Abstract
The surface of the airway epithelium represents a battleground in which the host intercepts signals from pathogens and activates epithelial defenses to combat infection. Wound repair is an essential function of the airway epithelium in response to injury in chronic airway diseases, and inhaled pathogens such as Pseudomonas bacteria are implicated in the pathobiology of several of these diseases. Because epidermal growth factor receptor (EGFR) activation stimulates wound repair and because LPS activates EGFR, we hypothesized that LPS accelerates wound repair via a surface signaling cascade that causes EGFR phosphorylation. In scrape wounds of NCI-H292 human airway epithelial cells, high concentrations of LPS were toxic and decreased wound repair. However, lower concentrations of LPS accelerated wound repair. This effect was inhibited by treatment with a selective inhibitor of EGFR phosphorylation (AG 1478) and by an EGFR neutralizing Ab. Metalloprotease inhibitors and TNF-α-converting enzyme (TACE) small interfering RNA inhibited wound repair, implicating TACE. Additional studies implicated TGF-α as the active EGFR ligand cleaved by TACE during wound repair. Reactive oxygen species scavengers, NADPH oxidase inhibitors, and importantly small interfering RNA of dual oxidase 1 inhibited LPS-induced wound repair. Inhibitors of protein kinase C isoforms αβ and a TLR-4 neutralizing Ab also inhibited LPS-induced wound repair. Normal human bronchial epithelial cells responded similarly. Thus, LPS accelerates wound repair in airway epithelial cells via a novel TLR-4→protein kinase C αβ→dual oxidase 1→reactive oxygen species→TACE→TGF-α→EGFR phosphorylation pathway.
When bacteria are inhaled and deposit in the airways, the epithelium responds, initiating a dynamic interaction between host and pathogen (1, 2). Thus, the surface of the airway epithelium can be considered a battleground in which the host intercepts signals from pathogens and activates epithelial defenses to combat infection. Airway injury damages the protective barrier of the airway epithelium, assisting invasion by pathogens. The host innate immune response involves a complex signaling cascade that releases cytokines to activate inflammation involving the following: 1) the recruitment of neutrophils that kill bacteria, 2) mucin production and subsequent mucociliary clearance, and 3) repair of epithelial cell damage (3, 4).
Gram-negative bacteria, especially Pseudomonas aeruginosa, are important in the pathophysiology of several chronic airway diseases, including cystic fibrosis (5), bronchiectasis (6), and chronic obstructive pulmonary disease exacerbations (7). An important virulence factor for this bacterium is P. aeruginosa LPS, which has a receptor on the airway epithelial surface, TLR-4 (8, 9). TLR-4 is a member of a family of receptors that identify pathogen-associated molecular patterns to signal a variety of innate immune responses (10, 11). Damage to the airway epithelium occurs in various chronic airway diseases (12, 13), and repair of the injured epithelium has been shown to involve activation of the epidermal growth factor receptor (EGFR)3 (14, 15), which is located on the airway epithelial surface (16). P. aeruginosa and its virulence factors are known to be toxic and to damage the airway epithelium (17, 18, 19), inhibiting wound repair (20, 21, 22). However, because the epithelium is capable of responding to inhaled pathogens to produce innate immune signaling, we hypothesized that the airway epithelium responds to the presence of nontoxic levels of Gram-negative bacterial product Pseudomonas LPS by accelerating wound repair. We describe a novel surface pathway that stimulates wound repair. This signaling pathway includes TLR-4, protein kinase C (PKC) αβ, dual oxidase 1 (Duox1), reactive oxygen species (ROS), TNF-α-converting enzyme (TACE), TGF-α, and EGFR phosphorylation.
Materials and Methods
Materials
AG-1478, AG-1295, TNF-α proteinase inhibitor-1 (TAPI-1), EGFR neutralizing Ab (Ab 3), epidermal growth factor (EGF) neutralizing Ab, TGF-α neutralizing Ab (Ab 3), GM-6001, TGF-α, apocynin, diphenyleneiodonium chloride (DPI), Gö6976, Gö6983, NG-monoethyl-l-arginine (NMEA), and rottlerin were purchased from Calbiochem. Amphiregulin and heparin-binding EGF (HB-EGF) neutralizing Abs were purchased from Santa Cruz Biotechnology. DMSO, n-propyl gallete (nPG), P. aeruginosa LPS serotype 10, and allopurinol were obtained from Sigma-Aldrich. TLR-4 neutralizing mAb (HTA-125) and TLR-3 neutralizing mAb were purchased from eBioscience.
Cell culture
Cells from the human pulmonary mucoepidermoid carcinoma cell line, NCI-H292, were plated at 1.0 × 105 cells/cm2 in 24-well plates (BD Falcon). Cells were grown in RPMI 1640 medium (Cell Culture Facility, University of California, San Francisco) containing 10% FCS, penicillin (100 U/ml), streptomycin (100 μg/ml), and HEPES (25 mM) at 37°C in a humidified 5% CO2 water-jacketed incubator. Because cell lines such as NCI-H292 cells show some variability in their response to stimuli and inhibitors at different passages (23), all experiments were performed in passages 80–90.
