|Home | About | Journals | Submit | Contact Us | Français|
House dust mite (HDM) allergens are important factors in the increasing prevalence of asthma. The lung epithelium forms a barrier that allergens must cross before they can cause sensitization. However, the mechanisms involved are unknown. Here we show that the cysteine proteinase allergen Der p 1 from fecal pellets of the HDM Dermatophagoides pteronyssinus causes disruption of intercellular tight junctions (TJs), which are the principal components of the epithelial paracellular permeability barrier. In confluent airway epithelial cells, Der p 1 led to cleavage of the TJ adhesion protein occludin. Cleavage was attenuated by antipain, but not by inhibitors of serine, aspartic, or matrix metalloproteinases. Putative Der p 1 cleavage sites were found in peptides from an extracellular domain of occludin and in the TJ adhesion protein claudin-1. TJ breakdown nonspecifically increased epithelial permeability, allowing Der p 1 to cross the epithelial barrier. Thus, transepithelial movement of Der p 1 to dendritic antigen-presenting cells via the paracellular pathway may be promoted by the allergen’s own proteolytic activity. These results suggest that opening of TJs by environmental proteinases may be the initial step in the development of asthma to a variety of allergens.
Sensitization to allergens such as house dust mite fecal pellet (HDMFP) proteins is a primary risk factor for the development of allergic sensitization and asthma (1–7). Whereas the events that occur in established asthma are understood in increasing detail, the process of allergen delivery that initiates the disease is unresolved.
When inhaled, HDMFPs contact the airway epithelium; they become hydrated and discharge their allergenic contents onto a mucosal barrier that is developmentally specialized to prevent the ingress of foreign proteins. To interact with dendritic antigen-presenting cells, potential allergens must cross the lung epithelium (6, 8). How allergens cross the airway lining is, therefore, central to the development of asthma, but the mechanism is unknown.
Paracellular channels, normally sealed by tight junctions (TJs) (9–11), offer a potential route for antigen penetration. TJs are macromolecular assemblies of proteins that form contiguous rings at the apices of epithelial cells (11). The TJ proteins ZO-1, ZO-2, ZO-3, AF-7, symplekin, 7H6, and cingulin (9, 12, 13) are cytoplasmically localized. However, occludin and claudins are transmembrane proteins (14–18) involved in TJ adhesion and sealing (19, 20). Exogenous perturbants might disrupt this seal, increasing epithelial permeability. If allergens promoted such disruption, it might explain how they encounter antigen-presenting cells.
The group-1 HDM allergen Der p 1 is related to papain and could act enzymatically when inhaled (6, 21). If so, it might either disrupt the airway epithelial barrier by direct activity against TJs or, indirectly, through nonjunctional integral membrane proteins that activate TJ-regulating transduction pathways.
To investigate this fundamental step in allergic sensitization, we examined the effects of solubilized HDMFP and Der p 1 on intercellular junctions and epithelial barrier function, using Madin-Darby canine kidney (MDCK) and 16HBE14o– human bronchial epithelial cell lines, 2 well-defined paradigms for intercellular junctional studies (22, 23). Experiments were conducted using numbers of fecal pellets or amounts of enzyme activity that attempted to model the exposure of human airways to these agents.
Dermatophagoides pteronyssinus were cultured in liver powder, and their fecal pellets (diameter 10-40 μm) were separated by sieving. To mimic the behavior of pellets in the domestic environment, they were “aged” for 8 months at room temperature and humidity. Pellets were solubilized in MEM with Earle’s salts (EMEM) and applied to cell monolayers (24). The depth of solution over the cells was ~150 μm, compared with the ~5–10 μm of fluid that normally covers the lung epithelium (25). The number of pellets was chosen as follows. Given the dimensions of the bronchial tree of adults (26) and the fact that HDMFPs are unlikely to penetrate beyond third-generation bronchi, the area exposed would be 143 cm2. Because an individual inhales >200 pellets per day (7) and the area of our cell monolayer was 1.3 cm2, this would suggest application of >200 pellets × 1.3 cm2/143 cm2 × 150 μm/10 μm, or >30 pellets per coverslip. However, because a higher rate of inertial impaction will occur in lower-generation airways, 40 was chosen as a first-order approximation.
