Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Airway epithelial cells perform many important functions, serving as an interface between environmental stimuli and the lung parenchyma. Normally the lower airways are pristine, free of bacterial flora or inflammatory cells, and are well protected by several layers of defenses including antimicrobial peptides, mucin, and ciliary action. There is a brisk epithelial response to airway injury caused by many different mechanisms (1, 2). Bacterial penetration of airway defenses and binding to epithelial receptors promptly results in epithelial cytokine expression (3). An alternative response may be the expression of the signaling cascade leading to apoptosis (programmed cell death). Activation of epithelial proinflammatory signaling cascades is mediated by nuclear factor (NF)-B, a transcriptional activator, which, in addition to its effects in promoting inflammation, is also known to have major antiapoptotic effects (4). Infection of mucosal surfaces can have significant effects on the turnover of epithelial cells. Whereas viral infections of the respiratory tract, notably those caused by Epstein- Barr virus, have a major antiapoptotic effect attributed to the need for ongoing host cell replication for efficient viral infection, effects of bacterial infection on the induction of apoptosis in airway cells are less well characterized. At other mucosal sites, a progression from epithelial cytokine activation to apoptosis, degeneration, and cell replacement has been described for the lumenal cells that interact with potential pathogens (7). However, there appears to be relatively little ongoing cell turnover and apoptosis in cells lining the airways in the absence of injury (8). Apoptosis of airway epithelial cells as a general response to bacterial infection has not been examined in detail.

The response of the airway epithelium to excessive bacterial stimuli is reflected in the pathology associated with cystic fibrosis (CF) lung disease. Through recognition of asialoGM1 receptors on airway epithelial cells, pathogens such as Pseudomonas aeruginosa stimulate translocation of NF-B and transcription of interleukin (IL)-8, the major polymorphonuclear leukocyte (PMN) chemokine, to eliminate bacteria, a process that is upregulated in cells with CF transmembrane conductance regulator (CFTR) mutations (2, 9). Inasmuch as NF-B is a key transcriptional activator, the association between increased NF-B activity and excessive inflammation in CF airways is of potential clinical significance. In normal airways this process is usually self-limited, but in CF it appears to be exaggerated (10). Although NF-B-dependent IL-8 expression has a critical role in eliciting a PMN response to infection, activation of apoptosis may be a mechanism for regulating inflammation, either by eliminating the cells which persist in signaling or by inhibiting NF-B (11). Activation of NF-B not only initiates an inflammatory response but also can block the activation of cell signaling pathways that promote apoptosis in other cell types (12).

There is extensive literature examining the consequences of specific proinflammatory cytokines, particularly tumor necrosis factor (TNF)-, on the induction of apoptosis. An important biologic effect of TNF- is the induction of apoptosis through its interaction with its receptor, TNFR. However, the same TNF-TNFR ligand-receptor interaction can also activate NF-B (5). Activation of the Fas ligand is another major signaling pathway that leads to apoptosis. In bronchial epithelial cells, ligation of Fas stimulates both apoptosis and expression of IL-8 (13). Thus, the activation of proinflammatory signaling pathways can elicit apoptotic responses as well as the expression of cytokines. The response of the normal airway epithelial cell to bacterial infection may consist of activation of both apoptotic and proinflammatory cascades (5, 6). However, in diseases such as CF, in which there appears to be an excessive inflammatory response, the balance between the activation and suppression of these potentially opposing pathways may be abnormal.

The effects of CFTR dysfunction on apoptosis have been studied both in vitro and in clinical specimens. Bronchial and intestinal biopsy specimens from CF patients had more apoptotic cells detected by terminal deoxyribonucleotidyl transferase-mediated deoxyuridine triphosphate (dUTP)-biotin nick-end labeling (TUNEL) assay as compared with controls, and apoptosis was particularly pronounced in areas with high CFTR expression, such as the submucosal cells (14). However, a mouse mammary epithelial cell line (C127) transfected with mutant CFTR was found to be resistant to apoptosis, as compared with cells expressing the wild-type gene (15). The difference in rates of apoptosis was suggested to be due to consequences of Cl channel dysfunction and lack of acidification in the cytoplasmic compartments required for caspase function and proteolytic activity. Although these studies may provide insights into the contribution of CFTR channel activity to apoptosis, murine mammary cells stimulated with cycloheximide may not be an appropriate model to reflect the responses of more highly differentiated, polarized cells that are also activated to express proinflammatory cytokines.

