C5393 or LMG 14294. The K56-2-specific clones were sequenced and compared to those from the B. cenocepacia J2315 sequencing project at the Sanger Institute (http://www.sanger.ac.uk/projects/B_cenocepacia/). B. cenocepacia strains J2315 and K56-2 belong to the same clonal group (38). Characteristics of the clones generated from each of the four SSH libraries were similar, with an average size of approximately 500 bp and a G+C content of 54%, with the exception of the library generated using B. multivorans C5393 as driver DNA with a hybridization temperature of 85°C, in which the average size of each insert was 1,240 bp and the G+C content was 65%.

Identification of sequences present in B. cenocepacia K56-2 and absent from B. multivorans C5393 and/or B. stabilis LMG 14294.

Plasmid DNA from 384 clones from the B. multivorans C5393 subtracted libraries and 301 clones from the B. stabilis LMG 14294 subtracted libraries were analyzed, and 205 were shown to be specific to B. cenocepacia K56-2 by dot blot hybridization for an efficiency of approximately 30%. Of these 205 clones, 102 were from the B. multivorans C5393 subtracted libraries and 103 from the B. stabilis LMG 14294 subtracted libraries. Sequencing of these clones led to the identification of 89 different ORFs. Genes identified from these 205 unique clones are shown in Tables 2 and 3. These ORFs were assigned to one of nine classes: DNA modification/phage-related/insertion sequences, cell membrane/surface structures, resistance, transport, metabolism, regulation, secretion systems, and unknown functions. Of the 89 different genes identified by SSH, 82 were present in a single copy (Table 2) and seven had multiple copies (Table 3) within the B. cenocepacia J2315 genome. All of the multiple copy genes coded for insertion sequence (IS) elements and were identified in both subtraction libraries (Table 3). Seven different IS elements were 95 to 100% identical at the nucleotide level to the B. cenocepacia J2315 genome and were located in 153 different locations (Table 3). For example, IS407A and transposase and inactivated derivatives (COG2801) are present in at least 13 copies each in the genome (Table 3).

TABLE 2.

B. cenocepacia K56-2 sequences identified by SSH libraries that are present in single copy in the B. cenocepacia J2315 genome

