|Home | About | Journals | Submit | Contact Us | Français|
Epithelia express oxidative anti-microbial protection that uses lactoperoxidase (LPO), hydrogen peroxide (H2O2) and thiocyanate to generate the reactive hypothiocyanite. Duox1 and Duox2, found in epithelia, are hypothesized to provide H2O2 for use by the LPO. To investigate regulation of oxidative LPO-mediated host defense by bacterial and inflammatory stimuli, LPO and Duox mRNA were followed in differentiated primary human airway epithelial cells, challenged with Pseudomonas aeruginosa flagellin or IFNγ. Flagellin upregulated Duox2 mRNA 20-fold, but only upregulated LPO mRNA 2.5-fold. IFNγ increased Duox2 mRNA 127-fold and upregulated LPO mRNA 10-fold. DuoxA2, needed for Duox2 activity, was also upregulated by flagellin and IFNγ. Both stimuli increased H2O2 synthesis and LPO-dependent killing of Pseudomonas aeruginosa. Reduction of Duox1 by siRNA showed little effect on basal H2O2 production, while Duox2 siRNA markedly reduced basal H2O2 production and resulted in an 8-fold increase in Nox4 mRNA. In conclusion, large increases in Duox2 mediated H2O2 production appear to be coordinated with increases in LPO mRNA and, without increased LPO, H2O2 levels in airway secretion are expected to increase substantially. The data suggest that Duox2 is the major contributor to basal H2O2 synthesis despite the presence of greater amounts of Duox1.
Mucosal surfaces that communicate with the outside environment are protected from infection by the presence of secreted antibacterial proteins and enzymes such as lysozyme, defensin peptides and lactoperoxidase (LPO). LPO is unique among these in that it protects epithelial surfaces not only from bacteria [for review see 1] but also from viruses  and fungi [3, 4]. LPO uses hydrogen peroxide (H2O2) to oxidize thiocyanate (SCN−) and thereby generate the nonspecific oxidative compound hypothiocyanite (OSCN−). Thus, LPO activity requires the presence of both SCN− and H2O2 to effectively act as an antibiotic defense system.
Thiocyanate is present at relatively high concentrations in milk, saliva and airway secretions but H2O2 is present at very low concentrations and, based on a computational model of LPO activity in the airway , it has been proposed that H2O2 synthesis rates control LPO activity rather than secretion of the enzyme itself as expected for lysozyme and defensins. Regulation of LPO activity by H2O2 availability may prevent damage or inappropriate signaling through high levels of H2O2 in the normal airway lumen.
The major source of H2O2 in the airway is provided by Dual oxidases (Duox1 and Duox2), that are homologues of gp91phox/Nox2 found in neutrophils. The Nox family comprises a group of multi-pass membrane flavoproteins, called NAPDH oxidases, which use NADPH as an electron donor to generate superoxide and H2O2. Duox1 and Duox2 are the major NADPH oxidases in both airway and intestinal epithelial cells and are believed to provide H2O2 as substrate for use in peroxidase mediated bacterial defense [6–9]. Duox1 plays an important role in repair of the epithelium  and is activated following stimulation of HBE1 (human bronchial epithelia 1) cell line with anti-ASGM1 (asialo Ganglioside-M1) antibody that binds and activates the asialoGM1 receptor . Other members of the Nox family are also expressed in epithelial cells but at much lower levels compared to Duox1 and 2 . Thus, regulation of Duox levels may be an important regulator of epithelial peroxidase defense.
Bacterial challenge of epithelia results in a pleiotropic response to limit infection. This response is communicated in part through interaction of bacterial molecules with toll-like receptors (TLRs) on the epithelial surface . Both flagellin and lipopolysaccharide (LPS) stimulate TLRs to alter gene expression through activation of the NF-κb signaling pathway and thus mediate the epithelial response to bacterial encounters. For example, flagellin binds to TLR5 and stimulates release of human β-defensin as an antibacterial response . Direct effects of bacterial product binding to TLRs on Duox and LPO expression have not been studied. Bacterial infection results in the release of proinflammatory cytokines that assist in modulating immune responses to the infection. Interferon-gamma (IFNγ) for example, increases Duox2 mRNA and enzyme activity . To understand how the LPO system activity is regulated during bacterial challenge, we examined changes in the expression of the LPO and Duox in airway epithelial cells following stimulation with Pseudomonas aeruginosa flagellin or proinflammatory cytokines and assessed the effects of increased Duox activity on bacterial killing.