In this report, the effects of P. aeruginosa on the induction of apoptosis in airway epithelial cells were studied. We were especially interested in determining what effects CFTR dysfunction might have on the induction of apoptosis, whether the postulated effects of CFTR mutation on apoptotic pathways are a consequence of a general lack of Cl channel activity or due to the effects of specific types of CFTR mutation, such as accumulation of the F508 mistrafficked CFTR in the endoplasmic reticulum (ER) and subsequent endogenous activation of NF-B. In addition, because bacterial interactions with airway cells are fundamental to the pathogenesis of CF lung disease, we compared several well characterized P. aeruginosa mutants for their ability to induce apoptosis to determine whether the same virulence factors associated with the induction of inflammation are also required to stimulate apoptosis.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Cell Culture

9HTEo cells, a human tracheal epithelial cell line, were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum and were stably transfected with either pCep (a control vector) or pCep-R (providing constitutive expression of the regulatory domain [R-domain] of CFTR), which results in typical CF-like physiology, i.e., failure of Cl secretion in response to cyclic adenosine monophosphate (cAMP) (16) and altered superficial sialylation (9). IB-3 cells (F508/ W1282X), a human bronchial epithelial cell line, and C-38 cells, the corrected line which expresses an episomal copy of a truncated form of CFTR and exhibits normal physiology but lacks the first extracellular domain of CFTR (17), were obtained from P. Zeitlin (Johns Hopkins University, Baltimore, MD). The IB-3 cells used in these studies contain the empty vector. CFT1-LCFSN cells and the CFT1-LC3 cells were obtained from James Yankaskas (University of North Carolina, Chapel Hill, NC). They were grown in serum-free Ham's F-12 media with supplements. CFT1 is a homozygous F508 cell line transformed with a retrovirus vector encoding a full-length wild-type CFTR (LCFSN) or the empty vector (LC3) (18). Human nasal polyp epithelial cells (NHNP) from normal persons were isolated using the protease method and grown in primary culture as described previously (3). Reagents were obtained from Sigma Chemical Co. (St. Louis, MO) unless otherwise specified.

Construction of 16HBE-14o Cells, Sense, and Antisense Cell Lines

16HBE-14o cells (16HBE), a human bronchial epithelial cell line, was provided by D. Gruenert (University of California San Francisco, San Francisco, CA). These cells form tight junctions when grown on vitrogen-coated wells (19). The first 131 nucleotides of human CFTR were isolated by polymerase chain reaction (PCR) and cloned into the pBK-cytomegalovirus vector in the sense and antisense orientations, as confirmed by restriction enzyme analysis and DNA sequencing. 16HBE cells were transfected with these plasmids by calcium phosphate coprecipitation and selected using G418. Colonies were screened for the presence of the vector using the sense primers TTGTAATACGACTCACTATAGGGC and GGCTTTTCCAGAGGCGACCTCTG (fragment 157 base pairs [bp]) and antisense TTGTAATACGACTCACTATAGGGC and AGAGGTCGCCTCTGGAAAAGGCC (fragment 211 bp). PCR-positive clones were tested for the presence of transfected DNA by Southern hybridization of extracted DNA cleaved with PvuII. Clones selected for further expansion had 1 to 10 copies of the plasmid per cell. Transfected cells were subjected to [³⁶Cl] efflux assays, with and without reagents to stimulate cAMP production (1 mM theophylline plus 10 µM forskolin), and monitored for 400 s. Cells transfected with the sense construct had a significant cAMP-stimulated increase in [³⁶Cl] efflux, whereas the cells transfected with antisense did not. Monolayers of antisense cells studied in an Ussing chamber did not display chloride secretion in response to amiloride followed by forskolin (data not shown).

Bacterial Strains and Culture Conditions

P. aeruginosa strains PAO1 (wild-type); PAO/NP (Pil); PAO/ NPfliA (PilFla) (20); PAO1exsA, lacking a type III secretion system (21); and PAO1algC, a mutant lacking lipopolysaccharide (LPS) side chains and a fully glycosylated core (20), were used. PAO1/GFP which constitutively expresses green fluorescent protein (GFP), was constructed by cloning a DNA fragment expressing enhanced GFP (Clontech, Palo Alto, CA) into the polylinker of the P. aeruginosa vector pUCP19 and electroporating PAO1 with the resulting plasmid. Bacteria were grown in luria broth overnight at 37°C with aeration. Aliquots were resuspended in phosphate-buffered saline (PBS) and diluted to give a concentration of 1 × 10⁶ colony-forming units (cfu)/ml to stimulate the epithelial monolayers.

