Comparative Transcriptomic Analysis of the Burkholderia cepacia Tyrosine Kinase bceF Mutant Reveals a Role in Tolerance to Stress, Biofilm Formation, and Virulence

The strains and plasmids used in this work are listed in Table 1. E. coli was grown in LB medium at 37°C. Unless otherwise stated, B. cepacia strains were grown in LB, pseudomonas isolation agar (PIA), or in EPS-producing S medium at 30°C (22). Growth media were supplemented with antibiotics when required to maintain the selective pressure at the following concentrations (in μg/ml): for B. cepacia, trimethoprim at 100, chloramphenicol at 300, and tetracycline at 400; for E. coli, trimethoprim at 100, chloramphenicol at 25, ampicillin at 100, tetracycline at 10, and kanamycin at 50.

Total DNA was extracted from bacterial cells by using the DNeasy blood and tissue kit (Qiagen), following the manufacturer's instructions. Plasmid DNA isolation and purification, DNA restriction/modification, agarose gel electrophoresis, and E. coli transformation were carried out using standard procedures (30). B. cepacia electrocompetent cells, prepared as previously described (28), were transformed by electroporation using a Bio-Rad Gene Pulser II (200 Ω, 25 μF, 2.5 kV) and grown overnight before being plated in selective medium. Plasmids for complementation experiments and pIN25 expressing the green fluorescence protein (GFP) were mobilized into Burkholderia cepacia strains by triparental conjugation using plasmid pRK2013 as the helper.

To obtain a B. cepacia IST408 bceE disruption, a 2,411-bp KpnI/EcoRI fragment containing the bceE coding region and flanking regions was amplified using primers BceE-up (5′-CGAGGTACCCGTGATCGTCA; restriction sites are shown in italics) and BceE-low (5′-GATGAATTCGACGTCGGCCACTT) and cloned into the same restriction sites of pBCSK. The plasmid obtained (pLM56-1) was further digested with XhoI, which has a single recognition site within the bceE coding region. The trimethoprim (Tp) resistance cassette was obtained from pUC-TP and was cloned into pLM56-1, resulting in the plasmid pLM56-4, which transcribed the cassette in the same orientation as the bceE gene. This plasmid was introduced into B. cepacia IST408 by electroporation, and transformants were selected by growth on PIA medium with trimethoprim. The colonies obtained were then screened in the presence of chloramphenicol. Colonies that did not grow in the presence of chloramphenicol but were trimethoprim resistant were considered candidates for allelic exchange of bceE by the bceE::Tp construct. The candidate insertion mutant was further characterized by PCR amplification and DNA sequencing.

PCR was used to amplify the coding region of bceE (primers bceE-fw, 5′-GAACATATGCTGAAACGCCCGATG, and bceE-rev, 5′-TGATCTAGAGGAGCAGCTGGCCGAGGA; restriction sites are in italics); bceF (primers bceF-fw, 5′-GAACATATGGTGAACACGCAAGCGAAA, and bceF-rev, 5′-TTATCTAGAATGCGGATCAGGCGCTCA), and rpoH (BCAL0787 in B. cenocepacia J2315; primers rpoH-fw, 5′-TTTCATATGAGCAACGCCCTGACCCTC, and rpoH-rev, 5′-CCGTCTAGAAACCGGTGGAAAAAATTG). Following restriction with NdeI and XbaI, each gene was cloned in the same restriction sites of pDA17. The plasmids obtained were pLM127-4, pLM127-9, and pLM127-11, carrying the rpoH, bceF, and bceE genes, respectively (Table 1). The nucleotide sequences of the cloned genes were confirmed by DNA sequencing.

