Bordetella bronchiseptica Flagellin Is a Proinflammatory Determinant for Airway Epithelial Cells

Cell culture, bacteria, and other reagents.Normal human primary bronchial epithelial (NHBE) cells were obtained from Clonetics (Walkersville, MD) and grown in bronchial epithelial supplemented growth medium (Clonetics). BEAS-2B immortalized human bronchial epithelial cells were obtained from the American Type Culture Collection (ATCC) (Manassas, VA) and grown in the same medium (Clonetics). The human lung carcinoma cell lines Calu-3 and A549 (ATCC) were routinely maintained in RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS). All cells were routinely grown in the absence of antibiotics. Human (HEK293-hTLR-5), and mouse (HEK293-mTLR-5) TLR-5-expressing HEK293 cells and vector-transfected (HEK293-null) cells were obtained from InvivoGen (San Diego, CA) and grown in Dulbecco modified Eagle medium (DMEM) supplemented with 10% FBS and 10 μg/ml blasticidin (InvivoGen). Stably transfected RAW 264.7 cells expressing a murine TLR-5-enhanced yellow fluorescent protein (mTLR-5-EYFP) (63) were kindly provided by S. Mizel (Wake Forest University School of Medicine, Winston-Salem, NC) and grown in DMEM supplemented with 10% FBS and 200 μg/ml G-418 (Sigma). B. bronchiseptica strains used in this work were kindly provided by Jeff Miller (UCLA School of Medicine, Los Angeles, CA) (Table 1). B. bronchiseptica RB50, a wild-type strain (isolated from a New Zealand White rabbit) which alternates among different phases in response to environmental stimuli (9), is the parental strain of the B. bronchiseptica strains used in this work. Strain RB54 (bvgSΔ54) is locked in the Bvg⁻ phase as a result of an in-frame deletion in bvgS (9). The isogenic flagellin mutant RB54ΔflaA contains a deletion in the flagellin structural gene flaA (10). Strain REV1 (frl^r [superscrpt “r” designates reversal of expression of frl locus; reference 1]) expresses frlAB (an activator of flagellar genes) from the fhaB promoter resulting in constitutive expression of flagella in the Bvg⁺ phase (1). The nonflagellated strain RBA5 (frl^r-ΔflaA) has an in-frame deletion in the flaA gene (1). The nonmotile strain RBA2 (ΔfrlAB) has a deletion in frlAB (1). Strain RB53 (bvgS-C3) is locked in the Bvg⁺ phase, and thus, it does not express flagellin (9). Strain WD3 (ΔbscN) is a type III secretion mutant, with an in-frame deletion in bscN, a putative ATPase required for the secretion process (68, 69). B. pertussis Bp536 and its Bvg⁻ phase-locked derivative strain Bp537 (52) and B. parapertussis 12822 (22) have been previously described. B. bronchiseptica, B. pertussis, and B. parapertussis were maintained on Bordet-Gengou (BG) agar, containing 7.5% defibrinated sheep blood for the determination of colony morphology and hemolytic activity. Liquid cultures were grown in Stainer-Scholte broth, supplemented with heptakis(2,6-di-O-methyl-)-β-cyclodextrin (Sigma) for B. pertussis. BvgAS activity in B. parapertussis was modulated by 40 mM MgSO₄. Salmonella enterica serovar Typhimurium 14028 and Pseudomonas aeruginosa 10145 were obtained from ATCC and were routinely grown in 3% tryptic soy broth. Gentamicin, polymyxin B, fetal bovine serum, and chemicals were obtained from Sigma Chemical Company (St. Louis, MO). The monoclonal antibody 15D8 (IGEN International, Gaithersburg, MD), directed against an epitope in E. coli flagellin, was used to detect Bordetella flagellin (2, 3).