Der p 1 was purified as described (27, 28) to >98% (Figure (Figure1).1). Enzymatic activity was measured spectrophotometrically using N-benzoyl-FVR-p-nitroanilide (single-letter amino acid notation). The activity of the aged fecal pellets used in Figures Figures22–4 was 17 nmol substrate degraded per minute; Der p 1 enzyme activity in Figures Figures55 and and66 was 2.1 nmol degraded per minute, with 3.1 nmol degraded per minute in Figures Figures77–9 (amounts decreased deliberately to mimic airway exposures more closely), and 17 nmol degraded per minute in Figure Figure10.10. Preparations were declared free of endotoxin by Limulus amebocyte lysis assay (ICN Pharmaceuticals Ltd., Basingstoke, United Kingdom).
Cell culture and immunocytochemistry were performed as described (24, 29). Permeability of D-[14C]mannitol (Du Pont NEN, Brussels, Belgium) was measured (24) in confluent monolayers grown on a Matrigel undercoat in Transwells (0.1-μm pore diameter; Corning-Costar Ltd., High Wycombe, United Kingdom). Allergen preparations were diluted in EMEM for addition to cells. Controls comprised EMEM alone. Transepithelial movement of Der p 1 was measured by 2-site ELISA using mAb 5H8 for coating and biotinylated mAb 4Cl (Indoor Biotechnologies, Chester, United Kingdom) for detection (30).
TJs and desmosomes were viewed by fluorescent antibody staining using fluorescein dichlorotriazine (DTAF) or TRITC-labeled secondary antibodies (Vector Laboratories, Peterborough, United Kingdom; Chemicon International Ltd., Harrow, United Kingdom). ZO-1 was stained using mAb R40.76 (22); occludin was stained with mAb MOC37 (31); and desmoplakin was stained with mAb 11-5F (32). The 2-photon molecular excitation microscopy (2PMEM) images were obtained using an LSM410 microscope (Carl Zeiss Ltd., Welwyn Garden City, United Kingdom) with titanium/sapphire mode-locked laser (Coherent Laser Group, Santa Clara, California, USA) excitation at 810 nm, using 80- to 100-femtosecond pulses (33).
Quantification was possible because 2PMEM imaging eliminated photobleaching of the specimens outside the focal plane, thereby providing a linear response. Measurements were made using programs written in IDL (Research Systems, Inc., Boulder, Colorado, USA). Three-dimensional isosurface reconstructions were prepared with IRIS Explorer (NAG, Oxford, United Kingdom). Analyses were performed on SGI Indigo2 and Indy workstations (Silicon Graphics, Reading, United Kingdom).
TJ proteins (ZO-1 or occludin) were quantified as fluorescence intensity (staining density) of the label at constant illumination power from the focal volume excited by the laser. Image stacks were read into computer memory as a 3-dimensional array, and the coordinates of cell boundaries were identified. The computer then searched for the most intense fluorescence near that X-Y coordinate in the image stack. Having found a start point, the staining pattern throughout the 3-dimensional data set was traced, voxel by voxel, to produce the coordinates that defined the periphery of the identified cell. The staining pattern normal to the traced coordinates was fitted by a Gaussian function to identify the peak fluorescence and its width in each cell. Breaks were defined as the number of times the staining intensity dropped below twice the average background intensity in adjacent voxels in the image stack along the cell periphery. Data were reduced to measures of average fluorescence present in the periphery of each cell, the background fluorescence present, the number of breaks in the peripheral staining pattern, and the mean length of the break.
For desmosomal puncta, discrete staining sites were counted with a standard region-counting algorithm (IDL) from a binary mask constructed as follows. An unsharp mask filter was used to reject large-area variations in background intensity. The binary mask was then constructed by accepting regions of 3 or more connected pixels whose individual intensities were more than the mean + 1 SD of the surrounding area (10 × 10 pixels).