For expression profiling, overnight cultures of wild-type B. cepacia IST408 and the isogenic mutants bceE::Tp and bceF::Tp were grown in LB medium and diluted into S medium to an initial optical density at 640 nm (OD₆₄₀) of 0.1. Triplicate 250-ml Erlenmeyer flasks containing 100 ml of medium were inoculated with each strain and cultured at 30°C with agitation at 250 rpm for 12 h. These bacterial cells were resuspended in RNAprotect bacteria reagent (Qiagen), and total RNA extraction was carried out using the RNeasy minikit (Qiagen), followed by DNase treatment (RNase-free DNase set; Qiagen) according to the manufacturer's recommendations. RNA integrity was checked on an Agilent 2100 Bioanalyser using an RNA Nano assay. RNA was processed for use on Affymetrix custom dual-species Burkholderia arrays (31), according to the manufacturer's prokaryotic target preparation assay instructions. Briefly, 10 μg of total RNA containing spiked poly(A) RNA controls [GeneChip poly(A) RNA control kit; Affymetrix, Santa Clara, CA] was used in a reverse transcription reaction with random primers (Invitrogen Life Technologies) to generate first-strand cDNA. After removal of RNA, 2 μg of cDNA was fragmented with DNase and end labeled with biotin by using terminal polynucleotidyl transferase (GeneChip WT terminal labeling kit; Affymetrix). The size distribution of the fragmented and end-labeled cDNA was assessed using the Agilent 2100 Bioanalyzer. Two micrograms of end-labeled fragmented cDNA was used in a 200-μl hybridization cocktail containing hybridization controls, and the mixture was hybridized on arrays for 16 h at 50°C. Modified posthybridization wash and double-stain protocols (FLEX450_0005; GeneChip HWS kit; Affymetrix) were used on a GeneChip fluidics station 450. Arrays were scanned on an Affymetrix GeneChip scanner 3000 7G. Biological triplicates of RNA from each bacterial culture were processed and analyzed.

For DNA analysis, a total of 1.5 μg of genomic DNA from B. cepacia IST408 was labeled using the Bioprime DNA labeling system (Invitrogen, Paisley, United Kingdom) following a strategy for genomic DNA hybridization to GeneChips developed by Hammond and coauthors (32). Cleanup was performed using the MinElute PCR purification kit (Qiagen, Hilden, Germany), and quality was checked on an Agilent 2100 Bioanalyser using a DNA 1000 assay. Five micrograms of labeled DNA was analyzed on an Affymetrix custom dual-species Burkholderia array following the protocol described above for RNA samples.

For microarray analysis, DNA-based probe selection was performed using Xspecies DNA hyb CDF batchmaker v3.2 (32; http://affymetrix.arabidopsis.info/xspecies/). An optimal signal cutoff of 80 and minimum number of probes per probe set of 7 were determined empirically, resulting in the masking of 24,223 of the original 226,576 probes on the array and the exclusion of 156 probe sets. The new cdf file created (Bcc1sa520656F_Xspecies.CEL80.cdf) was used in a DNA Chip Analyzer 2008 for the subsequent gene expression analyses. The 9 expression arrays were normalized to a baseline array with median CEL intensity by applying an invariant set normalization method (33). Normalized CEL intensities of the arrays were used to obtain model-based gene expression indices based on a perfect match (PM)-only model (34). Replicate data (triplicates) for each bacterial isolate were weighted gene-wise by using the inverse squared standard errors as weights. All genes compared were considered differentially expressed if the 90% lower confidence bound of the fold change (LCB) between the experimental value and baseline was above 1.2. The lower confidence bound criterion meant that we could be 90% confident that the fold change was a value between the lower confidence bound and a variable upper confidence bound. Li and Wong have shown that the lower confidence bound is a conservative estimate of the fold change and therefore more reliable as a ranking statistic for changes in gene expression (33).

This comparison resulted in 630 differentially expressed transcripts with a median false-discovery rate (FDR) of 1.7% when the bceF::Tp mutant was compared with the wild-type IST408. The comparison of the bceE::Tp mutant transcriptome with the one of IST408 showed 5 genes differentially expressed (≥1.6-fold change lower confidence bound, with a resulting FDR of 0%).

DNA microarray data were validated by real-time quantitative reverse transcription-PCR (qRT-PCR) as previously described (24). Total RNA was used in a reverse transcription reaction with TaqMan reverse transcription reagents (Applied Biosystems). qRT-PCR amplification of each gene (for primer sequences, see Table S1 in the supplemental material) was performed with a model 7500 thermocycler (Applied Biosystems). The expression ratio of the target genes was determined relative to the reference gene lepA, which showed no variation in transcript abundance under the conditions tested. Relative quantification of gene expression by real-time qRT-PCR was determined using the ΔΔC_T method (35).

