NEU1 Sialidase Regulates Membrane-tethered Mucin (MUC1) Ectodomain Adhesiveness for Pseudomonas aeruginosa and Decoy Receptor Release*♦

Background: Pseudomonas aeruginosa flagellin binds to the membrane-tethered mucin, MUC1. Results: Flagellin drives NEU1 to desialylate MUC1, thereby increasing its adhesiveness for Pseudomonas aeruginosa and its shedding. Conclusion: P. aeruginosa hijacks host NEU1 through its flagellin. Significance: P. aeruginosa mobilizes NEU1 to enhance its pathogenicity, but the host retaliates by releasing MUC1 as a hyperadhesive decoy receptor. Airway epithelia express sialylated receptors that recognize exogenous danger signals. Regulation of receptor responsiveness to these signals remains incompletely defined. Here, we explore the mechanisms through which the human sialidase, neuraminidase-1 (NEU1), promotes the interaction between the sialoprotein, mucin 1 (MUC1), and the opportunistic pathogen, Pseudomonas aeruginosa. P. aeruginosa flagellin engaged the MUC1 ectodomain (ED), increasing NEU1 association with MUC1. The flagellin stimulus increased the association of MUC1-ED with both NEU1 and its chaperone/transport protein, protective protein/cathepsin A. Scatchard analysis demonstrated NEU1-dependent increased binding affinity of flagellin to MUC1-expressing epithelia. NEU1-driven MUC1-ED desialylation rapidly increased P. aeruginosa adhesion to and invasion of the airway epithelium. MUC1-ED desialylation also increased its shedding, and the shed MUC1-ED competitively blocked P. aeruginosa adhesion to cell-associated MUC1-ED. Levels of desialylated MUC1-ED were elevated in the bronchoalveolar lavage fluid of mechanically ventilated patients with P. aeruginosa airway colonization. Preincubation of P. aeruginosa with these same ex vivo fluids competitively inhibited bacterial adhesion to airway epithelia, and MUC1-ED immunodepletion completely abrogated their inhibitory activity. These data indicate that a prokaryote, P. aeruginosa, in a ligand-specific manner, mobilizes eukaryotic NEU1 to enhance bacterial pathogenicity, but the host retaliates by releasing MUC1-ED into the airway lumen as a hyperadhesive decoy receptor.

Pseudomonas aeruginosa is a Gram-negative, flagellated, and opportunistic human pathogen that typically colonizes and/or infects debilitated and immunocompromised patients (1). In the respiratory tract, P. aeruginosa is one of the most common and lethal pathogens responsible for acute ventilator-associated pneumonia with directly attributable mortality rates of 40% (2). P. aeruginosa infections worsen the prognosis for bronchiectasis and chronic obstructive pulmonary disease patients (3). P. aeruginosa also adheres to and invades extrapulmonary epithelia (4 -8). Despite its recognized clinical impact, the molecular mechanisms that underlie P. aeruginosa pathogenesis and the host response to P. aeruginosa infection remain incompletely understood.
Bacterial adhesion to epithelial cells (EC) 2 is prerequisite to establishment of invasive infection and is mediated through interactions between microbial adhesins and their cognate host cell receptors (9). One P. aeruginosa adhesin, flagellin, is the structural protein that forms the major portion of the flagellar filament. Flagellin contributes to the virulence of pathogenic bacteria through increased motility, adhesion, and invasion (10). P. aeruginosa flagellin engages Toll-like receptor (TLR) 5 (11) and the transmembrane mucin 1 (MUC1) (12), and each receptor-ligand interaction is coupled to intracellular signaling. MUC1 consists of a Ͼ250-kDa ectodomain (ED), with a variable number of highly sialylated tandem repeats, which is proteolytically processed and shed from the EC surface (13). Three MUC1 sheddases have been identified, including matrix metalloproteinase (MMP) 14, a disintegrin and metalloproteinase (ADAM) 17, and ␥-secretase (14 -16).
Glycoprotein receptors for bacteria often contain glycan chains terminating with sialic acid (Sia). Here, Sia residues are strategically positioned to influence cell-cell and intermolecular interactions (17). Sia residues can mask binding sites for pathogens, their toxins, endogenous lectins, and protease recognition sites through protein conformational changes, electrostatic repulsion, and/or steric hindrance (18). The sialylation state of glycoconjugates is dynamically and coordinately regulated through the opposing catalytic activities of sialyltransferases and neuraminidases (NEU). NEUs constitute a large family of prokaryotic and eukaryotic glycolytic enzymes that hydrolyzes the linkages between Sia and its subterminal sugars (19).