Gene (homologue) or putative functionb	Clonea	Organism	e valuec	Accession no.	CDS no. in the J2315 genome	Chromosome location	K56-2 vs C5393d	K56-2 vs LMG 14294
DNA modification, phage-related, and IS elements
Bacteriophage P4 integrase	SH4-73 1F12	Ralstonia solanacearum GMI1000	e-115	NP_519992	BCAL1118	1		x
Hypothetical protein, 24K insertion sequence IS402	SH2-73 2C4 SH4-85 2G5	Burkholderia cepacia	6e-93	JQ1133	BCAL1125	1	x	x
gp21, phage BcepMu	SH-73 2B2	Burkholderia cenocepacia	1e-09	YP_024694	BCAL1598	1	x
COG0806: RimM protein, required for 16S rRNA processing	SH2-85 1A5	Burkholderia cepacia R18194	1e-99	ZP_00217269	BCAL2927	1	x
COG0228: ribosomal protein S16		Burkholderia cepacia R18194	3e-41	ZP_00217270	BCAL2928	1
Possible helicase	SH4-85 2H11	Rhodopseudomonas palustris CGA009	e-148	CAE27655	BCAM0780	2		x
Probable 5-methylcytosine-specific restriction enzyme A	SH4-85 2G2	Desulfotalea psychrophila LSv54	2e-57	YP_065769	BCAM1520	2		x
Phage-related repressor protein	SH2-85 2C6	Xylella fastidiosa 9a5c	1e-13	NP_298947	BCAM1053a	2	x
Putative bacteriophage protein	SH4-73 1B1	Salmonella enterica subsp. enterica serovar Typhi strain CT18	1e-24	NP_455540	BCAM1079	2		x
Putative bacteriophage protein		Salmonella enterica subsp. enterica serovar Typhi strain CT18	8e-16	NP_455541	BCAM1080	2		x
Phage-related integrase	SH2-73 2C3	Xanthomonas campestris pv. campestris strain ATCC 33913	8e-39	NP_636833	BCAM1923	2	x
IS1328 transposase	SH2-73 2E4	Shigella flexneri 2a	e-131	AAL08483	BCAM0771	2	x
Virulence-associated protein E (integrase)	SH4-73 1D9	Bacillus cereus ATCC 10987	8e-26	NP_976712	BCAM1877	2		x
Cell membrane/surface structure
COG4655: predicted membrane protein	SH4-85 2B5	Burkholderia cepacia R18194	0	ZP_00211424	BCAL1535	1		x
Putative glycosyl transferase (wbiF)	SH2-85 2E7	Burkholderia pseudomallei	1e-87	AAD05465	BCAL3122	1	x
COG0463: Glycosyltransferases involved in cell wall biogenesis	SH2-73 1A1	Trichodesmium erythraeum 1MS101	3e-25	ZP_00324655	BCAL3124	1	x
	SH2-85 1A9
O-antigen ABC transporter, permease protein (wzm)	SH2-73 1B2	Pseudomonas syringae pv. tomato strain DC3000	3e-54	NP_790910	BCAL3131	1	x
Putative dTDP-4-keto-l-rhamnose reductase (rmlD)		Burkholderia mallei	9e-83	AAK27394	BCAL3132		x
fkbH	SH4-73 1A1	Streptomyces hygroscopicus var. ascomyceticus	4e-25	AAF86387	BCAL3229	1		x
Glycosyltransferase	SH4-73 1C11	Lactobacillus plantarum WCFS1	1e-12	NP_784880	BCAL3234	1		x
UDP-galactopyranose mutase		Klebsiella pneumoniae	e-144	BAD03951	BCAL3235			x
Capsule polysaccharide export inner-membrane (ctrC) wcbD	SH4-73 1D6	Neisseria meningitidis MC58 Burkholderia pseudomallei	5e-38 5e-82	NP_273137AAK49797	BCAL3241 BCAL3242	1		x x
COG5295: autotransporter adhesin	SH2-85 2F11	Burkholderia fungorum LB400	1e-63	ZP_00283251	BCAM0224	2	x
COG5295: autotransporter adhesin		Burkholderia fungorum LB400	3e-20	ZP_00281431	BCAM0225		x
COG5486: predicted metal-binding integral membrane protein	SH2-85 1D10	Burkholderia cepacia R1808	1e-94	ZP_00221602	BCAM1601	2	x
Putative outer membrane usher	SH2-85 2D6	Burkholderia cepacia	0	AAM56039	BCAM2760	2	x
Major pilin (cblA)		Burkholderia cepacia	5e-68	AAM56038	BCAM2761	2	x
COG5295: autotransporter adhesin	SH2-85 1G8	Burkholderia cepacia R18194	0	ZP_00214964	BCAS0236	3	x
Resistance
COG1566: multidrug resistance efflux pump	SH2-85 1A2	Burkholderia cepacia R18194	e-111	ZP_00211749	BCAL1176	1	x	x
Fusaric acid resistance protein	SH4-85 1F2	Burkholderia mallei ATCC 23344	e-148	AAU49448	BCAL1177	1	x	x
Putative antibiotic resistance membrane protein	SH2-85 2E12	Burkholderia pseudomallei K96243	e-112	YP_108570	BCAL1178	1
Transport
COG1629: outer membrane receptor proteins, mostly Fe transport	SH2-85 1G6	Burkholderia cepacia R1808	0	ZP_00222248	BCAL1783	1	x
COG0848: biopolymer transport protein	SH2-85 2D7	Burkholderia cepacia R1808	2e-47	ZP_00222251	BCAL1787	1	x
COG2113: ABC-type proline/glycine betaine transport systems, periplasmic components	SH2-85 1H7	Burkholderia cepacia R18194	e-162	ZP_00213820	BCAM2312	2	x
Metabolism
Ethanolamine ammonia-lyase light chain (eutC)	SH2-85 1H12	Burkholderia pseudomallei K96243	e-116	YP_109966g	BCAL0060	1	x
Hydrolase, alpha/beta fold family	SH2-73 1D11	Burkholderia mallei ATCC 23344	4e-39	AAU49935	BCAL1021	1	x
Succinylornithine transaminase	SH2-73 1D6	Burkholderia pseudomallei K96243	0	YP_108982	BCAL1059	1	x
COG1021: peptide arylation enzymes	SH4-85 2B7	Polaromonas sp. strain JS666	0	ZP_00363238	BCAL1159	1		x
COG2303: choline dehydrogenase and related flavoproteins	SH2-85 1E11	Burkholderia fungorum LB400	e-150	ZP_00282303	BCAL1185	1	x