Plasmids pDA17, pLM127-9 (+pbceF), and pLM127-11 (+pbceE) were mobilized into the parental strains IST408, IST408 bceF::Tp, and IST408 bceE::Tp, respectively, by triparental conjugation. To determine their ability to restore cepacian biosynthesis in vivo these strains were grown in liquid S medium over 3 days at 30°C. EPS quantification was based on the dry weight of EPS recovered after ethanol precipitation of culture supernatants, as described before (28).

Survival of Burkholderia strains to heat stress was tested by inoculation of overnight cultures into 4 ml of S medium to a final OD₆₄₀ of 0.1. Strains were allowed to grow until the OD₆₄₀ reached 0.6 before being submitted to a static incubation at 50°C in a water bath for 20 min. At 4- or 8-min intervals, an aliquot of 100 μl was serially diluted and plated onto LB agar. After 48 h of incubation at 30°C, CFU were counted. To test tolerance to UV irradiation, 500-μl aliquots of the bacterial cell suspensions prepared in S medium (OD₆₄₀, 0.1) were transferred into PMMA cuvettes (VWR) and immediately exposed to UV irradiation (λ, 254 nm) for different times. Aliquots from each cell suspension were serially diluted and plated onto LB agar, and plates were incubated for 2 days at 30°C before CFU counting. Each experiment was repeated at least three times.

The method for detection of cyclic-di-GMP from B. cepacia IST404 and its derivative mutants bceF::Tp and bceE::Tp was adapted from that reported by Ryan and colleagues (36). Briefly, cells (300 μg) were harvested from agar plates or shaking cultures grown at 30°C in S medium for 12, 24, and 48 h. Nucleotides were extracted by resuspending the cells in water and heating at 100°C for 5 min. The extract was lyophilized, resuspended in 500 μl of water certified for high-performance liquid chromatography (HPLC) analysis, and filtered (0.2-μm pore size). Extracts equivalent to 150 μg (wt/wt) of cells were adjusted to 500 μl in 0.1 M triethylammonium acetate (TEAA) buffer (pH 4.5) and subjected to HPLC separation. HPLC was performed on a 250- by 4.6-mm reverse-phase column (Hypersil ODS 5 μm; Hypersil-Keystone) at room temperature, with detection at 260 and 280 nm, on an ICS-U3000 IC system (Dionex, United Kingdom). Running conditions were optimized by using synthetic cyclic-di-GMP (Biolog). Runs were carried out in 0.15 M TEAA buffer (pH 4.5) at 1 ml/min, using a multistep gradient of acetonitrile. Relevant fractions of 1 ml were collected, lyophilized, and resuspended in 10 μl H₂O. For quantification, a standard curve was established whereby synthetic cyclic-di-GMP was added to relevant cell extracts. The area of the peak was used to estimate the amount of cyclic-di-GMP in a sample in reference to the wet cell weight. Quantification was further validated by mass spectrometry using matrix-assisted laser desorption ionization–time of flight analysis.

For the motility test, swimming plates with 0.3% (wt/vol) Bacto agar (Difco) and swarming plates with 0.5% (wt/vol) Noble agar were prepared using S medium and LB with 0.5% (wt/vol) glucose, respectively. For estimation of motility, overnight bacterial cultures (5 μl) were inoculated onto an agar surface and incubated at 30°C for 48 h, followed by colony diameter determinations.

For the pellicle assay, overnight cultures were inoculated in tubes containing 5 ml of S medium at an OD₆₄₀ of 1.0 and kept static for 3 days at 30°C. Pellicles were assayed by visual inspection of the air-liquid interface, and the complete coverage of the surface with a cell layer was considered the extent of pellicle formation.

Biofilm formation assays in microtiter plates were performed as previously described (28, 37). Biofilm formation was also studied under continuous flow conditions using the sterile Stovall 3-chamber flow cell system (Fisher Scientific). Briefly, S medium was pumped into the system for 3 h prior to inoculation of bacteria to wash and prevent system bubbles, using an Ismatec peristaltic pump. The flow chambers were inoculated by injecting in each flow channel 300-μl aliquots of the overnight cultures of B. cepacia IST408, bceF::Tp mutant, or bceE::Tp mutant harboring pIN25 expressing the gfp gene to a dilution resulting in an OD₆₄₀ of 0.2. After inoculation, the flow channels were left without flow for 3 h, after which S medium flow was restarted at a rate of 0.4 ml/min. The flow cell system was kept at 30°C during the experiments, except during confocal microscopy analysis.