To confirm the results seen in NCI-H292 cells, limited experiments were also performed in normal human bronchial epithelial (NHBE) cells, which were purchased from Cambrex. Cells were passaged, seeded (in 24-well plates (BD Falcon) at 1.0 × 105 cells/cm2), and grown to confluence. Cells were grown in bronchial epithelial growth medium (Cambrex) supplemented with defined growth factors contained in the Single-Quot kit (Cambrex) at 37°C in a humidified 5% CO2 water-jacketed incubator.
Wound repair assay
NCI-H292 and NHBE cells were cultured in 24-well plates to confluence. A linear scratch wound was made with a 10-μl pipette tip in each well and washed with serum-free medium to remove cell debris, as described previously (14). The wounds were treated with LPS (10 μg/ml) with or without inhibitors, or with serum-free medium alone (control). Selected images were recorded continuously using an inverted microscope and digital video camera (Nikon TE2000-E) at 37°C and 5% CO2. SimplePCI 5.1 software (Technical Instruments) converted individual images into videos. Images, at 1-h intervals, were converted to TIFF files (Adobe Photoshop 8.0) and analyzed with image analysis software (NIH ImageJ 1.33u). Wound area was measured at 2-h intervals for 12 h. Initial experiments established an orderly rate of repair, so subsequent experiments analyzed the initial wound area compared with the area at 12 h. Wound repair was expressed as a percentage of the initial wound area measured at the start of the experiment ((initial wound area − 12-h wound area)/initial wound area × 100 = % wound repair).
Cytotoxicity assay
Lactate dehydrogenase (LDH) activity was measured in the supernatants of all wound repair samples. LDH release was measured at 12 h, using a cytotoxicity detection kit (Roche), according to the manufacturer’s protocol. There were no differences in LDH between control and treated wounds (data not shown), except at high concentrations of LPS (see Results; Fig. 1⇓). Results are represented as percentage of LDH released into the supernatant ((supernatant/supernatant + lysate) × 100 = % LDH release).
Concentration-dependent effects of LPS on wound repair. Wound repair was evaluated at 12 h with serum-free medium plus increasing concentrations of LPS (0–500 μg/ml: columns, y-axis, left). Percentage of LDH released into the supernatant was measured ((supernatant/supernatant + lysate) × 100 = % LDH release) (solid line, y-axis, right). Data are means ± SD (n = 3; #, p < 0.05 vs 300, 500 μg/ml; ∗, p < 0.01 vs 0, 100, 300, 500 μg/ml).
Release of soluble TGF-α in the wound repair assay
NCI-H292 cells were cultured in 24-well plates, as described above. After reaching confluence, cells were serum starved for 24 h before wounding. In these studies, cells were pretreated with inhibitors for 30 min in addition to EGFR neutralizing Ab (4 μg/ml), which prevents soluble TGF-α from binding to EGFR. To maximize the release of soluble EGFR ligands, multiple wounds were made in each well: a pipette tip was used to make five parallel linear scratch wounds, and then each well was rotated 90° and another five parallel linear scratch wounds were made. The wounds were washed to remove debris, and then an EGFR neutralizing Ab (4 μg/ml) and LPS (10 μg/ml), with or without inhibitors, were added to serum-free medium. After 2 h, cell supernatants were collected, and TGF-α was measured with a TGF-α ELISA kit (Oncogene Research Products), according to the manufacturer’s protocol. Total protein in cell lysates was measured using the bicinchoninic acid protein assay kit (Pierce Biotechnology), and results are represented as TGF-α (pg)/total protein (μg) to normalize for possible differences in cell number.
TACE and Duox1 small interfering RNA (siRNA) preparation and transfection
Predesigned human TACE siRNAs (104029, 104030, and 104031) were purchased from Ambion. In preliminary studies, TACE siRNA 104029 (300 nM) inhibited LPS-induced wound repair significantly. Therefore, this siRNA was selected for subsequent studies. The 21-nt sequences for TACE (siRNA 104029) were GGUUUUAAAGGCUAUGGAAtt (sense) and UUCCAUAGCCUUUAAAACCtg (antisense). Predesigned human Duox1 siRNAs (24873 and 24969) were purchased from Ambion. In preliminary studies, Duox1 siRNA 24873 (100 nM) inhibited LPS-induced wound repair significantly. Therefore, this siRNA was selected for subsequent studies. The 21-nt sequences for Duox1 (siRNA 24873) were GGACCAU GUGUUGGUUGAAtt (sense) and UUCAACCAACACAUGGUCCtc (antisense). Silencer Negative Control 1 siRNA (Ambion) was used as a nonspecific siRNA. siRNA transfection into NCI-H292 cells was conducted using Lipofectamine 2000 (Invitrogen Life Technologies) when cells were ∼40–50% confluent. When cells reached confluence, linear wounds were made and wound repair was measured, as described above. To determine siRNA transfection efficiency, FAM-labeled GADPH siRNA (Ambion) was transfected, as described. At the time wounds would have been made, the number of FAM-positive cells was counted by fluorescence microscopy. Transfection efficiency was >60% ((FAM-positive cells/total cells) × 100 = % transfection efficiency). Specific silencing of TACE and of Duox1 was confirmed using RT-PCR.