Whole-cell extracts were prepared by washing cells (3 ×5 × 106) twice with PBS containing 1 mM EGTA, 1 μg/mL aprotinin, 0.2 mM 4-(2-aminoethyl)-benzenesulphonyl fluoride hydrochloride (AEBSF), and 20 μM E-64 [L-trans-epoxysuccinyl-leucylamide-(4-guanidino)-butane] (34). Boiling sample buffer (0.125 M Tris-HCl [pH 6.8], 4% SDS, 20% glycerol, 0.1 mg/mL bromophenol blue, 1.44 M 2-mercaptoethanol) was added to the cells, and the lysate was scraped into an Eppendorf tube and boiled for 10 minutes. After SDS-PAGE, proteins were transferred electrophoretically to nitrocellulose membranes. Nonspecific binding was blocked with 5% wt/vol nonfat milk and 0.1% vol/vol Tween-20 in Tris-buffered saline (TBS), followed by incubation with primary antibodies — rabbit polyclonal anti–ZO-1 or anti-occludin (Zymed Laboratories Inc., San Francisco, California, USA) for 16HBE14o– cells, or mAb MOC37 for occludin in MDCK cells — diluted in TBS containing 2% wt/vol BSA and 0.1% vol/vol Tween-20. Detection was by enhanced chemiluminescence (ECL) technique (Amersham Pharmacia Biotech, Ltd., Little Chalfont, United Kingdom). To confirm that occludin could be degraded by Der p 1, whole-cell protein extracts were separated electrophoretically. A section of gel containing high- and low-mass occludin (60–80 kDa) was excised and treated (35) with Der p 1 for 30 minutes during stacking, before a second round of electrophoresis and immunoblotting.
Peptides (650 μg/mL) and Der p 1 were incubated in 50 mM Tris-HCl buffer/0.6mM DTT (pH 7.2) at 37°C. Samples (15 μL) were taken at predetermined time points, mixed with 5 μL 10% vol/vol trifluoroacetic acid (TFA) and injected onto a Lichrospher RP-18 HPLC column (150 mm × 3 mm). Elution was at 0.5 mL/min for 30 minutes in a linear gradient of 10–60% acetonitrile in water containing 0.04% TFA (Waters 616 pump with 486 tunable absorbance detector monitoring A280/260; Waters Ltd., Watford, United Kingdom). After a 4-minute flow diversion to remove the solvent front, the eluate was directed with a 10:1 stream split into the electrospray source of a VG Quattro II mass spectrometer (Micromass Ltd., Manchester, United Kingdom). Positive-ion spectra were acquired over the mass range 220–1,350 kDa every 3 seconds. Masses of the protonated species of potential cleavage fragments were used to identify reaction products. Peptides were 8- to 12-mer sequences from the extracellular loops of occludin and claudin-1, and were custom synthesized by Genosys Biotechnologies (Pampisford, United Kingdom).
Statistical significance, calculated using the Statistica program suite (Statsoft Inc., Tulsa, Oklahoma, USA), was taken as P < 0.05.
Application of 40 hydrated HDMFPs to confluent MDCK epithelial monolayers altered the staining pattern of ZO-1 (Figure (Figure2).2). Normally, TJs, as identified by ZO-1 immunostaining, form a continuous ring that circumscribes each cell. TJ rings became broken after 1 hour of HDMFP exposure. After 4 hours, TJ disruption was extensive, and some cells detached from the substratum, creating “holes” in the monolayer. Changes were more marked for ZO-1 than for desmoplakin, whose general appearance was not obviously altered (Figure (Figure2a).2a). Figure Figure2b2b depicts the 3-dimensional staining pattern contained within the extended-focus images of Figure Figure2a.2a. Clearly, 4-hour exposure to HDMFPs produced TJ breakage. Similar results were obtained in every experiment (n = 8), suggesting that a major disruption of TJ integrity had occurred. Digital image analysis (Figure (Figure2,2, c and d) showed significant increases in the number of TJ breaks per cell and a significant reduction in the density of ZO-1 staining, suggesting a loss of protein from TJs. The number of desmoplakin staining sites decreased only after a 4-hour treatment, without an obvious trend in the size of the desmoplakin puncta. Identical data were obtained in 16HBE14o– cells (not shown). In addition to the gross disruption of TJs (Figure (Figure2,2, a and b), some ZO-1 was displaced basally from its normal position apical to desmosomes (11), retaining its peripheral localization. This displacement of ZO-1 is suggested in the 3-dimensional reconstruction (Figure (Figure2b)2b) and by the X-Z sections (Figure (Figure33).
Permeability measurements were made to determine whether the effects of HDMFP on TJs might be associated with a diminished epithelial barrier. Figure Figure44 shows that increasing permeability of MDCK monolayers reflected the progression of TJ disruption. Sham-treated monolayers were unaffected (Figure (Figure44).