Apoptosis Assays

Confluent epithelial cell monolayers were incubated with 1 × 10⁶ cfu/ml of PAO1 for 1 to 8 h, washed with PBS, and further incubated for 12 to 18 h in the presence of gentamicin (100 to 200 µg/ml). Epithelial cells were then treated with commercially obtained fluorescent markers of apoptosis (22) as indicated, and analyzed with a Becton Dickinson Flow Activated Cell Sorter (FACS Calibur) equipped with a 15-mW, 488-nm, air-cooled argon laser using Cell Quest software.

Effects of the protease inhibitors, N--p-tosyl-L-lysine chloromethyl ketone (TLCK) (10 µM) or N-tosylamido-L-phenylalanine chloromethyl ketone (TPCK) (10 µM) in dimethyl sulfoxide were assayed by preincubating confluent monolayers for 60 min before the addition of PAO1 (1 × 10⁶ cfu/ml).

The effects of ethyleneglycol-bis-(-aminoethyl ether)-N,N'-tetraacetic acid (EGTA) were tested by incubating the monolayers with 50 µM EGTA in Ca²⁺-free epithelial Ringer's solution, pH 7.4, for 1 h before the addition of PAO1 and continuing incubation with bacteria in the presence of EGTA for increasing periods of time.

TUNEL Assay

DNA strand breaks were detected by in situ terminal deoxynucleotidyl transferase (TdT) assay and TUNEL assay (Promega, Madison, WI). After incubation with bacteria, epithelial cells were fixed in 1% paraformaldehyde for 20 min on ice, washed with 1% bovine serum albumin in PBS, and resuspended in 0.1% Triton for 20 min. The cells were then incubated in a mixture containing TdT and dUTP conjugated to fluorescein isothiocyanate for 1 h at 37°C and analyzed by flow cytometry.

Mitochondrial Permeability

Altered mitochondrial membrane permeability was detected using ApoAlert Mitochondrial Membrane Sensor (Clontech), following the manufacturer's instructions. In healthy cells, MitoSensor dye taken up by mitochondria forms aggregates that exhibit red fluorescence. In apoptotic cells, MitoSensor dye remains cytoplasmic and the monomeric form produces green fluorescence. Washed monolayers were trypsinized, suspended in 500 µl of DMEM/F12 plus 5 µg/ml of the mitosensor reagent, incubated at 37°C in 5% CO₂ for 20 min, and analyzed by flow cytometry as described earlier.

Statistical analysis of the results was performed by analyzing the data expressed as a mean ± standard deviation (SD) of the treatment or control groups and using either an unpaired t test with Welch correction or a -square analysis. A two-tailed P value was obtained using Graphpad InStat software, version 3.0.

Confocal and Fluorescence Microscopy

Cells were grown to confluence in a polarized fashion at an air-liquid interface on transwells (Costar, Corning, NY). Media was changed to DMEM/F12 plus 10% fetal calf serum with no antibiotics before stimulation with PAO/GFP (1 × 10⁸ cfu/ml) for 3 h followed by PBS washes. Adherent bacteria were labeled with rabbit antisera to PAO1 outer membrane proteins which were visualized with a secondary goat antirabbit immunoglobulin (Ig) G F(ab')₂ conjugated with Texas Red (1:40). The cells were washed with PBS, fixed with 2% paraformaldehyde, and permeabilized in 0.1% Triton in 5% donkey serum. Tight junctions were labeled with rabbit anti- ZO-1 (1:200) (Zymed, South San Francisco, CA) in 5% donkey serum for 1 h at room temperature, washed, and then incubated with donkey antirabbit IgG conjugated to Cy-5 or Texas Red (Jackson ImmunoResearch, West Grove, PA) (1:100) (23, 24). Cell membranes were also labeled with mouse antifodrin (1:200) (Chemicon, Temecula, CA) for 1 h at room temperature, washed, and incubated with donkey antimouse IgG conjugated to Cy-5 (1:100). Control 16HBE monolayers were compared with monolayers treated with 50 µM EGTA in Ca²⁺-free epithelial Ringers solution, pH 7.4, for 1 h before infection and throughout the incubation period.