Prokaryotic NEUs are established virulence factors for viral and bacterial pathogens (18). P. aeruginosa NEU, referred to as NanPs, contributes to bacterial pathogenesis and its expression has been linked to biofilm formation and airway colonization (20). Although much is known about prokaryotic NEUs as virulence factors, a role for mammalian host NEUs in bacterial pathogenesis, to our knowledge, has never been considered. Of the four known mammalian NEUs, NEU1 is the predominant sialidase expressed by human airway ECs, and the second most abundant, NEU3, is expressed at much lower levels (12). NEU1 is localized both to lysosomes and the cell surface (19) and is only active in association with its chaperone/transport protein, protective protein/cathepsin A (PPCA) (21). PPCA is a multipurpose protein that targets NEU1 to the lysosome and is absolutely required for proper folding, stability, oligomerization, and activation of NEU1 (21). We previously demonstrated intense NEU1 immunostaining at the superficial surface of the human airway epithelium, including the brush border of the trachea and bronchus (12). This NEU1 expression pattern closely correlated with that known for MUC1 in these same tissues (22,23). Furthermore, we established that forced NEU1 overexpression increases MUC1-ED desialylation and MUC1-ED-dependent P. aeruginosa adhesion to airway ECs in vitro (12). To extend these findings to a physiologically relevant context, we asked whether the MUC1 ligand, P. aeruginosa flagellin, might promote NEU1-mediated MUC1-ED desialylation and/or P. aeruginosa adhesion to and invasion of airway ECs. We now present evidence, for the first time, that a bacterial pathogen, P. aeruginosa, exploits the human sialidase, NEU1, to amplify its own pathogenicity through desialylation of its receptor, MUC1. At the same time, NEU1-mediated desialylation of MUC1-ED increases its shedding into the airway lumen to generate a soluble, hyperadhesive decoy receptor for P. aeruginosa. That NEU1 can act on a range of sialylated glycans, not only mucin-type, O-linked glycans abundant in MUC1 but also sialylated N-linked glycan chains, may implicate other sites through which NEU1 might influence the P. aeruginosa-airway EC interaction.
Bacterial Adhesion Assays-Human airway ECs and ECs infected with Ad constructs, transfected with siRNAs, and/or stimulated with flagellin (2.0 ϫ 10 5 ECs/well) were washed with PBS, fixed for 10 min with 2.5% glutaraldehyde, and washed, and bacterial adhesion was assayed as described (12). The ECs were incubated for 40 min at 37°C with bacteria (multiplicity of infection (m.o.i.) ϭ 100) and washed, and adherent bacteria were released with 0.05% trypsin, and colony forming units (CFUs) in the releasate were enumerated. Alternatively, the ECs were incubated with GFP-expressing P. aeruginosa (m.o.i. ϭ 100) and washed, and bound bacteria were examined by immunofluorescence microscopy. In other experiments, bacteria were preincubated for 30 min with culture supernatants from ECs infected with Ad constructs, transfected with siRNAs, and/or stimulated with flagellin or with human bronchoalveolar lavage fluids (BALFs) harvested from noncolonized patients, P. aeruginosa-colonized patients, or patients colonized with microorganisms other than P. aeruginosa, prior to adhesion assays. In still other experiments, EC culture supernatants or BALFs containing 2.5 g/ml MUC1-ED were preincubated overnight with 100 g/ml anti-MUC1-ED antibody or a species-and isotype-matched nonimmune IgG, after which Igs were immobilized on protein G-agarose for 2 h and removed by centrifugation prior to adhesion assays.
Knockdown of NEU1 and MUC1-Airway ECs (5.0 ϫ 10 5 ) were centrifuged, and the cell pellet was resuspended in 100 l of Amaxa Nucleofector solution (Lonza, Walkersville, MD) with 2.7 g of NEU1-targeting or control siRNAs as described (12). The EC-siRNA mixture was transferred to an Amaxa-certified cuvette and subjected to programmed electroporation, and the transfected ECs were cultured for 24 -72 h. In other experiments, airway ECs (5.0 ϫ 10 5 ) were transfected with 1.5 g of MUC1-targeting or control siRNAs using Lipo-fectamine2000 (Invitrogen) according to the manufacturer's instructions.
Immunoblotting Assays-Airway ECs infected with Ad constructs or transfected with siRNAs were lysed, and equal protein aliquots of lysates were resolved by SDS-PAGE and transferred to PVDF membranes. The membranes were probed with primary antibody followed by HRP-conjugated secondary antibody and enhanced chemiluminescence reagents as described (12).