All chemicals and reagents were purchased from Sigma Aldrich (St. Louis, MO) unless otherwise noted.
Flagellin was purified from Pseudomonas aeruginosa, ATCC 27,853 (Type A flagellin) as described by Brimer and Montie  or from PAO1(gift from K. Mathee, Florida International University) (Type B flagellin) as described by Adamo et al. . Both preparations were homogeneous by SDS gel analysis (Figure 1a) and gave similar results in stimulation experiments. All experiments shown used Type A flagellin. Type B flagellin gave indistinguishable increases in Duox2 mRNA levels.
All experiments used normal human bronchial epithelial (NHBE) cells re-differentiated at the air-liquid interface. Briefly, lungs that were not acceptable for transplantation were obtained by the Life Alliance Organ Recovery Organization according to Institutional Review Board approved consent and protocol. Bronchi were dissected and airway epithelial cells were isolated by protease digestion as described previously [17, 18]. Cells were expanded for one or two passages by growth in submerged culture and then plated on Corning 24-mm or 12-mm Transwell-clear culture inserts coated with human collagen type IV. Confluent cultures on inserts were exposed to air and used for experiments after differentiation as noted by the presence of cilia and mucus (2 – 4 weeks). Medias and methods used for isolation, growth and differentiation of cells have been previously described [18, 19]. Air-liquid interface (ALI) cultures were stimulated apically with Dulbecco’s PBS (DPBS) alone or containing 2 μg/ml flagellin for 4 h at 37°C. Apical solutions were then removed but not washed and cultures were returned to the incubator for different times of exposure to flagellin. NHBE cultures were stimulated for different times of exposure to 100 ng/ml interferon-γ (IFNγ) (BioSource International. Inc., Camarillo, CA) in the basolateral media. When not otherwise noted, flagellin stimulation was for 24 h and IFNγ stimulation was for 48 h. For NF-κb inhibition experiments, cells were pre-treated for 30 minutes with 100 μg/ml caffeic acid phenethyl ester (CAPE) or 50 μM pyrrolidine dithiocarbamate (PDTC) prior to and during stimulation with flagellin. For TLR blocking experiments, functionally blocking polyclonal rat anti-human TLR5 antibodies (PAb hTLR5, Invivogen Corp, San Diego CA) or non-immune control IgG were added to the apical and basolateral compartments at 20 μg/ml. After 1 h, flagellin was added to apical solution containing anti-TLR5 IgG and stimulation performed as described above for 6 h.
Total RNA was isolated using QIAshreder and the Qiagen RNeasy Plus Mini kit (Qiagen, Valencia, CA). cDNA was synthesized from 1 μg of total RNA using the iScript cDNA Synthesis Kit (BioRad, Hercules, CA). PCR amplification was performed using ICycler IQ (Biorad, Hercules, CA) and TaqMan Universal Assays including gene-specific primer and probe sets designed against the targeted molecules listed in Table 1 (Applied Biosystems, Branchburg, NJ). The relative quantification method (ΔΔCt) was calculated for analysis of gene expression assay using GAPDH for normalization.
NHBE cells were treated with flagellin (24 h) or IFNγ (48 h) as described above. DPBS alone or supplemented with either 100 μM ATP, 20 μM diphenyleneiodonium (DPI) or 100 μM ATP plus 20 μM DPI were added to the surface of ALI epithelial cells for 2 min at 37°C. Cells subjected to DPI inhibition were pre-incubated for 30 min at 37°C in the basolateral media. Samples (25 μL) were collected from the surface and mixed with an equal volume of DPBS containing 50 μM 7-dihydroxyphenoxazine (Amplex red, Molecular Probes, Eugene, OR) and 0.1 U/ml horseradish peroxidase (HRP) in a 96-well plate. Fluorescence was read (Ex 530nm/Em 590nm, 9 nm bandwidth) using a SpectraMax Gemini EM fluorometer (Molecular Devices, Sunnyvale, CA). In other experiments, apical H2O2 production was measured after 2 min of incubation at 37°C with exogenous lactoperoxidase (3 μg/ml) alone or thiocyanate (0.4 mM) alone or both together at the cell surface. For each experiment performed, the amount of apical H2O2 production was determined using a H2O2 standard curve. Data from the H2O2 production measurements were expressed as mean ± S.E.M.
