ABSTRACT
Burkholderia cenocepacia is an opportunistic bacterial species capable of causing life-threatening respiratory tract infection in persons with cystic fibrosis (CF). Unlike most other pathogens in CF, which typically remain confined to the endobronchial spaces, B. cenocepacia can traverse airway epithelium to cause bacteremia and sepsis. The mechanisms by which this occurs, however, are unknown. We examined the transmigration of B. cenocepacia through polarized respiratory epithelium. Representatives of three “epidemic” lineages common among CF patients in North America were able to traverse polarized 16HBE14o− cells in vitro. Transmigration of bacteria was associated with significant perturbations in epithelial permeability, as measured by a loss of transepithelial electrical resistance and increased flux of bovine serum albumin across the cell layer. Terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling and trypan blue exclusion assays, as well as lactate dehydrogenase levels, did not indicate excessive cytotoxicity or cell death in infected cell layers. Rather, confocal fluorescence microscopy demonstrated the loss of occludin from tight junctions. In contrast, zonula occludens 1 was well preserved along intercellular borders. Western blot analysis showed a shift in the major occludin isoforms from high- to low-phosphorylation states during infection. These observations suggest that B. cenocepacia traverses polarized respiratory epithelium by the dephosphorylation and dissociation of occludin from the tight-junction complex.
Burkholderia cepacia has emerged during the past 2 decades as an important opportunistic pathogen in persons with cystic fibrosis (CF) (11). Although B. cepacia was previously regarded as a single species, recent taxonomic studies have identified at least nine closely related species in what is now referred to as the B. cepacia complex (Bcc). This group includes B. cepacia, Burkholderia multivorans, Burkholderia cenocepacia, Burkholderia stabilis, Burkholderia vietnamiensis, Burkholderia dolosa, Burkholderia ambifaria, Burkholderia anthina, and Burkholderia pyrrocinia (4-6, 25, 28-32). Although each of these species has been recovered from respiratory secretions of persons with CF, B. multivorans and B. cenocepacia account for the great majority of isolates (12, 24). Mounting evidence also suggests that B. cenocepacia is somewhat more pathogenic in CF than are the remaining species (1, 8).
A feature that distinguishes Bcc from other pathogens in CF is the occurrence in a significant proportion of infected patients of a rapidly progressive necrotizing pneumonia and septicemia. Such extrapulmonary dissemination rarely occurs with Pseudomonas aeruginosa or other CF pathogens, and the microbial factors that account for this invasive capacity remain largely undefined. Histopathologic studies of Bcc-infected CF lung have shown substantial numbers of organisms between bronchial epithelial cells in regions with relatively undamaged bronchioles (17). In vitro studies using well-differentiated primary human epithelial cells provide evidence that Bcc can traverse the respiratory epithelium by both paracytosis (passage between adjacent cells) and transcytosis (invasion of and passage through individual cells) (20). However, the mechanisms by which this occurs are poorly understood.
A prerequisite for paracytosis of bacteria through differentiated epithelium is compromise of intercellular apical tight-junction complexes (2, 3). These arrays, composed of cytoplasmic and integral membrane proteins, have both regulatory and scaffolding functions. Among the major tight-junction constituents is zonula occludens 1 (ZO-1), a cytoplasmic-face protein that is critical for tight-junction stability and linkage of the complex to the actin cytoskeleton. Occludin, another major tight-junction component, is an integral membrane protein that appears to have important regulatory functions in tight-junction development. Mucosal pathogens utilize a number of strategies to target one or both of these proteins as a means of disrupting epithelial-barrier integrity (9).
In this study, we investigated the ability of B. cenocepacia to alter the permeability of and migrate through polarized respiratory epithelium. We further sought to characterize the potential roles of the tight-junction proteins ZO-1 and occludin in this process.