Purification of P. aeruginosa Flagellin-An overnight culture of P. aeruginosa strain PA01 was centrifuged at 5,000 ϫ g for 30 min, resuspended in Krebs-Ringer buffer, and incubated for 1 h at 37°C as described (26). The bacteria were pelleted by centrifugation, and the supernatant was filtered through a 0.22-m pore membrane and the filtrate boiled for 20 min. The filtrate was concentrated by centrifugal ultrafiltration, adjusted to pH 6.0, and flagellin was purified by sequential ion exchange chromatography using Macro-Prep High S and Macro-Prep High Q support (Bio-Rad). Aliquots of column fractions were resolved by SDS-PAGE and stained with Coomassie Blue to detect the 50-kDa flagellin protein band. Other column aliquots were processed for flagellin immunoblotting and for P. aeruginosa pilin immunoblotting to confirm the absence of pilin contamination.
Flagellin-containing fractions were incubated with polymyxin B-agarose (Pierce) to remove lipopolysaccharide (LPS), after which less than 0.1 endotoxin unit/g of protein was detected by the Limulus amebocyte lysate assay.
Flagellin Binding Assays-Airway ECs infected with Ad constructs (2.0 ϫ 10 5 ECs/well) were exposed to Alexa Fluor 594labeled flagellin (26). The ECs were washed; bound flagellin was determined by fluorometry ( ex ϭ 591 nm, ex ϭ 615 nm), and Scatchard analysis of the binding data was performed.
qRT-PCR-Total cellular RNA was extracted from SAECs using TRIzol reagent (Invitrogen). RNA purity was established with the 260:280 nm absorption ratio (Ͼ1.90). Total RNA (1.0 g) was treated with DNase I (Invitrogen) for 15 min and reverse-transcribed using avian myeloblastosis virus-reverse transcriptase and poly(T) primer (Promega). The resulting cDNA was amplified by PCR using pre-validated RT 2 qPCR primers (Qiagen, Frederick, MD) for detection of transcripts encoding human MMP14, ADAM17, ␥-secretase (PSEN1, PSEN2), and 18S rRNA. All qPCR reagents were from Qiagen. Thermal cycling was performed using the StepOnePlus qPCR system (ThermoFisher Scientific). The 2 Ϫ⌬⌬Ctt method was used to assess the amplitude of changes in gene expression (12).
Lectin Blotting Assays-Airway ECs infected with Ad constructs or transfected with siRNAs were lysed and the lysates immunoprecipitated with anti-MUC1 antibody. The MUC1 immunoprecipitates were analyzed by MAL, SNA, or PNA lectin blotting as described (12). In other experiments, airway ECs and ECs transfected with NEU1-targeting or control siRNAs were incubated with flagellin or medium alone, washed, and lysed. The lysates were incubated with PNA-agarose and bound proteins processed for MUC1 immunoblotting.
Enzyme-linked Immunosorbent Assays (ELISAs)-A549 cells were incubated for 6 h with 10 ng/ml flagellin or medium alone, and culture supernatants were processed for IL-8 ELISA as described (28). For MUC1-ED ELISA, airway EC supernatants and BALFs were added to ELISA plates, and wells were blocked with PBS, pH 7.0, containing 10 mg/ml BSA, and washed with PBS containing 0.05% Tween 20 (PBS-T). The wells were incubated with anti-MUC1 antibody, washed with PBS-T, and incubated with peroxidase-conjugated goat anti-mouse IgG antibody. Bound antibodies were detected with tetramethylbenzidine substrate at A 450 . MUC1 levels were calculated from a standard curve of serial dilutions of known concentrations of purified MUC1.
BAL-BAL was performed on mechanically ventilated patients who underwent standard of care diagnostic bronchoscopy using a modification of our previously described procedure (29). The study was approved by the University of Maryland Institutional Review Board (protocol number HP-00059183). After conscious sedation with fentanyl and midazolam and local anesthesia with 2% lidocaine, the bron-choscope was wedged in a 3rd or 4th order bronchus, after which 125 ml of sterile, pyrogen-free 0.9% NaCl was injected in 25-ml aliquots. The BALF was retrieved with gentle suction and pooled; the volume was recorded and transported to the laboratory on ice. The BALF was filtered through sterile gauze, centrifuged at 450 ϫ g to remove cells, and the supernatants concentrated 25-fold by passage through membrane filters (pore size, 100 kDa) mounted in Centricon tubes (Millipore). Single use BALF aliquots were stored at Ϫ70°C. To identify microbial cultures, the BALF was plated onto culture media in the following order as described (30): CHROMagar Acinetobacter, trypticase soy agar with 5% sheep's blood, chocolate agar, and Mac-Conkey agar. BALF MUC1-ED levels were quantified by ELISA as described above.