Antibacterial activity was assayed essentially as described previously . Pseudomonas aeruginosa (ATCC, 27853) were grown overnight in LB broth at 37°C in a rotary shaker. Bacteria were collected in the stationary phase diluted into LB broth with 15% glycerol and stored at −80°C. NHBE cultures were pretreated with either IFNγ or flagellin and without antibiotics in the medium. Secretions were collected in a single 500 μl DPBS wash. Anti-bacterial activity was assayed in 500 μl containing 300 μl of airway secretions, ± LPO (6.5 μg/ml), ± SCN (5 × 10−4 M), ± catalase (400 U/ml) and 2,400 – 3,600 bacteria. Mixtures were sampled before and after 4 h incubation at room temperature on a rotating vertical platform. CFUs were determined by plating 25 μl in triplicate on LB agar plates and overnight incubation at 37°C. Anti-bacterial activity was expressed as a ratio of CFU after 4 h to the starting CFU in the no LPO control. Antibacterial activity in this assay was previously shown to be SCN− dependent .
For Duox quantification, total protein extracts from NHBE cultures, treated with flagellin or IFNγ, were obtained by lysis at 100°C in 2% SDS, 10 mM Tris pH 8.3, 5 mM EDTA, followed by sonication and centrifugation at 14,000g. Protein was determined using the BCA protein assay (Pierce, Rockford, IL) and 20 μg were diluted with sample buffer (100 mM Tris pH 6.8, 4% SDS, 20% glycerol and 100 mM DTT), separated on a 7.5% Ready Gel Precast SDS-polyacrylamide gels (Biorad, Hercules, CA) and transferred by electrophoresis to Immobilon-P membranes (Millipore, Billerica, MA). The membranes were blocked at room temperature overnight with 1% gelatin prepared in 20 mM Tris Cl, pH 7.5, 140 mM NaCl (TBS) and supplemented with 0.1% Tween-20 (TTBS) in constant agitation. The membranes were incubated for 2 h at room temperature with either a rabbit non-immune serum prepared in 1% gelatin/TTBS or a rabbit polyclonal antibody raised against the Arg618-His1044 intracellular fragment of human Duox1/ThOx1 (1/10 000 dilution) that recognizes both Duox1 and 2 .
For IκBα and phospho IκBα detection, cells were harvested before and at times after treatment with either IFNγ or flagellin using 1% Triton X-100, 20 mM sodium phosphate, 150 mM NaCl, 5 mM EDTA, 50 mM HEPES, pH 7.8, 50 mM NaF, 1 mM sodium orthovanadate, 5 mM phenylmethylsulfonyl fluoride, 10 μg/ml leupeptin, and10 μg/ml aprotinin, for 30 minutes at 4 °C with constant agitation. Cell extracts (50 μg) were separated on SDS 4–15 % gradient PAGE gels and transferred to membranes as described above. Membranes were blocked at room temperature for 1 h with 1 % gelatin in 0.1 % TTBS and incubated overnight with either anti-IκBα or anti-phospho IκBα (Ser32/36), diluted 1:1000 (Cell signaling, Boston MA). The membranes were then incubated for 1 h with HRP-coupled goat anti-rabbit IgG. The immune complex was detected using chemiluminescence and intensity signals were quantified using ChemiDocXRS and Quantity One software (Bio-Rad, Hercules, CA).