Biofilm formation was examined at 48 h by scanning confocal laser microscopy on a Leica TCS SP5 inverted confocal microscope (DMI6000; Leica Microsystems CMS GmbH, Mannheim, Germany) with an argon laser for excitation. A 63× apochromatic objective with a numerical aperture of 1.2 (Zeiss, Jena, Germany) was used for all experiments. Strains under study expressed GFP, which allowed fluorescent visualization of bacterial cells. Data acquired were analyzed using the ImageJ program (http://rsbweb.nih.gov/ij/). Biomass thickness and the roughness coefficient were calculated using COMSTAT2 (http://www.comstat.dk) (38; M. Vorregaard, B. K. Ersbøll, L. Yang, J. A. J. Haagensen, S. Molin, and C. Sternberg, personal communication). The results shown were obtained using the three-dimensional (3D) viewer function and are representative of 3 independent experiments.

B. cepacia strains were screened for virulence in a G. mellonella wax moth larvae infection model as previously described (39). Larvae were injected with approximately 2 × 10⁶ bacterial cells suspended in 10 mM MgSO₄ with 1.2 mg/ml ampicillin and incubated in the dark at 30°C. Control larvae were injected with 10 mM MgSO₄ with 1.2 mg/ml ampicillin. Larvae were scored as dead or alive for a period of 2 days. Duplicates of 10 larvae were used in each experiment, and three independent assays with different larvae generations were performed.

Microarray data were deposited in the Gene Expression Omnibus (GEO) repository at NCBI under accession number GSE38183 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE38183).

Since the bceF::Tp mutant is unable to produce exopolysaccharide while the wild-type strain B. cepacia IST408 is (28), it was necessary to exclude that the possible differences for the strains were caused by the EPS directly and not by the presence/absence of the BceF kinase. Therefore, we constructed an insertion mutant for the bceE gene (located upstream of bceF) that encoded the predicted auxiliary membrane protein BceE, which is required for cepacian export. This insertion mutant, B. cepacia IST408 bceE::Tp, showed the same growth behavior as the wild-type B. cepacia IST408 strain and the bceF::Tp mutant (data not shown) and, as expected, was unable to produce cepacian (Fig. 1). In trans complementation of bceE::Tp and bceF::Tp mutants with the constitutive pDA17 vector harboring bceE (+pbceE) or bceF (+pbceF), respectively, restored EPS production to the wild-type level (Fig. 1).

To gain insight into the broader role of BceF tyrosine kinase, a transcriptomic approach using a custom Burkholderia GeneChip array was followed. Since we used B. cepacia IST408 as a model organism and the species represented on the array are B. cenocepacia J2315 and B. multivorans ATCC 17616 (31), there was the need to determine which individual probes representing a transcript hybridized efficiently to B. cepacia IST408 genomic DNA. Therefore, the genomic DNA of this bacterium was hybridized to the microarray. Setting a threshold for minimum intensity of 80 and for at least 7 probes per probe set within the software Xspecies (32), we retained 14,752 probe sets. Under these conditions, the comparison of the gene expression profile of the bceE::Tp mutant with that of IST408 showed 1 gene with statistically significant increased expression and 4 genes with statistically significant decreased expression (≥1.6-fold change lower confidence bound with a resulting false-discovery rate of 0%). The gene with increased expression encodes a putative metallo-beta-lactamase, and the genes with decreased expression are predicted to be involved in acetoin metabolism (see Table S2 in the supplemental material). These results confirmed that after 12 h of growth in S medium, the transcriptome of B. cepacia IST408 and of the bceE::Tp mutant derivative are very similar, and any differences between the bceF::Tp and IST408 transcriptomes are due to the presence/absence of the BceF kinase. The comparison of the bceF::Tp mutant transcriptome with that of the wild-type B. cepacia IST408 strain (≥1.2-fold change lower confidence bound with a resulting false-discovery rate of ≤1.7%) showed 304 genes with statistically significant increased expression and 326 genes with statistically significant decreased expression. Almost half (43%) of the identified genes belonged to the category of poorly characterized genes or had no COG classification. Figure 2 shows the major categories of the remaining differentially expressed genes. Globally, our results indicated that in the absence of BceF tyrosine kinase, the cell's transcriptome showed a significant decrease in the expression of genes related to cell motility, stress response (universal stress proteins, heat shock proteins, proteases, etc.), and DNA replication, recombination, and repair (Table 2). In contrast the expression levels of genes encoding transcriptional regulators or lipid metabolism and genes related to cyclic-di-GMP signaling were increased in the bceF::Tp mutant transcriptome (Table 2). (The complete lists of differentially expressed genes are presented in Tables S3 and S4 in the supplemental material.)