MATERIALS AND METHODS
Bacterial strains and growth conditions.Representatives of three epidemic B. cenocepacia strains common among Bcc-infected CF patients in North America were selected for study. These include PC8, PC184, and AU0355, representing the PHDC, Midwest, and ET12 clonal lineages, respectively (4). Enteropathogenic Escherichia coli E2348/69 was kindly provided by James Kaper (University of Maryland, Baltimore). E. coli strain JM109 was used as a noninvasive control. All bacteria were stored in skim milk at −80°C. For use in epithelial-transmigration assays, bacteria from frozen stock were grown on Luria-Bertani (LB) agar at 34°C for ∼36 h. A single colony was inoculated into LB broth and grown to mid-log phase (optical density at 600 nm, ∼0.6) at 37°C in a shaking incubator. Bacteria were harvested by centrifugation and resuspended to the desired concentration in antibiotic-free minimal essential medium with Earle's salts without phenol red, supplemented with 10% fetal bovine serum and 2 mM l-glutamine (supplemented MEM; Invitrogen, Carlsbad, CA).
Epithelial cell cultures.The simian virus 40 large T antigen-transformed human bronchial epithelial cell line 16HBE14o− (16HBE) was kindly provided by Dieter Gruenert (7). Cells were recovered from frozen stocks and maintained in supplemented MEM containing penicillin (100 U/ml) and streptomycin (100 μg/ml). To obtain a polarized epithelial layer, cells were grown at an air-liquid interface on semipermeable membranes as described previously (19). Briefly, ∼5 × 105 16HBE cells (between passages 45 and 55) were seeded onto Transwell (Costar, Cambridge, MA) cell culture inserts fitted with either 3.0-μm- or 0.4-μm-pore-size membranes (for transmigration assays or microscopy, respectively) coated with bovine collagen type I (BD Biosciences, San Diego, CA), human fibronectin (Sigma, St. Louis, MO) and bovine serum albumin (BSA; Sigma) and grown submerged in supplemented MEM with antibiotics. After 48 h, medium from the apical chamber was removed to create an air-liquid interface, and the basolateral chamber was refreshed with medium every other day thereafter.
To confirm the functional integrity of the epithelial cell layers, transepithelial electrical resistance (TER) was measured with an EVOM epithelial volt-ohmmeter and an EndOhm 6 tissue resistance measurement chamber (World Precision Instruments, Sarasota, FL). Epithelial cell layers grown as described above consistently showed TER values of >350 Ω/cm2 within a week of seeding.
Epithelial transmigration assays.Polarized 16HBE cell layers with TERs of >350 Ω/cm2 were shifted to antibiotic-free supplemented MEM 24 h prior to infection. B. cenocepacia (106 log-phase CFU in 200 μl of antibiotic-free supplemented MEM) were inoculated onto the apical surface of the cell layer to yield a multiplicity of infection (MOI) of 20:1 and incubated at 37°C. Culture medium samples from the basolateral chamber were collected at various time points and cultured on LB agar to determine bacterial density. TER was measured at the same time points. Negative controls included cell layers that were mock infected with cell culture growth medium alone or infected with E. coli JM109 at an MOI of 20:1.
Epithelial permeability.The permeability of cell layers was quantified by measuring the flux of fluorescein isothiocyanate (FITC)-labeled BSA across the epithelial layer, as described previously (34). In brief, bacteria were grown as described above, suspended in antibiotic-free supplemented MEM containing 0.1% (wt/vol) FITC-labeled BSA (Sigma), and inoculated onto the apical surface of the cell layer at an MOI of ∼20:1 (34). The medium in the basolateral chamber was initially free of FITC-BSA. At specific intervals after infection, 50 μl of medium was withdrawn from the basolateral chamber, and fluorescence was measured in an SLM 8000 spectrofluorimeter (SLM Instruments, Urbana, IL). Permeability coefficients (Papp) were calculated as described previously (26).
Cytotoxicity assays.The trypan blue exclusion assay was done as previously described (27). Measurement of lactate dehydrogenase (LDH) in apical and basal media was performed by using the Cytotox 96 kit (Promega, Madison, WI) according to the manufacturer's instructions. Epithelial cell death was also assayed by using the DeadEnd Fluorometric TUNEL System (Promega) according to the manufacturer's instructions.