pLKO.1-puro plasmids encoding non-targeted (NT), Duox1 or Duox2 shRNA (SHC002, TRCN0000045976 and TRCN0000045963, Sigma Aldrich, St. Louis MO) were amplified and inserts sequenced. To produce virus, HEK293T cells were co-transfected using 3rd generation, replication-defective lentiviral vectors and packaging plasmids as described previously  and viral yield was quantified using a p24 ELISA assay (Perkin Elmer, Boston MA). Submerged and subconfluent NHBE cells from three lung donors were infected with Duox1 or Duox2 shRNA lentiviruses, selected with 0.5 – 1 μg/ml puromycin until uninfected cells were dead, typically 3–5 days. Date and lung donor matched cultures that were either uninfected (not puromycin treated) or infected with non-targeted (NT) shRNA lentiviruses (selected with puromycin) were used as controls. Following 1 – 2 weeks of growth to confluency, cultures were trypsinized, replated onto culture inserts, regrown to confluency (1– 2 weeks) and then re-differentiated at the ALI (3–6 weeks). When cells were fully re-differentiated as demonstrated by mucus production and beating cilia, mRNA levels were assayed by real-time PCR using TaqMan kits listed in Table 1 (Applied Biosystems, Foster City, CA) according to manufacturer’s instructions, apical H2O2 production was measured using Amplex red, and Western blotting was performed as described above.
Results were compared by ANOVA and, if significantly different, by the Tukey Kramer honestly significant difference (HSD) test for differences between groups. In some cases, data were log transformed to normalize the distribution and then compared by ANOVA and Tukey Kramer HSD. For comparison of mRNA fold changes relative to untreated samples, transcript number relative to GAPDH was used for analysis.
To examine regulation of the epithelial oxidative LPO antibiotic system, fully differentiated airway epithelial cells, cultured at the air-liquid interface, were used as a model system and exposed apically to either Pseudomonas aeruginosa (PA) LPS or type A flagellin followed by quantification of LPO, Duox1, Duox2, DuoxA1, DuoxA2, Muc5ac, and Muc2 mRNA levels. LPS treatment (50 ng/ml, 24 h) showed little effect on levels of these mRNAs. In addition, no stimulation of IL-8 secretion could be measured (data not shown) consistent with reports by others using NHBE cultures. This could be explained by a lack of serum accessory proteins necessary for LPS action in the culture media . Stimulation with purified PA flagellin (18 h) however, correlated with strong upregulation of Duox2 mRNA (20.1 ± 7.6 fold, n = 6 lung donors, 24 cultures) (Figure 1b), while LPO mRNA was only slightly increased (2.7 ± 1.0 fold, n = 5 donors, 15 cultures,) and Duox1 mRNA was unchanged (1.1 ± 0.2 fold, n = 3 donors, 8 cultures). In time course experiments, Duox2 mRNA increased rapidly following treatment with flagellin (Figure 1c). Thus, flagellin increased Duox2 mRNA much more than LPO mRNA consistent with the hypothesis that LPO system activity is regulated by H2O2 production .
As a positive control for flagellin stimulation, changes in Muc2 mRNA, previously reported to increase in tumor-derived airway epithelial cell lines , was measured. In normal differentiated airway epithelia in ALI culture, flagellin upregulated Muc2 (5.2 ± 1.2, n = 3 donors, 6 cultures) and also Muc5ac mRNA (4.6 ± 1.0, n = 3 donors, 6 cultures, Figure 1b). As expected for flagellin-mediated TLR5 activation of the NF-κB pathway , pretreatment with CAPE (100ug/ml) or PDTC (50 μM), inhibitors of the NF-κB pathway, blocked flagellin-mediated increases in Duox2 mRNA. Since these inhibitors may have nonspecific actions as well, flagellin-induced IκB phosphorylation was confirmed by Western blots using phospho-IκB antibodies (Figure 1f). Preincubation with functionally blocking anti-TLR5 antibodies reduced induction of Duox2 mRNA (Figure 1g). Since Duox1 and Duox2 require the accessory proteins DuoxA1 and DuoxA2 respectively for activity [25–27], mRNA encoding these proteins was also measured. DuoxA2 mRNA was initially present at approximately one-third that of Duox2 (0.5 ± 0.1 × 10−3 DuoxA2 transcripts per GAPDH transcript vs. 1.6 ± 0.6 ×10−3 for Duox2; n = 5 lung donors). DuoxA2 was upregulated after flagellin stimulation to 10-fold higher transcript levels compared to Duox2, but with similar kinetics, while mRNA for the Duox1 accessory protein DuoxA1 was not altered (Figure 1h). After 24 h, both Duox2 and DuoxA2 declined and were near pre-stimulation values at 48 h.