To confirm the data obtained by microarray analysis, expression levels of 11 representative genes from COGs C, E, K, N, R, T, and U were analyzed by qRT-PCR. The results obtained were in good agreement with the microarray data (Table 3).

Metabolic changes accounted for one-third of differentially expressed genes between the bceF::Tp mutant and the wild-type strain. In particular, genes involved in energy conversion, coenzyme metabolism, and amino acid metabolism had decreased expression in the bceF::Tp mutant, while genes related to lipid metabolism and nutrient uptake had increased expression. Among the genes with decreased expression in the bceF::Tp mutant, there were genes encoding glycolytic enzymes (gapA and BCAM0311) and the gene pckG, which encodes the phosphoenolpyruvate carboxykinase enzyme that removes oxaloactetate from the citrate cycle into the gluconeogenesis pathway (Table 2). The genes with decreased expression required for coenzyme metabolism are involved in the metabolism of porphyrin, thiamine, folate, ascorbate, and biotin. In the category of amino acid metabolism, we observed a 1.7- to 1.9-fold decrease in expression of the genes gcvP and gcvT, which encode two of the enzymes of the glycine cleavage system that is involved in the synthesis of one-carbon (C₁) units used in the biosynthesis of purines, methionine, and thymine and for other cellular methylation reactions.

A category of genes with increased expression in the bceF mutant encodes ABC transporters involved in monosaccharide, amino acid, peptide, and mineral and organic ion transport (Table 2). With regard to lipid metabolism, we observed a 1.5-fold increase in expression of the cls gene, which encodes cardiolipin synthetase. Cardiolipin is an anionic tetraacylphospholip found in the cytoplasmic membrane of bacteria, and it is important for optimal assembly of protein complexes, such as the electron transport complexes (40). We also observed increased expression of genes involved in fatty acid biosynthesis (BCAM2001 and BCAM2283) and decreased expression of genes for their catabolism (BCAL3191 and BCAM0142). It is thus possible that in the absence of BceF kinase, bacteria need to restructure the membrane lipid composition, perhaps to stabilize some protein complexes and/or alter permeability.

Regarding cell wall biosynthesis, the gene dadX, which encodes a protein that catalyzes the reversible racemization of l-Ala and d-Ala, showed 2.0-fold decreased expression in the bceF::Tp mutant. As d-Ala is an essential component of the cell wall peptidoglycan, we cannot exclude that the bceF::Tp mutant may have differences in its peptidoglycan structure. Another important component of the cell wall is the lipopolysaccharide. Three genes involved in O-antigen biosynthesis and one in lipid A modification showed 1.7- to 2.4-fold increased expression in the bceF::Tp mutant strain. One of the genes, wbiF, encodes a glycosyltransferase involved in the transfer of a sugar residue to the O-antigen core oligosaccharide. The other two genes were wzm and wzt, which encode a predicted two-component ABC transporter required for O-antigen export across the plasma membrane. The last gene, with a 1.8-fold increase in expression, was BCAM1214, which encodes a protein homologue of LpxO dioxygenase enzymes from other bacteria; these enzymes are required for hydroxylation of lipid A. It is thus possible that the bceF::Tp mutant strain produces a different amount of and/or a modified lipopolysaccharide.

Microarray data also showed 1.6-fold decreased expression of ompR, which encodes the response regulator OmpR, in the bceF::Tp mutant. Together with the histidine kinase EnvZ, OmpR is involved in regulating outer membrane composition in several bacteria, is required for curli expression and for biofilm formation in E. coli, and is also involved in resistance to phagocytosis and in survival within macrophages by Yersinia pestis (41, 42). Although the genes regulated by EnvZ/OmpR in Burkholderia are unknown, we observed decreased expression of at least 9 genes encoding putative lipoproteins and outer membrane proteins in the bceF::Tp mutant, while only 4 showed increased expression. In addition, the secB gene, which encodes the preprotein translocase of the Sec pathway necessary for the export of proteins to the periplasm, outer membrane, or extracellular milieu, also showed reduced expression in the bceF::Tp mutant. Taken together, these observations suggest that in addition to the absence of exopolysaccharide, the bceF::Tp mutant must have other differences in its cell wall that affect permeability or interactions with the surrounding environment.