COG0673: predicted dehydrogenases and related proteins	SH2-85 1A12	Burkholderia cepacia R18194	e-152	ZP_00212578	BCAL1598	1	x
Putative hydrolase	SH2-73 1D11	Burkholderia pseudomallei K96243	7e-52	YP_107562	BCAL2851	1	x
COG0589: universal stress protein UspA and related nucleotide-binding proteins	SH2-85 2F12	Burkholderia cepacia R1808	5e-48	ZP_00222386	BCAM0276	2	x
COG5517: small subunit of phenylpropionate dioxygenase	SH4-85 2A2	Burkholderia cepacia R18194	9e-87	ZP_00217599	BCAM0647	2		x
COG4638: phenylpropionate dioxygenase and related ring-hydroxylating dioxygenases, large terminal subunit		Burkholderia cepacia R18194	0	ZP_00217598	BCAM0648			x
Probable isoquinoline 1-oxidoreductase (alpha subunit)	SH2-85 2D12	Ralstonia solanacearum GMI1000	9e-69	NP_520011	BCAM2097	2	x
Probable transmembrane isoquinoline 1-oxidoreductase (beta subunit)		Ralstonia solanacearum GMI1000	0	NP_520010	BCAM2098		x
Dienelactone hydrolase	SH2-73 1H11	Pseudomonas resinovorans	e-138	NP_758553	BCAM2127	2	x	x
	SH4-73 1H7
COG2856: predicted Zn peptidase	SH2-85 1G4	Burkholderia cepacia R1808	3e-84	ZP_00229177	BCAM2275	2	x
COG0665: glycine/d-amino acid oxidases (deaminating)		Burkholderia cepacia R18194	0	ZP_00213860	BCAM2276		x
COG0491: Zn-dependent hydrolases, including glyoxylases	SH2-85 1F11	Burkholderia cepacia R18194	e-153	ZP_00217927	BCAS0034	3	x
Regulatory
Transcriptional regulatory protein	SH2-85 2B9	Bradyrhizobium japonicum USDA 110	4e-64	NP_774624	BCAM0514	2	x
COG0464: ATPases of the AAA+ class	SH4-85 1A6	Mesorhizobium sp. strain BNC1	e-149	ZP_00197048	BCAM0524	2		x
Secretion systems
Type IV secretory pathway (virB4)	SH2-73 1D10	Brucella melitensis biovar abortus	0	AAF73897	BCAM0327	2	x
COG3843: type IV secretory pathway (virD2)	SH2-73 2C12	Burkholderia fungorum LB400	e-106	ZP_00278500	BCAM0340	2	x
traW	SH2-85 1A11	Yersinia pestis	4e-28	YP_093977	pBCA032	Plasmid	x
COG4959: type IV secretory pathway, protease (traF)		Dechloromonas aromatica RCB	2e-12	ZP_00151021	pBCA033	Plasmid	x
traB	SH2-85 1C6	Providencia rettgeri	4e-30	AAM08009	pBCA044	Plasmid	x
Unknown function
Hypothetical protein Yfj1	SH2-73 1C9	Escherichia coli CFT073	9e-66	NP_755527	BCAL1122	1	x	x
	SH4-85 1C4
COG1479: uncharacterized conserved protein	SH4-85 1E12	Trichodesmium erythraeum IMS101	e-124	ZP_00324546	BCAL1172	1		x
Hypothetical protein Bcepa03005069	SH4-73 1B9	Burkholderia cepacia R18194	4e-29	ZP_00213706	BCAL1291	1		x
Hypothetical protein Paer03000370	SH4-73 1B7	Pseudomonas aeruginosa PA14	1e-14	ZP_00136093	BCAL1292	1		x
Conserved hypothetical protein	SH2-73 1C1	Burkholderia pseudomallei K96243	e-132	CAH34721	BCAL1293	1	x	x
	SH4-85 2C8
Hypothetical protein XCC1451	SH2-85 1E7	Xanthomonas campestris pv. campestris strain ATCC 33913	4e-24	NP_636824	BCAL1302	1	x
Hypothetical protein Bcepa03006404	SH2-85 1H4	Burkholderia cepacia R18194	1e-42	ZP_00212357	BCAL2463	1	x
Hypothetical protein Raeut03003273	SH2-73 1E10	Ralstonia eutropha JMP134	e-99	ZP_00168249	BCAL2969	1	x	x
	SH4-85 1C12
Hypothetical protein SAV5825	SH4-85 2A9	Streptomyces avermitilis MA-4680	3e-83	NP_827002	BCAM0066	2		x
Hypothetical protein CC1419	SH2-73 1G8	Caulobacter crescentus CB15	1e-19	NP_420232	BCAM0073	2	x	x
Hypothetical protein Mdeg02003270	SH4-73 1B12	Microbulbifer degradans 2-40	2e-38	ZP_00315468	BCAM0074		x	x
Hypothetical protein XAC2421	SH4-85 1H9	Xanthomonas axonopodis pv. citri strain 306	4e-31	NP_642737	BCAM0163	2		x
Putative lipoprotein		Burkholderia pseudomallei K96243	0	YP_110763	BCAM0164			x
Hypothetical protein Bucepa03003024	SH2-85 2F12	Burkholderia cepacia R1808	8e-41	ZP_00222387	BCAM0275a	2	x
Hypothetical protein Bucepa03003022		Burkholderia cepacia R1808	3e-38	ZP_00222385	BCAM0277		x
COG4430: uncharacterized protein conserved in bacteria	SH2-85 2B9	Ralstonia metallidurans CH34	4e-76	ZP_00274169	BCAM0513	2	x
Hypothetical protein Raeut03003279	SH2-73 1D3	Ralstonia eutropha JMP134	1e-20	ZP_00350986	BCAM1087	2	x
Putative lipoprotein	SH2-85 1D10	Burkholderia mallei ATCC 23344	6e-53	YP_102742	BCAM1598	2	x
Putative exported protein		Burkholderia pseudomallei K96243	3e-19	YP_106937	BCAM1600		x
Hypothetical protein Bcepa03003586	SH4-85 2G5	Burkholderia cepacia R18194	5e-55	ZP_00214872	BCAM2101	2		x
Putative VGR-related protein	SH2-73 2A10	Ralstonia solanacearum GMI1000	0	NP_522190	BCAM0148	2	x	x
Probable phospholipase protein	SH4-73 1D12	Ralstonia solanacearum GMI1000	e-171	NP_522191	BCAM0149		x	x
	SH4-85 2D2
Hypothetical protein lpg1269	SH4-85 1A10	Legionella pneumophila subsp. pneumophila strain Philadelphia 1	1e-46	YP_095300	BCAS0661A	3		x

