Polarization of protease-activated receptor 2 (PAR-2) signaling is altered during airway epithelial remodeling and deciliation

Protease-activated receptor 2 (PAR-2) is activated by secreted proteases from immune cells or fungi. PAR-2 is normally expressed basolaterally in differentiated nasal ciliated cells. We hypothesized that epithelial remodeling during diseases characterized by cilial loss and squamous metaplasia may alter PAR-2 polarization. Here, using a fluorescent arrestin assay, we confirmed that the common fungal airway pathogen Aspergillus fumigatus activates heterologously-expressed PAR-2. Endogenous PAR-2 activation in submerged airway RPMI 2650 or NCI–H520 squamous cells increased intracellular calcium levels and granulocyte macrophage–colony-stimulating factor, tumor necrosis factor α, and interleukin (IL)-6 secretion. RPMI 2650 cells cultured at an air–liquid interface (ALI) responded to apically or basolaterally applied PAR-2 agonists. However, well-differentiated primary nasal epithelial ALIs responded only to basolateral PAR-2 stimulation, indicated by calcium elevation, increased cilia beat frequency, and increased fluid and cytokine secretion. We exposed primary cells to disease-related modifiers that alter epithelial morphology, including IL-13, cigarette smoke condensate, and retinoic acid deficiency, at concentrations and times that altered epithelial morphology without causing breakdown of the epithelial barrier to model early disease states. These altered primary cultures responded to both apical and basolateral PAR-2 stimulation. Imaging nasal polyps and control middle turbinate explants, we found that nasal polyps, but not turbinates, exhibit apical calcium responses to PAR-2 stimulation. However, isolated ciliated cells from both polyps and turbinates maintained basolateral PAR-2 polarization, suggesting that the calcium responses originated from nonciliated cells. Altered PAR-2 polarization in disease-remodeled epithelia may enhance apical responses and increase sensitivity to inhaled proteases.

We hypothesized basolateral polarization of PAR-2 may limit its activation except during times of epithelial barrier breakdown during infection or inflammation. Alteration of epithelial polarization or cell-type composition occurs in obstructive airway diseases such as asthma and COPD (9,10). Loss of airway cilia may occur with viral or bacterial infection (11)(12)(13)(14), type-2 inflammation-driven remodeling (15), or smoking (16,17). Squamous metaplasia occurs after bacterial infection, particularly in cystic fibrosis-related CRS (18), and is frequently reported in tissue samples from CRS patients (19,20).
Much of the literature on airway remodeling focuses on barrier dysfunction, but airway cells in vitro can maintain tight junctions and barrier function despite substantial morphological changes, including loss of cilia (12). Less is known about how signaling is altered before large-scale barrier breakdown in severe disease. We hypothesized that loss of epithelial polarization may affect PAR-2 signaling independently of barrier disruption, perhaps sensitizing cells to protease allergens.
We first tested activation of PAR-2 by Aspergillus fumigatus, the most prevalent fungal cause of pulmonary allergic disease, including allergic bronchopulmonary aspergillosis associated with chronic lung injury in patients with cystic fibrosis (CF) or chronic asthma (21). We next examined PAR-2 signaling in airway squamous cells using pharmacology and live-cell imaging. We then tested PAR-2 signaling in well-differentiated airliquid interface (ALI) cultures of primary nasal cells exposed to disease-related modifiers, including IL-13, CSC, and retinoic acid deficiency, all of which alter epithelial morphology. We also examined isolated tissue from inflamed nasal polyp and compared it with control middle turbinate. The data below reveal important alterations of the polarity of PAR-2 signaling in squamous cells and primary epithelial cells de-differentiated through disease-relevant mechanisms.

A. fumigatus conditioned media can directly activate PAR-2
Although Alternaria proteases are well-documented to activate PAR-2 (22,23), evidence exists that A. fumigatus also activates PAR-2 in the airway (7) and cornea (24). A. fumigatus is a common cause of allergic bronchopulmonary aspergillosis or allergic fungal rhinosinusitis (21) and can cause fatal invasive aspergillosis (25). We sought to first confirm whether A. fumigatus produces PAR-2-activating proteases, as this is highly important for considering the role of PAR-2 in aspergillosis and fungal sinusitis.
We utilized a fluorogenic assay (Trio assay) to directly visualize PAR-2 activation by its interaction with ␤-arrestin (26), which associates with activated and phosphorylated GPCRs and mediates further signal transduction and/or internalization to endosomes (Fig. 1a) (27). This assay is based on a tripartite GFP, with the 11th ␤-strand of GFP fused to the C terminus of PAR-2, the 10th ␤-strand of GFP fused to ␤-arrestin, and ␤-strands 1-9 of GFP expressed as a soluble protein (26). When PAR-2 is activated, association of PAR-2 with ␤-arrestin allows a complete GFP to form and increases fluorescence (Fig. 1b). This approach eliminates other effects that may impinge on downstream signaling (e.g. toll-like receptor activation during exposure to fungus-conditioned media) by allowing us to look The receptor is co-expressed with ␤-arrestin tagged with the ␤10-strand of GFP along with soluble GFP ␤-strands 1-9. Association of the PAR-2 with ␤-arrestin allows the formation of a complete fluorescent GFP molecule; soluble mCherry is included on the plasmid as a transfection control (not shown in the model). c, representative images of A549 cells expressing PAR-2 Trio components (plus mCherry as transfection control) stimulated with 10 M 2FLI for 0 -120 min. Scale bars are 15 m. d, normalized GFP fluorescence in cells stimulated with 2FLI or buffer alone (HBSS) at time points taken over 2 h. Data points are independent experiments imaged on three different days (n ϭ 5-9). a and b were created with Biorender.com.

