Studies on the Involvement of the Exopolysaccharide Produced by Cystic Fibrosis-Associated Isolates of the Burkholderia cepacia Complex in Biofilm Formation and in Persistence of Respiratory Infections

Abstract

Bacteria belonging to the Burkholderia cepacia complex (BCC) are important opportunistic pathogens that lead to respiratory infections in patients with cystic fibrosis (CF). The clinical outcome following colonization with BCC bacteria is highly variable, and so far, unpredictable. A large percentage (80 to 90%) of BCC isolates from CF patients produce the exopolysaccharide (EPS) cepacian, which has been hypothesized to play a role in the colonization and persistence of these bacteria in the CF lung. In this work, we demonstrate that although it is not required for the initiation of biofilm formation, cepacian plays a role in the establishment of thick biofilms. This conclusion was based on a comparison of the abilities of EPS-defective mutants derived from a B. cepacia mucoid CF isolate by random plasposon insertion mutagenesis and the ability of the parental strain to form biofilms. However, the systematic characterization of 108 CF isolates, corresponding to 15 distinct strains, indicated that other strain-dependent factors are also involved in the development of thick, mature biofilms. The isolates examined belonged to the species B. cepacia, B. multivorans, B. cenocepacia, and B. stabilis and were obtained during a 7-year period of surveillance from 21 CF patients receiving care at the major Portuguese CF center. Most of them (90%) were serial isolates from 12 persistently infected patients. In spite of the concept that bacteria growing in biofilms display more resistance to antibiotics and to host phagocyte killing than do planktonically growing cells, no clear correlation could be established between the ability of the various strains examined to produce EPS and/or to form biofilms in vitro and the persistence or virulence of the respiratory infections they caused in different patients.

Bacteria belonging to the Burkholderia cepacia complex (BCC) have become problematic opportunistic pathogens in patients with cystic fibrosis (CF) (13, 14, 17, 23). These bacteria are highly resistant to multiple antibiotics and pose the risk of spread among patients with CF by social contact (12). Furthermore, in approximately 20% of BCC-infected CF patients, BCC bacteria can lead to a rapid decline in lung function and to a fatal necrotizing pneumonia accompanied with septicemia, called the cepacia syndrome (22). In most cases, infection with BCC bacteria occurs in patients colonized with Pseudomonas aeruginosa (13). Biofilm formation by P. aeruginosa has been recognized as an important clinical problem since bacteria growing in biofilms display more resistance to antibiotics and more protection from the immune response of the host than do planktonically growing cells (18, 36, 39). Biofilm formation has also been observed for the BCC (6, 20), and B. cepacia and P. aeruginosa are thought to coexist as mixed biofilms in the lungs of patients suffering from CF (33, 34).

Exopolysaccharide (EPS) production has been implicated in biofilm formation (19, 39). Studies with Vibrio cholerae and Escherichia coli indicated that EPS is essential for the development of mature biofilms because it stabilizes the three-dimensional biofilm architecture (9, 37, 40). The initiation of biofilm formation is, however, dependent on surface pili and flagella for the interaction with the surface during the attachment process and for moving along in two dimensions, allowing the formation and enlargement of microcolonies (37). The overproduction of alginate by clinical isolates of P. aeruginosa is considered to contribute to biofilm formation and to the long-term survival of the bacterium in the lungs of CF patients (10, 13). Although mucoid colony morphotypes have been considered rare in both clinical and environmental isolates of the BCC (13), approximately 80 to 90% of the BCC isolates involved in respiratory infections among the CF patients receiving care at the major Portuguese CF center are EPS producers (30; also see the results of this study). It was therefore hypothesized that the EPS may play a role in the colonization and persistence of BCC bacteria in the CF lung (30), as has been ascribed to P. aeruginosa alginate. The EPS produced by different BCC isolates obtained from various CF centers in different countries is composed of a branched acetylated heptasaccharide repeating unit made up of d-glucose, d-rhamnose, d-mannose, d-galactose, and d-glucuronic acid (1:1:1:3:1) (1, 2, 3, 4, 30, 32). This suggested that this EPS, recently denominated cepacian (35), is specific to BCC bacteria, and the pathway leading to the nucleotide sugar precursors that are necessary to its biosynthesis was proposed (31).