RNA isolation, reverse transcription, and RT-PCR
Total RNA was isolated and reverse transcription was performed, as previously described (24, 25). Total RNA was isolated using RNAqueous-4PCR kit (Ambion), and 2 μg was primed with oligo(dT) and reverse transcribed using a RETROscript kit (Ambion) in a final volume of 20 μl (RT reaction), according to the manufacturer’s instructions. Two microliters of the reverse-transcriptase reaction was PCR amplified in a 50-μl reaction. The following primers were used: Duox1, 5′-GCCCTGTACAACCAGGACTT-3′ (forward) and 5′-CGCACAAATTGTTCAAGGAC-3′ (reverse); TACE, 5′-ACCTGAAGAGCTTGTTCATCGAG-3′ (forward) and 5′-CCATGAAGTGTTCCGATAGATGTC-3′ (reverse). As internal controls, primers for Rig/S15 rRNA, a housekeeping gene that is constitutively expressed, were used. They were 5′-TTCCGCAAGTTCACCTACC-3′ (forward) and 5′-CGGGCCGGCCATGCTTTACG-3′ (reverse). The PCR mixture was denatured at 94°C for 5 min, followed by 38 cycles at 94°C for 30 s, 57°C for 45 s, and 72°C for 45 s. After PCR, 10-μl aliquots were subjected to 1.5% agarose gel electrophoresis and stained with ethidium bromide.
Immunocytochemical staining
Cells grown on four-chamber slides (Nunc) were serum starved for 24 h before experiments to maintain low basal levels of phosphorylation. For inhibition studies, cells were pretreated with a PKC αβ inhibitor (Gö6976; 70 nM; Calbiochem) for 30 min before making a scratch wound. The wounds were washed to remove cellular debris, and LPS (10 μg/ml), Gö6976, or serum-free medium (control) was added. After 30 min, wounds were washed, fixed, and stained, as described previously (26), using rabbit mAbs to PKC isoforms αβ (1/100; Cell Signaling Technology). Consecutive cells at the wound margin were analyzed at ×40, positive staining cells were counted, and the percentage of PKC αβ-positive cells was calculated (27).
Statistical analysis
Data are presented as means ± SD (n = 3). ANOVA was used to determine statistically significant differences (p < 0.05).
Results
Effect of LPS concentration on wound repair
P. aeruginosa products such as LPS are known to cause pulmonary epithelial damage (19). We hypothesized that the outcome of LPS-induced repair in epithelial wounding depends on the balance between epithelial protective responses and potential epithelial damage by the bacterial products. We found that a low concentration of LPS (10 μg/ml) accelerated wound repair (Fig. 1⇑), but higher LPS concentrations resulted in decreased wound repair; at 500 μg/ml LPS, wound repair was inhibited completely (Fig. 1⇑). At higher concentrations of LPS, the decrease in wound repair was associated with increasing LDH release (Fig. 1⇑), an indication of epithelial cell damage.
Addition of EGFR ligand (TGF-α) or Pseudomonas bacterial product (LPS) accelerates wound repair via EGFR phosphorylation
In the control state, wound area decreased in an orderly fashion over 12 h; the EGFR ligand, TGF-α (10 ng/ml), accelerated wound repair (Fig. 2⇓A). The addition of a selective EGFR tyrosine kinase inhibitor, AG 1478 (10 μM), inhibited wound repair both in the control state and with the addition of TGF-α (Fig. 2⇓A). From these results, we conclude that EGFR activation modulates wound repair in NCI-H292 cells.
Effects of TGF-α and LPS on wound repair. Confluent cultures of NCI-H292 cells were wounded, and wound closure was recorded continuously by time-lapse videomicroscopy. A, Wound repair was measured over 12 h with serum-free medium alone (control, •) or with the addition of TGF-α, an EGFR ligand (10 ng/ml, ▪). Control- and TGF-α-treated wounds were treated with an EGFR phosphorylation inhibitor, AG 1478 (10 μM) (control + AG 1478, ○; TGF-α + AG1478, □). Results are expressed as percentage of wound repair. Data are means ± SD (n = 3; ∗, p < 0.05 vs control and TGF-α + AG 1478; #, p < 0.05 vs control + AG 1478). B, Wound repair was measured over 12 h with serum-free medium alone (control, •) or with the addition of LPS (10 μg/ml, ▪). Control- and LPS-treated wounds were treated with AG 1478 (10 μM) (control + AG 1478, ○; LPS + AG 1478, □). Results are expressed as percentage of wound repair. Data are means ± SD (n = 3; ∗, p < 0.05 vs control and LPS + AG 1478; #, p < 0.05 vs control + AG 1478).