PAR-2 polarization changes with loss of airway cilia
directly at the binding of arrestin to the receptor after receptor activation. We visualized PAR-2 activation via appearance of GFP fluorescence in A549 alveolar-like cells (Fig. 1c). Stimulation with PAR-2 peptide agonist 2FLI resulted in the appearance of punctate GFP fluorescence over the course of 2 h (Fig. 1, c and d).
Similarly, in PAR-2 Trio-transfected BEAS-2B immortalized bronchial cells, we also observed activation of GFP fluorescence by 2FLI but not AY-NH 2 (Fig. 3, a-f), supporting results observed in A549 cells. We likewise observed PAR-2 activation by A. fumigatus but not A. niger in BEAS-2B cells (Fig. 3, g-k).
Controversy exists over whether certain host and pathogen proteases activate PAR-2 by cleavage at the activating site or inhibit PAR-2 by cleavage downstream, removing the tethered ligand and "disarming" it (28). Neutrophil and Pseudomonas aeruginosa elastase have both been suggested to either activate (4,29,30) or disarm PAR-2 (31,32). Differential results may be due to altered protein processing in different cell types or the different assays used. We tested whether several proteases implicated in allergy can activate PAR-2 expressed in airway cells using this assay. We observed PAR-2 activation with human lung tryptase, neutrophil elastase, and dust mite protease Der p 3 in BEAS-2B cells (Fig. 3, l-r). Together, these data verify activation of PAR-2 by A. fumigatus as well as common lung proteases.
To test whether A. fumigatus activates PAR-2 G-protein signaling as well as arrestin binding, we imaged A549s loaded with the calcium indicator Calbryte 590. A red calcium indicator was used to reduce effects of autofluorescence of the fungal media during acute addition. We and others previously showed that PAR-2 is linked to calcium in airway epithelial cells (8). It is important to note that arrestin binding and G-protein signaling are not mutually exclusive, and our goal was not to quantify the relative contributions of these two pathways (e.g. agonist biasing) in this study. Although the Trio assay above allowed examination of heterologous PAR-2 directly (via arrestin binding), measuring calcium responses as a readout of GPCR-driven G-protein signaling allows examination of signaling of endogenous PAR-2.
Stimulation of untransfected A549s with dilute (25%) A. fumigatus media activated a calcium response that was reduced ϳ50% by incubation with PAR-2 antagonists FSLLRY-NH 2 or AZ3451 or by heat treatment (Fig. 4, a and b). This is likely because A. fumigatus media contain multiple secreted products and metabolites that can impinge on calcium, including proteases that appear to activate PAR-2. When A. fumigatus CM was dialyzed overnight against HBSS to remove small molecules and equilibrate salts but to retain larger proteins (6 -8 MWCO dialysis tubing), we observed calcium signals that were more completely inhibited by FSLLRY-NH 2 or heat treatment (Fig. 4, c and d). These results confirm activation of PAR-2 by A. fumigatus (7,24) and support a potential role for this protein in the airway response to fungal pathogens beyond Alternaria. We next sought to examine how PAR-2 signaling might change in airway squamous cells, as squamous epithelial de-differentiation is commonly observed in airway diseases, as described above.
We also observed activation of PAR-2 by human lung tryptase and Der p 3 (Fig. S4). PAR-2 activation also had a small but significant effect of acutely-reducing squamous cell line metabolism and/or proliferation through lowering of cAMP, but did not activate apoptosis (Fig. S5). PAR-2 activation increased secretion of TNF␣ and GM-CSF from RPMI 2650 cells and IL-6 and GM-CSF from NCI-H520 cells, likely involving dual coupling to G q and G i G-protein isoforms (Fig. S6). Thus, PAR-2 in airway squamous cells may regulate inflammation. To understand whether this is reflective of squamous phenotype or the tumor origin of these cells, we next tested how PAR-2 signaling and polarization change with epithelial de-differentiation and/or remodeling using primary cells.

Primary airway epithelially-ciliated ALI cultures can be dedifferentiated/remodeled by exposure to a type 2 cytokine or submersion
To begin to understand how epithelial remodeling changes polarization of PAR-2 in primary sinonasal cells, we initially de-differentiated/remodeled primary ciliated ALI cultures by two methods. The first was exposure to IL-13 (10 ng/ml, basolateral), which induces cilia loss and goblet cell metaplasia (33,34). The second was apical submersion, which induces a more squamous phenotype (33,35). ALIs were cultured for 21 days to establish a well-ciliated baseline (Fig. 5a). IL-13 or apical submersion both resulted in a decrease in the number of ciliated cells over 6 days, shown by a reduction in ␤-tubulin IV immunofluorescence (Fig. 5, a and b). Although IL-13 increased goblet cell-specific Muc5AC intensity (Fig. 5, a and c), submersion only minimally increased Muc5AC at 6 days ( Fig. 5, a and c). Submersion instead increased expression of transglutaminase 1 (TG-1; measured via ELISA), previously reported to reflect squamous differentiation (Fig. 5d) (36,37). Transepithelial electrical resistance (TEER) remained intact up to 4 days (Fig.  5e), and maintenance of epithelial barrier function was confirmed by measurement of FITC-conjugated dextran flux at day 4 as well as ability of cells to maintain low apical glucose levels PAR-2 polarization changes with loss of airway cilia (Fig. S7a). Evidence of cell death (LDH release) was not observed (Fig. S7b).
We thus chose the 4-day exposure time to IL-13 or submersion (21 days of differentiation plus a further 4 days of IL-13 or submersion, here termed Day 21 ϩ 4) for further experiments as it was a time point when the epithelium was remodeled but barrier function remained intact. At this time point, we saw no differences in expression of PAR-2 by qPCR (Fig. S7c), which mimicked a lack of differences observed in nasal polyp tissue (frequently characterized by type 2 inflammation) and control middle turbinate tissue (Fig. S7c), in contrast to another study (38). As a control, we did see increased expression of PAR-2 in ALIs after stimulation with lipopolysaccharide, agreeing with previous results from bronchial fibroblasts (Fig. S7c) (39).  (26, 29, 30, 74 -78). Using this assay allows us to directly look at PAR-2 activation in the absence of any other signals downstream of other targets of proteases and/or CM that could impinge on the PAR-2 signaling pathway. a, after 90 min stimulation with PAR-2 agonist 2FLI, GFP fluorescence was increased ϳ25-fold compared with buffer (HBSS) only or PAR-4 -activating peptide AY-NH 2 . The GFP fluorescence increase, signaling PAR-2 activation, was lost when cells were stimulated with 2FLI in the presence of 10 M PAR-2 antagonist FSLLRY-NH 2 . b, bar graph showing data from experiments as in a with alternative PAR-2 agonists AC 55541 and SLIGRL-NH 2 plus scrambled LRGILS-NH 2 and cholinergic agonist CCh, histamine, and purinergic agonist ATP (100 M each). c, representative images of PAR-2 activation with A. fumigatus (A. fum.) CM. Cells were exposed to conditioned media for 10 min followed by washing and resuspension in HBSS for 90 min. Bar graph on right shows normalized GFP fluorescence. Significance was determined by one-way ANOVA with Dunnett's post-test comparing values to media only control; **, p Ͻ 0.01. d, A549 cells were transfected with ␤2AR Trio. A. fumigatus CM did not activate ␤2AR but 100 nM isoproterenol did. Bar graph shows quantification of results. All bars show data points from Ն7 independent experiments imaging fields from independent transfected wells on separate days. Significance was determined by one-way ANOVA with Dunnett's post-test comparing all values to HBSS alone; **, p Ͻ 0.01. Scale bars are 15 m.