For the present work, we examined the putative role of cepacian in biofilm formation and in the persistence and virulence of BCC respiratory infections. This study was based on a comparison of the abilities of EPS-defective mutants recently derived by a random plasposon insertion mutagenesis strategy from the mucoid CF isolate B. cepacia IST408 (27) and the ability of the parent strain to form biofilms. Three of these mutants carried plasposon insertions in the genes bceF (IST408-SS1 and IST408-SS2) and bceI (IST408-SS3), which belong to the EPS biosynthetic gene cluster bce (27). BceF is a putative tyrosine kinase, while BceI is presumably involved in the polymerization of the cepacian repeat unit (27). The results of this study indicate that although the EPS is not essential for the initiation of biofilm formation, the mucoid wild-type strain forms a thicker biofilm than the isogenic EPS-defective mutants. To elucidate the possible role of cepacian and biofilms in the persistence and virulence of CF respiratory infections, we characterized 108 clinical BCC isolates according to their abilities to synthesize EPS and to form biofilms in vitro. The isolates were obtained from 21 Portuguese CF patients who received care at the Santa Maria Hospital in Lisbon during a 7-year period of surveillance. These 108 isolates corresponded to 15 different strains belonging to B. cepacia (formerly genomovar I), B. multivorans (formerly genomovar II), B. cenocepacia (formerly genomovar III) recA subgroups A and B, and B. stabilis (formerly genomovar IV) (8). Most of them (90%) were serial isolates from 12 different patients who were persistently colonized (8). The results of this study, together with information on the clinical outcomes of the infected patients, indicate that (i) the development of thick biofilms involves other strain-dependent factors besides the EPS and (ii) the persistence and virulence of respiratory infections caused by BCC bacteria depend on other determinants besides the abilities to produce EPS and to form mature biofilms.

MATERIALS AND METHODS

Bacterial strains, plasmids, and culture conditions.

B. cepacia IST408, a high-EPS-producing isolate obtained from a Portuguese CF patient (30), and the EPS-defective mutants IST408-SS1, IST408-SS2, and IST408-SS3, derived from IST408 by plasposon mutagenesis (27), were used for this work. A collection of 108 BCC isolates, described previously by Cunha et al. (8), was also examined. These isolates were obtained from the respiratory secretions of 21 patients with CF who attended the Santa Maria Hospital CF Center in Lisbon, Portugal, from January 1995 to June 2002. In general, sputum samples were obtained every 2 months, when patients were subjected to a periodic consultation to monitor their clinical status. Cultures were taken more often from patients showing clinical deterioration.

When in use, bacteria were maintained on Pseudomonas isolation agar (Difco) plates supplemented with 600 μg of kanamycin/ml to keep selective pressure on the EPS-defective mutants IST408-SS1, IST408-SS2, and IST408-SS3. S medium, described by Richau et al. (30), was used to quantify EPS production by the BCC strains. When S medium was used as a solid medium, 20 g of agar (Iberagar)/liter was added. Unless otherwise stated, liquid cultures were carried out with orbital agitation (250 rpm) in Lennox broth supplemented with kanamycin when appropriate. Growth was monitored by measuring the optical density at 640 nm (OD₆₄₀).

Biofilm formation assays.