In our data set, the gene rpoH encoding the sigma 32 alternative sigma factor showed 1.8-fold decreased expression in the bceF::Tp mutant compared with the IST408 wild-type strain. RpoH mediates the heat shock response, but it is also activated under conditions that destabilize folded proteins or make correct nascent protein folding more difficult (43). The comparison of our data with genes that are known to be regulated by RpoH in E. coli (43) revealed many with decreased expression in the bceF::Tp mutant (Table 2). These genes encode chaperones GroES1 and HSP21, the dsbC genem which encodes a protein involved in disulfide bond formation, several genes that encode universal stress-related proteins, and genes encoding putative proteases. The RpoH regulon also encodes proteins involved in cofactor biosynthesis and iron-sulfur assembly (43) and, as we described above, several genes involved in cofactor biosynthesis showed decreased expression. Another function of the RpoH regulon is the protection of DNA and RNA. Hence, we observed a decreased expression of genes involved in the repair of DNA lesions (recR and mutL) in the bceF::Tp mutant, as well as a predicted Holliday junction resolvase-like protein (BCAL3141), a specific single-stranded DNA-specific exonuclease (BCAL2188), and the cold shock protein gene cspA. Also, gene nusB, whose product binds to RNA polymerase to alleviate the effects of supercoiling on transcription, and the RNA chaperone-encoding genes hfq and hfq2 showed decreased expression in the bceF::Tp mutant. To determine whether the decreased expression of the putative RpoH and its regulon members is relevant under heat shock conditions, wild-type IST408 and the bceE::Tp and bceF::Tp mutants and the complemented mutants were incubated at 50°C for 20 min, and CFU were counted at several time points (Fig. 3A). The results obtained showed a more sensitive phenotype for the bceF::Tp mutant, with the cells having a lower survival rate under heat stress. In trans complementation of the bceF::Tp mutant with the pDA17 vector containing the bceF or rpoH genes rescued its heat-sensitive phenotype (Fig. 3A). When cells were exposed to UV light for a period of time, we also observed a decrease in the survival rate of the bceF::Tp mutant compared to the bceE::Tp and wild-type B. cepacia, which was rescued by the introduction of bceF or rpoH genes carried by pDA17 (Fig. 3B).

Although the RpoH regulon is also induced by oxidative stress, the only two genes with a putative role in antagonizing this type of stress and that showed decreased expression in the bceF::Tp mutant were BCAM0703, which encodes a putative glutathione S-transferase, and BCAL2410, which encodes a protein with a rhodanese domain (Table 2). Despite that, no differences were found between the wild-type strain and the mutants with respect to growth inhibition in the presence of oxidative stress agents, such as H₂O₂ or cumene hydroperoxide (data not shown).

Analysis of gene expression levels in the flagellar regulon indicated 1.5-fold decreased expression in the bceF::Tp mutant of the cheY, cheB3, cheZ, and tar genes, whose products are involved in chemotaxis. In addition, the filament cap-encoding gene fliD1, the flagellar basal body-encoding genes flgABCE1, and flgM, which encodes the negative regulator of flagellin synthesis also displayed decreased expression in the bceF::Tp mutant (Table 2). To test whether differences in the expression of these genes led to a phenotype in semisolid medium, the swimming and swarming abilities of the strains were performed. Consistent with the gene expression data, we observed that after 24 h of incubation the bceF::Tp mutant displayed lower swimming and swarming motilities than the wild-type strain and the bceE::Tp mutant, but these phenotypes were complemented by the expression of the bceF gene from pDA17 (Fig. 4).