PAR-2 polarization changes with loss of airway cilia Epithelial remodeling is associated with altered polarization of PAR-2 signaling
In primary ALIs at 5 days after seeding, when ALIs have electrical resistance but cilia have not yet developed, both apical or basolateral 2FLI or trypsin induced calcium responses (Fig. 6, a and f). After a saturating dose of apical 2FLI, another increase in apical 2FLI did not elicit a further response, whereas application of basolateral 2FLI did (Fig. 6a, left). This suggests that two separate pools of PAR-2 receptors exist, separated by the apical-to-basolateral tight-junction diffusion barrier. By day 21, responses to apical 2FLI or trypsin were lost, whereas responses to basolateral stimulation were intact (Fig. 6, b and f).  (26). c-f, after 90 min stimulation, 2FLI (c) but not HBSS alone (d) or AY-NH 2 (e) resulted in ϳ20-fold increased GFP fluorescence, quantified in f. g-k, cells were stimulated for 10 min with 25% A. fumigatus (A. fum.) CM (g, strain 13073 left and 1022 right), heat-inactivated A. fumigatus CM (h; 20 min; 100°C; strain 13073 left and 1022 right), A. niger CM (i), or media only (j) diluted in HBSS, followed by washing with HBSS and incubation for 90 min. GFP quantification is shown in k. l-r, cells were stimulated for 10 min with 25 nM trypsin (l), human lung tryptase (m), neutrophil elastase (n), Der p 3 (o), thrombin (p), or heat-inactivated trypsin (q), followed by washing with HBSS and incubation for 90 min. Quantification of GFP fluorescence is shown in r. Data points in b, f, k, and f are independent experiments performed on different days (n Ն 5). Significance was determined by one-way ANOVA with Dunnett's post-test comparing all values to HBSS alone; **, p Ͻ 0.01. Scale bars are 15 m.

PAR-2 polarization changes with loss of airway cilia
If primary cells were incubated for a further 4 days with the apical side submerged (Day 21 ϩ 4 submerged) to promote squamous de-differentiation, the response to apical 2FLI was again observed (Fig. 6, c and f). Basolateral application of antagonist FSLLRY-NH 2 blocked the response to basolateral but not apical 2FLI (Fig. 6c, right), ruling out 2FLI diffusion across the epithelial barrier to the basolateral side. Similar results were seen at ALIs after four subsequent days of basolateral IL-13 ( Fig.  6d). If primary ALIs were carried out to day 25 with no IL-13 and no apical submersion, only basolateral 2FLI responses were observed (Fig. 6f). RPMI 2650 squamous cells cultured at ALI also had 2FLI responses with application to either the basolateral or apical side; basolateral addition of antagonist blocked basolateral but not apical responses (Fig. 6, e and f). No significant differences in TEERs were observed (Fig. 6g).

Altered PAR-2 polarization increases epithelial fluid secretion during apical exposure to a PAR-2 agonist
We measured airway-surface liquid (ASL) height in primary ALIs. Physiological ASL was labeled with Texas Red dextran sonicated in perfluorocarbon (PFC), and ASL height was measured from orthogonal confocal slices. Primary ALIs were compared at differentiation day 25 (Fig. 7a) or after 21 days ϩ 4 subsequent days IL-13 (Fig. 7b). Baseline ASL height was not different. Basolateral 2FLI caused an increase in ASL height in both groups (Fig. 7, a-c) that was inhibited by basolateral application of the Na ϩ K ϩ 2Cl Ϫ co-transporter inhibitor bumetanide (Fig. 7c), showing that this reflected fluid secretion. However, when 2FLI was added to the apical side in PFC, IL-13-treated cultures, but not control cultures, exhibited increased ASL height (Fig. 7, a-c) that was blocked by calcium-activated Cl Ϫ channel (CaCC) inhibitor CaCC inh -A01 but not CFTR inh -172 ( Fig. 7, b and c).
Similar responses to apical 2FLI were observed after 21 days of differentiation ϩ 4 days of submersion (Fig. 7c). We performed experiments using cells from CF patients homozygous for ⌬F508 CFTR. CF ALIs robustly secreted fluid in response to PAR-2 stimulation (Fig. 7d). The responses to 2FLI were inhibited by the PAR-2 antagonist FSLLRY-NH 2 (Fig. 7e), confirming they were indeed due to PAR-2. These data suggest that apical PAR-2 agonist exposure does not result in epithelial fluid secretion in well-differentiated cultures, but remodeling or dedifferentiation results in apical PAR-2-activated fluid secretion via CaCC.