Like TGF-α, LPS (10 μg/ml) accelerated wound repair (Fig. 2⇑B). This effect was also inhibited by the addition of AG 1478 (10 μM) (Fig. 2⇑B), but the addition of the selective tyrosine kinase inhibitor of platelet-derived growth factor (AG 1295, 10 μM) was without effect (data not shown). These results implicate EGFR activation in wound repair by both the EGFR ligand TGF-α and the P. aeruginosa product LPS.
Ligand-dependent EGFR activation and the EGFR ligand TGF-α mediate LPS-induced wound repair
Treatment of wounds with an EGFR neutralizing Ab (4 μg/ml), which blocks ligand binding sites on EGFR and inhibits subsequent EGFR phosphorylation, prevented LPS-accelerated wound repair (Fig. 3⇓A), implicating ligand-dependent EGFR phosphorylation in the response to LPS. Treatment with a TGF-α neutralizing Ab (4 μg/ml) (Fig. 3⇓A) also inhibited LPS-induced wound repair. This inhibition was not seen with the addition of neutralizing Abs (4 μg/ml) to EGF, amphiregulin, or HB-EGF (data not shown). In addition, measurement of TGF-α released into cell supernatant showed a significant increase in the presence of LPS compared with control (59.3 ± 9.5 vs 31.5 ± 0.8 pg/μg total protein; p < 0.05; n = 3). These results implicate the EGFR ligand, TGF-α, in LPS-accelerated wound repair in NCI-H292 cells.
LPS-induced wound repair involves ligand-dependent EGFR activation, the EGFR ligand TGF-α, and the metalloprotease TACE. A, Wound repair was analyzed at 12 h with serum-free medium (control, left column) or with the addition of LPS (10 μg/ml, second column). A neutralizing EGFR Ab (EGFR Ab, 4 μg/ml, third column) or a neutralizing Ab to TGF-α (TGF-α Ab, 4 μg/ml, fourth column) was added to LPS-treated wounds. Data are means ± SD (n = 3; ∗, p < 0.05 vs control; #, p < 0.05 vs LPS + EGFR Ab; +, p < 0.05 vs LPS + TGF-α Ab). B, A general metalloprotease inhibitor, GM 6001 (10 μM, third column), or a more selective TACE inhibitor, TAPI-1 (TAPI, 10 μM, fourth column), was added to LPS-treated wounds. Data are means ± SD (n = 3; ∗, p < 0.05 vs control; #, p < 0.05 vs LPS + GM6001; +, p < 0.05 vs LPS + TAPI). C, NCI-H292 cells were transfected with or without TACE siRNA and cultured for 72 h, and then analyzed for TACE mRNA expression by RT-PCR. Cells were transfected with Lipofectamine (Invitrogen Life Technologies) alone (vehicle), TACE siRNA (300 μM), nonspecific siRNA (300 μM), or no treatment (control). Rig/S15 was used as an internal marker. D, NCI-H292 cells were transfected with TACE siRNA and wounded. Wound repair was analyzed at 12 h with serum-free medium (control, left three columns) plus Lipofectamine (vehicle, 1.5 μl/500 μl total volume) and TACE siRNA (300 μM), or a nonspecific siRNA sequence (300 μM). These results were compared with wounds treated with LPS (10 μg/ml, right three columns) plus Lipofectamine (vehicle, 1.5 μl/500 μl total volume) and TACE siRNA (300 μM), or a nonspecific siRNA sequence (300 μM). Data are means ± SD (n = 3; #, p < 0.05 vs TACE; ∗, p < 0.01 vs TACE).
Inhibition of metalloprotease activation prevents LPS-induced wound repair
The metalloprotease TACE is known to cleave and release the EGFR proligand TGF-α from the epithelial surface in airway epithelial cells (25), allowing the cleaved ligand to bind to and phosphorylate EGFR. In this study, we found that a general metalloprotease inhibitor, GM6001 (10 μM), and a more selective TACE inhibitor, TAPI-1 (10 μM), prevented LPS-accelerated wound repair (Fig. 3⇑B). In addition, release of TGF-α into the cell supernatant was prevented by pretreatment with TAPI-1 (10 μM) compared with LPS alone (18.5 ± 0.6 vs 59.3 ± 9.5 pg/μg total protein; p < 0.05; n = 3). These results implicate a metalloprotease, perhaps TACE, and the cleavage and release of the EGFR ligand TGF-α in wound repair induced by LPS.
siRNA knockdown of TACE prevents LPS-induced wound repair
Because TACE cleaves pro-TGF-α (28), we examined whether TACE is involved in LPS-induced wound repair. We used siRNA to knockdown TACE expression in NCI-H292 cells. In the control state, TACE siRNA (300 μM) decreased TACE mRNA expression (Fig. 3⇑C) and suppressed wound repair (Fig. 3⇑D). LPS-accelerated wound repair was also markedly decreased (Fig. 3⇑D). A nonspecific control siRNA (300 μM) was without effect in both the control and LPS-treated wounds (Fig. 3⇑D). These results implicate TACE in LPS-accelerated wound repair.