Biofilm formation assays were based on a methodology described by O'Toole and Kolter (29). Overnight liquid cultures of the CF isolate IST408 and three EPS-defective mutants isolated in a previous work (27) were transferred to Lennox or S liquid medium and grown at 30°C with orbital agitation until the mid-exponential phase was reached. The cultures were subsequently diluted to a standardized culture OD₆₄₀ of 0.5, and 20 μl of this cell suspension was used to inoculate the wells of a 96-well polystyrene microtiter plate (Greiner Bio-One) containing 180 μl of Lennox or S medium. Wells containing sterile growth medium were used as negative controls. Plates were incubated at 30 or 37°C for 24 or 48 h without agitation. Biofilm formation in the polystyrene microtiter plates by the 108 clinical isolates was examined under standardized conditions, specifically after 24 h of growth in S medium at 30°C without agitation. Growth was assessed by measuring the absorbance of cultures in the wells at 640 nm with a VERSAmax tunable microplate reader (Molecular Devices). For biofilm quantification, the culture media and unattached bacterial cells were removed from the wells by careful rinsing with water (three times, 200 μl for each rinse). Adherent bacteria were stained with 200 μl of a 1% crystal violet solution for 15 min at room temperature (50 ml of the solution was prepared by adding 1% [wt/vol] crystal violet in 10 ml of 95% ethanol to 40 ml of water containing 0.4 g of ammonium oxalate). After three gentle rinses with 200 μl of water each time, the dye associated with the attached cells was solubilized in 200 μl of 95% ethanol and the biofilm was quantified by measuring the absorbance of the solution at 590 nm (A₅₉₀) in a microplate reader.

Motility and chemotaxis assays. (i) Swimming assay.

Swim agar plates (1% tryptone [Difco, Detroit, Mich.], 0.5% NaCl, and 0.3% agar [Oxoid]) were point inoculated by the use of sterile toothpicks with bacteria from colonies growing on Pseudomonas isolation agar plates and were incubated at 30°C for 24 h. Motility was assessed by examining the circular turbid zone formed by the bacterial cells migrating away from the point of inoculation.

(ii) Swarming assay.

Swarm plates (8 g of nutrient broth [Difco]/liter, 0.5% agar [Oxoid], and 5 g of glucose/liter) that had been dried overnight at room temperature were point inoculated by the use of sterile toothpicks and were incubated at 30°C for 24 h.

(iii) Chemotaxis assay (swarm plate assay).

Swarm medium without glucose was inoculated with 10 μl of a washed cell suspension (OD₆₄₀ = 0.3) at the center of the plate. On the left side of the plate, 3 drops (10 μl each) of phosphate-buffered saline was used as a negative control, and on the right side, 3 drops (10 μl each) of a 10% Casamino Acids (Difco) solution was used as a chemoattractant. The plates were incubated overnight at 30°C. The formation of rings in the direction of the substrate was considered a positive result.

EPS quantification.

EPS production was assessed by confluent growth of the different isolates in S solid medium. Plates were inoculated with 100 μl of a suspension of cells harvested during the exponential phase of growth, resuspended to obtain a standardized OD of 0.2 ± 0.02 (mean ± standard deviation), and incubated for 5 days at 30°C. After the incubation period, the plates were scraped and the material obtained was resuspended in 0.9% NaCl (wt/vol) by vortexing. The bacterial cells present in these suspensions were separated by centrifugation at 20,000 × g for 15 min. The EPS was precipitated from the cell-free supernatant by the addition of 2.5 volumes of cold ethanol and then was air dried and redissolved in distilled water. The total sugar content was assessed by the phenol-sulfuric acid method (11), with the EPS produced at 30°C by isolate IST408 used as a standard. For this purpose, the EPS solution was further dialyzed against distilled water at 4°C for 24 h and then recovered by freeze-drying. The cell pellets obtained from each plate were washed once with 0.9% NaCl, and the protein content was quantified by the biuret method. EPS production was expressed in grams of total sugars per biomass. Biomass was associated with the amount of protein present (in grams). The results are means of at least three independent cultivations and three determinations of the total sugar and protein contents for each sample.

RESULTS

Effect of assay growth conditions on the biofilm formed.

Biofilm formation by the mucoid CF isolate IST408 was assessed in polystyrene microtiter dishes with either Lennox broth or S medium and an incubation temperature of 30 or 37°C. Consistent with previous reports for other bacterial species (29), the size of the biofilms formed was dependent on the growth medium composition and the growth temperature (Fig. 1), although biofilms formed by strains of the BCC were previously considered to be virtually independent of the medium composition and incubation temperature (6, 20). Remarkably, the culture conditions leading to thicker biofilms in our study, specifically growth in S medium at 30°C (Fig. 1), were those that were reported before to lead to higher EPS production (30).