Infection of epithelial cells for chemokine and cytokine analyses.Epithelial cells were seeded onto six-well plates and grown to about 95% density. Monolayers were infected with 2 × 10⁸ CFU of each bacterial strain (corresponding to a multiplicity of infection [MOI] of 100) in 1 ml of RPMI 1640 medium without serum or antibiotics and incubated at 37°C for a period of 60 min. Conditioned medium samples from the 60-min infection were collected, centrifuged at 10,000 × g for 10 min to remove debris, concentrated 10-fold by lyophilization, and analyzed by Western blotting with flagellin-specific antibodies as described below. Following the 1-h infection, cells were washed extensively with phosphate-buffered saline and further incubated in RPMI 1640 medium supplemented with 10% FBS and 100 μg/ml gentamicin for 24 h. Conditioned media were then collected for the analysis of cytokine and chemokine secretion by enzyme-linked immunosorbent assays (ELISAs) with Quantikine reagents (R&D Systems, Minneapolis, MN) according to the manufacturer's instructions. Data are reported as means ± standard deviations. All determinations of cytokine and chemokine secretion were done in duplicate and repeated in at least three independent experiments. Cytotoxicity was determined as described previously (8), using a lactate dehydrogenase-based in vitro toxicology kit (Sigma) according to the manufacturer's instructions.

Host defense gene expression analysis.Calu-3 human lung epithelial cells were seeded onto six-well plates and grown to about 95% density. Epithelial monolayers were infected for 60 min and then washed as described above. Cells were subsequently incubated in RPMI 1640 medium containing 10% FBS and 100 μg/ml gentamicin, and total RNA was prepared with RNAzol B (Tel-Test, Inc., Friendswood, TX) at 6 h postinfection. Northern hybridization for matrilysin and glyceraldehyde-3-phosphate dehydrogenase mRNAs was done as described previously (32). For the expression analysis of human β-defensins, total RNA samples were reverse transcribed using random hexamer primers (Perkin-Elmer, Branchburg, NJ), and cDNAs were then amplified by PCR as described previously (31). The sizes of the amplified products for hBD2 and hBD1 are 241 and 258 bp, respectively. Reactions were analyzed on 6% acrylamide gels.

Purification of flagellin and protein sequencing.Flagellin was purified from overnight culture supernatants of B. bronchiseptica RB54, P. aeruginosa (ATCC 10145), and S. enterica serovar Typhimurium (ATCC 14028), as described previously (32). Flagellin was analyzed by electrophoresis on 12% sodium dodecyl sulfate (SDS)-polyacrylamide gels and stained with Coomassie blue. Flagellin was further purified by using polymyxin B beads (Sigma), and the removal of endotoxin was verified as described previously (8). For the N-terminal sequencing of flagellin, purified protein was visualized by staining with 0.1% Coomassie brilliant blue, excised, and sequenced by automated Edman degradation using a PE Applied Biosystems 492 sequencer as described previously (31).

Immunoblotting.Conditioned medium samples were separated on 12% SDS-polyacrylamide gels, transferred by semidry electrophoretic transfer to nitrocellulose membranes (Hybond ECL; Amersham Pharmacia Biotechology, Buckinghamshire, United Kingdom), and probed with the antiflagellin monoclonal antibody 15D8 as described previously (31).

Statistical analysis.For statistical comparisons, data on chemokine and cytokine expression were analyzed by analysis of variance and Bonferroni-type multiple t test. A P value of <0.01 was considered significant.