ROS mediate LPS-induced wound repair
ROS are reported to activate TACE (24). In this study, we found that the ROS scavengers, nPG (100 μM) and DMSO (1%), prevented LPS-induced wound repair (Fig. 4⇓). TGF-α released into cell supernatant was inhibited by pretreatment with nPG (100 μM) compared with LPS alone (31.7 ± 14.0 vs 59.3 ± 9.5 pg/μg total protein; p < 0.05; n = 3). These results implicate ROS in response to LPS.
Effect of ROS scavengers on LPS-induced wound repair. Wound repair was measured at 12 h with serum-free medium (control, left column) or with the addition of LPS (10 μg/ml, second column). ROS scavengers nPG (100 μM, third column) or DMSO (1%, fourth column) were added to the LPS-treated wounds. Data are means ± SD (n = 3; ∗, p < 0.05 vs control; #, p < 0.05 vs LPS + nPG; +, p < 0.05 vs LPS + DMSO).
NADPH oxidases (Nox) mediate LPS-induced wound repair
Nox are known to generate ROS in airway epithelial cells by a core component homolog Duox1 (29, 30). In this study, we found that general Nox inhibitors (DPI (1.0 μM) or apocynin (1.0 mM)) prevented LPS-induced wound repair (Fig. 5⇓A). To exclude the involvement of other oxidases, we investigated the effect of inhibitors of xanthine oxidases (allopurinol (100 μM), NO synthase, and NMEA (100 μM)). These inhibitors had no significant effect on LPS-induced wound repair (Fig. 5⇓A). These results implicate Nox in LPS-induced wound repair.
LPS-induced wound repair involves Nox and Duox1. A, Wound repair was analyzed at 12 h with serum-free medium (control, left column) or with the addition of LPS (10 μg/ml, second column). A general Nox inhibitor DPI (1.0 μM, third column) or apocynin (APO, 1.0 mM, fourth column) was added to LPS-treated wounds. A NO inhibitor, NMEA (100 μM, fifth column), or a xanthine oxidase inhibitor, allopurinol (ALP 100 μM, sixth column), was added to LPS-treated wounds. Data are means ± SD (n = 3; ∗, p < 0.05 vs control; #, p < 0.05 vs LPS + DPI; +, p < 0.05 vs LPS + APO). B, NCI-H292 cells were transfected with or without Duox1 siRNA and cultured for 72 h, and then analyzed for Duox1 mRNA expression by RT-PCR. Cells were transfected with Lipofectamine (Invitrogen Life Technologies) alone (vehicle), Duox1 siRNA (100 μM), nonspecific siRNA (100 μM), or no treatment (control). Rig/S15 was used as an internal marker. C, NCI-H292 cells were transfected with Duox1 siRNA and wounded. Wound repair was analyzed at 12 h with serum-free medium (control, left three columns) plus Lipofectamine (vehicle, 1.5 μl/500 μl total volume), Duox1 siRNA (100 μM), or a nonspecific siRNA sequence (100 μM). These results were compared with wounds treated with LPS (10 μg/ml, right three columns) plus Lipofectamine (vehicle, 1.5 μl/500 μl total volume), Duox1 siRNA (Duox1, 100 μM), or a nonspecific siRNA sequence (100 μM). Data are means ± SD (n = 3; #, p < 0.01 vs Duox1; ∗, p < 0.01 vs Duox1).
Duox1 mediates LPS-induced wound repair
Recently, Duox1, a gp91phox homolog in airway epithelial cells, has been shown to activate TACE via ROS production (24). We studied the effects of siRNA knockdown of Duox1 expression in NCI-H292 cells. In the control state, Duox1 siRNA (100 μM) inhibited Duox1 mRNA expression (Fig. 5⇑B) and noticeably inhibited wound repair (Fig. 5⇑C). LPS-induced wound repair was also markedly inhibited by Duox1 siRNA (Fig. 5⇑C). A nonspecific control siRNA (100 μM) was without effect in both control and LPS-treated wounds (Fig. 5⇑C). These results implicate Duox1 in control- and in LPS-induced wound repair.
Calcium-dependent PKC isoforms mediate LPS-induced wound repair
PKC isoforms have been described to activate Nox (31). The PKC isoforms can be divided into two general categories: Ca2+ dependent (α, β, γ) or Ca2+ independent (δ, ε, λ, ζ, θ). LPS has been shown to stimulate activation of both calcium-dependent and calcium-independent isoforms, including PKC α, β, δ, and ε (32, 33, 34). In this study, we found that a general PKC inhibitor Gö6983 (70 nM) prevented LPS-induced wound repair (Fig. 6⇓A). A PKC αβ (Ca2+-dependent)-selective inhibitor, Gö6976 (70 nM), also inhibited LPS-induced wound repair, whereas a PKC δθ (Ca2+- independent)-selective inhibitor, rottlerin (3 μM), had no significant effect (Fig. 6⇓A). PKC αβ phosphorylation at the wound margin was evaluated by immunocytochemistry. The addition of LPS (10 μg/ml) increased PKC αβ phosphorylation at the wound margin compared with control (Fig. 6⇓, B and C), an effect that was decreased in wounds pretreated with PKC αβ inhibitor, Gö6976 (Fig. 6⇓, B and C). These results implicate Ca2+- dependent PKC αβ in LPS-induced wound repair.