FIG. 1.

Comparison of the amounts of biofilm formed after 24 h of cultivation, without shaking, by the mucoid CF isolate IST408 in the wells of polystyrene microtiter dishes containing either Lennox broth (white bars) or S medium (black bars) at 30 or 37°C. The absorbance at 590 nm (A₅₉₀) quantifies the amount of crystal violet associated with the biofilm after staining, as described in Materials and Methods.

EPS-defective mutants form smaller biofilms than does strain IST408.

The time course of biofilm development by the mucoid isolate and the nonmucoid isogenic mutants was compared for 48 h of incubation under the most favorable growth conditions, as determined above (Fig. 2). Although the three EPS-defective strains were also capable of producing biofilms, the amounts of biofilms formed were remarkably less than the amount of biofilm formed by strain IST408. At the end of the incubation time, the thin biofilms formed by the three EPS⁻ mutants were rapidly dispersed, whereas the biofilm developed by IST408 was much more stable, as it came off in large cell clumps and was more difficult to disperse. The decreased amounts of biofilms formed by the EPS-defective mutants cannot be attributed to growth defects, since the specific growth rates and biomass yields of the four strains examined, calculated during batch cultivation in shake flasks, were identical (Fig. 3). The results from chemotaxis and motility assays (not shown) also indicated that the differences detected in biofilm formation cannot be attributed to an impaired ability of the mutants to move across the surface. This is consistent with the apparent identical capacity exhibited by the four strains examined to initiate surface attachment. Altogether, these results clearly indicate that biofilm development was similarly affected in the three EPS-defective mutants and that although it is not required for biofilm formation, the EPS produced contributes to the development of thick and stable biofilms in BCC bacteria.

FIG. 2.

Comparison of biofilms formed in the wells of polystyrene microtiter dishes with S medium by B. cepacia IST408 and the EPS-defective plasposon insertion mutants IST408-SS1, IST408-SS2, and IST408-SS3 after 48 h of growth at 30°C without shaking. The amounts of biofilm formed (A₅₉₀) by B. cepacia IST408 (•), IST408-SS1 (▿), IST408-SS2 (⋄), and IST408-SS3 (○) were assessed for 48 h. Error bars represent the standard deviations of the mean values for three independent experiments.

FIG. 3.

Comparison of growth curves (S medium at 30°C with orbital agitation [250 rpm]) of the EPS-producing isolate IST408 (□) and the three EPS-defective mutants IST408-SS1 (○), IST408-SS2 (▴), and IST408-SS3 (▪) examined for biofilm formation as shown in Fig. 2.

Comparison of the ability of 108 CF isolates to produce EPS and to form biofilms in vitro.

The comparison of the ability of BCC isolates from CF patients to produce EPS in S solid medium and to form biofilms in polystyrene microplates was extended to a collection of 108 BCC isolates. Ninety-seven of these isolates were serial isolates from 12 persistently infected CF patients among a total of 21 infected CF patients receiving care during a 7-year period at the major Portuguese CF center, the Santa Maria Hospital in Lisbon (8). A characterization of the genomovar status of these isolates revealed that 26 isolates (24%) were B. cepacia, 11 isolates (10.2%) were B. multivorans, 24 isolates (22.2%) were B. cenocepacia recA subgroup A, 29 isolates (26.9%) were B. cenocepacia recA subgroup B, and 18 isolates (16.7%) were B. stabilis (8). Since a ribotyping analysis of these 108 BCC isolates by the use of EcoRI generated 15 distinct ribopatterns, we assumed that there were solely 15 distinct strains (8). The examined isolates belonging to the B. multivorans (11 isolates) and B. stabilis (18 isolates) species generated single ribopatterns (ribopatterns 9 and 1, respectively), suggesting that they are clonal variants of only two strains of the referred species. However, isolates belonging to B. cepacia, B. cenocepacia subgroup A, and B. cenocepacia subgroup B generated three, four, and six distinct ribopatterns, respectively. The amount of EPS produced and the size of the biofilm formed in vitro were determined for each isolate. Remarkably, the amounts of EPS produced and the sizes of the biofilms formed by serial isolates obtained from the same patient and exhibiting the same ribopattern varied. This variation was not due to the low reproducibility level of the assays, as exemplified in Fig. 4. In this figure, the amounts of EPS produced and the sizes of the biofilms formed by all of the sequential isolates with ribopatterns 12 and 15, obtained for different isolation times from patients N and R, respectively, are compared. This phenotypic variability with respect to EPS production among BCC isolates corresponding to the same strain is documented and may result from differential gene expression (25). Moreover, it was also recently observed by Head and Yu that despite having identical genomic profiles, sequential CF isolates of P. aeruginosa produce variable amounts of biofilms (15).