Statistical Analysis-All values were expressed as means Ϯ S.E. Differences between means were compared using the Student's t test and considered significant at p Ͻ 0.05.
Flagellin Is Required for NEU1-responsive P. aeruginosa Adhesion and Invasion-Flagellin is necessary for P. aeruginosa adhesion to airway ECs (26), and NEU1 promotes this adhesion ( Fig. 1, B, D, E, and G) (12). Because P. aeruginosa expresses multiple EC adhesins (9), we asked whether flagellin is required for NEU1-responsive P. aeruginosa adhesion, and if so, might NEU1 increase the binding of flagellin itself to airway ECs. NEU1 overexpression of airway ECs increased the adhesion of flagellin-expressing P. aeruginosa but not that of the flagellindeficient P. aeruginosa fliC Ϫ isogenic mutant ( Fig. 2A). Because P. aeruginosa flagellin is an established ligand for TLR5 (11), we examined whether TLR5 contributes to the NEU1 effect on P. aeruginosa adhesion. Given that primary SAECs and A549 cells both co-express MUC1 and TLR5 (13), HEK293T cells that express extremely low or undetectable endogenous MUC1 or TLR5 were individually transfected for MUC1 expression ( Infection of MUC1-expressing HEK293T cells with Ad-NEU1 increased P. aeruginosa adhesion compared with Ad-GFP-infected cells, whereas Ad-NEU1 infection of TLR5-expressing HEK293T cells did not (Fig. 2C). Because flagellin is required for NEU1-regulated P. aeruginosa adhesion to airway ECs ( Fig.  2A), we tested whether flagellin is also required for NEU1-responsive P. aeruginosa invasion. NEU1 overexpression increased invasion of flagellin-expressing P. aeruginosa but not P. aeruginosa fliC Ϫ (Fig. 2D). Finally, we investigated whether NEU1 promotes the binding of purified flagellin to airway ECs. Our flagellin preparation migrated as a single band on Coomassie Blue-stained SDS-polyacrylamide gels (  (28). Infection of airway ECs with Ad-NEU1 increased the binding of Alexa Fluor 594-labeled flagellin compared with Ad-GFP-infected controls (Fig. 2H). Scatchard analysis of the binding data revealed a 2.9-fold reduction in the dissociation constant (K d ) for flagellin binding to Ad-NEU1-infected ECs (K d ϭ 2.54 nM) compared with Ad-GFP-infected ECs (K d ϭ 7.35 nM) (Fig. 2I). The sialylation state of a cell surface receptor can regulate its surface residency (35). However, the number of flagellin-binding sites on ECs infected with the two Ad constructs were comparable as indicated by extrapolation of the best fit binding curves to the abscissa (Ad-NEU1, 158,000 sites/EC; Ad-GFP, 168,000 sites/EC) (Fig. 2I, dashed lines). These data indicate that flagellin is the P. aeruginosa adhesin that participates in NEU1-responsive P. aeruginosa adhesion to and invasion of MUC1-expressing ECs and that NEU1 promotes flagellin binding to the ECs through increased binding affinity of its receptor, MUC1, not through increased receptor number. Although the binding affinity of monomeric flagellin for airway ECs was clearly increased by NEU1 expression, the magnitude of this effect with P. aeruginosa flagella is likely greater than that measured here given the clustering/density-related polyvalent binding interaction between the multimeric P. aeruginosa flagellum and the multiple tandem repeats of MUC1.