Immunohistochemistry

Mice homozygous for the CFTR S489X (null) mutation or wild-type littermate controls (cftr / and +/+) were inoculated with P. aeruginosa impregnated in agar beads as previously described (10). Lungs were harvested under sterile conditions 3 d after inoculation and frozen immediately in liquid nitrogen. Paraffin-embedded 5-µm sections were obtained and treated as described earlier to detect TUNEL-positive cells.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Detection of PAO1-Induced Apoptosis in 9HTEo Cells

Apoptotic 9HTEo cells in confluent monolayers after exposure to PAO1 were identified by labeling with fluorescent markers for cellular events that occur during apoptosis. By flow cytometric analysis, similar populations of cells were identified by the mitochondrial permeability marker after either ultraviolet (UV) irradiation for 60 min or exposure to PAO1 for 6 h (Figure 1A). The apoptotic cells differed in size and complexity from either a control population of normal cells or frankly necrotic cells. The apoptotic cells were unable to exclude propidium iodide but had evidence of nuclear condensation and fragmentation, which differentiated them from the necrotic cell population. Apoptotic cells were also readily visualized by TUNEL labeling and fluorescence microscopy (Figure 1B).

View larger version (13K): [in this window] [in a new window]   View larger version (97K): [in this window] [in a new window]   Figure 1.   Analysis of apoptotic 9HTEo cells by flow cytometry. (A) Using the mitochondrial sensor assay system, apoptotic and control cells were distinguished by flow cytometry. Equivalent wells of cells were exposed to UV light for 60 min to induce apoptosis (solid line), incubated with PAO1 for 6 h (dashed line), or left untreated (control, shaded gray). Apoptotic cells, which are fluorescently labeled by the mitochondrial sensor, are shifted to the right. (B) TUNEL labeling to identify apoptotic cells after exposure to PAO1 for (upper panels) 3 and (lower panels) 6 h. The same field of the 9HTEo monolayers is shown under visible light (left panels) and under fluorescence microscopy (right panels). Original magnification: ×40.

Susceptibility to Apoptosis and the Presence of Tight Junctions

The number of apoptotic 9HTEo cells, a transformed cell line, or NHNP cells, growing in primary culture, were quantified using the TUNEL assay after exposure to PAO1 for increasing periods of time. Although more than 50% of the 9HTEo cells were TUNEL-positive after 8 h of exposure to PAO1, less than 10% of the cells in primary culture had evidence of apoptotic changes (Figure 2). We postulated that the presence of tight junctions, an expected property of the NHNP cells but not the 9HTEo cells, may provide a barrier that prevents bacterial access to epithelial receptors that mediate the apoptotic response. Using confocal microscopy, we assessed the barrier function provided by confluent monolayers of the 9HTEo cells and the NHNP cells, grown in a polarized fashion on semipermeable membranes and exposed to identical inocula of PAO1 (Figure 3A). Significantly greater numbers of bacteria were associated with the 9HTEo monolayer, penetrating below the apical surface and intercalating between adjacent cells, as seen in a confocal image of a section 8 µm below the apical surface of the monolayer. Bacteria were virtually excluded from the NHNP monolayers. ZO-1 was localized along the plasma membrane of the NHNP cells consistent with the presence of tight junctions, whereas the distribution of this marker in the 9HTEo cells was diffuse (23, 24) (Figure 3B).

View larger version (12K): [in this window] [in a new window]   Figure 2.   Kinetics of apoptosis in NHNP and 9HTEo cells. TUNEL labeling and quantification by flow cytometry was used to follow the fraction of epithelial cells that became apoptotic after exposure to PAO1 for increasing amounts of time, as indicated.

View larger version (136K): [in this window] [in a new window]   Figure 3.   NHNP and 9HTEo cells imaged by confocal microscopy. (Top panels) P. aeruginosa association with monolayers. NHNP and 9HTEo cells grown on semipermeable supports are shown after exposure to 1 × 108 cfu/ml of PAO1/GFP for 3 h and imaged 8 µm from the apical surface. The NHNP cells were also treated with anti-ZO1 and a fluorescent Cy-5 tag to delineate the cell boundaries. The 9HTEo cells were treated with antifodrin, which was visualized with Cy-5. Original magnification: ×100. (Lower panels) Demonstration of tight junctions. Monolayers grown under the same conditions as shown in top panels were treated with antibody to ZO-1, a marker for tight junctions. The images shown were 3 µm from the apical surface. Original magnification: ×100.

To further establish that tight junctions are associated with resistance to apoptosis, we tested the susceptibility of 16HBE cells, which have tight junctions when grown in polarized monolayers on vitrogen-coated supports (19). These cells were resistant to PAO1-induced apoptosis, similar to the NHNP cells, under control conditions. However, upon disruption of their tight junctions with 50 µM EGTA, up to 35% of the 16HBE cells were apoptotic in comparison with less than 10% of the cells exposed to the bacteria alone for 6 h (Figure 4A). Loss of tight junctions that accompanied the EGTA treatment was demonstrated by the diffuse pattern of ZO-1 reactivity as compared with the localization of ZO-1 at the apical borders of cells not exposed to EGTA (Figure 4B).

