Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

In cystic fibrosis (CF), defects in the CF transmembrane conductance regulator (1) cause chronic severe respiratory and gastrointestinal disease, resulting in the death of most patients by about the age of 30. Most of the mortality associated with CF results from obstruction of the airways, infection, and chronic inflammation, leading to bronchiectasis and respiratory failure (2). Neutrophils are the prominent inflammatory cell in the airways of patients with CF, even in individuals with minimal pulmonary disease (3). This influx of neutrophils, in response to the major CF airway pathogen Pseudomonas aeruginosa, is mediated by a number of chemoattractants produced by respiratory epithelial cells (4, 5). Neutrophils, in an attempt to clear the bacteria, release lytic enzymes such as neutrophil elastase (NE), which degrades the outer membrane of Pseudomonas but also damages host proteins such as elastin and fibronectin (6). Both mechanical disruption of cells and exposure to even low concentrations of NE increase the release of proinflammatory mediators from the respiratory epithelium, further enhancing the inflammatory response (7, 8). Moreover, NE is a potent secretogogue in airway epithelium (6) and promotes secretory-cell hyperplasia, thus contributing to airway obstruction. All of these effects of NE on epithelial cell functions, however, can be blocked by inhibition of the enzyme activity of the protease.

The lungs are protected against destruction by NE by two endogenous antiproteases, secretory leukoprotease inhibitor, produced by respiratory epithelial cells, and ₁-antitrypsin (A1AT) (9, 10). A1AT, the major serum antiprotease, is produced mainly by hepatocytes and mononuclear phagocytes, and diffuses into the alveoli and pulmonary interstitium where it acts as the primary inhibitor of NE (10). The importance of this inhibitory function is observed in hereditary A1AT deficiency, which causes early onset emphysema (10). Although the level of A1AT is elevated in the epithelial lining fluid of patients with CF, the abundance of NE overwhelms the antiproteases in the lung, resulting in the inadequate protection of the lower respiratory tract (6).

Improving the protease/antiprotease balance in the CF airway may prevent structural damage and interrupt the inflammatory cycle. In a rat model of chronic endobronchial Pseudomonas infection, delivery of A1AT by aerosol limits the lung inflammatory response and actually improves bacterial clearance (11). Unfortunately, the results of clinical studies examining the use of A1AT in patients with CF have been disappointing. Systemic administration of A1AT is inefficient, resulting in airway concentrations below the level necessary to inhibit the massive amounts of NE present in the epithelial lining fluid (12). A1AT administered by this route is delivered predominantly to the alveoli and not the airway, which is the primary site for CF involvement. Moreover, epithelia are generally impermeable to macromolecules, and such agents are prevented from diffusing across by the tight junctions that connect adjacent epithelial cells. Human A1AT can also be delivered by inhalation, but penetration of the mucus layer is difficult and the aerosolized antiprotease is distributed to favor the better-ventilated, less-compromised regions of the lung (12, 13).

We propose a novel therapeutic strategy to circumvent these difficulties and deliver A1AT directly to the respiratory epithelial surface by targeting the polymeric immunoglobulin (Ig) receptor (pIgR). The human pIgR is expressed at high levels in the mucosal epithelium of the airway and gastrointestinal tract (14). This receptor is specifically adapted for the uptake and nondegradative transfer of large molecules such as dimeric IgA (dIgA) and pentameric IgM from the basolateral to apical surface. At the apical surface the receptor is cleaved, releasing the ligand bound to the extracellular domain, or secretory component (SC), of the receptor (15). These properties make the pIgR an ideal target for directing therapeutic proteins into the lumen. Because access is from the basolateral surface via the bronchial circulation, delivery through this receptor would be expected to be greatest to inflamed regions of the airway that have the greatest blood flow. In addition, pulmonary inflammation may increase the expression of this receptor in the airway. The expression of pIgR is augmented by various cytokines, including tumor necrosis factor-, interferon-, and interleukin (IL)-4 (16).

To target the pIgR for A1AT delivery to the airway, we produced a recombinant fusion protein consisting of an antihuman pIgR single-chain Fv (scFv) linked to human A1AT. We demonstrated that the fusion protein was transported from the basolateral to apical surface by a pIgR- expressing epithelial cell monolayer, and that it was secreted into the apical medium as an active antiprotease.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Construction of the MDCK-pIgR Cell Line

Madin-Darby canine kidney (MDCK) cells were transfected with the human pIgR complementary DNA (cDNA) using the lipofectin reagent (Life Technologies, Grand Island, NY). The cells were cultured in Eagle's minimum essential media (BioWhittaker, Walkersville, MD) supplemented with 10% fetal bovine serum (FBS), 2 mM L-glutamine, 15 mM N-2-hydroxyethylpiperazine-N'-ethane sulfonic acid, 100 units/ml penicillin, and 100 µg/ml streptomycin. Stably transfected cells were selected for neomycin resistance, and positive cells were sorted repeatedly for the highest level of pIgR by fluorescence-activated cell sorting, using a goat polyclonal antibody against human SC (Sigma, St. Louis, MO), followed by a fluorescein isothiocyanate-conjugated secondary antibody.