Epithelial remodeling and altered PAR-2 polarization enhances cytokine secretion during apical exposure to PAR-2 agonists
To further examine the effects of PAR-2 signaling alterations above, we examined secretion of TGF-␤2, an important epithelium-derived cytokine in type 2 airway diseases (40). In RPMI 2650 ALIs, TGF-␤2 secretion was activated by basolateral 2FLI and inhibited ϳ50% by the G i inhibitor pertussis toxin (PTX; Fig. 8a). In contrast, activation of the TNF␣ receptor, which is not a GPCR, increased TGF-␤2 that was not inhibited by PTX (Fig. 8a). In primary differentiated ALIs, apical 2FLI had no effect, but basolateral 2FLI increased TGF-␤2 that was inhibited by basolateral but not apical FSLLRY-NH 2 (Fig. 8b). Similarly, apical trypsin or lung tryptase had little effect, whereas basolateral trypsin and tryptase increased TGF-␤2 secretion (Fig. 8b).

PAR-2 polarization changes with loss of airway cilia
the apical response to PAR-2, we tested TGF-␤2 secretion and found both basolateral responses and apical responses were blocked by PAR-2 antagonist FSLLRY-NH 2 (Fig. 8d).

Impaired ciliogenesis resulting from CSC exposure or retinoic acid deficiency also correlates with altered polarization of PAR-2
The above data suggest that epithelial de-differentiation or remodeling affects the polarization of PAR-2 and may sensitize epithelial cells to protease allergens by allowing apical PAR-2 activation. We tested whether other alterations of epithelial development and ciliogenesis cause this phenotype. We used CSC, shown to inhibit ciliogenesis in nasal ALIs (17). Human ALIs cultured in the presence of 10 -30 g/ml CSC exhibited impaired ciliogenesis at day 20 ( Fig. 9, a and b). Higher concentrations increased epithelial permeability, but barrier function remained intact up to 30 g/ml CSC (Fig. 9, c and d). CSC exposure did not affect baseline-resting intracellular calcium concentration ([Ca 2ϩ ] i , Fig. 9e).
On day 20, human nasal ALIs cultured in 10 -30 g/ml CSC, apical 2FLI, Der 3 p, and tryptase induced calcium responses, in cultures (starting at day 21) exposed to subsequent IL-13, apical submersion, or no treatment (control). Ten fields from one ALI from one patient were imaged and averaged for an independent experiment; results shown are mean Ϯ S.E. of 4 -6 independent experiments (4 -6 patients). d, squamous marker TG-1 quantified by ELISA at day 25 (21 days at air and then 4 subsequent days of IL-13, submersion, or no treatment). Each data point is an independent ALI from a different patient (n ϭ 5 total ALIs). Significance was determined by one-way ANOVA with Dunnett's post-test; **, p Ͻ 0.01. e, TEER at time points as in b and c. Significance was determined by one-way ANOVA with Bonferroni post-test; *, p Ͻ 0.05, and **, p Ͻ 0.01.

PAR-2 polarization changes with loss of airway cilia
whereas control cells (0 g/ml CSC) did not respond (Fig. 9f). The apical responses observed in CSC-treated cultures were blocked by the PAR-2 antagonist FSLLRY-NH 2 (Fig. 9g). As PAR-2 is partially coupled to inhibitory G␣ proteins (G␣ i ) that inhibit adenylyl cyclase, we looked at the ability of apical 2FLI exposure to reduce cellular cAMP levels. However, as baseline cAMP levels are low, we co-stimulated ALIs with a low level of adenylyl cyclase-activating compound forskolin (1 M) to elevate cAMP. We found cAMP was ϳ50% reduced in CSCtreated but not untreated ALIs with apical exposure to 2FLI or PAR-2 agonist AC 55541, fitting with apical activation of PAR-2 signaling only in CSC-treated cells (Fig. 9h). Reductions in cAMP were inhibited by FSLLRY-NH 2 (Fig. 9h), demonstrating PAR-2 specificity.
Apical PAR-2 activation also induced TGF-␤2 and GM-CSF in human nasal ALIs cultured in 10 -30 g/ml CSC, whereas control cells responded only to basolateral PAR-2 stimulation (Fig. 9i). Similar results were observed in mouse ALIs and par-2 Ϫ/Ϫ mice, again confirming that responses to 2FLI were indeed due to PAR-2 (Fig. S9). Together, these data suggest that cigarette smoke-induced alterations of epithelial morphology may change the polarity of PAR-2 in airway epithe-lial cells well before the onset of full epithelial breakdown. This may contribute to inflammation and disease progression. We next tested another disease-relevant modifier of epithelial differentiation.

PAR-2 polarization changes with loss of airway cilia
2FLI response observed in low R.A. cultures was blocked by FSLLRY-NH 2 (Fig. 10f), demonstrating PAR-2 dependence of this effect. Calcium and cAMP responses correlated with GM-CSF secretion at 24 h (Fig. 10g). Thus, impairment of ciliogenesis by CSC or reduced R.A. can both alter epithelial polarity of PAR-2 signaling.

PAR-2 polarization changes with loss of airway cilia Patient tissue explants from polyps, but not control turbinate, exhibit apical PAR-2 responses
To further test whether epithelial remodeling observed in airway diseases translates to altered PAR-2 responses, we took tissue explants from nasal polyps, removed from CRS patients during functional endoscopic sinus surgery, and control middle turbinate tissue, removed from patients during surgery for other indications (e.g. trans-nasal approaches to the skull base to remove pituitary tumors). Tissue was mounted, imaged by confocal microscopy, and stimulated apically with PAR-2 agonists. Polyp tissue exhibited apical responses to 2FLI, trypsin, or Der p 3, but not thrombin (Fig. 11, a and b). Control uninflamed middle turbinate tissue did not (Fig. 11c). Data are summarized in (Fig. 11d). The apical responses to PAR-2 agonists 2FLI or SLIGRL-NH 2 in polyp tissue were inhibited by PAR-2 antagonist FSLLRY-NH 2 (Fig. 11e). Scrambled agonist peptide LRG-ILS-NH 2 had no effect on Ca 2ϩ (Fig. 11e).
Immunofluorescence revealed that PAR-2 was nonetheless basolaterally expressed in ciliated cells from nasal polyps (Fig.  11f), as we previously reported for nonpolyp tissue (8). Moreover, the calcium responses to PAR-2 stimulation were not enhanced in isolated polyp versus turbinate ciliated cells (Fig.  11g). We thus hypothesized that apical PAR-2 responses do not originate from within the ciliated cells themselves, but rather may occur through de-differentiation and/or remodeling of the epithelium toward a more squamous phenotype.