^aClones in bold indicate those which were used to screen the Bcc strain collection for their distribution within the Bcc by dot blot hybridization and/or PCR.

^bFragments identified by SSH were analyzed by using the unpublished annotation from the B. cenocepacia J2315 sequencing project at the Sanger Institute (http://www.sanger.ac.uk/projects/B_cenocepacia/). Each ORF predicted by the Sanger Institute containing homologous sequences was identified based on sequence homology using BLASTX (http://www.ncbi.nlm.nih.gov/BLAST/). Clones may overlap more than one gene.

^ce value is based on the BLASTX result of the entire predicted ORF.

^dx, SSH library and/or libraries in which the clone was identified.

TABLE 3.

B. cenocepacia K56-2 sequences identified by SSH that are present in multiple copies in the B. cenocepacia J2315 genome

Homologue or putative functionb	Clonea	Organism	e valuec	Accession no.	No. of copies in the J2315 genome	K56-2 vs C5393d	K56-2 vs LMG 14294
Insertion sequences
Putative transposition helper protein	SH2-73 1D12	Salmonella enterica subsp. enterica serovar Cubana	5e-87	AAM10642	4	x	x
	SH4-73 1D1
	SH4-85 2A2
ISRS011-transposase OrfA protein	SH2-73 1A4	Ralstonia solanacearum GMI1000	2e-38	NP_521211	5	x
Transposase	SH4-73 1A9	Klebsiella pneumoniae	9e-76	NP_943536	12		x
ISGsu6-transposase OrfA	SH2-73 1D12	Geobacter sulfurreducens PCA	6e-83	NP_953639	2	x	x
	SH4-73 1D1
	SH4-85 2A2
IS407A-transposase OrfA	SH2-85 1F2	Burkholderia mallei ATCC 23344	8e-35	YP_102805	4	x
IS407A-transposase OrfA	SH2-85 1F2	Burkholderia mallei ATCC 23344	6e-35	YP_103788	9	x
IS407A-transposase OrfB	SH2-85 1F2	Burkholderia mallei ATCC 23344	e-143	YP_102806	9	x
IS407A-transposase OrfB	SH2-85 1F2	Burkholderia mallei ATCC 23344	e-143	YP_105547	4	x
COG2801: transposase and inactivated derivatives	SH2-73 2B6	Ralstonia metallidurans CH34	e-100	ZP_00272805	13	x	x
	SH4-73 1C5
	SH4-85 1C6
COG4584: transposase and inactivated derivatives	SH2-73 2D1	Burkholderia cepacia R1808	e-148	ZP_00219761	2	x	x
	SH4-73 1D1
	SH4-85 2B4

^aClones in bold indicate those which were used to screen the Bcc strain collection for their distribution within the Bcc by dot blot hybridization and/or PCR.

^bFragments identified by SSH were analyzed by using the unpublished annotation from the B. cenocepacia J2315 sequencing project at the Sanger Institute (http://www.sanger.ac.uk/projects/B_cenocepacia/). Each ORF predicted by the Sanger Institute containing homologous sequences was identified based on sequence homology using BLASTX (http://www.ncbi.nlm.nih.gov/BLAST/).

^ce value is based on the BLASTX result of the entire predicted ORF.

^dx, SSH library and/or libraries in which the clone was identified.

Although fewer genes were identified by using B. stabilis LMG 14294 as driver DNA than B. multivorans C5393, the distribution in classes of genes between the two libraries was almost identical (Table 4). More than half of the genes identified were DNA modification/phage-related/IS elements. With the exception of a few enzymes related to DNA replication and repair (Table 2), most of the genes in that category were phage-related genes (Table 2) and multicopy IS elements (Table 3). These genes were all located in low G+C regions, demonstrating the preference of SSH in identification of A+T rich regions.

TABLE 4.

Distribution of genes identified by suppression-subtractive hybridization between B. cenocepacia and B. multivorans or B. stabilis

Class of genes	% of each class of genes identified
	K56-2 vs C5393	K56-2 vs LMG 14294
DNA modification/phage-related/IS elements	55	59
Cell membrane/surface structure genes	9	9
Resistance-associated genes	3	3
Transport-associated genes	3	0
Regulatory genes	1	1
Metabolic genes	11	6
Secretion systems	5	0
Unknown function	13	22

Several genomic regions were identified containing multiple SSH clones within the same gene cluster (Table 2). For example, three clones were identified within a 7.7-kb low G+C content cluster on chromosome 1 including a bacteriophage P4 integrase (BCAL1118), a hypothetical yfjI (BCAL1122), and an IS402-like element (BCAL1125). Two clones were identified in a fusaric acid resistance cluster containing three genes: BCAL1176, BCAL1177, and BCAL1178. An outer membrane receptor (BCAL1783) gene and a biopolymer transport gene (BCAL1787) were located in a putative iron transport cluster. Two different type IV secretion systems, one on the chromosome 2 and one on the resident plasmid (19) were identified in multiple clones. A lipopolysaccharide (LPS) synthesis cluster as well as a putative capsular polysaccharide cluster previously described (50) were identified by the isolation of three clones containing portions of genes in each of these clusters.

Distribution of selected B. cenocepacia K56-2 clones identified by SSH across the B. cepacia complex.

Twenty of the 89 different clones identified in B. cenocepacia K56-2 by the four SSH libraries were selected to determine their distribution within nine species of the Bcc (11, 38). Criteria chosen for the selection of these genes were based on their clustering in the genome mapping and their possible roles in virulence. One of the clones contained the cblA pili gene which has previously been shown to be present primarily in the B. cenocepacia ET12 lineage (38); therefore, we used it as a positive control for the dot blot hybridization approach. PCR was used to validate the dot blot hybridization data for yfjI, the Vgr-related protein gene, and the phospholipase-like gene.