Immunodetection of ZO-1, occludin, B. cenocepacia, and filamentous actin in infected cultures.Polarized epithelial cells incubated with medium alone or infected with B. cenocepacia for 24 h were washed, fixed in methanol at −20°C, and blocked with 1% BSA for 1 h. The fixed cells were thereafter incubated overnight at 4°C with a mouse monoclonal antibody to either ZO-1 or occludin (BD Biosciences). In addition, B. cenocepacia was colocalized with these tight-junction proteins by probing the cell layers with a previously described polyclonal rabbit antibody (10, 16). After the cell layers were washed to remove unbound antibody, they were incubated with Alexa Fluor 488-conjugated goat anti-mouse immunoglobulin G (Molecular Probes Inc., Eugene, OR) to detect ZO-1 and occludin or Alexa Fluor 594-conjugated goat anti-rabbit immunoglobulin G (Molecular Probes) to visualize B. cenocepacia. In parallel, infected cell layers were first immunostained for B. cenocepacia and then stained with Alexa Fluor 594-conjugated phalloidin (Molecular Probes) to visualize filamentous actin or DAPI (4′,6-diamidino-2-phenylindole; Sigma) to visualize cell nuclei. Immunostained cell layers were examined with a Zeiss Confocal Laser 510 scanning microscope.
To investigate the possible roles of secreted bacterial factors and the necessity of direct contact between bacteria and epithelial cells, 16HBE cells were grown to 90% confluence on glass coverslips, which were coated in the same manner as the cell culture inserts, and submerged in supplemented MEM with antibiotics in a 12-mm well (33). B. cenocepacia (107 CFU) was added to a transwell cell culture insert fitted with a 0.4-μm-pore-size membrane and suspended in culture medium above the cell layer to prevent direct contact between the bacteria and the epithelial cells. After 24 h of incubation, the cell layer was fixed, immunostained to detect ZO-1 or occludin, and examined by confocal microscopy as described above.
Western blot analyses.Detergent-soluble (cytoplasmic and cell membrane) and -insoluble (cytoskeletal and cytoskeleton-associated) protein fractions were prepared from cell layers incubated with medium alone, E. coli, or B. cenocepacia as described previously (14), and total protein was quantified by using a modified Lowry assay (Bio-Rad Laboratories, Hercules, CA). Protein extracts (containing 10 μg of protein) were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and proteins were transferred to Immobilon (Millipore, Bellerica, MA) membranes. The membranes were blocked with 5% skim milk and then further incubated overnight at 4°C with antibody to either ZO-1 or occludin. Bound antibody was detected by using a horseradish peroxidase-conjugated second antibody and visualized by enhanced chemiluminescence (Pierce Biotechnology, Rockford, IL).
Statistical analyses.Data were analyzed by using GraphPad (San Diego, CA) software. To compare multiple groups, a standard parametric one-way analysis of variance (ANOVA) with Tukey-Kramer posttest analysis was used.
RESULTS AND DISCUSSION
Bacterial migration across polarized bronchial epithelial cell layers.Bacteria were recovered from the basal medium 4 h after infection of apical cell surfaces with either enteropathogenic E. coli E2348/69 or B. cenocepacia PC8, PC184, or AU0355 (Fig. 1). The bacterial density increased steadily thereafter, and by 24 h postinfection equaled the density in the apical medium, indicating ongoing transmigration of organisms through the cell layer and/or bacterial replication in the basal medium. In contrast, E. coli JM109 was not detected in the basal medium 24 h after infection of the apical cell surface, although, as with the three B. cenocepacia strains, it too increased to a density of approximately 108 CFU/ml in the apical medium by 24 h postinfection. Although significant differences were noted in the concentrations of the three B. cenocepacia strains in basal medium at 8 h postinfection (P < 0.05 between AU0355 or PC184 and PC8), by 12 h the bacterial concentrations were comparable. In preliminary work, we noted that these three strains had similar growth rates in supplemented MEM (data not shown). These observations suggest that B. cenocepacia strains may vary with respect to their ability to alter epithelial permeability and transmigrate through polarized epithelium early in the course of infection; however, further work will be required to more precisely define potential differences in the kinetics of invasion between specific strains.