We also compared cell-to-cell adhesion in wild-type bacteria and in the bceE::Tp and bceF::Tp mutants by using static broth cultures. The results showed that broth cultures of the wild-type strain remained visibly clear after 3 days of incubation, with clumping of the bacterial cells at the bottom of the culture tube (Fig. 5A). For this property, the bceE::Tp mutant showed an intermediate phenotype, and the broth culture of the bceF::Tp mutant showed mainly bacterial cells in suspension (Fig. 5A). Pellicle formation at the air-liquid interface was mainly observed in the wild-type IST408 and, to a lesser extent, in both mutants (Fig. 5A). Since pellicle formation is indicative of the production of an adhesive matrix, we first investigated biofilm formation under static conditions by measuring cell adhesion to an abiotic surface based on crystal violet staining. The bceF::Tp mutant biofilm at 48 h retained 2.0- and 1.3-fold less crystal violet than the biofilm of wild-type B. cepacia IST408 and the bceE::Tp mutant, respectively (Fig. 5B). Both mutant phenotypes were restored by expressing in trans from pDA17 the mutant strain defective gene. We also tested biofilm formation in continuous flow cells by constitutively expressing GFP from a plasmid in the three strains. After a 48-hour period, B. cepacia IST408 biofilms showed well-differentiated microcolonies with shapes resembling early cauliflower-like structures (Fig. 5C). The bceE::Tp mutant formed cell aggregates that covered more than 50% of the optical field, but without forming 3D macrocolonies as thick as the ones observed for the wild-type strain, with maximum thicknesses of 7.76 and 4.12 μm for the wild-type and bceE::Tp mutant, respectively. In contrast, the bceF::Tp mutant seemed to be unable to form mature biofilm forms, presenting small and scarce colonies throughout the chamber surface with a maximum thickness for this biofilm of 2.74 μm (Fig. 5C). The biomass contained in each strain's biofilm was also evaluated using COMSTAT2, with the parental IST408 biofilms having a biomass volume-to-surface ratio of 6.61 μm³/μm², the bceE::Tp mutant with 3.45 μm³/μm², and the bceF::Tp mutant with 2.53 μm³/μm², which is consistent with the biofilm formation data obtained using crystal violet staining.

Microarray data showed the altered expression of genes BCAL1069, BCAL0430, and BCAM1670, which are putatively involved in cyclic-di-GMP signaling (Table 2). Genes BCAL0430 and BCAM1670, encoding putative cyclic-di-GMP diguanylate cyclases with a GGDEF domain, showed 1.7- to 2.0-fold increased expression in the bceF::Tp mutant. In contrast, gene BCAL1069, which encodes a putative cyclic-di-GMP phosphodiesterase, showed 1.5-fold decreased expression. As an overview, the genome of B. cenocepacia J2315 has at least 25 genes involved in cyclic-di-GMP production and turnover. Therefore, it is difficult to predict whether changes in the expression of genes encoding digualylate cyclases and phosphodiesterases will have an effect on the cyclic-di-GMP concentration. To assess whether BCAL1069, BCAL0430, and BCAM1670 gene products are relevant in the regulation of intracellular cyclic-di-GMP levels in B. cepacia, this secondary messenger was extracted from cells of the wild-type IST408 and bceE::Tp and bceF::Tp mutants at 12, 24, and 48 h of growth in S medium (Fig. 6). The intracellular level of cyclic-di-GMP in the wild-type IST408 and bceE::Tp mutant did not change over the 48-h test period, with the mutant showing a slightly higher concentration. In contrast, the bceF::Tp mutant showed at least a 4-fold increase of cyclic-di-GMP levels above the wild-type B. cepacia IST408 in the first 24 h, with the concentration dropping to levels similar to the bceE::Tp mutant but still 2-fold higher than in the IST408 strain (Fig. 6).

Mutation in the bceF gene affected diverse phenotypes, such as motility, resistance to stress, and biofilm formation, among others. Due to the importance of these phenotypes in B. cepacia virulence, we determined whether the bceF::Tp mutant could also be attenuated in virulence by using Galleria mellonella as an acute infection model. This was accomplished by injecting approximately 2 × 10⁶ CFU of each bacterial strain into larvae. The results indicated virulence attenuation by the bceF::Tp mutant, with 35% of the larvae being alive at 48 h postinfection, while in the presence of the bceE::Tp mutant and the wild-type IST408 almost all larvae were dead (Fig. 7). Complementation of the bceF::Tp mutant with pDA17 expressing the bceF gene restored virulence to the wild-type level. These data regarding virulence attenuation of the bceF::Tp mutant are consistent with the decreased expression of genes putatively involved in virulence as well as the observed phenotypic properties.