NEU1 Regulates P. aeruginosa Adhesion to Airway Epithelia
Levels of Desialylated MUC1-ED Are Increased in BALF from P. aeruginosa-colonized Patients-Because P. aeruginosa flagellin increases both MUC1-ED desialylation (Fig. 4, D-I) and its shedding (Fig. 6, F and G), we investigated whether MUC1-ED levels might be increased in BALF from patients with P. aeruginosa airway colonization. MUC1-ED levels were quantified by ELISA and normalized to total BALF protein, which ranged from 69 to 101 g/ml. Mean MUC1-ED levels were 9.9-fold higher (range, 5.7-17.2-fold) in BALF from P. aeruginosa-colonized patients compared with levels in BALF from noncolonized subjects, and 5.5-fold greater (range, 2.5-9.6-fold) compared with levels in BALF from patients colonized with non-P. aeruginosa microorganisms (S. aureus, K. pneumoniae, Klebsiella oxytoca, Proteus mirabilis, or Candida albicans) (Fig. 7A). MUC1-ED in BALF from P. aeruginosa-colonized patients was desialylated compared with MUC1 ED in BALF from these same two control groups (Fig. 7, B and C). These data indicate that our in vitro results can be extended to in vivo human pathophysiology and that the increased MUC1-ED shedding is specific for P. aeruginosa airway colonization. Although LPS and other bacterial constituents reportedly alter NEU1 activity (37), our studies with the LPS-expressing P. aeruginosa fliC Ϫ mutant (Figs. 2, A and D, and 3, A and B), purified polymyxin B preadsorbed flagellin (Figs. 3, C-F, 4, D-I, 5, 6, H and I, and 8B), and BALFs harvested from patients colonized with bacteria other than P. aeruginosa (Figs. 7, A-C, and 9A) indicate that contribution of bacterial components other than P. aeruginosa-derived flagellin is unlikely. Given the dilutional effect inherent to the BAL procedure, these results may underestimate the true levels of desialylated MUC1-ED in the BALF of P. aeruginosacolonized patients.
Desialylated MUC1-ED Shed in Response to Flagellin Competitively Inhibits P. aeruginosa Adhesion to EC-associated MUC1-ED-NEU1 desialylates the MUC1-ED, rendering it hyperadhesive for P. aeruginosa at 30 min (Figs. 1, B, D, E, and  G, 2, A and C, and 5, A-E), yet after 24 h it increases MUC1-ED shedding (Fig. 6, C, F, and G). To establish whether the shed, hyperadhesive MUC1-ED competitively inhibits P. aeruginosa adhesion to EC-associated MUC1-ED, P. aeruginosa was preincubated with supernatants of airway ECs infected with Ad-NEU1 prior to assaying for adhesion. A MUC1-ED dose-dependent reduction of bacterial adhesion to fresh EC monolayers was observed (Fig. 8A). Preincubation of P. aeruginosa with supernatants of ECs infected with Ad-GFP did not diminish adhesion. Preincubation with supernatants of flagellin-stimulated ECs blocked P. aeruginosa adhesion in a dose-dependent manner (Fig. 8B). Prior NEU1 silencing abolished the ability of supernatants of flagellin-stimulated ECs to reduce P. aeruginosa adhesion compared with supernatants of control siRNA-transfected ECs (Fig. 8B). Preincubation of supernatants of either Ad-NEU1-infected ECs or ECs stimulated by FIGURE 7. Levels of desialylated MUC1-ED are increased in BALF from P. aeruginosa-colonized patients. A, MUC1-ED levels in BALF from noncolonized patients, P. aeruginosa-colonized patients, or patients colonized with non-P. aeruginosa microorganisms were quantified by ELISA and normalized to total BALF protein. B, equal protein aliquots (100 g) of BALF from noncolonized patients, P. aeruginosa-colonized patients, or patients colonized with non-P. aeruginosa microorganisms were incubated with PNA-agarose and the PNA-binding proteins processed for MUC1-ED immunoblotting. C, densitometric analyses of the blots in B. Bars represent mean Ϯ S.E. MUC1-ED signal (n ϭ 3). The n for each condition is indicated. *, increased mean MUC1-ED levels or signals in BALF from P. aeruginosa-colonized patients compared with levels in BALF from noncolonized patients or patients colonized with non-P. aeruginosa (Non-Pa) microbes at p Ͻ 0.05. The results are representative of Ն2 independent experiments. flagellin, with anti-MUC1-ED antibody, depleted the samples of detectable MUC1-ED (Fig. 8C) and reduced their ability to inhibit P. aeruginosa adhesion compared with supernatants preincubated with the same concentration of a species-and isotype-matched nonimmune IgG (Fig. 8, D and E). Transfection of airway ECs with MUC1 siRNA also profoundly diminished the ability of their supernatants to inhibit P. aeruginosa adhesion compared with supernatants from control siRNAtransfected ECs (Fig. 8D). These data indicate that MUC1-ED shed in vitro in response to flagellin competitively inhibits P. aeruginosa adhesion to MUC1-ED-expressing airway ECs.