Infection leads to inflammation. The inflammatory cytokine IFNγ was reported to increase Duox2 mRNA . Therefore, IFNγ-stimulated cultures were examined for changes in LPO, Duox1 and 2 mRNAs (Figure 2). In contrast to the modest increase following flagellin treatment, LPO mRNA was upregulated by IFNγ after 48 h of stimulation (10.8 ± 2.3 fold, n = 15 donors, 79 cultures). After 48 h, IFNγ increased Duox2 mRNA to very high levels (127 ± 38 fold, n = 12 donors), greater than those reported by Harper et al. . IFNγ did not alter expression of Duox1. The mRNA level of Duox maturation factors, DuoxA1 and DuoxA2, was also measured. DuoxA2 was highly increased in cells stimulated with IFNγ to levels equal to or higher than Duox2 mRNA (Figure 2b). DuoxA1 was unchanged, suggesting that DuoxA2 is coordinately regulated with Duox2 (Figure 2b). Longer periods of incubation showed a decrease in both Duox2 and DuoxA2 mRNA (Figure 2b).
Although LPO protein is not highly expressed in these cultures and is difficult to quantify, Duox protein and H2O2 generation activity can be measured. There are several caveats to the protein measurements, however. Untreated cultures express 4–5 fold higher levels of Duox1 mRNA compared to Duox2 mRNA. Thus, a flagellin-induced 20-fold increase in Duox2 will only result in about 4 times more Duox2 mRNA than Duox1 mRNA relative to GAPDH and possibly protein levels. In addition, there are no antibodies available that can discriminate between Duox 1 and Duox 2 (see below).
To confirm that the increased expression of Duox2 mRNA was reflected in increased protein, flagellin and IFNγ treated cultures were assayed by Western blotting. Western blots used an antiserum raised against a 426 amino acid peptide of Duox1 that is 62% identical to the Duox2 sequence; thus, the antiserum reacts preferentially with Duox1 but cross-reacts with Duox2 [20, 27]. Flagellin treated cultures showed an increase of 1.4 ± 0.22 (n = 8 donors, 18 cultures) in total Duox expression reflecting the increase in Duox2 mRNA to levels near that of Duox1 (Figure 3). Western blots of IFNγ-treated cultures also demonstrated an increase in total Duox protein (2.11 ± 0.23 fold, n = 4 donors, 1 culture per donor). These values are less than expected based on Duox2 mRNA changes. However, the antibodies preferentially recognize Duox1. Since Duox1 appears to be expressed at 3–4 fold higher levels than Duox2 in resting cells, these antibodies are not able to quantitatively assess changes in Duox2 expression in the presence of Duox1. For this reason, biological activity of Duox was examined after stimulation.
Application of either flagellin or IFNγ to the apical surface of cultures did not elicit an immediate (< 1 h) increase in H2O2 production (data not shown). However longer treatments that increased Duox2 mRNA (flagellin, 24 h, and IFNγ, 48 h) resulted in dramatically increased H2O2 production after addition of PBS to the apical surface of cultures and this increase was sensitive to inhibition by DPI (Figure 4). Flagellin treated cells showed a 5-fold increase in H2O2 production and IFNγ treated cells increased basal H2O2 production 35-fold. Purinergic stimulation with ATP, previously shown to increase Duox activity via increases in intracellular [Ca2+] [6, 28, 29], further stimulated H2O2 production in both IFNγ and flagellin treated cultures (Figure 4). DPI inhibited basal and ATP stimulated H2O2 production, consistent with NADPH oxidase activity. Increases in H2O2 production were quantitatively not identical to the changes in mRNA for several reasons. One of these is that addition of DPBS to the apical surface for H2O2 measurements dilutes [H2O2]. Despite this dilution, IFNγ-treated and ATP-stimulated cultures still exhibited high levels of H2O2 (10 μM).