Apical PAR-2 stimulation requires intercellular communication to increase ciliary beat frequency
To test whether ciliated cells autonomously respond to apical PAR-2 stimulation, we measured CBF to apical PAR-2 agonists in control and IL-13-treated cultures. We combined CBF measurement with two structurally-distinct gap junction inhibitors, steroid-like small molecule inhibitor carbenoxolone (CBX) and peptide inhibitor Gap 27 (46). Control cultures exhibited CBF increases with basolateral but not apical 2FLI; basolateral responses were unaffected by CBX (Fig. 12, a, b and  g). IL-13-treated cultures exhibited CBF increase with apical 2FLI (Fig. 12c) inhibited by CBX or Gap 27 (Fig. 12, d-g). Basolateral 2FLI responses were not inhibited by CBX (Fig. 12g). Apical ATP responses were inhibited by neither CBX nor Gap 27 (Fig. 12g). These data support a model by which well-differentiated airway epithelial tissue (Fig. 12h) expresses basolateral PAR-2 within ciliated cells, as CBF increases do not require intercellular gap junction communication (Fig. 12h). However, dedifferentiation of the epithelium results in apical expression of PAR-2 within nonciliated cells that can increase CBF in ciliated cells only in the presence of functional gap junctions, likely to allow diffusion of calcium and/or inositol 1,4,5-trisphosphate (Fig. 12i).

Discussion
Prior literature on PAR-2 in the airway has focused on increased PAR-2 expression in inflamed airway mucosa (2,38,47). However, we show that a potentially important overlooked consideration is the polarity of PAR-2 signaling, which can change in the absence of PAR-2 expression level changes and in the absence of epithelial barrier breakdown. This may represent a novel mechanism by which the airway epithelium can become responsive to apical exposure to inspired/inhaled proteases such as dust mite or Alternaria proteases. Our study suggests that changes in airway epithelial composition can have marked effects on polarity even before the basolateral membrane becomes accessible with epithelial breakdown. Such processes may contribute to the early pathogenesis of inflammatory airway diseases even before chronic inflammation and reduced barrier function.
We also confirm previous observations that PAR-2 is a player in detection of A. fumigatus using an optical assay that can detect PAR-2 activation independent of other effects A. fumigatus conditioned media may have. Furthermore, we defined mechanisms of PAR-2 inflammation and fluid secretion in well-differentiated sinonasal cells cultures at ALI. Although previous studies have suggested PAR-2 activation can stimulate fluid secretion that involves CFTR (48,49), we only observed CaCC-dependent secretion, suggesting this pathway would not be altered in patients with CF due to loss of CFTR function.
Our data suggest that such alterations of epithelial morphology and composition due to inflammation, exposure to toxic compounds, or other environmental factors can all alter PAR-2 polarity (Fig. 13). This may occur through increased Th 2 cytokines, cigarette smoke exposure, or vitamin A deficiency. Exposure to inhaled allergens like dust mite or fungal proteases may normally be tolerated in healthy airway tissue if these allergens are cleared by mucociliary clearance before they can degrade the epithelial barrier and activate basolateral PAR-2. Moreover, airway submucosal gland secretions contain a variety of protease inhibitors that could neutralize pathogen proteases upon contact with mucus (50,51). However, in some obstructive airway diseases, altered polarization of PAR-2 may combine with impaired protease inhibitor secretion due to gland duct plugging often observed (52)(53)(54). These phenotypes may result in epithelial sensitization to proteases at the apical membrane, increasing inflammatory responses. . c, peak ASL heights from independent experiments, each the average of 10 ASL measurements from one ALI; there were six experiments per condition. Note no inhibition of secretion with CFTR inh 172 (ϳ10 M), but there was inhibition with NKCC1 inhibitor bumetanide (100 M) or PAR-2 antagonist FSLLRY-NH 2 (ϳ10 M). Drugs applied to the apical (ap.) side were sonicated in perfluorocarbon to not disturb the aqueous ASL layer; thus, apical concentrations are approximate. d, data experiments using ⌬F508/⌬F508 homozygous CF ALIs. Note no response in CF cultures to basolateral (bl.) forskolin (20 M), which elevates cAMP and activates CFTR (non-CF cultures responded in c). Each point is an independent experiment using cells from one patient (five patients total). e, experiments similar to c were performed Ϯ PAR-2 antagonist FSLLRY-NH 2 to demonstrate PAR-2 specificity of the response in both apical and basolateral 2FLI exposure. Each point is an independent experiment using an ALI culture from one patient (four to six ALIs from four to six patients per condition). Significance was determined in c-e by one-way ANOVA with Dunnett's post-test; **, p Ͻ 0.01 versus control (no stim) for each condition.

PAR-2 polarization changes with loss of airway cilia
Squamous de-differentiation is a commonly-observed histological finding in CRS patients (19,20), CF patients (55), or smokers (16,56). Squamous metaplasia also occurs during epithelial injury and repair (15). Prior studies have focused on the loss of ciliated cells and resulting reduced mucociliary clearance, but we show here that introduction of squamous cells into the epithelium can alter airway epithelial signaling, including an increase in apical PAR-2 responsiveness. This suggests disruption of the normal epithelial composition during disease has physiological consequences beyond reduced mucociliary clearance.
RPMI 2650 cells were proposed as a model of nasal permeability for drug transport studies (57, 58) and cytokine release in

c, bar graph
showing TGF-␤2 secretion with apical versus basolateral 2FLI and combinatorial effects with TNF␣. RPMI 2650 squamous ALIs and primary ALIs exposed to 4 days of submersion to induce squamous differentiation exhibited apical 2FLI responses but control ALIs did not. d, bar graph showing TGF-␤2 secretion with apical versus basolateral 2FLI as in b Ϯ FSLLRY-NH 2 (50 M). Each point is an independent experiment using an ALI culture from one patient (four to six ALIs from four to six patients per condition). Significance was determined in a-d by one-way ANOVA with Dunnett's post-test comparing values to the respective control; **, p Ͻ 0.01. a, ##, p Ͻ 0.01 between basolateral 2FLI and 2FLI with PTX by one-way ANOVA with Bonferroni post-test. c, ##, p Ͻ 0.01 versus TNF␣ alone by one-way ANOVA with Bonferroni post-test. allergy (59) or viral infection (60). Our data here suggest that RPMI 2650 PAR-2 polarization is similar to squamous epithelial cells but not the same as well-differentiated ciliated cells, so interpretation of RPMI 2650 studies should be done cautiously. Altered polarization may extend to multidrug resistance proteins or other transporters or channels relevant for epithelial absorption. RPMI 2650 studies might not reflect healthy sinonasal epithelium but may reflect inflamed remodeled epithelium such as polyp tissue.