Desialylated MUC1-ED in BALF from P. aeruginosa-colonized Patients Competitively Inhibits P. aeruginosa Adhesion to EC-associated MUC1-ED-Because MUC1-ED shed in vitro in response to P. aeruginosa flagellin inhibits P. aeruginosa adhesion to EC-associated MUC1-ED (Fig. 8, B, D, and E), we examined whether BALF from P. aeruginosa-colonized patients containing increased levels of desialylated MUC1-ED might inhibit bacterial adhesion. P. aeruginosa adhesion was tested following preincubation of the bacteria with BALF from the same three groups of patients analyzed for desialylated MUC1-ED BALF levels (Fig. 7, A-C). Preincubation of P. aeruginosa with BALF from P. aeruginosa-colonized patients inhibited bacterial adhesion to airway ECs by 52.9%, compared with preincubation with BALF from noncolonized subjects, and by 42.1%, compared with preincubation with BALF from patients colonized with non-P. aeruginosa microorganisms (Fig. 9A, panel i). We next assessed the ability of these same BALFs to inhibit adhesion of the same non-P. aeruginosa bacteria tested for NEU1-responsive adhesion (Fig. 1B) to airway ECs. None of the BALFs tested inhibited the adhesion of these non-P. aeruginosa bacteria (Fig.  9A, panels ii-vi). Finally, preincubation of BALF from P. aeruginosa-colonized patients with anti-MUC1-ED antibody, which depleted the samples of detectable MUC1-ED (Fig. 9B), greatly reduced their ability to inhibit P. aeruginosa adhesion compared with preincubation of the same BALFs with the IgG control (Fig 9, C and D). These combined data indicate that desialylated MUC1-ED in BALF from P. aeruginosa-colonized patients acts as a soluble decoy receptor to competitively inhibit P. aeruginosa adhesion to MUC1-ED-expressing airway ECs i. ϭ 250) and cultured for 24 h, and EC culture supernatants were collected and protein concentrations measured. P. aeruginosa was preincubated for 30 min with supernatant from Ad-NEU1-infected ECs containing the indicated concentrations of MUC1-ED or supernatant from Ad-GFP-infected ECs containing an equivalent amount of total protein and assayed for adhesion to fresh A549 cell monolayers. B, A549 cells were transfected with NEU1-targeting or control siRNAs and incubated for 24 h. The ECs were incubated for 30 min with 30 ng/ml P. aeruginosa flagellin, cultured for 24 h, and EC culture supernatants collected, and protein concentrations were measured. P. aeruginosa was preincubated for 30 min with supernatant from control siRNA-transfected ECs containing the indicated concentrations of MUC1-ED, or supernatant from NEU1 siRNA-transfected ECs containing an equivalent amount of total protein, and assayed for adhesion to fresh A549 cell monolayers. **, decreased P. aeruginosa adhesion following preincubation with supernatants from Ad-NEU1-infected compared with Ad-GFP-infected ECs, or control siRNA-transfected compared with NEU1targeting siRNA-transfected ECs at p Ͻ 0.05. C, supernatants of A549 cells infected with Ad-NEU1 or stimulated by flagellin, each containing 2.5 g/ml MUC1-ED, were incubated with 100 g/ml anti-MUC1-ED antibody or a nonimmune IgG control. Immunoglobulins were immobilized on protein G-agarose and removed by centrifugation, and the supernatants were processed for MUC1-ED immunoblotting (IB). D, P. aeruginosa was preincubated for 30 min with the supernatants in C, or supernatants of A549 cells transfected with MUC1-targeting or control siRNAs, and assayed for adhesion to A549 cells. *, increased P. aeruginosa adhesion following MUC1-ED depletion compared with preincubation with IgG control or MUC1-targeting siRNA-transfected compared with control siRNA-transfected ECs at p Ͻ 0.05. A, B, and D, each bar represents mean Ϯ S.E. CFUs/well (n ϭ 6). E, supernatants of A549 cells infected with Ad-NEU1 or stimulated by flagellin were incubated with the IgG control (panels i and ii) or depleted with anti-MUC1-ED antibody (panels iii and iv). Immunoglobulins were immobilized on protein G-agarose and removed by centrifugation. GFP-P. aeruginosa were preincubated for 30 min with the supernatants, washed, and incubated with A549 cells. The ECs were washed, counterstained with DAPI, and processed for fluorescence microscopy. Scale bar, 25 m. and that MUC1-ED inhibitory activity for bacterial adhesion is specific for P. aeruginosa. Because the levels of total protein in the recovered BALFs reached 0.1 mg/ml, and the in vivo concentrations of shed MUC1-ED in BALF from P. aeruginosacolonized subjects, normalized to total protein, reached 4.3 g/mg (Fig. 7A), the levels of MUC1-ED in these same BALFs, up to 0.43 g/ml, approached those MUC1-ED concentrations capable of inhibiting P. aeruginosa adhesion in vitro (Յ0.63 g/ml) (Fig. 8, A and B). Again, given the dilutional effect inherent to the BAL procedure, combined with the known sequestration of a fraction of MUC1-ED within the fluid lining the bronchoalveolar compartment (38), these data likely underestimate the true levels of shed MUC1-ED in the BALF from P. aeruginosa-colonized patients.