LPS-induced wound repair involves Ca2+-dependent PKC αβ. A, Wound repair was measured at 12 h with serum-free medium alone (control; left column) or with the addition of LPS (10 μg/ml; second column). A general PKC inhibitor, Gö6983 (PKC gen, 70 nM, third column), or a more selective PKC αβ (Ca2+-dependent) inhibitor, Gö6976 (PKC αβ, 70 nM, fourth column), was added to the LPS-treated wounds. These results are compared with a PKC δθ (Ca2+-independent)-selective inhibitor rottlerin (PKC δθ, 3 μM, fifth column). Data are means ± SD (n = 3; ∗, p < 0.05 vs control; #, p < 0.05 vs LPS + Gö6983 (PKC gen); +, p < 0.05 vs LPS + Gö6976 (PKC αβ)). B, Photomicrographs of PKC αβ phosphorylation at the wound edge. Cells were stained for phospho-PKC αβ in wounds with serum-free medium alone (control), with the addition of LPS (10 μg/ml), or with LPS (10 μg/ml) plus pretreatment with PKC αβ inhibitor, Gö6976 (PKC αβ Inhib, 70 nM). Photomicrographs are shown at ×20 magnification and are representative of three separate experiments. Bar, 50 μm. C, Cells staining positive for phospho-PKC αβ at the wound margin were counted with serum-free medium (control; left column), the addition of LPS (10 μg/ml; middle column), or pretreatment with PKC αβ inhibitor, Gö6976 (PKC αβ, 70 nM, right column), and percentage of PKC αβ-positive cells was calculated. Data are means ± SD (n = 3 separate experiments; #, p < 0.01 vs control; ∗, p < 0.01 vs LPS + Gö6976 (PKC αβ)).
LPS-induced wound repair in NHBE cells
To confirm the results of studies with NCI-H292 cells, selected experiments targeting critical steps in LPS-induced wound repair were also performed with NHBE cells. The addition of LPS (10 μg/ml) to NHBE cells accelerated wound repair compared with control (Fig. 7⇓). This effect was inhibited by the addition of a selective EGFR tyrosine kinase inhibitor, AG 1478 (10 μM) (Fig. 7⇓). A metalloprotease inhibitor that shows some selectivity for TACE, TAPI-1 (10 μM), prevented LPS-accelerated wound repair in NHBE cells (Fig. 7⇓). A ROS scavenger, nPG (100 μM), inhibited LPS-induced wound repair (Fig. 7⇓), and the addition of a general Nox inhibitor apocynin (1.0 mM) prevented LPS-induced wound repair in NHBE cells (Fig. 7⇓). siRNA knockdown of TACE and Duox1 was not possible in NHBE cells. Finally, the addition of a PKC αβ (Ca2+-dependent)-selective inhibitor, Gö6976 (70 nM), also inhibited LPS-induced wound repair. These results confirm the studies in NCI-H292 cells.
LPS-induced wound repair in NHBE cells. Wound repair in NHBE cells was measured at 12 h with serum-free medium (control, first column) or with the addition of LPS (10 μg/ml, second column), and selected inhibitors were added to confirm results seen in NCI-H292 cells. An EGFR tyrosine kinase phosphorylation inhibitor, AG 1478 (10 μM, third column), a metalloprotease inhibitor with some selectivity for TACE, TAPI-1 (TAPI, 10 μM, fourth column), a ROS scavenger nPG (100 μM, fifth column), a general Nox inhibitor apocynin (APO, 1.0 mM, sixth column), or a selective PKC αβ (Ca2+-dependent) inhibitor, Gö6976 (PKC αβ, 70 nM, seventh column), was added to LPS-treated wounds. Data are means ± SD (n = 3; ∗, p < 0.05 vs control; #, p < 0.05 vs LPS + each inhibitor).
TLR-4 is required for LPS-induced acceleration of wound repair
LPS is a ligand for TLR-4 (8, 9). A neutralizing Ab to TLR-4 (0.1 μg/ml) prevented LPS-induced acceleration of wound repair, but a TLR-3 neutralizing Ab (0.1 μg/ml) was without effect (Fig. 8⇓). These results suggest that LPS binding to TLR-4 is involved in LPS-accelerated wound repair.
The role of TLR-4 in LPS-induced wound repair. Wound repair was evaluated at 12 h with serum-free medium (control, first column) or with the addition of LPS (10 μg/ml, second column). Neutralizing Abs to TLR-4 (0.1 μg/ml, third column) or TLR-3 (0.1 μg/ml, fourth column) were added to LPS-treated wounds. Data are means ± SD (n = 3; ∗, p < 0.05 vs control; #, p < 0.05 vs LPS + TLR-4 Ab).