PAR-2 polarization changes with loss of airway cilia
Topical long-lasting PAR-2 antagonists might be useful in CRS patients with fungal or dust mite-activated allergic rhinitis by limiting apical PAR-2 activation, particularly if antagonists

PAR-2 polarization changes with loss of airway cilia
could be designed to not cross the epithelial barrier and block basolateral PAR-2 (e.g. a dextran conjugate), and delivered via topical nasal rinse. Such antagonists delivered via inhaler may be similarly useful in patients with allergic asthma. This requires further investigation to test the clinical utility of this strategy. Alternatively, restoring a normal differentiated epithelial phenotype by reducing inflammation may lower the epithelial sensitivity to proteases.
Multiple PAR-2-activating agonists were used in this study to ensure results were not an artifact of any one peptide or protease. PAR-2 antagonist FSLLRY-NH 2 (catalog no. 4751/1), PAR-2 agonist SLIGRL-NH 2 (catalog no. 1468/1; EC 50  as stock 40 mg/ml in DMSO stored at Ϫ80°C. Unless indicated below, all other regents were from Sigma-Aldrich.

Cell line culture
Cell line culture was as described previously (8,61,62). Briefly, RPMI 2650 and NCI-H520 squamous cells as well as BEAS-2B and A549 cells were obtained from ATCC (Manassas, VA). 16HBE cells (SV-40 immortalized normal human bronchia) were obtained from D. Gruenert (University of California San Francisco, San Francisco). Cells were grown in submersion in Minimal Essential Media with Earle's salts (Gibco; Gaithersburg, MD) supplemented with 10% FetalPlex (GeminiBioproducts; West Sacramento, CA) and 1% penicillin/streptomycin mix (Gibco). RPMI 2650 and 16HBE cells were seeded for ALI culture onto Corning transwells (0.3-cm 2 surface area; 0.4-m pore size) (Corning, NY) and grown for 5 days in the above media to confluence. Upon apical exposure to air, the basolateral medium was switched to primary cell differentiation media indicated below. ALIs were fed three times weekly. Transfection of submerged BEAS-2B and A549 cells was carried out in 8-well chamber slides (CellVis) with Lipofectamine 3000 as described previously (61).
XTT assays were carried out as described in the manufacturer's instructions using phenol-free DMEM (5.5 mM Glc; Gibco). Changes in XTT fluorescence were measured in real time in a Tecan Spark 10M plate reader equipped with heating (37°C) and gassing (5% CO 2 ). Technical replicates were used to ensure the reliability of single values. Experimental replicates presented in the text are independent experiments performed on different days with different batches of cells. PAR-2 immunofluorescence, biochemistry, and molecular biology in squamous cell lines were carried out as described for primary cells (8).

Fungal culture
Fungal culture was carried out as described previously (63). Briefly, cultures of A. niger (strain WB326 (ATCC16888), A. fumigatus NIH5233 (ATCC 13073), and NRRL163 (ATCC1022)) were grown by the Clinical Microbiology Lab, Philadelphia Veterans Affairs Medical Center, in 40 ml of BACTEC Myco/F Lytic Culture Vials (BD Biosciences) at 30°C for 10 days. CM was then extracted and filtered sequentially through 0.45-and 0.2-m filters as described previously (63). Aliquots of CM were flash-frozen and stored at Ϫ80°C and thawed immediately before use. For calcium-imaging experiments, fungal CM was dialyzed overnight (Spectra/Por 1 regenerated cellulose Figure 9. Impaired ciliation by CSC exposure alters PAR-2 polarity. a, representative image showing reduction of cilia in ALIs exposed to CSC. b, quantification of ␤-tubulin IV immunofluorescence (green) and ELISA measurement of acetylated tubulin (gray). c, TEER from cultures exposed to CSC for 20 days of ALI differentiation; 30 g/ml reduced cilia but maintained TEER. d, FITC-dextran permeability and apical glucose concentration with CSC, confirming barrier integrity at Յ30 g/ml CSC. e, bar graph showing no statistically significant difference in resting intracellular calcium concentration [Ca 2ϩ ] i in cultures treated with 0, 10, and 30 g/ml CSC as in b-d. Significance was tested with one-way ANOVA with Bonferroni post-test. f, peak calcium responses to apical (ap.) and basolateral (bl.) 2FLI (25 M), Der 3 p (1 M), and tryptase (1 M) in ALIs with 0, 10, or 30 g/ml CSC. Note increase of apical 2FLI responses with increased CSC. g, bar graph showing apical responses to 2FLI or AC 55541 after CSC exposure were inhibited by PAR-2 antagonist FSLLRY-NH 2 . h, decreases in cAMP during apical PAR-2 stimulation (50 M 2 FLI or 20 M AC 55541 Ϯ 100 M PAR-2 antagonist FSLLRY-NH 2 for 10 min) in ALIs treated with 0 or 30 g/ml CSC as in b-g were quantified by ELISA. As baseline cAMP levels were too low to quantify by this method, cultures were treated concomitantly with 1 M adenylyl cyclaseactivating compound forskolin to elevate cAMP high enough for measurement. i, TGF-␤2 and GM-CSF were measured by ELISA after 24 h apical or basolateral stimulation as in e. Note increased cytokine secretion with apical 2FLI in ALIs exposed to CSC. Significance was determined in b-g and i determined by one-way ANOVA with Bonferroni post-test comparing each value to its respective 0 CSC control; **, p Ͻ 0.01. Data points are from 6 to 10 individual ALIs from three to five human patients.  MWCO tubing) at 4°C against an ϳ500ϫ volume of HBSS to remove small molecules and equilibrate salt concentrations.