Discussion
Invasive P. aeruginosa lung infections remain an important worldwide cause of morbidity and mortality in debilitated and immunocompromised patients despite advances in antibiotic therapy and supportive care. P. aeruginosa adhesion to airway epithelia is mediated through a direct interaction between its flagellin and the cell surface sialoprotein, MUC1. We now report that treatment of airway ECs with P. aeruginosa flagellin rapidly (Յ30 min) increases NEU1-dependent MUC1 desialylation, thereby promoting bacterial adhesion to and invasion of the ECs. We previously reported that under these identical experimental conditions, EC cytotoxicity in response to P. aeruginosa was not observed (31,32). Because flagellin is an FIGURE 9. Desialylated MUC1-ED in BALF from P. aeruginosa-colonized patients competitively inhibits P. aeruginosa adhesion to EC-associated MUC1-ED. A, indicated test bacteria were preincubated for 30 min with BALF from P. aeruginosa-colonized patients containing 2.5 g/ml MUC1-ED or BALF from noncolonized patients or patients colonized with non-P. aeruginosa microorganisms containing an equivalent amount of BALF total protein. The bacteria were washed and assayed for adhesion to A549 cells. **, decreased bacterial adhesion following preincubation with BALF from P. aeruginosacolonized patients compared with BALF from either noncolonized patients or patients colonized with non-P. aeruginosa microbes at p Ͻ 0.05. B, BALF from P. aeruginosa-colonized patients was incubated with anti-MUC1-ED antibody or a nonimmune IgG control. Immunoglobulins were immobilized on protein G-agarose and removed by centrifugation, and the BALF supernatants were processed for MUC1-ED immunoblotting. C, P. aeruginosa were preincubated for 30 min with the BALFs in B and assayed for adhesion to A549 cells. *, increased P. aeruginosa adhesion following MUC1-ED depletion compared with preincubation with IgG control. A and C, each bar represents mean Ϯ S.E. CFUs/well (n ϭ 6). The results are representative of Ն2 independent experiments. D, BALF from P. aeruginosa-colonized patients were incubated with the IgG control (panels i-iv) or depleted with anti-MUC1-ED antibody (panels v-viii). Immunoglobulins were immobilized on protein G-agarose and removed by centrifugation. GFP-P. aeruginosa were preincubated for 30 min with the BALFs, washed, and incubated with A549 cells. The ECs were washed, counterstained with DAPI, and processed for fluorescence microscopy. Scale bar, 25 m. established MUC1 ligand, these results support an amplification process through which P. aeruginosa increases its pathogenicity. However, at a later time (24 h), MUC1-ED desialylation also promotes its release from the EC surface. Although MUC1-ED shedding might be expected to reduce EC-associated MUC1-ED and P. aeruginosa adhesion, prior pulse-chase studies have demonstrated that shed MUC1-ED molecules are rapidly replaced by newly synthesized molecules exported to the cell surface (39). Shedding of the MUC1-P. aeruginosa complex into the airway lumen prior to bacterial invasion, together with liberation of unligated MUC1 as a soluble decoy receptor, might well reduce invasive infection in vivo. To our knowledge, this is the first report demonstrating that a prokaryotic organism, P. aeruginosa, uses its flagellin to hijack human eukaryotic NEU1 to increase MUC1-dependent P. aeruginosa adhesion and invasion, although the host responds by generating a hyperadhesive decoy receptor limiting invasive infection. Although MUC1 silencing clearly abrogated NEU1-mediated changes in P. aeruginosa adhesion and invasion, and MUC1 immunodepletion of BALFs greatly reduced their ability to inhibit P. aeruginosa adhesion, it is still conceivable that NEU1 might also desialylate O-and/or N-linked glycans of other soluble or membranebound mucins, or non-mucin sialoproteins, that in turn contribute to these changes.