Separation of HDMFP extracts into various fractions yielded a partially purified active principle enriched in Der p 1. This mimicked the events above in both MDCK and 16HBE14o– cells (not shown), and could be inhibited by the cysteine proteinase inhibitor E-64. Because the amino acid sequence of Der p 1 implies that it is a cysteine proteinase, these observations suggested that Der p 1 might be the active principle. To test this, we prepared pure Der p 1. Figure Figure5,5, a and b illustrate the effect of pure Der p 1 on 16HBE14o– cells immunostained for occludin. The response to Der p 1 is similar to that shown in Figure Figure2.2. Quantitative reductions in TJ continuity and occludin content were found (Figure (Figure5c).5c). No significant changes occurred in either the number or size of desmoplakin puncta (Figure (Figure5d).5d). Immunostaining of ZO-1 in MDCK cells was also disrupted by Der p 1, confirming the reciprocity of these effects on TJ proteins in the 2 cell lines (data not shown).
Figure Figure6a6a shows that the alteration in TJ structure after a 2.5-hour exposure to Der p 1 produced an ~12-fold increase in mannitol permeability, indicating that TJ breakage (Figure (Figure6a,6a, filled bars) is associated with a loss of TJ integrity. These changes were due to the enzymatic activity of Der p 1, because neither oxidized nor heat-inactivated Der p 1 had any effect. Figure Figure6b6b shows that TJ breakage led to a transepithelial migration of the allergen that was inhibitable by E-64. Furthermore, the ability of heat-inactivated Der p 1 to cross the monolayer was <1% that of catalytically active Der p 1 after a 7-hour treatment (Figure (Figure66b).
Occludin was detected in immunoblots at ~62–65 kDa, with less intense forms at ~70–75 kDa (Figure (Figure7a).7a). After exposure of cells to Der p 1, occludin was partially cleaved into products of decreased molecular mass (~53 kDa, ~31 kDa, and ~22 kDa). A 45-kDa band was detected in both control and Der p 1–treated cells. To address whether Der p 1 cleavage sites exist within extracellular domains of occludin, we used HPLC/electrospray mass spectrometry to analyze the products formed by reacting pure Der p 1 with synthetic peptide segments of the occludin loop sequences. 88AWDRGYGTSLLG99, of the first extracellular loop of human occludin, underwent facile cleavage of LL (product AWDRGYGTSL). Attack of GT (product TSLLG) and YG bonds (product GTSLLG) also occurred (Figure (Figure77b).
Figure Figure7c7c shows that ZO-1 also underwent proteolysis, indicating that the response to Der p 1 also involved activation of intracellular protein processing. The effects of Der p 1 were mimicked by apical application of papain (substrate-degrading activity 5.4 nmol/min) to MDCK cells for 2.5 hours, which increased mannitol permeability from 0.35 ± 0.11 × 10–6 cm/s to 9.30 ± 1.70 × 10–6 cm/s (n = 3, P < 0.05). Immunoblotting revealed similar profiles of cleavage in occludin and ZO-1 as in Der p 1 (not shown).
To support the view that Der p 1 degrades occludin, gel-enriched high- and low-mass occludin were treated with Der p 1. The ensuing loss of occludin suggested that degradation could be catalyzed directly (Figure (Figure77d).
Heat-inactivated Der p 1 had no effect on occludin in MDCK cells, whereas degradation was extensive with active Der p 1 (Figure (Figure8a).8a). The cysteine proteinase inhibitor antipain retarded the action of Der p 1 (Figure (Figure8a),8a), as did the occludin loop peptide (Figure (Figure8b).8b). Cysteine proteinase inhibitors and occludin peptide also inhibited the allergen-dependent loss of TJ immunostaining and the permeability change in epithelial monolayers (not shown). However, the cellular response to Der p 1 was not inhibited by AEBSF, pepstatin, and [4-(N-hydroxyamino)-2R-isobutyl-3S-(thiophen-2-yl-sulphonylmethyl)succinyl]-L-phenylalanine-N-methylamide (BB-250), (Figure (Figure8c),8c), suggesting that serine, aspartic, and matrix metalloproteinases play little role in the extracellular processing of occludin following Der p 1 exposure. The evidence presented in Figure Figure88 also supports degradation of the occludin peptide by Der p 1 being due to cysteine proteinase activity, because the facile cleavages of LL and GT were attenuated by antipain and E-64 (Figure (Figure8,8, d and e). These changes were accompanied by increases in the amounts of undegraded substrate.