PAR-2 Trio assay
PAR-2 Trio assay was modified from Ref. 26. Vectors for PAR-2 and ␤2 adrenergic receptor Trio assays were from Xiaokun Shu via Addgene. Cells were co-transfected with two pcDNA3 plasmids encoding the GFP ␤-strands 1-9 (Addgene catalog no. 121684) and a pcDNA3 plasmid expressing the receptor fused to GFP ␤-strand 11, an arrestin fused to GFP ␤-strand 10, and mCherry as a transfection marker (Addgene catalog no. 113609 or 113610) and imaged 24 h after transfection at room temperature using standard GFP and TRITC filters on an Olympus IX83 microscope with ϫ40 (0.75NA PlanFluor) objective. Sapphire CKAR or Sapphire CKAR(T/A) (64) was from Jin Zhang via Addgene (catalog nos. 118468 and 118469) and was imaged using the 380 nm channel of a fura-2 filter set (Chroma Technologies, Rockingham, VT).

Generation of primary sinonasal ALI cultures from residual surgical material
Primary human sinonasal cell culture was carried out as extensively described (8,61,62,65). Tissue acquisition was carried out in accordance with University of Pennsylvania guidelines regarding the use of residual clinical material, using tissue from patients Ն18 years of age undergoing surgery for sinonasal disease (CRS), or other procedures (e.g. trans-nasal approaches to the skull base). Institutional review board approval (no. 800614) and written informed consent was obtained in accordance with the United States Department of Health and Human Services code of federal regulation Title 45 CFR 46.116 and the Declaration of Helsinki. Clinical information regarding patient samples is listed in Table S1.

PAR-2 polarization changes with loss of airway cilia Generation of mouse nasal septum ALIs
Experiments using primary mouse nasal septal ALI cultures utilized residual tissue from mice euthanized for other experimental purposes according to the "Guide for the Care and Use of Laboratory Animals" (Institute of Laboratory Animal Resources, National Research Council) with institutional approval. Cell culture was carried out as described.
Nasal septum was removed from C57BL/6J Wt and PAR-2 knockout (B6.Cg-F2rl1 tm1Mslb /J; The Jackson Laboratory, Bar Harbor, ME) mice euthanized for other experimental purposes with IACUC approval according to National Institutes of Health guidelines, the principles of ARRIVE, and the Basel Dec-laration. No animals were sacrificed solely for the experiments in this study. Nasal septum was dissociated and cultured at air-liquid interface, as described previously (8), and used after 3 weeks' exposure to air for full differentiation in DMEM/F12K media containing 2% NuSerum (Corning), 100 units/ml penicillin, 100 g/ml streptomycin.

qPCR
RNA was isolated from ALI cultures or stripped turbinate or polyp epithelium, as described previously (46), and qPCR was performed using TaqMan primers (Thermo Fisher Scientific) for F2RL1 (Hs00608346_m1), ACTB (Hs01060665_g1), and , which activates apical purinergic receptors, was a positive control. d, peak change in calcium in tissue treated apically with PAR-2 agonists. Each data point is an independent experiment using tissue from separate patients (n ϭ 7 per condition). e, peak change in calcium in polyp tissue treated apically with PAR-2 agonist (50 M 2FLI or 120 M SLIGRL-NH 2 ) Ϯ antagonist (80 M FSLLRY-NH 2 ), confirming responses are due to PAR-2. Each data point is an independent experiment using tissue from separate patients (n ϭ 4 per condition). f, immunofluorescence showing lack of apical PAR-2 staining in polyp ciliated cells with Na ϩ K ϩ ATPase (NKP) as marker for the basolateral membrane and ␤-tubulin IV (␤TubIV) (cilia marker) shown in green. Pearson's correlation and Mander's overlap coefficients for NKP and PAR-2 in ciliated cells were both Ն0.95 in 10 images analyzed. Scale bars are 10 m. g, peak change in calcium in isolated ciliated cells; n.s., no significant difference was observed between polyp and turbinate. Significance in d and g was determined by one-way ANOVA with Bonferroni post-test and significance in e by one-way ANOVA with Dunnett's post-test comparing all values to HBSS only control; **, p Ͻ 0.01. Each data point is an independent ALI culture from a separate individual patient (n ϭ 5 per condition). Significance was determined by one-way ANOVA with Bonferroni post-test; **, p Ͻ 0.01 between bracketed values. h and i, schematic of PAR-2 regulation of ciliary function in well-differentiated epithelium (modeled by control cultures; h) versus remodeled epithelium (modeled by IL-13 stimulated cultures; i). Data support that basolateral PAR-2 is expressed under both conditions and can regulate CBF directly within ciliated cells. Apical PAR-2 is not expressed in ciliated cells, but apical PAR-2 stimulation can transmit signals (likely calcium, Ca 2ϩ ) to ciliated cells through gap junctions. h and i created with Biorender.com. Figure 13. Model of altered PAR-2 signaling in airway disease. We hypothesize that well-differentiated and highly-ciliated airway epithelium can tolerate brief exposure to fungal or dust mite proteases due to the mainly basolateral expression of PAR-2, which is able to activate inflammatory and other responses only if the epithelium is breached. However, airway remodeling through a variety of disease modifiers can change the polarity of epithelial PAR-2 responses and may sensitize the epithelium to proteases and thus enhance inflammatory responses. Created with Biorender.com.

PAR-2 polarization changes with loss of airway cilia
glyceraldehyde-3-phosphate dehydrogenase (Hs02786623_g1) in separate reactions, and relative expression was calculated by calculated by means of the 2 Ϫ⌬⌬Ct method.

Measurement of FITC-conjugated dextran permeability
FITC-conjugated dextran permeability was measured as described previously (66). Briefly, FITC-dextran (10 kDa) in phenol red-free DMEM was placed on the apical side of the culture, and basolateral solution (also phenol red-free DMEM) was collected from the basolateral side after 30 min of incubation at 37°C. Samples were read in a Tecan Spark 10M fluorescence plate reader at 485 nm excitation and 525 nm emission. Basolateral media from cells treated with apical media only (not FITC-dextran) were used for background subtraction.