Since these ALI cultures do not express significant levels of LPO protein (made primarily by submucosal glands), the addition of LPO and thiocyanate (SCN−) to the cultures was tested for its ability to reduce accumulated H2O2 and to demonstrate that extracellular LPO and SCN− can utilize the apical H2O2 produced by Duox. The product of LPO catalysis, OSCN−, does not oxidize Amplex red at 0.1 mM and thus a reduction in Amplex red fluorescence would be expected if LPO consumes the produced H2O2. Addition of LPO and SCN− reduced the measured H2O2 to low levels (Figure 4c) suggesting that the complete LPO system can control the increased H2O2 levels after stimulation. Thus, it is unlikely that H2O2 could accumulate to the levels observed in our assays when LPO and SCN− are present. However, increased H2O2 production was associated with increased LPO mediated bacterial killing capacity of cultures washes. Both flagellin and IFNγ treatment resulted in increased LPO/SCN− dependent killing of PA and this killing was reversed by catalase (Figure 5).
The roles of Duox1 and Duox2 in H2O2 production were directly assessed using siRNA. Undifferentiated cultures were infected with lentiviruses expressing either Duox1 or Duox2 targeted shRNA. Following infection and selection for virally encoded puromycin resistance, cultures were redifferentiated at the air-liquid interface. Duox1 and 2 shRNA specifically reduced levels of the targeted mRNA (Figure 6a). Western Blots with anti-Duox antiserum confirmed the reduction in Duox1 protein estimated from the decrease in Duox1 mRNA. If Duox1 mRNA predicts Duox1 protein expression, Duox1 is about 80% of the total Duox1 and Duox2; thus, with shRNA mediated reduction of 80% in Duox1 mRNA (to 20% of baseline), Duox1 protein expression detected by Western blot would be expected to reveal about 64% Duox protein reduction (Figure 6c). Reduction of Duox2 protein could not be reliably detected by Western blotting since the antibody recognizes Duox1 preferentially in addition to Duox2, and Duox2 mRNA is ~3.5-fold less than Duox1 (see Figure 7e).
Reduction of Duox1 mRNA by siRNA reduced baseline H2O2 production by only 27% (Figure 6b) while Duox2 mRNA reduction decreased baseline H2O2 production by 58% (p < 0.05 compared to control) suggesting that Duox2 may have a significantly higher baseline H2O2 production activity than Duox1 and may be the major element in unstimulated extracellular H2O2 production. In contrast, Duox1 siRNA lowered ATP stimulated activity by 70% (p < 0.05 compared to ATP stimulated control) while Duox2 siRNA had a lesser effect on ATP stimulated activity compared to Duox1 siRNA (46% reduction, p < 0.05 compared to ATP stimulated control). This result is expected if both enzymes are equally stimulated given their relative mRNA levels at baseline. Thus, these data suggest that Duox2 may be responsible for a large part of basal H2O2 production while Duox1 is responsible for the majority of ATP-stimulated H2O2 production.
Quantification of other NADPH oxidases mRNAs (Nox1 - 5) showed that only Nox4 mRNA was significantly upregulated (8-fold) in response to siRNA reduction of Duox2 expression in NHBE cells (Figure 7d). In Duox2 siRNA expressing cultures, Nox4 mRNA is increased from 0.1x to nearly 0.5x of Duox2 in control cells. Other Nox enzymes were not upregulated in Duox siRNA treated cells.
Although Nox1 mRNA was increased 4-fold after flagellin treatment (Figure 7a) as described in intestinal epithelial cell line T84 , the relative level of Nox1 mRNA to Duox1 and 2 mRNA suggests that it is not a significant contributor to H2O2 production. In addition, Nox2 mRNA increased 3.5-fold in response to flagellin (Figure 7a) and 3-fold following IFNγ (Figure 7b) as was reported previously for Nox2 in neutrophils [31, 32]. Nox3 mRNA was not detected in differentiated NHBE cultures. No changes were seen in NADPH quinone oxidoreductase (data not shown), an intracellular enzyme that can also produce H2O2. Thus, cells respond to changes in the basal level of H2O2 production by increasing other enzymes capable of compensation.
The data presented here suggest that Duox2 plays a role in LPO-mediated oxidative host defense and that LPO, Duox2, and DuoxA2 are coordinately regulated by both bacterial and inflammatory stimuli. Since H2O2 can be damaging to cells and since H2O2 also can serve as a regulator of intracellular signaling [e.g. 33], control of its levels is important during use as a substrate for host defense. The epithelium must coordinate H2O2 production to increase LPO activity in response to bacteria but protect against an imbalance of LPO and H2O2 that can arise later during inflammation. The changes in Duox2, DuoxA2 and LPO reported here suggest a mechanism by which epithelia manage oxidative host defense without allowing steady high levels of H2O2 in the airway lumen.