Construction of the Antihuman pIgR scFv

Mouse monoclonal antibodies against purified human SC were generated by standard methods (19) in the Cystic Fibrosis Monoclonal Core Facility at Case Western Reserve University. An scFv was constructed from clone 4121 by polymerase chain reaction (PCR) essentially as described by Nicholls and colleagues (20).

Construction and Expression of a Chimeric Gene Encoding the scFv-A1AT Fusion

The scFv and human A1AT cDNAs were amplified by PCR using primers with unique restriction sites to facilitate ligation and subcloning, with the scFv coding sequence upstream of the A1AT cDNA. In addition, the 5' scFv primer included the coding sequence for the mouse Ig- leader to direct secretion in eukaryotic cells (21). The leader-scFv-A1AT fusion construct was subcloned into pPac, containing the Drosophila actin 5C promoter for constitutive expression (22), and was transfected into Drosophila Schneider Line 2 (S2) cells using a modified protocol from Incardona and Rosenberry (23). Stable lines were generated by selection with hygromycin B followed by cloning in soft agar. The cells were cultured at 27°C in Schneider's Drosophila Medium (Life Technologies) supplemented with 10% FBS, 50 µg/ml gentamicin, 100 U/ml penicillin, and 100 µg/ml streptomycin. Maximal secretion of the fusion protein was 5 µg/ml/7 d in confluent culture flasks.

Affinity Purification of the scFv-A1AT Fusion Protein

Human A1AT-specific goat antibody was isolated from the IgG fraction of antiserum (INCSTAR, Stillwater, MN) by incubation with purified human A1AT (Sigma) immobilized on sepharose beads (Pharmacia Biotech, Uppsala, Sweden). After elution with 0.05 M glycine and 0.15 M NaCl, pH 2.5, the purified antibody was coupled to sepharose. Conditioned medium from S2 cells expressing the scFv-A1AT fusion protein was dialyzed against phosphate-buffered saline (PBS) before incubation with the sepharose-bound goat antihuman A1AT antibody. Nonspecific proteins were removed by washing with 0.05 M Na₂HPO₄ and 0.5 M NaCl, pH 6.3, and the scFv-A1AT fusion protein was eluted as described previously and dialyzed against PBS.

Transcytosis of Anti-pIgR scFv-A1AT Fusion Protein and Free A1AT

Purified antihuman SC scFv-A1AT fusion protein or purified human A1AT was added to culture medium on either the apical (0.3 ml) or basolateral (0.5 ml) side of confluent MDCK and MDCK-pIgR monolayers grown on 1-µm filters (Becton Dickinson, Franklin Lakes, NJ). MDCK culture media were placed on the opposite side. After incubation for 24 h at 37°C, the media were collected and analyzed by sandwich enzyme-linked immunosorbent assay (ELISA) for A1AT.

ELISA for A1AT

For analysis of A1AT, media samples were incubated in microtiter plates coated with goat antihuman A1AT (INCSTAR) as capture antibody, and bound protein was detected by incubation with rabbit antihuman A1AT (Sigma), followed by a peroxidase-conjugated secondary antibody. Purified human A1AT was used for standardization.

Association of A1AT and scFv-A1AT to NE

Association reactions were initiated by adding A1AT (Calbiochem, La Jolla, CA) or scFv-A1AT to 0.6 nM purified human sputum elastase (Elastin Products Co., Owensville, MO) in 0.1 M Tris and 0.5 M NaCl, pH 7.5, at 23°C. To ensure that binding was measured under pseudo first-order conditions, the concentration of inhibitors was adjusted to 1.7 to 4.2 times the concentration of NE. The concentrations of the inhibitors were determined by titrations with standard NE. At various times the reactions were stopped by adding substrate N-methoxysuccinyl-ala-ala-pro-val -nitroanilide (Sigma) to give a final concentration of 1.0 mM, and a continuous assay trace was immediately recorded at 410 nm on a Cary 3A spectrophotometer (Varian, Walnut Creek, CA). Background rates in the absence of NE were subtracted. Assay rates, v, from association reactions were divided by control assay rates in the absence of these inhibitors to give a normalized value (v_(N)). These values were fitted by nonlinear regression analysis (Fig P 6.0 for DOS; Biosoft, Ferguson, MO) to Equation 1, where v_(N)initial and v_(N)final are the calculated values of v_(N) at times zero and at equilibrium, respectively, and the observed pseudo first-order rate constant k for the approach to equilibrium is given by Equation 2 (24).