View larger version (15K): [in this window] [in a new window]   View larger version (103K): [in this window] [in a new window]   Figure 4.   EGTA treatment renders 16HBE cells susceptible to apoptosis. (A) Apoptotic cells were identified using the mitochondrial permeability assay quantified by flow cytometry after treatment of confluent 16HBE monolayers with 50 µM EGTA for 1 h and/or infection with PAO1 for 1, 3, or 6 h. Results shown indicate the mean ± SD of triplicate wells. Differences between the EGTA control and the EGTA-plus-PAO1 wells at 6 h were significant (P = 0.01). (B) Confocal images of ZO-1 staining to indicate tight junctions in 16HBE cells (left) under control conditions and (right) after EGTA treatment. Original magnification: ×40.

Inhibition of NF-B Translocation Increases Susceptibility to Apoptosis

Epithelial cell monolayers, with or without tight junctions, are readily stimulated by a number of P. aeruginosa gene products to activate NF-B and transcribe proinflammatory cytokines. Because the activation of NF-B inhibits apoptosis in other cell types, we tested this effect in respiratory epithelial cells. For NF-B translocation to occur, proteins of the IB family must be phosphorylated and proteolytically degraded in the proteosome. The protease inhibitors TLCK and TPCK act as inhibitors of NF-B translocation by blocking the breakdown of IB in the proteosome, an effect we previously demonstrated in these airway epithelial cells (3). The effects of TPCK and TLCK on the induction of apoptosis in epithelial cells stimulated with PAO1 were tested (Figure 5). TLCK and TPCK at 10 µM increased the number of apoptotic cells to 28 and 15% of the population, respectively, as compared with the 5.5% that were apoptotic in the absence of the protease inhibitors. These findings are consistent with data from other cell types, suggesting that inhibition of NF-B increases apoptosis.

View larger version (9K): [in this window] [in a new window]   Figure 5.   Protease inhibitors increase susceptibility to apoptosis. Apoptotic 9HTEo cells after treatment with 10 µM TLCK, TPCK, or a media control for 1 h followed by PAO1 stimulation for 1 h were detected by altered mitochondrial permeability detected by flow cytometry. Differences in apoptosis between the control group treated with PAO alone versus treatment with TLCK and TPCK were statistically significant, with P < 0.0001, as determined by -square analysis.

Susceptibility to Apoptosis Is Equivalent in Matched Cell Lines with and without CFTR Dysfunction

CFTR mutations have been postulated to confer antiapoptotic effects by increasing pH in specific cell compartments impairing the nucleolytic activity of caspases (15). Indirect effects of CFTR mutations associated with increased expression of proinflammatory cytokines and activated NF-B may also diminish apoptosis. To directly test the effects of CFTR mutations on apoptosis, several matched cell lines with well characterized defects in CFTR function were assayed after PAO1 exposure (Figure 6). 9HTEo cells, transfected with a control pCep vector, or pCep-R constitutively expressing the CFTR R-domain which causes a CF-like phenotype (lack of Cl secretion in response to cAMP stimulation), had equivalent rates of apoptosis, approximately 70% of the 10,000 cells screened by flow cytometry using the mitochondrial permeability assay. IB-3 (CF) and C-38 cells corrected with a functional truncated copy of CFTR were similar, with approximately 60% of the cells apoptotic after 3 h of PAO1 exposure. Over 70% of the CFT1 cells, whether complemented with a full-length functional CFTR or with the vector, were apoptotic after 3 h of PAO1 exposure and almost 90% were apoptotic after 6 h of incubation with bacteria. In contrast, 16HBE cells (which form tight junctions) transfected with plasmids expressing CFTR antisense complementary DNA (cDNA) or the control cDNA in the "sense" orientation were resistant to apoptosis, with less than 10% of the population labeled with the fluorescent marker, reflecting loss of mitochondrial permeability. The electrophysiologic properties of these cells reflected the absence of CFTR Cl transport after forskolin stimulation, as compared with the cells transfected with the "sense" control (Figure 7). An NHNP control with similarly negligible rates of apoptosis is shown for comparison. Thus, cells with CFTR dysfunction, whether due to the F508/W1282X or homozygous F508 mutations expected to accumulate in the ER, Cl channel dysfunction due to overexpression of the R-domain with normal CFTR localization on the cell surface, or lack of CFTR expression, were all similarly susceptible to the induction of apoptosis and did not differ significantly from control strains with normal CFTR activity. The 16HBE cells were uniformly resistant to PAO1 induction of apoptosis whether or not CFTR was expressed.