The high levels of H2O2 detected in assays of IFNγ treated cultures illustrate the importance of coordinate regulation. These NHBE cultures contained 10 μM H2O2 in 300 μl of assay volume on the apical surface. Extrapolation of 10 μM H2O2 in the assay fluid to the expected airway surface liquid volume before addition of PBS predicts ~0.6 mM H2O2 without dilution by the assay mixture. Such a high concentration is not likely encountered in vivo even during stimulation of cultures. Despite the presence of LPO mRNA, NHBE cultures do not secrete LPO protein and thus the major H2O2 scavenger in airway secretions  is absent during the measurements of H2O2. Application of exogenous LPO and thiocyanate to the apical culture compartment rapidly consumes H2O2 preventing its accumulation in vitro.
A computational model of LPO enzymatic activity , suggests that stimulation of epithelial cells by bacterial flagellin or IFNγ can provide an increase in LPO activity by increasing Duox2 mediated H2O2 substrate levels without a significant change in LPO expression (not shown) and supports the idea that the measured increases in H2O2 production after IFNγ would result in accumulation of damaging levels of this reactive oxygen species if the increase is not accompanied by increases in H2O2 scavenging for example by increased LPO enzyme activity. Thus, induction of LPO mRNA expression coincident with the significantly higher H2O2 concentration following IFNγ exposure may provide a mechanism to prevent accumulation of excess H2O2 via increased LPO consumption. Despite expression of LPO mRNA, NHBE cultures do not synthesize measurable levels of LPO enzyme, possibly due to inappropriate culture conditions.
Bacterial killing due to increased H2O2 levels was shown using washes of apical surfaces after treatment of cultures with IFNγ or flagellin. No killing was observed without treatment in contrast to assays performed directly on tracheal surfaces  or directly on surfaces of cultured epithelial cells . These latter assays used direct addition of bacteria in very small volumes (25 μl PBS) to the surface of rat airway epithelia or bovine tracheal surfaces. Thus, the concentration of H2O2 encountered is higher as very little dilution of airway surface liquid occurred. In our assays, culture surfaces were washed with 300 μl of PBS thus diluting H2O2 concentration and showing the difference in stimulated and unstimulated cultures. The low number of PA killed by the LPO system in our experiments reflects the strong dependence of killing on H2O2 production. PBS washes contain only the H2O2 present at the time of the wash, without any ability to replenish H2O2 in the absence of epithelial cell Duox activity, limiting the number of bacteria that can be killed in a PBS wash taken from cultures. Thus, the number of CFU reduced by the LPO system in the experimental conditions does not reflect the robustness of the system in vivo.
The lack of antibodies specific for Duox2 versus Duox1 has hampered studies of their individual contribution to cellular functions. In airway epithelia, Duox1 mRNA is the major NADPH oxidase message while Duox2 is the second most abundant NADPH oxidase and is present at 4–5 fold lower levels. That protein levels reflect mRNA abundance is suggested by Western blots showing siRNA reduction of Duox1 results in significant decreases in total Duox protein while Duox2 siRNA does not cause a measurable reduction in Duox protein by Western blots. This is consistent with previous data using ionomycin stimulation of submerged undifferentiated NHBE cells treated with Duox1 antisense oligonucleotides . It was surprising that Duox2 siRNA reduced basal H2O2 production more than that resulting from Duox1 siRNA. In the case of both Duox1 and 2 knock down, mRNA levels were reduced by 80% down to 20% of mean normal levels. These data are consistent with those of Moskwa et al  examining rat airway H2O2 production. That Duox2 contributes more to basal H2O2 production is supported by the upregulation of Nox4 mRNA in Duox2 knockdown cells while no similar increase in any NADPH oxidase mRNA was noted in Duox1 knockdown. Recently both Duox1 and Duox2 siRNAs were reported to each block nearly all extracellular H2O2 production in airway epithelial cell lines and primary human small airway epithelial cells in submerged undifferentiated state . Our results differ and suggest that differentiation of NHBE at the ALI results in different Duox activities when compared to submerged undifferentiated cultures.