Most of the genes analyzed were unique to the ET12 clone of B. cenocepacia (Table 5). Genes that were present only in the ET12 clone included yfjI (BCAL1122), the fusaric acid genes (BCAL1176 and BCAL1177), and the autotransporter adhesin genes (BCAM0224 and BCAM0225). The genes identified within the putative capsule polysaccharide cluster (BCAL3229, BCAL3234, BCAL3235, BCAL3241, and BCAL3242) were present in all ET12 strains and possibly one strain of B. cepacia, which hybridized weakly compared to the ET12 strains. Genes present on the resident plasmid were present in the ET12 clone and one strain of Burkholderia pyrrocinia only. A type IV secretion system gene, virB4, was shown to be present in 9 out of 11 strains of B. cenocepacia and absent from the other species of the Bcc. A Vgr-related protein gene and a phospholipase-like gene were detected in six of the B. cenocepacia strains including the ET12 strains. A putative membrane protein (BCAL1535) associated with or within a fimbriae operon was identified, and the adjacent cpaC (BCAL1528) gene that has homology to a pilus assembly protein was determined to be present only in B. cenocepacia. Interestingly, genes within a LPS cluster did not have the same distribution pattern; wbiF was widely distributed in the Bcc compared to the ABC-transporter (wzm) and rmlD which were present in the B. cenocepacia ET12 clone, and one strain of Burkholderia vietnamiensis and Burkholderia dolosa, respectively. Two of the multicopy IS elements were widely distributed, whereas the putative transposition helper protein, the transposase, and inactivated derivatives were present in the ET12 strains and in few other species (Table 5).

TABLE 5.

Distribution of selected B. cenocepacia K56-2 genes in the B. cepacia complex

Genea	No. of strains positive for gene by either dot blot hybridization or PCR analysis
	B. cepacia (4)b	B. multivorans (8)	B. cenocepacia		B. stabilis (4)	B. vietnamiensis (4)	B. dolosa (3)	B. ambifaria (3)	B. anthina (4)	B. pyrrocinia (2)
			ET12 (4)	Non-ET12 (7)
Inserts with a single copy in the genome
SH2-85 2D6	0	0	4	0	0	0	1	0	0	0
Outer membrane usher (BCAM2760)
cblA (BCAM2761)
SH4-85 1C4	0	0	4	0	0	0	0	0	0	0
Hypothetical yfjI (BCAL1122)**
SH4-85 1F2	0	0	4	0	0	0	0	0	0	0
Fusaric acid resistance (BCAL1176, BCAL1177)
SH4-73 1A1	1c	0	4	0	0	0	0	0	0	0
fkbH (BCAL3229)
SH4-73 1C11	1c	0	4	0	0	0	0	0	0	0
Glycosyltransferase (BCAL3234)
UDP-galactopyranose mutase (BCAL3235)
SH4-73 1D6	1c	0	4	0	0	0	0	0	0	0
ctrC (BCAL3241)
wcbD (BCAL3242)
SH2-85 2E7	1	5	4	2	0	3	1	1	0	2
wbiF (BCAL3122)
SH2-73 1B2	0	0	4	0	0	1	1	0	0	0
wzm (BCAL3131)
rmlD (BCAL3132)
SH2-85 2F11	0	0	4	0	0	0	0	0	0	0
Autotransporter adhesins (BCAM0223, BCAM0225)
SH2-85 1G6	0	3	4	0	1	3	0	0	1	1
Outer membrane receptor (BCAL1783)
SH2-73 2A10	0	0	4	2	0	0	0	0	0	0
Vgr-related protein (BCAM0148)**
Phospholipase (BCAM0149)**
SH2-85 1A11	0	0	4	0	0	0	0	0	0	1
traW (pBCA032)
traF (pBCA033)
SH2-85 1C6	0	0	4	0	0	0	0	0	0	1
traB (pBCA044)
SH2-73 1D10	0	0	4	5	0	0	0	0	0	0
virB4 (BCAM0327)**
cpaC*	0	0	4	5	0	0	0	0	0	0
Putative lipoprotein (BCAM1598)*	0	0	4	7	0	0	0	0	0	0
Inserts with multiple copies in the genome
SH2-85 1F2	0	4	4	0	3	3	1	0	1	1
IS407A OrfA and OrfB
SH4-73 1A9	0	1	4	2	0	0	0	0	0	0
Transposase, K. pneumoniae
SH2-73 1D12	0	0	4	0	0	1	0	0	0	1
Transposition helper protein
SH4-85 1C6	0	0	4	0	0	0	0	1	0	1
COG4584: transposase and inactivated derivatives

^aClones in bold were probes in dot blot hybridization. Distribution determined by dot blot hybridization unless noted otherwise. *, distribution determined by PCR using internal primers to the gene; **, distribution determined by both dot blot hybridization and PCR. Bcc strains used in this study were from the B. cepacia panel (11, 38).

^bNumber of strains tested.

^cWeak hybridization with B. cepacia Cep509.

Construction and characterization of mutations in genes unique to B. cenocepacia.

Deletion mutants were constructed in selected genes including the hypothetical yfjI (BCAL1122), the Vgr-related protein gene (BCAM0148), the phospholipase-like gene (BCAM0149), an autotransporter adhesin (BCAM0223), cpaC (BCAL1528) of the fimbriae cluster, and the virB1-virB2 genes of the chromosomal type IV secretion system (Table 1). These genes were selected based on their uniqueness to B. cenocepacia and/or to the ET12 clone (Table 5) and their possible roles in virulence.