That NEU1 overexpression enhanced airway EC adhesiveness for P. aeruginosa and its flagellin, and bacterial invasion of these same ECs, suggests that removal of terminal Sia residues unmasks a cryptic binding site(s) within the MUC1-ED comprised of one or more sugars subterminal to Sia and/or its protein backbone. A549 airway ECs express the sialyl-Thomsennouveau antigen (40), and creation of the Thomsen-nouveau antigen through NEU1-mediated desialylation might expose a binding site for P. aeruginosa and its flagellin. MUC1-ED exhibits a ␤-turn helix resulting from its high proline content (41). This extended rod-like conformation allows MUC1 to protrude higher than most membrane-associated proteins above the EC surface where it is strategically positioned to interact with P. aeruginosa in the airway lumen. The degree of MUC1 extension into the lumen depends upon the number of tandem repeat units, a genetically inherited allelic polymorphism varying from 20 to 120 repeats (13). Whether upon desialylation, MUC1 molecules with more tandem repeats exhibit greater adhesiveness for P. aeruginosa and/or superior decoy receptor function, or whether individuals expressing different repeat numbers have altered susceptibility to invasive P. aeruginosa lung infection, is unknown.
Analysis of the mouse NEU1 gene promoter has identified potential positive and negative cis-acting elements that might regulate NEU1 expression (42). Although flagellin failed to increase NEU1 expression or catalytic activity, it did promote MUC1-NEU1 association. That flagellin promoted this association within 30 min of stimulation diminishes the likelihood of, but does not entirely exclude, de novo NEU1 mRNA or protein synthesis as a source for the increased NEU1 associated with MUC1. Although LPS reportedly promotes NEU1 association with TLR4 (37), the fact that our flagellin preparation was pre-adsorbed with polymyxin B (26) reduces the possibility that LPS contamination of flagellin is responsible for MUC1-NEU1 association. It is possible that undetectable amounts of LPS might synergize with other unknown mediators to provoke this effect. Although we previously reported that flagellin engagement of MUC1-ED activates ERK1/2 (34), pharmacologic blockade of ERK1/2 activation failed to prevent flagellin-stimulated NEU1 association with MUC1. Others have reported that c-Src and EGF receptor each regulate MUC1-CD-dependent intracellular signaling (43,44). However, pharmacologic inhibition of either c-Src or EGF receptor activation did not block flagellininduced increases in NEU1-MUC1 co-immunoprecipitation (data not shown). It is conceivable that flagellin activates one or more signaling pathways, possibly through TLR5, that promotes the association of preformed NEU1 with MUC1. Because NEU1 exists in a complex with PPCA and ␤-galactosidase (21), MUC1-NEU1 association might be indirectly mediated through either of these or other NEU1-binding partners. Alternatively, MUC1-binding partners (e.g. glycogen synthase kinase 3␤, ␤-catenin) (13) may participate in the MUC1-NEU1 interaction.
P. aeruginosa NanPs might contribute to basal levels of P. aeruginosa adhesion seen in airway ECs in which NEU1 has been silenced (12). However, P. aeruginosa expression of NanPs has not been associated with increased bacterial adhesion to airway ECs (20). In contrast to other bacterial NEUs, which contain a catalytic pocket structurally similar to that of canonical NEUs, crystallographic analysis of NanPs has revealed a more open, atypical structure (45). We hypothesize that P. aeruginosa commandeers the human sialidase, NEU1, to compensate for its structurally and functionally distinct NanPs, to desialylate MUC1-ED, and to increase P. aeruginosa adhesion to and invasion of human airway ECs. Although manipulation of host cell pathways by bacteria to enhance their virulence has been recognized as a central tenet of bacterial pathogenesis for several decades (46), to the best of our knowledge these results are the first to implicate a eukaryotic NEU in bacterial pathogenesis. Interventions that target influenza-expressed NEU have entered clinical practice. Whether these NEU inhibitors interfere with the catalytic activity of human NEU1 or whether human NEU1-specific inhibitors can be designed are clinically important questions. Our results further suggest that the host responds to P. aeruginosa hijacking of human NEU1 by releasing desialylated MUC1-ED from the EC surface to competitively inhibit interaction between flagellin-expressing P. aeruginosa and EC-associated MUC1. It is tempting to speculate that the balance between these opposing processes, i.e. P. aeruginosa adhesiveness versus decoy receptor function, and how they are impacted by the activities of other NEUs, sialyltransferases, and/or proteases ultimately dictate the clinical outcome of an individual airway EC exposure to P. aeruginosa. Future studies are needed to establish whether similar pathogenic mechanism(s) might be extended to other flagellin-expressing airway pathogens not examined here, mucin gene products other than MUC1, additional adhesin-receptor interactions, and/or extrapulmonary mucosal surfaces.