Measurement of airway-surface liquid glucose
Glucose permeability was measured as described previously (67). Briefly, glucose permeability was measured by adding 30 l of glucose-free, phenol red-free DMEM into the apical compartment of 0.33-cm 2 transwells with normal differentiation media placed on the basolateral side. After 6 h of incubation at 37°C to allow glucose to equilibrate, 25 l of apical solution was removed to measure glucose concentration using a colorimetric kit (Cayman Chemical) as indicated by the manufacturer's instructions and as described previously (67).

Measurement of CBF
CBF was measured using the Sisson-Ammons Video Analysis system (68) as described previously (8,62). Briefly, cultures were imaged on a Leica microscope with 20 ϫ 0.8 NA objective and Basler A602f camera running at 100 frames/s. Experiments were carried out at 28°C. ALIs were washed into HEPES-buffered HBSS containing 1.8 mM Ca 2ϩ .

Measurement of cAMP levels by ELISA
Primary sinonasal ALIs were transferred to basolateral HBSS and stimulated with forskolin in 30 l of apical HBSS Ϯ 2FLI Ϯ FSLLRY-NH 2 as indicated. After 10 min, cells were lysed and assayed for cAMP using cAMP Biotrak Enzyme Immunoassay (EIA) System (GE Healthcare) according to the manufacturer's instructions normalized to mg/ml protein levels. Because we found that baseline cAMP levels were below the limit of detection of this ELISA, we quantified changes in stimulated cAMP, as reported previously (61).

Imaging of ASL height
Imaging of ASL height was performed as described previously (69,70). Briefly, cultures were first washed twice with 25 M DTT (5-min prewash) to remove large clumps of airway mucus, followed by at least 5-7 washes with PBS alone to remove residual DTT. ASL was labeled with Texas Red dextran (10,000 M r ; 1 mg/ml) sonicated in perfluorocarbon (PFC-77) Ϯ 2FLI (0.5 mg/ml) and placed in a humidified, heated, and gassed (5% CO 2 ) stage-top chamber (Tokai Hit; Tokyo, Japan). The apical side of the cultures was otherwise unmodified; any further drug additions were made basolaterally. Spinning disk confocal images were taken using a ϫ60 (1.0 NA) water immersion objective with 0.2-m step size. ASL height was determined, as described previously (69), using 70% of maximum fluorescence threshold to define ASL boundaries. For each culture, three separate fields were analyzed, with the averaged value was treated as a single data point.

Imaging of Ca 2؉ in cells and tissue explants
Calcium imaging was performed as described previously (8,46,61,62). Submerged RMPI 2650 and NCI-H520 cells were loaded with 5 M Fluo-4 for 45 min, and Ca 2ϩ was imaged as described previously (8,61,62). Primary human ALIs were loaded, as described previously (8,46,61,62), with 10 M Fluo-4 acetoxymethyl ester (AM) for 90 min on the apical side. Imaging was performed using a Nikon TS-100 microscope with equipped with 10 ϫ 0.3 NA PlanFluor objective (for ALIs) or 20 ϫ 0.8 NA PlanApo objective (for submerged cells) and standard FITC/GFP filter set. Mouse ALIs were loaded similarly with Calbryte 590 to avoid green autofluorescence and visualized with TRITC filter set. Background was estimated by imaging unloaded ALIs or off cell region for submerged cells at identical settings as described (46). Experiments utilized HBSS buffered with 20 mM HEPES at pH 7.4 with 1.5 mM Ca 2ϩ and either 5.5 mM glucose (basolateral side) or 0 glucose (apical side). Experiments performed in the absence of Ca 2ϩ (0-Ca 2ϩ ) utilized cultures loaded with BAPTA-AM (10 M; 15 min preincubation) and extracellular solutions containing no added Ca 2ϩ and 1 mM EGTA.
Flat thin sections of mucosa (Ն1 mM thick) were cut from middle turbinate or nasal polyp tissue and mounted in an oblong EasyMount Ussing Chamber insert (Harvard Instruments) to isolate the apical and basolateral surfaces. The tissue and insert were then placed apical side up in a Petri dish containing HBSS containing 1ϫ MEM amino acids, 1ϫ MEM vitamins, 2 mM L-glutamine. The apical side was loaded with 15 M Calbryte 590-AM in the HBSS solution as above containing 0.4% pluronic F127 for 2 h. Insert was then washed and secured apical side down with vacuum grease on top of a perfusion chamber (Warner Instruments, RC-26GLP) with a low channel and glass coverslip bottom held on a heated stage (Warner PH-1) and perfused with 37°C Tyrode's buffer, containing (in mM): 140 NaCl, 5 KCl, 2 CaCl 2 , 1 MgCl 2 , 10 glucose, 5 Na pyruvate, 10 HEPES, pH 7.4. Imaging was performed on an inverted microscope (Olympus IX83) with a standard TRITC filter set using a 10 ϫ 0.3 NA objective and spinning disk confocal unit (Olympus DSU), XCite 120 LED Boost light source (Excelitas Technologies), Orca Flash 4.0 sCMOS camera (Hamamatsu), and Metafluor using 250-ms exposure and 4 ϫ 4 binning and 5 s sampling frequency to minimize dye bleaching due to the high excitation intensity needed due to the spinning disk.
For measurements of resting [Ca 2ϩ ] i in cultures Ϯ CSC, we loaded cultures with 10 M fura-2-AM in HBSS for 90 min, followed by washing in fura-2-free HBSS. Each culture was imaged for 2 min (one frame every 4 s) to obtain an average baseline fura-2 340/380 emission ratio (R). Cultures were then washed into ionomycin ϩ thapsigargin (10 g/ml each)-containing HBSS with no added calcium plus 2 mM EGTA on both apical and basolateral sides and imaged after 5 min to obtain R min . Cultures were then washed into buffer containing ionomycin ϩ thapsigargin and 10 mM calcium and imaged after 5