View larger version (25K): [in this window] [in a new window]   Figure 6.   Apoptosis in cell lines with CFTR mutations. Apoptosis was detected by altered mitochondrial permeability and quantified by flow cytometry in the cell lines indicated under control 0 (shaded bar) conditions, or after PAO1 infection for 3 (filled bars) or 6 h (patterned bars).

View larger version (22K): [in this window] [in a new window]   Figure 7.   Properties of 16HBE cells transfected with cftr sense or antisense constructs. Chloride efflux was measured in cells transfected with (A) cftr-sense construct or (B) antisense construct and stimulated with theophylline plus forskolin or ionomycin, as indicated.

Apoptosis in cftr / and +/+ Murine Lungs

To determine whether the results obtained with cell lines correlate with an in vivo situation, we compared lung histology in cftr / and control mice infected with PAO1 in agar beads. TUNEL-positive cells in comparable sections of lung were detected by in situ immunocytochemistry (Figure 8). A positive control, a section of normal lung treated with deoxyribonuclease (DNase) to produce fragmented DNA, is shown to indicate the sensitivity of the immunofluorescent antibody used and illustrate the normal architecture of the lung. The TUNEL-positive columnar epithelial cells lining the bronchi were well delineated in the longitudinal section (Figure 8A). In infected mice, patchy areas of inflammation were evident in the lungs of both wild-type and cftr / lungs. Inflammatory material in the airway lumen and alveolar spaces was diffusely TUNEL-positive, presumably due to the fragmented DNA of degrading PMNs. The columnar epithelial lining of the airways seen in longitudinal section was notably free of TUNEL-positive cells (Figure 8B). We detected TUNEL-positive mucus in the infected normal control mice, but this was limited chiefly to the bronchus and not the alveolar spaces (Figure 8C). Bronchial epithelial cells were not TUNEL-positive. Equivalent amounts of TUNEL-positive lymphoid cells were evident in sections of the spleens of all the animals studied (data not shown). Although more inflammatory material was evident in the airways of the cftr / mice, there were few, if any, apoptotic cells lining the airways visualized.

View larger version (175K): [in this window] [in a new window]   Figure 8.   Immunohistologic screening for apoptosis in infected mice. Paraffin-embedded sections from normal control (cftr +/+) or cftr / mouse lungs were stained for TUNEL-positive cells; images on the left were viewed with visible light and images on the right were viewed by fluorescence microscopy. (Top panels) Positive control sections from a normal control mouse (uninfected) were treated with DNAse and stained for TUNEL-positive cells. (Middle panels) PAO1-infected cftr / murine lung demonstrating fluorescent material within the lumen of a bronchus and in alveolar spaces, but no labeling of the columnar epithelial cells lining the bronchus. (Lower panels) PAO1-infected control cftr +/+ (wild-type) murine lung with inflammatory material within a bronchus that is TUNEL-positive, but no obvious epithelial TUNEL-positive cells. (Original magnification: ×20.)

Expression of Multiple Virulence Factors Is Required to Induce Apoptosis

Having found that susceptibility to apoptosis is related to the resistance of the monolayers to bacterial penetration, we postulated that the expression of specific P. aeruginosa virulence factors, particularly those important in invasion, may induce apoptosis. 9HTEo cells were exposed for 6 h to standardized inocula of PAO1 mutants that lack expression of specific virulence factors involved in the pathogenesis of airway infection (20), and were assayed for markers of apoptosis 18 h later (Figure 9). A nonpiliated mutant (PAO/NP) caused apoptosis in less than 10% of the cells, as compared with 50% of the cells which were apoptotic by either TUNEL or the mitochondrial permeability marker after exposure to wild-type PAO1. Lack of a functional type III secretion system (PAOexsA), or lack of mature LPS (PAOalgC) resulted in minimal rates of apoptosis equivalent to control cells not exposed to bacteria. A superficial stimulus which elicits NF-B translocation, antibody to asialoGM1, did not induce apoptosis by 18 h, nor did TNF-. The secreted extracellular products (culture supernatant) of PAO1 and each of the mutant strains accumulated after 18 h of growth were insufficient to stimulate apoptosis (data not shown).