FIG. 4.

Comparison of EPSs produced after 5 days of incubation at 30°C in S medium plates (white bars) and of biofilms formed in microtiter plates after 24 h of incubation at 30°C (black bars) by sequential isolates obtained from the same patient and exhibiting the same ribopattern. Bars represent the standard deviations of the means for at least three independent assays for EPS production and for sizes of the biofilms formed.

Despite the variation registered for the amounts of EPS produced and the sizes of the biofilms formed in vitro by different sequential isolates of the same strain obtained from the same patient, the results obtained are presented in Fig. 5 and were associated with specific strains and patients. Specifically, the data in Fig. 5 are the median values of the amounts of EPS produced and the sizes of the biofilms formed by all of the serial isolates corresponding to the same strain (with the same ribopattern) and isolated from a specific patient. Confirming previous observations indicating that the majority of the BCC isolates from CF patients produce EPS (30), 96 of the 108 isolates examined (89%) produced EPS under the standard conditions tested (Fig. 5). Among the 12 isolates that did not produce detectable amounts of EPS, 10 belonged to B. cepacia (9 of them were among the 13 serial isolates obtained from patient N and 1 was the single B. cepacia isolate obtained from patient O) and the other 2 belonged to B. stabilis (2 of the 5 serial isolates obtained from patient B).

FIG. 5.

Amounts of EPS produced after 5 days of incubation in S medium plates at 30°C (white bars) and sizes of biofilms formed in microtiter plates after 24 h of incubation at 30°C (black bars) by different BCC strains. The corresponding ribopattern (arabic numbers), the infected CF patient (capital letters), and the bacterial species are also indicated. In cases of serial isolates, numbers above the bars indicate the numbers of isolates with the same referred ribopattern that were obtained from the same indicated patient. The results are given as the median values of the EPS produced and the amounts of biofilm formed by the different strains (same ribopattern) obtained from the same patient. For each isolate, the results are the means of at least three independent assays for EPS production and for the size of the biofilm formed. Bars represent the standard deviations of the median values, calculated for all of the isolates belonging to the same strain and obtained from the same patient.

All of the clinical isolates examined were capable of forming biofilms, although they did so in variable amounts, independent of their species or genomovar status or ability to produce EPS (Fig. 5). The thinner biofilms observed were produced by a B. cepacia isolate, obtained from patient O, which was not capable of producing EPS under the tested conditions; two other isolates, belonging to B. cenocepacia subgroup B and obtained from patient G, formed a very thin biofilm and also produced small amounts of EPS (Fig. 5). These observations are in agreement with the indications obtained before by comparing the amount of the biofilm formed by the mucoid isolate IST408 with those produced by the respective EPS-defective mutants. Although a similar correlation was observed for several other isolates (for example, for isolates with ribopatterns 2, 5, 14, 15, and 16), a significant number of exceptions were also registered (Fig. 5). This was the case for the 13 B. cepacia serial isolates obtained from patient N, which produced biofilms of a significant size although 9 of these isolates did not produce EPS and the other 4 isolates produced small amounts of EPS. The isolates obtained from patient Y (belonging to B. multivorans), the isolates from patients C and I (belonging to B. cenocepacia subgroup A), and the isolates from patients A and D (belonging to B. cenocepacia subgroup B) all formed thick biofilms but only produced small (or null) amounts of EPS (Fig. 5). These results indicate that no clear correlation can be established between the amounts of EPS produced and the thicknesses of the biofilms formed when nonisogenic strains are compared, suggesting that there are other strain-dependent factors that play a role in determining the thickness of biofilms.