Virulence studies were conducted using the rat agar bead model to determine if these genes contributed to lung damage and/or to the persistence of chronic infections. Rats infected with SB002 (yfjI) had a significant decrease in lung pathology compared to rats infected with K56-2 (P < 0.001) despite no significant difference in the number of bacteria being recovered from the lungs, suggesting an important role for the hypothetical YfjI protein in inflammation in this model (Fig. 1). Rats infected with SB004 (phospholipase-like gene), SB006 (autotransporter adhesin), and SRC2 (chromosomal type IV secretion system) did not have significant differences in either lung pathology or bacterial persistence (Fig. 1). One of the animals infected with wild-type B. cenocepacia K56-2 had cleared the bacteria from its lungs. Although the number of bacteria recovered from lungs infected with K56-2 were less than the mutant strains, the differences were not statistically significant (Fig. 1A). None of these mutants, including SB003 (Vgr-related protein) and SB007 (cpaC) (Table 1), were attenuated in virulence in the alfalfa infection model (data not shown).

FIG. 1.

Virulence studies in a chronic lung infection model. A. Means ± standard deviations of CFU recovered 14 days postinfection from the lungs of four to five rats per bacterial strain. B. Means ± standard deviations of the percent of the lung with inflammatory exudates. Four or five rats were analyzed per group at 14 day postinfection. SB002 is significantly different than K56-2 (P < 0.001) by ANOVA.

Using SSH, we identified genes that were previously shown to be required for survival in the rat agar bead model using a STM approach including genes involved in LPS synthesis and synthesis of a putative polysaccharide capsule (31). B. cenocepacia K56-2 mutant strains with transposon insertions in LPS or capsule genes (Table 1) were used together with our deletion mutants to identify phenotypic properties which might be important for survival and/or the increased virulence of B. cenocepacia compared to both B. multivorans C5393 and B. stabilis LMG 14294.

B. cenocepacia has previously been shown to form biofilms on abiotic surfaces (13, 30). Five transposon mutants with insertions in genes of the LPS and the putative capsule polysaccharide clusters, 32D2 (wbiG) and 34D8 (rmlD) from the LPS cluster and 10F1 (hypothetical protein), 17D8 (fkbH), and 36B4 (cpxA) from the capsule cluster (31) (Table 1), and six mutants with deletions in genes unique to B. cenocepacia and/or the ET12 clone (SB002, SB003, SB004, SB006, SB007, and SRC2) (Table 1) were tested for their ability to form biofilms. Strains SB003 (Vgr-related protein), SB006 (autotransporter adhesin), as well as the LPS mutants 32D2 and 34D8 produced significantly more biofilm compared to the wild-type strain K56-2 (P < 0.001) as shown in Fig. 2. Strain 36B4, with a mutation in the capsule cluster, produced slightly more biofilm compared to its parent strain (P < 0.05). None of the mutants tested were defective in biofilm formation in the conditions used.

FIG. 2.

Biofilm formation. Strains were grown in LB broth for 24 h at 37°C on a rocking platform in a 96-well polystyrene microtiter plate with a pegged lid. Biofilms formed on polystyrene pegs were stained with 1% crystal violet. Values shown are the means ± standard deviations for four pegs. This experiment was repeated at least three times with similar results. 36B4 is significantly different than K56-2 (*, P < 0.05) by ANOVA. SB002, SB006, 32D2, and 34D8 are significantly different than K56-2 (**, P < 0.001) by ANOVA.

B. cenocepacia was previously shown to differentiate into swarmer cells and migrate over agar (30, 34). Swarming and swimming phenotypes of selected mutants are displayed in Table 6. B. cenocepacia K56-2 was able to swarm over an agar surface, and differentiation from vegetative cells into typical elongated swarmer cells (23) was observed by microscopy (data not shown). Mutants SB003, SB006, SB007, 10F1, and 34D8 exhibited major swarming defects as demonstrated by much smaller zones of swarming on agar surface compared to the wild-type strain B. cenocepacia K56-2 (Table 6). With the exception of 36B4, which was slightly reduced in swarming motility, the other mutants were similar to the wild type (Table 6).

TABLE 6.

Comparison of swarming and swimming motility between mutants and wild-type B. cenocepacia K56-2

Strain	Swarminga	Swimming
		Soft agara	Liquidb
K56-2	++++	++++	+
SB002 (ΔyfjI)	++++	++++	+
SB003 (ΔVgr-related protein)	+	++	−
SB004 (Δphospholipase-like gene)	++++	++++	+
SB006 (Δautotransporter adhesin)	++	++++	+
SB007 (ΔcpaC)	++	++	+
SRC2 (ΔvirB1 and virB2)	++++	++++	+
10F1 (hypothetical protein)	++	++	+
17D8 (fkbH)	++++	++	+
32D2 (wbiG)	++++	++++	+
34D8 (rmlD)	++	++++	+
36B4 (cpxA)	+++	+++	+

^a+, ≤25% of the K56-2 zone of swarming or swimming; ++, 26-50% of the K56-2 zone of swarming or swimming; +++, 51-75% the K56-2 zone of swarming or swimming; ++++, 76-100% the K56-2 zone of swarming or swimming.

^bOvernight cultures in NB + 0.5% glucose were examined for motility by microscopy. +, motile; −, nonmotile.

The swimming motility phenotype of these mutants was also determined. SB003, SB007, 10F1, and 17D8 were defective in swimming compared to the parent strain using soft agar plates (Table 6). Swimming motility from overnight cultures was analyzed by microscopy, and only SB003 was nonmotile (Table 6), which could explain its defect in both swarming and swimming motility using swarm and swim agar plates.