View larger version (10K): [in this window] [in a new window]   Figure 9.   P. aeruginosa virulence factors required to induce apoptosis. (A) 9HTEo cells were incubated for 6 h with an inoculum of 1 × 106 cfu/ml of each of the PAO1 mutants indicated, with antibody to asialoGM1, or with TNF- (10 ng/ml) for 12 h. Apoptotic cells were enumerated 18 h later by flow cytometry using either mitochondrial permeability (black bars) or TUNEL labeling (grey bars). (For descriptions of cells examined, see RESULTS.)

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Mucosal epithelial cells can respond to bacteria by several different mechanisms (25, 26). Airway epithelial cells have been characterized extensively for their proinflammatory response to specific bacterial virulence factors, and necrosis of epithelial cells as a consequence of bacterial toxins is also well established (27). The ability of P. aeruginosa to induce apoptosis in airway epithelial cells was found to be dependent upon properties of both host epithelial cells and the bacteria. Normal polarized airway epithelial cells in primary culture were resistant to apoptosis, even after prolonged (6 h) exposure to bacteria. This may reflect the inherent properties of the airway epithelium, which is a stable mucosal surface with relatively low rates of cell proliferation under normal conditions (8). Human airway epithelial cells are capable of proliferation after damage, and regeneration occurs promptly, usually within 48 h (26). However, exposure of airway epithelial cells to an inoculum of P. aeruginosa sufficient to activate proinflammatory cytokine expression did not activate apoptosis in cells that form confluent monolayers with tight junctions.

Airway epithelial cells appear to differ in this respect from the epithelial cells lining other mucosal surfaces, such as the genitourinary (28) or gastrointestinal (GI) epithelium. The gut mucosa is regularly regenerated through an orderly progression of cell differentiation from basal cells deep within crypts to the fully differentiated cells at the tips of the villi which are shed and replaced (7). Although many pathogenic bacteria trigger apoptosis in GI cells (25), this process seems much more limited in the lung. This may reflect the normal functions of these respective mucosae, as GI epithelial cells are constantly exposed to solid lumenal contents whereas the airway cells line an empty conduit for gas exchange. Apoptosis may be a normal component of gut regeneration but represent a pathologic response in the lung.

The susceptibility of respiratory epithelial cells to apoptosis was at least partially dependent upon the integrity of tight junctions. The 9HTEo cells that lack tight junctions were readily susceptible to apoptosis after exposure to wild-type P. aeruginosa PAO1. In contrast, 16HBE cells and the NHNP airway cells, which retain tight junctions, were highly resistant to apoptosis unless these junctions were disrupted. Confocal images of the bacterial-epithelial interaction suggest that many more organisms gain access to the relevant epithelial binding sites on 9HTEo cells and interact with basolateral as well as apical receptors. Other investigators have demonstrated significantly increased P. aeruginosa-epithelial interactions in cell culture systems that do not form tight junctions (29).

The resistance of airway epithelial cells to apoptosis may be due, in part, to the stimulation of NF-B by adherent P. aeruginosa. NF-B is an important transcriptional activator that, in addition to initiating transcription of proinflammatory cytokines, also regulates expression of genes that inhibit cell death (4, 5, 11, 13). NF-B appears to have an antiapoptotic effect in respiratory cells. By blocking proteolytic degradation of IBs in the proteosome, the rate of apoptosis in the 9HTEo cells was increased. Inasmuch as apical stimulation of airway cells with several different P. aeruginosa ligands is sufficient to activate the NF-B response (3), this is likely to have an antiapoptotic effect in respiratory epithelial cells. This is teleologically appealing as airway cells express cytokines to induce a PMN response to eradicate bacteria. Induction of apoptosis would limit further PMN signaling, and loss of these barrier cells would expose basolateral surfaces of the epithelium and increase the risk of bacterial invasion. An alternative consequence of epithelial cytokine production would be the activation of apoptosis through an autocrine circuit. However, TNF-, the cytokine often associated with apoptotic responses and present in the milieu of infected airways, did not stimulate apoptosis, suggesting that pulmonary epithelial cells have a highly selective response to cytokine signals.