Effect of flagellin in the development of proinflammatory responses in human lung epithelial cells exposed to Bordetella bronchiseptica.To examine the role of B. bronchiseptica flagellin in the development of proinflammatory and host defense responses in the airways, we used a model of coinfection of epithelial cells and bacteria (8, 31, 32) and studied the behavior of selected markers of inflammation regulated by flagellin (48). In particular, we compared cytokine and chemokine secretion by human lung epithelial cells exposed to isogenic flagellated and nonflagellated B. bronchiseptica strains. To determine the role of flagellin in the absence of Bvg⁺ phase-specific factors, we exposed cells to the Bvg⁻ phase-locked motile strain RB54 and the isogenic flagellin mutant RB54ΔflaA (9, 10). As shown in Fig. 1A, secretion of the chemokine macrophage inflammatory protein 3α (MIP-3α) (CCL20) by BEAS-2B cells was up-regulated compared to uninfected cells in response to exposure to RB54 but not RB54ΔflaA. A similar result was obtained when we determined IL-8 (CXCL8) secretion by A549, NHBE, BEAS-2B, and Calu-3 cells exposed to B. bronchiseptica (Fig. 1B and C; see Fig. 3C; also data not shown), suggesting that flagellin expression by the bacterium induces these inflammatory markers. Secretion of IL-8 by NHBE cells was reduced compared to the other cell lines examined (Fig. 1), in agreement with previous reports (64), but a similar significant reduction in IL-8 secretion was observed in response to nonmotile bacteria (Fig. 1C). To determine the effect of flagellin expressed ectopically in the Bvg⁺ phase, we used REV1 (frl^r), a strain engineered to express flagella constitutively in a wild-type background (1). As shown in Fig. 1A and B, REV1 induced the secretion of proinflammatory markers by human lung epithelial cells to an extent similar to RB54, while exposure to the isogenic flagellin mutant RBA5 (frl^r::ΔflaA) had no significant effect (Fig. 1A and B; also data not shown). No significant induction of cytokines and chemokines was observed in response to exposure to RBA2 (a nonmotile ΔfrlAB deletion mutant [1]) or the nonflagellated Bvg⁺ phase-locked strain (RB53) in any of the human lung epithelial cell lines mentioned above (Fig. 1 and data not shown). Acute exposure to nonflagellated B. pertussis and B. parapertussis did not induce significant IL-8 secretion by A549 cells (Fig. 1B), consistent with previous reports (5). Finally, we used strain WD3 (ΔbscN), a B. bronchiseptica type III secretion mutant (69). Type III secretion has been shown to modulate signaling pathways in host cells (51, 68). As shown in Fig. 1A and B, WD3 induced the secretion of MIP-3α and IL-8 by human lung epithelial cells to an extent similar to the parental strain RB50, suggesting that factors secreted by the type III secretion system are not critical for these responses. Note that when RB50 and its isogenic derivative WD3 are grown in liquid culture, these strains can switch to the Bvg⁻ phase and express flagellin (Fig. 2), resulting in the induction of proinflammatory mediators in lung epithelial cells (Fig. 1). In contrast, we determined that RB50 and WD3 grown on solid agar media did not express flagellin and did not induce proinflammatory markers (data not shown). Altogether, these data show that flagellin expression by B. bronchiseptica results in the induction of a proinflammatory response in a variety of human lung epithelial cell lines and human primary airway cells. In this experimental design, no significant cytotoxicity due to B. bronchiseptica infection was observed at 6 and 24 h postinfection, the time points at which host responses were analyzed (data not shown).

• Open in new tab
• Download powerpoint

FIG. 1.

Bordetella bronchiseptica flagellin is a critical determinant of MIP-3α and IL-8 secretion by human lung epithelial cells. (A) BEAS-2B human bronchial epithelial cells were exposed to B. bronchiseptica RB54, RB54ΔflaA, REV1, RBA5, RBA2, RB53, WD3, and RB50 for 1 h at a multiplicity of infection of 100. After extensive washes and the addition of fresh media containing antibiotics, cells were incubated for 24 h. Conditioned media were collected at 24 h postinfection, and the secretion of MIP-3α was examined by an ELISA. Cntl, control (uninfected cells). The response to flagellated and nonflagellated strains was statistically significantly different (P < 0.001) (*); the response to RB50 and WD3 was not significantly different (**). (B) A549 cells were exposed to B. bronchiseptica, B. pertussis Bp536 and Bp537 (isogenic Bvg⁻ derivative), and B. parapertussis 12822 (Bpp) grown in the absence or presence of 40 mM MgSO₄ (Bvg⁻ phase condition), as described above for panel A. Conditioned media were collected at 24 h postinfection, and the secretion of IL-8 was examined by an ELISA. Differences between the response to motile and nonmotile strains were statistically significant (P < 0.001) (*); the response to RB50 and WD3 did not differ significantly (**). (C) Normal human bronchial epithelial cells were exposed to B. bronchiseptica as described above for panel A. Conditioned media were collected at 24 h postinfection, and the secretion of IL-8 was examined by an ELISA. The response to RB54 and RB54ΔflaA was significantly different (P < 0.01) (*).

• Open in new tab
• Download powerpoint

FIG. 2.