Translocation of bacteria across polarized 16HBE cell layers. Cell layers were uninfected (○) or infected with E. coli JM109 (•), E. coli E2348/69 (⋄), B. cenocepacia PC8 (⧫), B. cenocepacia PC184 (□), or B. cenocepacia AU0355 (▴). The data points represent the mean (± standard error of the mean) concentrations of bacteria in basolateral media of four replicate experiments at each time point. The data were transformed to log10 and analyzed by one-way ANOVA with Tukey-Kramer posttest analysis. At the 8-h time point, concentrations of AU0355 and PC184 were significantly greater than that of PC8 (P < 0.05).
Transepithelial electrical resistance and epithelial permeability.To evaluate the integrity of infected polarized cell layers, we measured both the TER and the transepithelial flux of FITC-BSA. The maintenance of TER and the impermeability of FITC-BSA, which is not actively transported across 16HBE cell monolayers (34), provide complementary measures of epithelial cell barrier integrity.
Infection of apical cell surfaces with any of the three B. cenocepacia strains resulted in a steady decline in TER (Fig. 2A). This decrease was most pronounced between 4 h and 8 h after infection. Again, differences were observed between the three B. cenocepacia strains, although they did not reach statistical significance. In contrast, significantly (P < 0.001) smaller reductions in TER (most likely a result of shifting cultures from air/liquid to submerged conditions during infection) were observed in E. coli JM109- or mock-infected cell layers even after 24 h of incubation.
Integrity of infected and uninfected polarized 16HBE cell layers. (a) TER across cell layers. Cell layers were uninfected (○) or infected with E. coli JM109 (•), B. cenocepacia PC8 (⧫), B. cenocepacia PC184 (□), or B. cenocepacia AU0355 (▴). Uninfected and JM109-infected cell layers show significantly less reduction in TER than B. cenocepacia-infected cell layers (P < 0.001). The data are presented as percent change (± standard error of the mean) from the initial TER at each time point indicated. (b) Permeability of cell layers. Papp were calculated by measuring the flux of FITC-BSA from apical to basolateral media. Statistically significant differences (determined by using one-way ANOVA with Tukey-Kramer posttest analysis) between uninfected (control) cell layers or cell layers infected with E. coli JM109 and B. cenocepacia-infected cell layers are indicated by asterisks. *, P < 0.05; ***, P < 0.001.
To further define B. cenocepacia-induced epithelial barrier compromise, we measured the diffusion of FITC-BSA from apical to basal compartments during infection with each of the three B. cenocepacia strains. By 4 h postinfection, significantly more FITC-BSA was detected in the basal media of B. cenocepacia-infected cell layers than was observed in E. coli- or mock-infected control cell layers. By 20 h postinfection, differences were observed between the three B. cenocepacia strains tested, with AU0355- and PC184-infected layers showing significantly greater FITC-BSA flux than strain PC8 (P < 0.01), again suggesting that B. cenocepacia strains vary with respect to their capacity to disrupt epithelial integrity. The magnitude of these differences was more apparent when Papp were calculated for each strain (Fig. 2B). This calculation takes into account the total transepithelial flux of FITC-BSA during the 24-h time course of infection. Once again, PC8 showed significantly less capacity to disrupt the epithelial barrier than did strains PC184 and AU0355. Throughout 24 h, E. coli JM109-infected cultures never allowed more transepithelial flux of FITC-BSA than did mock-infected cell layers, indicating that our findings were not merely due to a nonspecific effect of bacterial infection.