DISCUSSION

SSH has been used successfully in the identification of virulence determinants and genetic diversity of several bacterial pathogens (reviewed by Winstanley) (71). A capsule polysaccharide cluster was identified in both Burkholderia pseudomallei (52) and Burkholderia mallei (17) by SSH with the nonpathogenic Burkholderia thailandensis and was demonstrated to be an important virulence determinant of both pathogens in animal infection models (17, 52). Genomic diversity was also observed between clinical isolates of B. pseudomallei with the identification of a B. mallei-specific prophage in B. pseudomallei 1026b by comparison to the sequenced B. pseudomallei K96243 (16). SSH has been used previously to analyze differences between strains of B. cenocepacia (36, 50). IS1363 identified by SSH between a B. cenocepacia PHDC strain and a non-PHDC strain of B. cenocepacia is present almost exclusively in two B. cenocepacia epidemic clones, ET12 and PHDC, which were implicated in outbreaks within the CF population (36). B. cenocepacia is divided into four groups based on the recA sequence, and a putative capsule polysaccharide cluster, a Vgr-related protein gene, and a phospholipase-like gene were previously identified by SSH between a highly transmissible B. cenocepacia recA type A and a nontransmissible recA type B (50).

Previous studies using SSH in different Burkholderia species used a hybridization temperature of 73°C (16, 17, 50, 52). In the present study, these conditions resulted in isolation of fragments with a G+C content averaging 54%, which is similar to previous SSH studies involving Burkholderia species (16, 17, 50, 52). Increasing the hybridization temperature to 85°C when using B. multivorans C5393 as driver DNA made it possible to identify genes with an average G+C content corresponding to that of the genome; although this temperature did not result in a SSH library with an increased G+C content with B. stabilis LMG 14294 as the driver.

Although we may have been conservative in our screening of the libraries by dot blot hybridization with labeled genomic DNA, the 20 clones chosen for distribution studies within the Bcc were always absent from either or both driver strains (Table 5), demonstrating the efficiency of our approach. Although not confirmed by hybridization or PCR, the remaining genes identified in Tables 2 and 3 are highly likely to be absent from either C5393 or LMG 14294. Several independent clones contained genes localized within the same cluster or operon, which demonstrates the effectiveness of the SSH screening approach used. Not all differences between the compared species were identified, however, since some genes known to be unique to B. cenocepacia ET12 including the genes of the cci genomic island (2) and IS1363 (36) were not identified. Because of the extent of variation between these species, it would be very unlikely that this method could ever approach identification of all differences.

Many ET12-specific genes or clusters identified in this study had previously been shown to be implicated in B. cenocepacia virulence, including genes that were required for survival in the rat agar bead model using a STM approach (31). Our study showed that genes within the capsule polysaccharide cluster and the wzm-rmlD genes of the LPS cluster were not widely distributed within the Bcc (Table 5).

Virulence studies in the rat agar bead model were performed only on a selected number of our B. cenocepacia K56-2 deletion mutants (Fig. 1). Natural colonization steps are bypassed in this infection model, therefore mutation of genes involved in adherence is unlikely to alter virulence. cpaC is predicted to be part of an operon encoding for a fimbriae structure. Therefore, strain SB007 with a mutation in cpaC was not tested in the agar bead model. Strain SB003 (Vgr-related protein) was also not tested in this model since the phospholipase-like gene mutation (SB004) is located 3 bp downstream of the Vgr-related protein gene and probably in the same operon.

Although there is no clinical evidence that Bcc forms biofilms in CF lungs, there is evidence that P. aeruginosa forms biofilms in chronically infected CF lungs (49, 59). Flagella-driven motility like swarming was demonstrated to facilitate colonization of the urinary tract by Proteus mirabilis (1) and to upregulate virulence gene products such as hemolysin, urease, and protease (23). Phenotypes such as biofilm formation and swarming and swimming motility may correlate with virulence, and therefore we determined that mutations in selected genes identified by SSH influenced these phenotypes.

Genes wbiF and rmlD, which are part of an O-antigen synthesis cluster, were identified in both studies. Mutations in rmlD and wbiG, which is downstream of wbiF, affect the lipid A core and sensitivity to human serum (47). Interestingly, we have shown that these genes also affect biofilm production, suggesting that the LPS phenotype influences biofilm formation. The swarming motility phenotype was also reduced for the rmlD mutant (34D8) (Table 6), which is consistent with some observations made in Salmonella enterica serovar Typhimurium, where O-antigen mutants were rescued by surfactin for swarming, indicating that LPS O antigen improves surface fluidity for colonies to swarm (64). Surprisingly, the distribution of genes within this O-antigen cluster in the Bcc varies depending on their location in the cluster. wbiF is present in most of the Bcc species, whereas wzm-rmlD are present in the ET12 clone of B. cenocepacia and one strain of both B. vietnamiensis and B. dolosa each (Table 5). Ortega et al. (47) showed that this O antigen cluster has several transcriptional units and wbiF and wzm-rmlD are located in two different transcriptional units. Further, the G+C content seems to change at the limits of these two transcriptional units, suggesting that the genes rmlBACD-wzm-wzt-vioA-wbxFED (47) may have been acquired at different times in evolution.