Several lines of experimental evidence imply that CFTR mutations should prevent apoptosis. Activation of NF-B, which is antiapoptotic, is a common characteristic of cells with CFTR mutation attributed to both exogenous stimulation by adherent bacteria and to cell stress, perhaps associated with the accumulation of mutant CFTR in the ER (2). In addition, Gottlieb and Dosanjh suggested that Cl channel dysfunction and impaired acidification inhibit caspase activity and protect mammary epithelial cells with CFTR dysfunction from apoptosis (15). Moreover, the induction of apoptosis in GI cells is dependent upon expression of nitric oxide (7), which is impaired in cells with CFTR mutations (30); all factors which might inhibit apoptosis in CF cells. The few clinical studies of apoptosis in biopsies from patients with CF are descriptive and lack matched normal controls, making it difficult to sort out effects of chronic infection, inflammation, and epithelial destruction from those due to mutant CFTR (14, 31). However, our studies using respiratory epithelial cell lines with CFTR dysfunction due to several different mechanisms did not indicate differences in apoptosis in CF cells as compared with normal controls. Mistrafficking of F508/ W1282X or F508 CFTR did not prevent apoptosis after PAO1 infection, suggesting that localization of normal CFTR on the cell surface (32) is not involved in the induction of apoptosis. Lack of CFTR Cl channel function due to overexpression of the R-domain in 9HTEo cells which otherwise have normal trafficking of CFTR and the expected "CF-like" properties, including increased bacterial adherence and IL-8 expression (9), did not affect rates of apoptosis as compared with controls. Although the 9HTEo, CFT1, IB-3, and C-38 cells had sufficient rates of apoptosis to detect a protective effect conferred by CFTR mutation, the 16HBE cell lines were so resistant to apoptosis that any protection associated with CFTR would be negligible. Data obtained from infected cftr / and +/+ mice, while not quantitative, was consistent with the in vitro studies; few apoptotic cells were detected in the airway epithelia, and no appreciable differences were found between the cftr / and +/+ mice. Overall, our studies suggest that factors other than CFTR mutation predominate in determining what cellular events are activated after bacterial infection.

Airway epithelial cells are clearly capable of undergoing apoptosis when sufficiently stimulated by P. aeruginosa, and our data suggest that exposure of basolateral surfaces of the airway epithelial cell to bacteria is involved in this response. Cell dissociation is a potent stimulus for apoptosis inasmuch as the loss of anchoring to cell matrix components or interruption of cell-cell contact are important signals to regulate cell proliferation (33). In end-stage CF lung disease, epithelial integrity is disrupted and there are likely to be significant numbers of apoptotic cells, as found in CF lungs examined at the time of lung transplant (31). However, even cells that lack tight junctions but still maintain close cell-to-cell contacts are still relatively resistant to apoptosis. To induce apoptosis in 9HTEo monolayers, intact bacteria expressing pili, flagella, and mature LPS were required. In addition, as has been shown for the intracellular pathogens Yersenia and Shigella, the products of the P. aeruginosa type III secretion system were also necessary to initiate apoptosis. The P. aeruginosa exoenzymes S and T are delivered into eukaryotic cells by the type III apparatus after pilin-dependent binding (33). These cytotoxins are bifunctional adenosine diphosphate ribosylating enzymes that interact with small molecular weight guanidine triphosphatases such as Rho, Rac, and Cdc-42 (34), which are important in the maintenance of cytoskeletal components and tight junctions in polarized epithelia. Loss of cell-cell anchoring or contact with matrix elements is a major stimulus for apoptosis preventing metastatic growth of dissociated cells. The ability of these P. aeruginosa toxins to disrupt actin polymerization and dissociate cell-cell contacts is a likely cause of apoptosis (35). PAO1 culture supernatant, despite the presence of potent exoenzymes, was insufficient to activate apoptosis. The experimental data are consistent with a requirement for the type III secreted toxins and intimate bacterial-epithelial contact to initiate an apoptotic response to bacterial infection. Apoptosis is not simply a result of exoproduct toxicity.

P. aeruginosa induction of apoptosis in the airway epithelium appears to be a highly regulated process activated only by the coincidence of specific epithelial and bacterial factors. Fundamental properties of the respiratory tract, the maintenance of unobstructed airways to facilitate the gas exchange, mitigate against the routine sloughing of airway cells as a general response to lumenal bacteria. Unlike other mucosal barriers that activate proinflammatory signaling cascades and apoptotic pathways immediately after exposure to bacterial pathogens, airway cells are infrequently shed in response to P. aeruginosa infection. It remains to be established what role, if any, apoptosis may play in regulating NF-B-dependent activation of cytokine expression early in the course of infection. Instead, our data suggest that apoptosis occurs only after significant loss of epithelial integrity, and even with loss of tight junctions requires the additional insult of bacterial toxins that further disrupt the normal cytoskeletal architecture.