Reaction of Der p 1 with 64KVFDSLLNLNS74 from the first extracellular loop of murine claudin-1 yielded facile cleavage of LL (products KVFDSL and LNLNS) and NL (product KVFDSLLN). Scission of LN also occurred (product KVFDSLLNL) (Figure (Figure9).9). 138WYGNRIVQ144 (second extracellular loop) underwent facile cleavage of GN (products WYG and NRIVQ) (Figure (Figure9),9), with evidence of a minor reaction at YG (not shown).
Cells whose TJs had been extensively disrupted by Der p 1 recovered following its removal (Figure (Figure10,10, a and b). Within 1 hour, a reticular pattern of ZO-1 was forming (Figure (Figure10b).10b). Reestablishment of occludin rings was slower (Figure (Figure10a)10a) but did eventually progress to completion.
Our results suggest an entirely novel view of the sensitization to HDM allergens: the intrinsic proteolytic activity of the allergen leads to degradation of TJs in airway epithelium, thus increasing the accessibility of the sentinel dendritic antigen-presenting cells residing beneath the epithelial barrier. Specifically, the HDM allergen Der p 1 leads to cleavage of the TJ membrane protein occludin, suggesting a major role for the allergen in decreasing the effectiveness of the epithelial barrier.
That Der p 1 is a proteinase has been indicated previously (6, 21, 24, 36–42). We reasoned that its proteolytic effects in airway epithelium might be significant to sensitization because of the degree to which airway mucosal barrier is exposed to foreign proteins. Small amounts of HDMFP or pure Der p 1 caused TJs to break and become displaced from their normal position apical to desmosomes (11). Whereas the effects of HDMFPs on epithelial cells were mediated by 8–10 ng of Der p 1 protein (1 fecal pellet contains ~0.2 ng of Der p 1), the proteolytic effects of Der p 1 described in other cells require microgram amounts (40, 41). Moreover, because Der p 1 is a cysteine proteinase susceptible to oxidation, with irreversible loss of catalytic activity, our deliberate aging of HDMFPs means that the catalytically effective amount of Der p 1 delivered to the cell surface was probably somewhat less than 8–10 ng. Thus, our findings indicate that epithelial cells could be the primary proteolytic targets of Der p 1 in vivo, because they are the cells most likely to be exposed to the highest concentrations of inhaled allergens.
Another novel aspect of our study is the use of 2PMEM to quantify TJ proteins in situ. Using 2PMEM enabled us to correlate the time-dependent decrease of both occludin and ZO-1 with the increase in epithelial permeability and with cleavage of these proteins, as shown by immunoblotting. We thus established a link between allergen proteinase activity and diminished effectiveness of the epithelial barrier.
Occludin was expressed predominantly in a 62- to 65-kDa form, with possibly phosphorylated (31, 43) forms at ~70–75 kDa. The 70–75-kDa band was evident in 16HBE14o– cell extracts blotted with polyclonal antibody, but less visible in MDCK cell extracts blotted with MOC37. In 16HBE14o– cells exposed to pure Der p 1, both high- and low-mass forms were reduced in intensity. We can conclude that the cleavage products detected all contain the COOH-terminal of occludin, because the blotting antibody recognizes a 150-residue COOH-terminal segment. Inspection of the occludin sequence and prediction of its membrane topography using the program SOSUI (44) suggest that the 53-kDa occludin product arises by cleavage within the first external loop. The accessibility of this loop to exogenous macromolecules is demonstrated by occludin immunostaining in nonpermeabilized CaCo-2 cells (19), using antibodies against the segment of the first extracellular loop in which we report putative cleavage sites. Furthermore, as cleavage proceeds, steric hindrance provided by the TJ itself would minimize, so that the velocity of TJ disassembly should increase progressively. In contrast to the above, the 31-kDa and 22-kDa occludin products are probably derived by intracellular processing, because their masses are consistent with the COOH-terminal domain alone.
Two general possibilities could explain the extracellular cleavage of occludin. Either Der p 1 is envisaged to cleave occludin directly or Der p 1 might activate a cell surface zymogen that then degrades occludin. Experiments favor the more direct option: (a) the immunoblotting evidence and identification of putative Der p 1 cleavage sites in a synthetic peptide strongly suggest that the first extracellular loop of occludin may be considered a substrate of Der p 1; (b) an occludin loop peptide itself acted as an inhibitor of occludin degradation in situ; (c) the failure of serine, aspartic, and metalloproteinase inhibitors to block the in situ degradation of occludin by Der p 1 argues against the existence of an extracellular proteolytic cascade involving enzymes of these families.