Discussion
The airway epithelial surface, in addition to its role as a physical barrier, represents a battleground, in which the host intercepts signals from pathogens and activates epithelial defenses to combat infection. When microbes are inhaled, the innate immune system initiates a dynamic signaling cascade (10, 11). Wound repair is an essential function of the airway epithelium in response to inhaled pathogens in chronic airway diseases (15, 35). Gram-negative bacteria, especially P. aeruginosa, are implicated in the pathobiology of these diseases (5, 6, 7), and P. aeruginosa is known to cause toxicity and injury to the airway epithelium. However, because the epithelium is capable of responding to inhaled substances to produce innate immune signaling, we propose that the epithelial response depends on the concentration of the invading stimulus. We hypothesized that nontoxic concentrations of P. aeruginosa LPS activate EGFR and accelerate wound repair. In this study, we showed that P. aeruginosa LPS accelerates wound repair in human airway epithelial cells via activation of a cascade involving a series of molecules on the airway epithelial surface.
First, we showed that the effects of LPS are concentration dependent. Low concentrations of LPS increased wound repair, whereas higher concentrations decreased wound repair and were toxic to the airway epithelium. Because EGFR activation is known to regulate wound repair and because LPS, a secreted product of P. aeruginosa, causes EGFR activation (25), we investigated the role of EGFR phosphorylation in LPS-induced wound repair. Blockade of EGFR tyrosine kinase phosphorylation inhibited LPS-induced wound repair, implicating EGFR activation in the wound repair process. A family of growth factors (i.e., EGF, TGF-α, HB-EGF, and amphiregulin) activates the EGFR. In alveolar (36), airway (15), corneal (37), keratinocyte (38), and intestinal (39) cell studies, the addition of growth factors activates EGFR and accelerates wound repair. In this study, we found that an EGFR neutralizing Ab, which prevents EGFR ligands from binding to the receptor, inhibits LPS-induced wound repair, implicating ligand-dependent EGFR activation in LPS-induced wound repair. Therefore, we examined the EGFR ligand involved in this airway epithelial response to LPS, using neutralizing Abs to various EGFR ligands to prevent their binding to EGFR. The addition of a TGF-α neutralizing Ab inhibited LPS-induced wound repair, whereas neutralizing Abs to EGF, HB-EGF, and amphiregulin were without effect, suggesting that LPS induced the cleavage and release of TGF-α, allowing it to bind to and activate EGFR. Furthermore, we showed that LPS caused the release of TGF-α into the cell supernatant, confirming that TGF-α release occurred in response to LPS. TGF-α release has been measured in an intestinal wound model (40). In corneal (41, 42) and keratinocyte (43, 44) wound models, HB-EGF was implicated, suggesting that different EGFR ligands are involved in repair of various epithelia.
Next, we examined the mechanisms underlying LPS-induced TGF-α release. Metalloproteases are known to cleave EGFR proligands and to release soluble ligands, making them available for binding to the EGFR (45). We hypothesized that during LPS-stimulated wound repair, a metalloprotease on the surface of epithelial cells releases soluble TGF-α, allowing it to bind to and activate EGFR. Incubation of the wounds with a general metalloprotease inhibitor (GM 6001) prevented LPS-induced wound repair, implicating a metalloprotease in this process. A similar effect was seen using this general metalloprotease inhibitor in corneal (41, 42) and keratinocyte (46) wounds. TACE, a disintegrin and metalloprotease-17 family member present on the surface of airway epithelial cells, is known to cleave pro-TGF-α and release soluble TGF-α (28, 47). Addition of TAPI-1, a somewhat selective inhibitor of TACE, also prevented LPS-induced wound repair. This inhibitor was used in a model of intestinal wound repair (40). However, because TAPI-1 is not completely specific for TACE, we also examined the effect of siRNA knockdown of TACE on LPS-induced wound repair. siRNA is an effective means of silencing gene expression and protein production (48, 49). TACE siRNA prevented LPS-induced wound repair, implicating TACE as the metalloprotease involved in EGFR proligand cleavage.
Because ROS are known to activate TACE in the airway epithelium (24), we examined the role of ROS scavengers in wound repair. We showed that ROS scavengers inhibit LPS-induced wound repair. Next, we examined the mechanism of ROS production in this signaling cascade. Nox generate ROS production. The core component of Nox is the catalytic subunit glycoprotein p91phox, and several homologs have been identified in various cell types (50, 51). Recently, TLR-4 activation was shown to generate ROS (52), and Park et al. (53) reported that TLR-4 has a direct interaction with Nox 4, one of the gp91phox homologs. This interaction is essential for LPS-induced ROS production and activation of NF-κB. In mice given P. aeruginosa, ROS production and NF-κB activation required p47phox, a cytosolic component of Nox (54). Recently, a homolog of the gp91phox, Duox1, was identified in human airway epithelial cells and was found to generate ROS (30). In this study, we showed that selective inhibitors of Nox prevent LPS-induced wound repair. This effect was not seen with xanthine oxidase inhibitors nor with NO synthase inhibitors. Furthermore, because Duox1 siRNA decreased LPS-induced wound repair markedly, we conclude that Duox1 is involved in the LPS-induced generation of ROS.