Cytotoxicity.Although an increase in the flux of FITC-BSA across infected differentiated epithelial cell layers may result from the disruption of intercellular tight junctions (34), an increase in cell layer permeability would also be observed with more generalized epithelial cytotoxicity or cell death. In their study of epithelial invasion by Bcc, Schwab et al. (20) noted microscopic evidence of damage to superficial layers of well-differentiated primary human airway epithelial cell cultures in areas subjacent to B. cenocepacia biofilms in vitro. Cell damage and LDH release were also observed by Sajjan et al. (16) in squamous-differentiated primary airway epithelial cells after infection with B. cenocepacia. However, after 12 h and 24 h of infection, we detected no differences in trypan blue exclusion or LDH levels between B. cenocepacia-, E. coli-, or mock-infected cell cultures. Further, we found no differences between infected and mock-infected cell layers by terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling assay (data not shown), suggesting that a mechanism other than cell death accounts for the increased permeability of B. cenocepacia-infected polarized epithelial cell layers.
Alteration of tight-junction organization.The maintenance of TER and the relative impermeability of polarized epithelium to macromolecules require the association of transmembrane tight-junction proteins with cytoskeletal proteins at the apicolateral cell surface (2). To investigate the possible disruption of intercellular tight junctions by B. cenocepacia, we examined the tight-junction-associated cytoplasmic protein ZO-1 and the transmembrane protein occludin in infected cell layers by using confocal immunofluorescence microscopy. In uninfected control cell layers, both ZO-1 and occludin were observed at the margins of individual cells in a typical chicken wire-like pattern (Fig. 3A and B), indicating intact tight-junction complexes. In contrast, cell cultures infected with B. cenocepacia AU0355 revealed the colocalization of bacteria with occludin that appeared disrupted and translocated to cytoplasm (Fig. 3C). In contrast, ZO-1 showed no obvious disruption by confocal microscopy (data not shown).
Confocal immunofluorescence microscopy of infected polarized 16HBE cell layers. (a) Uninfected cell layer immunostained with antibody to ZO-1. (b) Uninfected cell layer immunostained with antibody to occludin. (c) Cell layer infected with AU0355 for 24 h and then immunostained with antibodies to occludin (green) and B. cenocepacia (red); the white arrows indicate occludin dislocated to cytoplasm. (d) Z-series cross-section of entire cell layer depicted in panel c; the black arrows point to the apical surface and indicate areas of colocalization of bacteria and occludin.
To determine the potential roles of secreted factors and whether bacterial-epithelial cell contact is necessary for disruption of occludin, polarized epithelial cells were grown on glass coverslips submerged in medium and overlaid with 0.4-μm-pore-size cell culture inserts containing B. cenocepacia AU0355. After 24 h, the cells were stained for occludin and examined by confocal microscopy. Tight-junction organization appeared undisturbed, suggesting the absence of a secreted toxin capable of tight-junction disruption, as is seen with certain enteric pathogens, such as C. difficile, Bacteroides fragilis, and enteropathogenic E. coli (22). The apparent lack of tight-junction disruption also indicates the absence of a global nonspecific effect of bacterial replication (e.g., nutrient depletion) in apical fluid that might account for the tight-junction changes observed in infected cell layers. Similar results were obtained with cell layers incubated with spent filter-sterilized bacterial culture supernatant (data not shown).
Infected cell cultures were also treated with Alexa Fluor 584-conjugated phalloidin and DAPI to examine actin filaments and cell nuclei, respectively. Despite the compromise in epithelial barrier integrity and the loss of tight-junction-associated occludin, phalloidin staining did not reveal large-scale alterations in the actin cytoskeleton, stress fiber formation, or contraction of the actomyosin ring. In contrast, Schwab et al. (21), using a higher infecting inoculum of B. cenocepacia, noted actin cytoskeletal rearrangements in well-differentiated airway epithelia associated with invading bacterial biofilm in vitro. In our model, although abundant bacteria were observed in infected cell layers, cell nuclei showed no signs of condensation or fractionation, again indicating little cytotoxicity or cell death.