We identified three clones within a putative capsule polysaccharide cluster as well as IS elements, ISRS011 that flanks one end of the cluster and IS407 elements that are present within the cluster. One gene of this putative capsule polysaccharide cluster, cpxA (wzt), had previously been identified by SSH (50). Using wcbB and cpxA as probes, Parsons et al. (50) reported that the cluster was present in B. cenocepacia ET12 strains, B. cepacia Cep509 and ATCC25416, and B. multivorans CF-A1-1. PCR analysis could only confirm the presence of both wcbB and cpxA in B. cenocepacia ET12. Amplification of only one of these two genes could be performed in B. cepacia strains while none were amplified in B. multivorans, suggesting that there were either sequence differences or possibly other genes that cross-hybridized (50). In our study, although there was weak hybridization with B. cepacia Cep509 for all three probes used, we have shown that this cluster is primarily found in the ET12 clone by dot blot hybridization using probes for fkbH (BCAL3229), glycosyltransferase (BCAL3234), UDP-galactopyranose mutase (BCAL3235), ctrC (BCAL3241), and wcbD (BCAL3242) (Table 5).

As previously reported by Parsons et al. (50), these genes have homology to the capsule polysaccharide genes of B. mallei (17) and B. pseudomallei (52). Although there are similarities between the clusters of the three species, there are several differences including insertions and deletions of several genes. Although the interruption of wcbO by IS407-like elements likely inactivates the wcbOPQRS operon, this inactivation was only observed in B. cenocepacia J2315 (50). Hunt et al. (31) identified three STM mutants with transposons inserted in this polysaccharide cluster (mutants 10F1, 17D8, and 36B4), indicating that these genes are important for survival in the lungs of a chronic lung infection model and suggesting a possible role in host resistance.

The hypothetical yfjI gene identified by SSH was also previously found to be important for survival (31). The presence of a bacteriophage P4 gene adjacent to yfjI and genes in a low G+C content region suggests that this region was acquired by horizontal transfer since phages have been implicated in the movement of virulence factors between bacterial species (67). Although yfjI homologues are present in other bacterial species, the role of this gene is unknown. It is predicted to encode a cytoplasmic protein (PSORTb of 8.96). Its contribution to lung pathology (Fig. 1A) and its uniqueness to B. cenocepacia ET12 warrant further studies on its role in virulence of B. cenocepacia ET12 and possibly other species with yfjI homologues.

Both our study and that of Parsons et al. (50) identified a cluster containing a Vgr-related protein and a phospholipase-like gene. Vgr-related genes are present in several gram-negative bacteria and contain repeated dipeptide motifs (valine-glycine repeats) that are often associated with Rhs elements (rearrangement hot-spots) in E. coli (69, 70). P. aeruginosa PAO1 contains 10 Vgr homologues (70), and six of them were found to be associated with genomic islands (20). Although their functions are yet to be defined, Vgr-related proteins are often associated with ligand-binding proteins at the bacterial surface or are secreted (69). In this study, we showed that the Vgr-related protein gene was involved in biofilm formation and swarming and swimming motility (Fig. 2 and Table 6), suggesting that a mutation in a Vgr-related protein gene affects phenotypes that are surface related. In P. aeruginosa, competition studies between a phospholipase D (PLD) mutant and its parent strain in the rat agar bead model demonstrated that PLD contributed to the ability of P. aeruginosa to persist in the lungs (70). In our study, the rats infected with the PLD mutant had slightly less bacteria recovered from the lungs although the difference was not significant. It is possible that a greater difference might have been observed in a competition assay with the wild-type strain. PLDs are widely distributed and found in both eukaryotic and prokaryotic cells (21). They have been associated in bacterial pathogenesis with the murine toxin of Yersinia pestis (54) as well as in the virulence of Corynebacterium ovis (44). PLD is a member of a superfamily that includes prokaryotic and eukaryotic PLDs, cardiolipin synthase, phosphatidylserine synthase, poxvirus envelope proteins, bacterial endonuclease, and helicase. All members of the superfamily have one or two copies of the HKD motif, which is a conserved active site [HxK(x)₄D or HxK(x)₄D(x)₆GSxN] (21, 43). The phospholipase-like gene in B. cenocepacia contains these two distinct HKD motifs.

Engledow et al. (19) demonstrated the presence of two different TFSSs in B. cenocepacia K56-2, one located on chromosome 2 and one on the resident plasmid. In this study, we also identified these two TFSSs (Table 2) and showed that the chromosomal one was unique to B. cenocepacia (Table 5). Some pathogenic bacteria use TFSS to translocate virulence factors, which are DNA or protein macromolecules, to a large array of target cells (26). The plasmid-encoded TFSS of B. cenocepacia K56-2, which is primarily found in B. cenocepacia ET12 (Table 5), is directly involved in plant pathology as demonstrated by plant tissue watersoaking symptoms. The chromosomal TFSS does not appear to be involved in plant virulence as demonstrated by our studies with the alfalfa model (data not shown) and those of Engledow et al. (19).

More than half of the genes identified were IS elements (Table 4). Most of the IS elements were present in multiple copies in the B. cenocepacia J2315 genome, and seven of these multiple copy IS elements were found in 153 locations. IS elements in B. cenocepacia ET12 suggest the presence of several genomic islands other than the described cci (2), since these islands are often characterized by regions of low G+C containing IS elements.

Although this study has not identified all the differences that may account for the increased prevalence and virulence of the B. cenocepacia ET12 clone versus the other species of the Bcc, we have demonstrated that genes unique to B. cenocepacia and/or ET12 may play a role in the increased virulence of the ET12 clone. We have also determined that the previously uncharacterized yfjI is important for virulence, and further studies are in progress to determine the function of this gene.