Expression of flagellin by Bordetella bronchiseptica. (A) Conditioned media from epithelial cells exposed for 1 h to B. bronchiseptica, B. pertussis, and B. parapertussis were collected, centrifuged to eliminate debris, and resolved by SDS-polyacrylamide gel electrophoresis (SDS-PAGE). Flagellin was detected by Western blotting with a monoclonal antibody directed against Escherichia coli flagellin. A single band of approximately 40 kDa was detected in conditioned media of cells exposed to B. bronchiseptica RB54, REV1, RB50, and WD3. No band was detected in conditioned media of cells exposed to B. bronchiseptica RB54ΔflaA, RBA5, RBA2, RB53, B. pertussis Bp536 and Bp537, and B. parapertussis (Bpp). The positions of molecular mass markers (in kilodaltons) are shown to the left of the gels. Cntl, control (uninfected cells). (B) Samples of a cell lysate and purified flagellin from the motile strain RB54 were separated by SDS-PAGE and stained with Coomassie blue. The positions of molecular mass markers (in kilodaltons) are shown to the left of the gel. A unique protein of approximately 40 kDa was present in the purified flagellin preparations. (C) Aliquots of RB54 cell lysate and purified flagellin shown in panel B were resolved by SDS-PAGE, transferred to nitrocellulose filters, and probed by Western blotting with a monoclonal antibody against E. coli flagellin. A unique reactive band of approximately 40 kDa was detected. The positions of molecular mass markers (in kilodaltons) are shown to the left of the gel.

Expression of flagellin by B. bronchiseptica during infection.Consistent with the hypothesis that flagellin triggers the inflammatory response observed in human lung epithelial cells, comparable amounts of flagellin were detected in the conditioned media of the 1-h period of infection by strains RB54, REV1, RB50, and WD3 (Fig. 2A). As expected, no signal was detected in conditioned media of cells exposed to the flagellin-deficient, nonmotile strains RB54ΔflaA, RBA5, RBA2, and RB53 and B. pertussis and B. parapertussis (Fig. 2A). The motility phenotype was confirmed in standard soft-agar motility assays for all B. bronchiseptica strains and correlated with the presence or absence of flagellin as detected by Western blotting of conditioned media and bacterial cell lysates (data not shown). As mentioned before, RB50 and WD3 grown on solid agar media did not express flagellin (data not shown).

B. bronchiseptica flagellin induces host defense gene expression in human lung epithelial cells.In order to examine the role of flagellin in the pathogenesis of B. bronchiseptica, we purified this protein from strain RB54 (32). A unique band migrating at about 40 kDa, consistent with the expected mass of B. bronchiseptica flagellin (1, 46), was detected by Coomassie blue staining in our purified preparations (Fig. 2B), and its identity was further confirmed as flagellin by Western blotting (Fig. 2C) and N-terminal sequencing. We exposed Calu-3 human airway epithelial cells to purified flagellin, as well as to strains RB54 and RB54ΔflaA. As shown in Fig. 3A, expression of the inducible antimicrobial peptide hBD2 was up-regulated by exposure to flagellin and RB54, but not to the flagellin mutant RB54ΔflaA. hBD1, which is constitutively expressed by epithelial cells (31, 44), was used as an internal control. Furthermore, the expression of matrilysin, a matrix metalloproteinase involved in host defense (66), which is specifically up-regulated by P. aeruginosa flagellin (32), was also induced by Bordetella flagellin and RB54, but not by RB54ΔflaA (Fig. 3B). In similar experiments, conditioned media were harvested at 24 h postinfection, and selected proinflammatory mediators were examined by ELISAs. The secretion of IL-8, granulocyte-macrophage colony-stimulating factor (GM-CSF), and MIP-3α was induced by treatment with purified flagellin to an extent similar to exposure to RB54 (Fig. 3C, D, and E). On the basis of the amount of flagellin detected by Western blotting in conditioned medium samples of infected cells, direct exposure to B. bronchiseptica RB54 for 1 h at an MOI of 100 was roughly equivalent to treatment with 10⁻⁷ M purified flagellin (data not shown). Altogether, these results indicate that B. bronchiseptica flagellin is an important determinant of expression of these host defense genes and a modulator of airway mucosal defense.