Dephosphorylation of occludin.Bacterial pathogens can alter epithelial tight-junction integrity through a variety of mechanisms, including down-regulation of tight-junction protein expression and dephosphorylation and dislocation of tight-junction protein constituents (13, 15, 18, 23, 36). To assess potential changes in the relative partitioning of occludin and ZO-1 to the tight-junction complex during B. cenocepacia infection, we prepared detergent-insoluble fractions from cell layers and examined them for the presence of occludin and ZO-1 by Western blot analysis. Reduction of these proteins from the detergent-insoluble phase suggests their dislocation from the tight junction to the cytoplasm (18, 22, 23). Similarly, an accumulation of relatively dephosphorylated occludin in the detergent-insoluble phase indicates its dissociation from the tight-junction assembly (35). Immunoblots of detergent-insoluble protein extracts from uninfected cell cultures or cells infected with E. coli JM109 showed a major protein band at 75 to 79 kDa, consistent with the hyperphosphorylated form of occludin, and a faint protein band at 65 to 71 kDa, representing occludin with reduced phosphorylation (35) (Fig. 4). In contrast, protein extracts from AU0355- or enteropathogenic E. coli E2348/69-infected cells showed a reduction in the high-molecular-weight form of occludin with a concomitant accumulation of the relatively dephosphorylated form in the detergent-insoluble fraction. This effect appeared to be time dependent and correlated with the number of bacteria recovered from the basolateral medium. Thus, a greater reduction in phosphorylated high-molecular-weight occludin was observed in cells at 24 h postinfection in B. cenocepacia-infected cultures and as early as 3 h to 6 h postinfection in E. coli E2348/69-infected cultures (Fig. 4A). We observed no changes in ZO-1 between control and infected cultures (Fig. 4B).
Western blot analysis of detergent-insoluble protein fractions from 16HBE cells. Equal amounts of protein were loaded in each lane, and the blots were probed with mouse monoclonal antibody to occludin (A), ZO-1 (B), or β-actin (C). Lane 1, molecular weight markers; lane 2, mock-infected control cells; lane 3, cells infected with noninvasive E. coli JM109 for 24 h; lanes 4 and 5, cells infected with B. cenocepacia AU0355 for 8 h and 24 h, respectively; lanes 6, 7, and 8, cells infected with enteropathogenic E. coli S2348/69 for 3, 6, and 12 h, respectively. All infections were done with an MOI of 20:1. The solid arrow indicates the position of hyperphosphorylated occludin; the open arrow indicates the position of relatively dephosphorylated occludin.
Summary.Taken together, our findings demonstrate the ability of B. cenocepacia to increase the permeability of and migrate across polarized respiratory epithelium expressing tight junctions. Our data suggest that this increased permeability does not result from disintegration of the cell layer due to cell injury or death. Rather, the dephosphorylation of occludin and its dislocation from focal points along the tight-junction axis appear to be involved in the disruption of tight-junction integrity, thereby allowing the transepithelial migration of bacteria. Confocal microscopy did not demonstrate large numbers of intracellular bacteria, suggesting that in our model epithelial invasion occurs primarily via paracytosis. However, we did not otherwise specifically investigate the role of epithelial cell internalization in this process, and our study does not exclude the possibility that tight-junction disruption might also permit bacteria to access as yet undefined basolateral epithelial receptors involved in intracellular invasion (13). Nevertheless, our study provides important insight into the mechanisms by which B. cenocepacia traverses respiratory epithelium. The precise roles of bacterial density and biofilm formation and whether all species of the Bcc utilize the same mechanisms for epithelial invasion require further investigation.
ACKNOWLEDGMENTS
This work was supported by grants (to J.J.L.) from the Cystic Fibrosis Foundation and NIH (AI054411-01). J. Y. Kim was supported by NIH grant T32-HD007513 and the C. S. Mott Children's Hospital Fund for Research. U. S. Sajjan was supported by the Carroll Haas Research Fund in Cystic Fibrosis.
FOOTNOTES
- Received 7 May 2005.
- Returned for modification 23 June 2005.
- Accepted 20 July 2005.
- Copyright © 2005 American Society for Microbiology