To determine the mechanism of Duox1 activation, we examined the role of PKC. PKC isoforms are generally grouped as follows: 1) classical or calcium dependent (α, β, γ), and 2) calcium independent (δ, ε). In this study, we showed that selective inhibition of the calcium-dependent PKC αβ, but not the calcium-independent PKC isoforms, prevents LPS-induced repair. In addition, LPS increased PKC αβ phosphorylation at the wound margin, further implicating PKC αβ in this signaling cascade. Because PKC signaling of mucin production is upstream of Duox1 (24), our results suggest that PKC αβ activates Duox1 in LPS-induced airway epithelial wound repair. Finally, to determine whether TLR-4 is required for LPS binding, we added a TLR-4 neutralizing Ab to the wounded epithelium. Addition of this Ab inhibited LPS-induced wound repair. Thus, binding of LPS to its receptor initiates an epithelial cell surface signaling cascade that activates the EGFR and stimulates wound repair.
In summary, present studies show that LPS, a bacterial product of Gram-negative bacteria, causes accelerated wound repair in airway epithelial cells via a TLR-4→PKC αβ→DUOX1→ROS→TACE→TGF-α→EGFR phosphorylation pathway (Fig. 9⇓). In addition, we showed that the effects of LPS are concentration dependent: low concentrations of LPS increased wound repair, whereas higher concentrations decreased wound repair and were toxic to the airway epithelium. The response to low concentrations of LPS suggests that the airway epithelium provides an important function by intercepting signals from pathogens and activating host defenses. In contrast, higher concentrations of LPS overcome this response and cause cytotoxicity that may initiate epithelial damage and facilitate invasion. In addition to wound repair, important host defenses include recruitment of neutrophils to kill pathogens and production of mucins to assist in their clearance. We suggest that high concentrations of pathogens may also overwhelm these innate immune defenses to cause pathology or to exacerbate chronic airway diseases.
Diagrammatic scheme of LPS-induced wound repair. LPS binds to its receptor, TLR-4, to activate PKC isoforms PKCα and PKCβ. Duox1 is then activated, generating ROS (represented by ○). ROS activate the latent form of TACE, which has an inhibitory prodomain (represented by the curved black line) covering its active domain (represented by scissors), removing the prodomain and exposing the active domain to cleave pro-TGF-α. Soluble TGF-α is released, which binds to and activates EGFR, initiating signaling for wound repair.
Activation of the innate immune system is essential in the host response to invading microbes, and TLRs provide sites for the rapid detection and cellular response to invaders (10, 11). TLRs have been shown to signal cytokine transcription (55, 56), to recruit and activate neutrophils to clear inhaled pathogens (57, 58), and to induce mucin production (25), which assists in mucociliary clearance of the neutrophils and the microbes (59). We describe a surface defense mechanism against P. aeruginosa that accelerates wound repair. This coordinated cascade activates TLR-4 to stimulate EGFR. EGFR activation can generate multiple epithelial cell effects that are important in innate immune defense. Stimulation of EGFR induces IL-8 expression (60), which causes the recruitment and activation of neutrophils (61, 62). The neutrophils, with engulfed bacteria, adhere to secreted mucins, whose production is also stimulated by EGFR activation (16), and are cleared from the airways by mucociliary clearance and cough. In addition, antimicrobial peptides such as the defensins and cathelicidins are activated by EGFR (63, 64) and contribute to the host defense against pathogens. Further study is required to determine whether the TLR-4 to EGFR signaling pathway is conserved within the TLR family and represents a common response to other pathogens (e.g., Gram-positive bacteria, viruses, fungi, and parasites).
Acknowledgments
We thank Drs. Michael Matthay, Joanne Engel, and Sasha Shafikhani (all at the University of California, San Francisco) for their generous advice regarding wound repair assays and access to videomicroscopy equipment, and Dr. Julien Hoffman (University of California, San Francisco) for assistance with statistical analysis.
Disclosures
The authors have no financial conflict of interest.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵1 This work was supported by Cardiovascular Research Institute funding.
↵2 Address correspondence and reprint requests to Dr. Jay A. Nadel, University of California, San Francisco, 505 Parnassus Avenue, Room S-1183, San Francisco, CA 94143-0130. E-mail address: jay.nadel{at}ucsf.edu
↵3 Abbreviations used in this paper: EGFR, epidermal growth factor receptor; DPI, diphenyleneiodonium chloride; Duox1, dual oxidase 1; EGF, epidermal growth factor; HB-EGF, heparin-binding EGF; LDH, lactate dehydrogenase; NHBE, normal human bronchial epithelial; NMEA, NG-monoethyl-l-arginine; Nox, NADPH oxidase; nPG, n-propyl gallete; PKC, protein kinase C; ROS, reactive oxygen species; siRNA, small interfering RNA; TACE, TNF-α-converting enzyme; TAPI-1, TNF-α proteinase inhibitor-1.
- Received June 23, 2006.
- Accepted September 28, 2006.
- Copyright © 2006 by The American Association of Immunologists