• Open in new tab
• Download powerpoint

FIG. 3.

Effect of Bordetella bronchiseptica flagellin on host defense gene expression. (A) Calu-3 human airway epithelial cells were exposed to purified B. bronchiseptica flagellin (10⁻⁷ M). Infection with B. bronchiseptica RB54 and RB54ΔflaA for 1 h at a multiplicity of infection of 50 was used as a control (Cntl). Total RNA was prepared at 6 h postinfection, and the expression of human β-defensins hBD2 and hBD1 was examined by reverse transcription-PCR. (B) Calu-3 cells were treated with purified B. bronchiseptica flagellin (10⁻⁷ M) or infected as described above for panel A, and expression of the matrix metalloproteinase matrilysin (MMP-7) was examined by Northern blotting with a specific probe at 6 h postinfection. MAT, matrilysin. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as a housekeeping control gene. (C, D, and E) Calu-3 cells were exposed to purified B. bronchiseptica flagellin (10⁻⁷ M) or infected with B. bronchiseptica RB54 and RB54ΔflaA as described above for panel A. Conditioned media were collected 24 h after treatment or infection, and the secretion of IL-8, GM-CSF, and MIP-3α was examined by an ELISA. Differences between the response to RB54 and RB54ΔflaA were statistically significant (P < 0.001) (*).

Bordetella flagellin is similar in potency to other bacterial flagellins.We compared the effects of flagellin purified from Bordetella, Pseudomonas, and Salmonella species on IL-8 secretion by human lung epithelial cells. In these experiments, Calu-3 cells were challenged for 24 h with different concentrations of endotoxin-free flagellin, and the accumulation of IL-8 in the conditioned media was determined by ELISA. As shown in Fig. 4, treatment with 10⁻⁸ M Bordetella flagellin resulted in an approximately 20-fold induction in IL-8 levels, while the same concentration of S. enterica serovar Typhimurium flagellin resulted in a 24-fold induction. Consistent with recent observations (4, 24), P. aeruginosa flagellin showed slightly reduced bioactivity (about 10-fold induction in IL-8 secretion with exposure to 10⁻⁸ M flagellin). Therefore, Bordetella flagellin bioactivity in human airway epithelial cells is comparable to that of flagellin from other gram-negative pathogens.

• Open in new tab
• Download powerpoint

FIG. 4.

Bioactivity of Bordetella bronchiseptica flagellin is comparable to those of other bacterial flagellins. Calu-3 human airway epithelial cells were challenged with the indicated concentrations (M) of purified B. bronchiseptica (Bb), P. aeruginosa (Pa), and S. enterica serovar Typhimurium (St) flagellin. Conditioned media were collected at 24 h postchallenge, and the secretion of IL-8 was examined by an ELISA. Cntl, control (uninfected cells).

Bordetella flagellin signals through human and mouse TLR-5. Bordetella spp. that are human pathogens (B. pertussis and B. parapertussis) are nonmotile, while the species with a broad host range, B. bronchiseptica, modulates motility through BvgAS activity. To determine whether the motility-host restriction dichotomy among bordetellae relates to flagellin receptor specificity, we used an in vitro system of TLR-5 receptors expressed on a human epithelial cell line with intrinsic low levels of expression of most TLRs. Thus, we infected human epithelial kidney cells (HEK293) stably transfected with human (HEK293-hTLR5) and mouse (HEK293-mTLR5) TLR-5 with B. bronchiseptica and determined the levels of IL-8 in conditioned media 24 h postinfection (Fig. 5A). For a control, we used HEK293 cells transfected with vector alone (HEK293-null). As shown in Fig. 5A, infection of HEK293-hTLR5 and HEK293-mTLR5 cells with motile strains RB54 and REV1 resulted in similar induction in IL-8 secretion compared to uninfected cells. Furthermore, there was a very limited response to the flagellin mutants RB54ΔflaA and RBA-5 and the nonmotile strain RBA-2 (Fig. 5A). In fact, IL-8 secretion in response to the flagellin mutants was not significantly different from the low level of IL-8 induction observed in HEK293-null cells exposed to B. bronchiseptica (Fig. 5A). This response of HEK293-null cells may be mediated by intrinsic low levels of expression of other receptors in the parental cells (57). In addition, transfected cells were treated for 24 h with 10⁻⁸ M endotoxin-free flagellin purified from B. bronchiseptica and S. enterica serovar Typhimurium, and secretion of IL-8 was examined by ELISA as before (Fig. 5B). Results show that treatment with Bordetella flagellin induces IL-8 secretion to a similar extent in cells expressing human and mouse TLR-5, suggesting that it signals effectively through both receptors. Furthermore, Bordetella flagellin bioactivity in these cells was comparable to Salmonella flagellin bioactivity.