Persistence of infections and clinical outcome in relation to EPS production and the size of the biofilm.

To elucidate whether or not there is a correlation between increased EPS production and/or the size of the biofilm formed in vitro and the persistence of BCC infection in CF patients, the results described above were analyzed in relation to the clinical outcomes of the infected patients. Patients were considered to be persistently infected if at least three positive cultures for BCC bacteria persisted for a 6-month period; according to this criterion, 12 of the 21 CF patients were persistently infected, specifically patients B, H, G, J, N, O, P, R, T, U, W, and Y (Fig. 5).

The isolates exhibiting the highest EPS biosynthetic ability generated ribotypes 2, 5, 13, 14, 15, and 16 (Fig. 5). Isolates with ribopattern 2 were obtained from patients A, E, F, and X, who were all transiently colonized, and from patient W, who was persistently infected for at least a 6-month period. The remaining highly mucoid isolates were obtained from patients G, P, O, Q, R, and T (Fig. 5). With the exception of patient Q, these patients were persistently infected (8). The remaining patients from whom BCC bacteria were sporadically isolated (patients C, D, I, and AB) harbored strains with a reduced ability to produce EPS but capable of forming thick biofilms (Fig. 5). Of the 12 persistently infected patients examined in this study, 9 harbored at least one highly EPS-producing strain (Fig. 5), while 3 (B, N, and Y) harbored strains producing null or very small amounts of EPS. The isolates forming the largest biofilms, with ribopatterns 1, 2, 3, 5, 7, 8, 9, 11, 15, and 16 (Fig. 5), were also obtained from transiently (A, E, F, X, C, I, D, and AB) and persistently (W, Y, G, J, O, R, T, and U) infected patients.

Three of the 21 CF patients examined in this study (patients G, H, and J) succumbed to the cepacia syndrome, but only patient G had bacteremia. This patient harbored three different strains over an 8 month-period; two of them, belonging to B. cenocepacia subgroup B, produced very small amounts of EPS and formed thin biofilms, while the other strain (B. cepacia) produced significant amounts of EPS and was capable of substantial attached growth in vitro (Fig. 5). Patient J simultaneously harbored strains belonging to B. multivorans (ribotype 9) and to B. cenocepacia subgroup A (ribotype 11), which both produced detectable amounts of EPS and were capable of forming thick or moderately sized biofilms in vitro. Notably, the third patient (patient H) whose death was related to the cepacia syndrome was also persistently infected with an EPS-producing B. stabilis strain, which formed biofilms of moderate size. Two other deceased patients (patients N and T) were registered during the surveillance period of this study. Patient N died after progressive deterioration during a persistent infection with a non-EPS-producing B. cepacia strain that was capable of forming biofilms of a moderate size in vitro, while patient T was infected with a highly EPS-producing and biofilm-producing B. cenocepacia subgroup B strain. Altogether, these results indicate that no clear correlation can be established between the persistence or virulence of BCC bacteria in CF respiratory infections and the ability of the bacteria to produce significant amounts of EPS and/or to form thick biofilms in vitro.

DISCUSSION

The ability of bacteria to form biofilms has been associated with their capacity to cause disease in the human host (7). It is well documented that the EPSs produced by E. coli and V. cholerae are essential for the development of mature biofilms, as strains producing null or small amounts of EPS only produce thin biofilms that are devoid of normal architecture (9, 40). Alginate production in P. aeruginosa has also been correlated with the ability of this bacterium to form thick and mature biofilms, where the bacteria exhibit higher resistance to antimicrobials (16, 28) and to host phagocyte killing. Therefore, it is considered that these cell responses contribute to persistent infections which lead to the characteristic chronic destruction of CF patients' airways (13). However, there are apparently conflicting observations described in the recent literature with respect to eventual differences in the structural architecture or antibiotic sensitivity of the biofilms formed by mucoid P. aeruginosa compared with nonmucoid mutants (16, 28, 38). These contrasting observations may at least partially be due to differences in the strain and experimental conditions examined, as suggested by the observations of Klausen et al. (24).