Until recently (18), occludin was the only known transmembrane protein of TJs. Despite the proposal that claudin-1 and -2 may be more significant in TJ structure (20), functional evidence indicates that disruption of occludin causes epithelial barriers to fail. Peptides mimicking extracellular loops of occludin open TJs (45) and prevent the gain of adhesiveness in occludin-null cell lines transfected with occludin (19). These results would not be expected if occludin were of secondary importance in TJ structure and function. Our data provide clear evidence that Der p 1 increases epithelial permeability and disrupts TJs. The inescapable conclusion is that occludin and all other TJ adhesion proteins that regulate paracellular permeability must be affected in this process, and that initiation of the response ultimately depends upon the proteolytic activity of Der p 1.
Potential Der p 1 cleavage sites exist in both extracellular loops of claudin-1. In 64KVFDSLLNLNS74 cleavage of LL was facile (as seen in occludin). This suggests that like papain (46), Der p 1 exhibits a P1′ preference for hydrophobic residues. Other cleavages, especially of YG in occludin and in loop 2 of claudin-1, were less facile. Interestingly, the facile LL cleavage site of claudin-1 is replicated in 5 of the 8 known claudins (47), whereas in claudins 4, 5, and 7 the equivalents of P1 69L are conservative substitutions (M,V,V) (47). Collectively, these observations suggest that occludin and claudins are all targets of Der p 1 and that there may be a preference for small, uncharged polar residues at P2 in its natural mammalian substrates.
In addition to cleavage of occludin, ZO-1 (22) was degraded following exposure of cells to HDMFPs or Der p 1. Because ZO-1 is intracellular (9, 16, 22), and therefore unlikely to have been directly degraded by Der p 1, we propose that its breakdown is a consequence of TJ disassembly. A similar explanation has been invoked to explain intracellular cleavage of the transmembrane protein N-cadherin following loss of intercellular contacts (48).
The possible modes of action of Der p 1 on TJs are shown schematically in Figure Figure11.11. To summarize, our data are consistent with direct extracellular cleavage of occludin, and possibly claudins (mechanism in cell A). This may trigger intracellular processing, or, alternatively, intracellular cleavage could arise through a receptor and transduction pathway that is decoupled from TJs (the cell surface receptor and intracellular pathway in cell B). However, it seems less likely that TJs are degraded by activation of a cell surface zymogen that then attacks TJs (extracellular pathway in cell B). Whatever the mechanism of intracellular proteolysis, the effect is to open the epithelial barrier (cell C).
TJ cleavage evoked by HDMFP or Der p 1 differs from other exogenous proteinases that cause aberrant TJ-like strands to be formed (49–52). However, aberrant TJ strands increase the plasticity of TJs because paracellular permeability increases during their formation (50). This difference in behavior between Der p 1 and certain other proteinases might explain why Der p 1 is an insidious allergen capable of promoting a high level of reactivity in individuals with a constitutional predisposition to allergy.
Our results indicate that epithelial permeability is changed nonselectively by HDMFPs and pure Der p 1. This disruption was demonstrated using mannitol, which permeates the epithelial barrier only paracellularly (53), providing a sensitive index of TJ function (29). Furthermore, and of significance for allergen presentation, the increased permeability following TJ cleavage facilitated the transepithelial permeation of Der p 1.The nonselectivity of the increase implies that the transepithelial permeation of all proteins would be enhanced. This would lead to an increased probability of any inhaled allergen being able to encounter antigen-presenting cells of the airway’s immune system at sites where epithelial permeability was increased. We note that allergy to Der p 1 is frequently associated with reactivity to unrelated allergens (54–56), as would be anticipated from our proposed mechanism.
Our observations suggest that allergic sensitization may be promoted by any environmental proteinase that attacks, directly or indirectly, the integrity of the epithelial barrier. We therefore suggest that prevention of TJ disruption (by inhibiting the environmental proteinase or even promoting TJ reassembly/assembly) could provide an entirely new approach to the treatment of asthma by limiting exposure of the immune system to allergens.
We are grateful to the National Asthma Campaign for financial support. We thank Bruce Stevenson, Tom Fleming, Shoichiro Tsukita, and the Commonwealth Serum Laboratory for reagents.