• Open in new tab
• Download powerpoint

FIG. 5.

Bordetella bronchiseptica flagellin signals through both human and mouse TLR-5. (A) HEK293-null, HEK293-hTLR5, and HEK293-mTLR-5 epithelial cells were exposed to B. bronchiseptica RB54, RB54ΔflaA, REV1, RBA5, and RBA2 as described in the legend to Fig. 1. After extensive washes and the addition of fresh media containing antibiotics, cells were incubated for 24 h. Conditioned media were collected at 24 h postinfection, and the secretion of IL-8 was examined by an ELISA. Cntl, control (uninfected cells). The response to flagellated and nonflagellated strains was statistically significantly different (P < 0.001) (*). (B) HEK293-null, HEK293-hTLR5, and HEK293-mTLR-5 epithelial cells were treated with 10⁻⁸ M endotoxin-free purified flagellin from B. bronchiseptica (Bb) and S. enterica serovar Typhimurium (St), and the secretion of IL-8 was examined by an ELISA in 24-h conditioned media. Cntl, control (untreated cells).

Proinflammatory effects of Bordetella exposure in macrophages.To investigate the contribution of TLR-5-dependent signaling to the proinflammatory effects of Bordetella exposure, we used a system of stably transfected RAW 264.7 murine macrophages expressing this receptor (63). Control RAW 264.7 cells (which respond to TLR-2 and -4 ligands but do not express TLR-5 and do not respond to flagellin [40]) and cells expressing TLR-5 (TLR-5-EYFP) were infected with B. bronchiseptica, and cytokine levels were determined in 4-h conditioned media. As shown in Fig. 6A and B, exposure of TLR-5-expressing RAW cells to motile strains RB54 and REV1 resulted in a three- to fivefold increase in tumor necrosis factor alpha (TNF-α) and MIP-2 secretion over the isogenic flagellin mutants RB54ΔflaA and RBA-5. By contrast, exposure of control RAW cells to B. bronchiseptica resulted in increased cytokine levels that were not significantly different between motile strains and flagellin mutants. For a control, cells were incubated with purified Bordetella flagellin, and the levels of TNF-α were determined in conditioned media (Fig. 6C). Treatment of TLR-5-expressing cells, but not control cells, resulted in a 10-fold increase in TNF-α. Altogether, these results confirm the importance of flagellin in the development of proinflammatory responses to B. bronchiseptica infection.

• Open in new tab
• Download powerpoint

FIG. 6.

Effects of Bordetella infection on mouse macrophages. (A and B) Control RAW 264.7 and TLR-5-expressing RAW 264.7 (TLR-5-EYFP) cells were infected with B. bronchiseptica at an MOI of 10 for 30 min. Following infection, cells were washed extensively and incubated in fresh media containing 100 μg/ml gentamicin. TNF-α (A) and MIP-2 (B) levels were determined by ELISAs in 4-h conditioned media and corrected by cell number. Data are expressed as the percentage of TNF-α and MIP-2 secreted by cells exposed to each of the motile strains. Cntl, control (uninfected cells). The response of TLR-5-expressing RAW 264.7 cells to flagellated and nonflagellated strains was significantly different (P < 0.05) (*). TNF-α secretion by control RAW 264.7 cells exposed to the different B. bronchiseptica strains did not vary significantly (**). (C) Cells were treated with 10⁻⁹ M purified flagellin, and TNF-α levels in 4-h conditioned media were analyzed by ELISAs. Data are expressed as the increase in induction over untreated cells. Cntl, control (no addition); Bb, B. bronchiseptica flagellin.