The availability of three EPS-defective mutants with no growth defects derived from a mucoid CF isolate belonging to the BCC enabled us to demonstrate that, for these bacteria also, the EPS is required for the development of thick and mature biofilms. In our study, we first optimized both the growth medium and the temperature for EPS production and biofilm formation. Although the laboratory S medium, the temperature of 30°C, and the abiotic surface used to compare biofilm development cannot be considered growth conditions identical to those established in the CF lung, they allow a clear differentiation of the ability of these bacteria to produce EPS in different amounts and to form biofilms of different sizes. Based on the experimental conditions established, we found that the amount of the biofilm formed is associated with the level of EPS produced. This correlation was definitively established when we compared isogenic strains that apparently only differed in their EPS biosynthetic abilities. However, this correlation failed in certain cases, when the comparison involved different strains producing variable levels of EPS. Indeed, several discrepancies were observed when 108 BCC CF isolates were compared with respect to the EPS production yield and the dimension of the biofilm formed in vitro. These isolates corresponded to only 15 different strains belonging to B. cepacia, B. multivorans, B. cenocepacia, or B. stabilis. The results suggest that other strain-related factors, besides the level of EPS produced, play a role in determining the thickness of the biofilm formed. In particular, differences in the growth kinetics may account for the differences observed in the sizes of the biofilms formed by the distinct strains examined, as suggested by the observations of Conway et al. (6). Remarkably, none of the genes belonging to the bce cluster, recently described as being involved in cepacian biosynthesis (27), were identified by a screen carried out by Huber et al. (21). This screen was biased exactly towards BCC mutants that were defective in the late stages of biofilm development. The inactivated genes identified in that study were genes encoding surface proteins, genes involved in biogenesis and the maintenance of an integral outer membrane, and genes encoding regulatory factors. The corresponding regulatory mutants produced highly reduced amounts of N-octanoyl homoserine lactone, the major signal molecule of the cep quorum-sensing system of the BCC, which has been proposed to be a major checkpoint for biofilm formation (21).

Although B. cenocepacia is the predominant species in the CF infected populations examined so far (26) and although it is considered to represent a significant clinical risk, other less represented species may also be associated with a serious outcome (8, 26). In particular, in the Portuguese CF population from whom the BCC isolates examined in the present study were obtained, B. cepacia and B. stabilis are significantly represented, after B. cenocepacia (8). These less frequent species were also associated with poor clinical outcomes, including the cepacia syndrome, and include strains that produce EPSs (8). Most of the persistently infected patients who were monitored for this study harbored BCC isolates capable of producing EPSs. However, some of the highest EPS-producing bacteria were also obtained from transiently infected patients. The adaptation of an isolate of B. cenocepacia that resulted in its persistence in a mouse model of infection was recently described (5). Significantly, this persistence was correlated with a change in the colonial morphology from matte to shiny, with EPS production possibly contributing to the shiny morphology (5). Chronic infections were predominant in this CF population, and 9 (75%) of the 12 persistently infected patients examined in this study harbored at least one highly EPS- and/or thick-biofilm-producing strain. However, the few exceptions observed limited definitive conclusions, rendering it impossible to clearly associate the ability of CF isolates to produce EPS and/or to form thick biofilms in vitro and their ability to lead to chronic infections and clinical deterioration.

The results presented in this work clearly indicate that although EPS is not required for biofilm formation, it is involved in the formation of thick and mature biofilms by BCC bacteria in vitro. Therefore, we hypothesized that by promoting the formation of mature biofilms, the EPS may enhance bacterial survival in the CF lung since sessile bacteria within the biofilm can more efficiently withstand host immune responses and antibiotic action than planktonic cells, thus contributing to the persistence of BCC bacteria in the CF airways. However, there are certainly other strain-dependent factors that play a role in the formation of thick biofilms. Moreover, there are other factors involved in the persistence and virulence of respiratory infections caused by these opportunistic pathogens in patients with CF, depending on both the bacterial strain and the host.