IFNγ was previously reported to increase Duox2 mRNA [8, 36]. In these studies, NHBE cells, exposed to the air-liquid interface for one week, upregulated Duox2 mRNA 19-fold. In our studies, complete differentiation (2–4 weeks) and the use of multiple lung donors showed Duox2 mRNA was upregulated 125-fold. Quantification of H2O2 production in these cultures showed much higher levels of H2O2 than the relative changes predicted by earlier studies using undifferentiated cultures and highlight the importance of using the fully differentiated culture model to investigate effects of cytokines. The mechanism by which IFNγ stimulates Duox2 and DuoxA2 mRNA levels is not known. Although flagellin stimulated TLR5 signaling works through NFκB , IFNγ is generally thought to work through STAT pathways, although it has been shown act in STAT-independent ways including to upregulate NFκB activity [e.g. 37]. Thus, it is possible that multiple pathways can be used to upregulate Duox2 mediated H2O2 production.
The increase in Nox4 after Duox2 reduction suggests that Nox4 activity may replace a portion of the basal H2O2 production by Duox2 following siRNA reduction. The Nox4 induction after Duox2 siRNA expression does not counter the decrease in H2O2 perhaps in part because its expression still is less than the normal levels of Duox2. It is not possible to assess the contribution of Nox4 to the remaining H2O2 production without performing a double siRNA expression study where Nox4 and Duox2 are both reduced. We speculate that the noted increase in NOX4 may relate to a needed basal level of H2O2 in the cytoplasm to maintain the appropriate level of H2O2 mediated signaling. Although Nox4 is thought to reside in the ER, H2O2 readily crosses cell membranes and thus may contribute to the redox state of the cytoplasm. In the absence of basal Duox2 expression, higher levels of Nox4 might serve to replace the needed H2O2 in the cytoplasm.
The measured relative abundance of Nox mRNAs differs significantly from that reported by Schwarzer et al.  who also used NHBE cultures. The data presented here show higher levels of the other Nox genes relative to Duox1 and Duox2, with Nox1 and Nox4 present at the highest levels. These data are mean values of triplicate cultures from three individual lung donors. The reason for this discrepancy is not apparent.
We previously showed that purinergic stimulation of airway epithelia results in acute increases in H2O2 production presumably via increases in intracellular [Ca2+] and increases in Duox activity via regulation through its EF-hand domains [6, 38]. Others have reported that flagellin-TLR5 interaction causes increases in intracellular Ca2+ ; however, we were unable to measure any acute (<10 min) increases in H2O2 production following flagellin stimulation (data not shown). Nevertheless, increases in Duox2 mRNA were detected within 6 h after flagellin application concomitant with increases in H2O2 production suggesting a rapid response to bacteria. In contrast to our data, Boots et al.  showed that Duox1 H2O2 production in HBE1 cells is acutely activated by stimulation with anti-ASGM1 that ligates the asialoGM1 receptor known to be involved in TLR5 activation. In this latter study, H2O2 production was assessed by peroxidase catalyzed formation of di-tyrosine that is highly sensitive and consumes H2O2 as it is formed and may measure H2O2 production not detectable during the 2 min assay period used here. Previous reports of flagellin stimulated Nox1 activity in intestinal epithelia suggested TLR5 mediated increases in Nox1 mRNA via the NF-κB pathway  although no quantitative measurements of mRNA were made in these latter studies. NHBE cultures also showed an increase in Nox1 mRNA (4-fold) after flagellin stimulation suggesting that similar regulation Nox1 and Duox2 occurred in NHBE cells.
The data presented here suggest that production of H2O2 is highly regulated. In addition reconstitution of an appropriate in vitro system is needed to accurately evaluate cellular responses to stimuli that alter the production of oxidants such as H2O2.
This work was supported by NIH grants HL066125 to GEC, HL-60644 and HL-89399 to MS, and DK076652 and DK057805 to PS. The authors thank Drs. Maria Elena Monzon and Marina Casalino Matsuda for helpful advice and Dr. Philip Whitney for IL-8 assays.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.