(1)

where

(2)

In Equation 2, k_assoc is the association rate and k_dissoc is the dissociation rate of the inhibitor binding to NE. Because the binding of A1AT to NE has been considered irreversible, for these calculations of association rate constants k_dissoc can be fixed to zero.

Analysis of Antiprotease Activity of Transcytosed Fusion Protein

Transcytosed fusion protein (0.8 pmol) was immunoprecipitated from the apical medium of MDCK-pIgR cells using a sepharose-bound goat antihuman SC antibody (Sigma). Fusion protein before transcytosis was also incubated with conditioned apical MDCK-pIgR media to allow the scFv domain to bind to SC then was similarly immunoprecipitated. As a control, purified human A1AT was added to mock immunoprecipitations after unbound protein had been removed by washing. Each immunoprecipitation was divided in half. To investigate the ability of the A1AT domain to form an inactivation complex with human NE, one sample of each immunoprecipitation was incubated with 0.4 pmol purified NE at room temperature for 20 min. All samples were boiled for 5 min in 2% sodium dodecyl sulfate (SDS) before electrophoresis on a 4 to 12% polyacrylamide gradient gel (Novex, San Diego, CA). A1AT was detected by Western blot with a rabbit antihuman A1AT antibody followed by a peroxidase-conjugated secondary antibody.

Statistical Analysis

Data are expressed as means ± standard error of the mean (SEM). Statistical comparisons between groups were made using paired Student's t tests.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Expression and Activity of a Recombinant Antihuman pIgR-A1AT Fusion Protein

We cloned and assembled an scFv (25) from a hybridoma line expressing a monoclonal antibody directed against the extracellular domain of the human receptor. The parental antibody and scFv recognizes both free human SC as well as secretory IgA. Thus, the antibodies bind to the pIgR outside of the natural ligand binding site and would not be expected to compete with dIgA for receptor binding. A chimeric gene consisting of an Ig- light chain leader sequence, linked to the antihuman pIgR scFv, linked to the human A1AT cDNA was then constructed and expressed by Drosophila Schneider Line 2 (S2) cells. The fusion protein was recognized by an antihuman A1AT antibody and migrated at the expected molecular weight of 78 kD in SDS polyacrylamide gel electrophoresis (PAGE) (Figure 1). The scFv domain of the protein retained antigen-binding activity, as demonstrated by its ability to bind to immobilized human SC in an ELISA (Figure 2, upper panel). The A1AT domain of the fusion protein inhibited NE with an association rate constant of 0.79 ± 0.13 × 10⁷ M¹s¹, similar to that of plasma A1AT (0.87 ± 0.05 × 10⁷ M¹s¹) (Figure 2, lower panel). Even bound to human SC, the association rate constant was not significantly different from the natural antiprotease (0.86 ± 0.11 × 10⁷ M¹s¹). These values are in reasonable agreement with that obtained for plasma A1AT by Beatty and associates using an indirect chymotrypsin competition assay (k_assoc = 6.5 ± 4.0 × 10⁷ M¹ s¹) (26).

View larger version (28K): [in this window] [in a new window]   Figure 1.   Expression and purification of the anti-pIgR scFv-A1AT fusion protein. (a) Recombinant scFv-A1AT fusion protein was detected in the culture medium of transfected S2 cells by immunoprecipitation and Western blot using antibodies directed against A1AT. Lane 1: purified human A1AT control; lane 2: conditioned medium from untransfected S2 cells; lane 3: conditioned medium from S2 cells transfected with the scFv-A1AT construct. (b) Control A1AT (lane 1) and affinity-purified scFv-A1AT fusion protein (lane 2) were analyzed by SDS-PAGE and Coomassie staining.

View larger version (28K): [in this window] [in a new window]   View larger version (13K): [in this window] [in a new window]   Figure 2.   Bifunctional activity of the scFv-A1AT fusion protein. (Upper panel) ELISA showing binding of the anti-pIgR scFv-A1AT fusion protein to purified human SC. Dilutions of conditioned medium from scFv-A1AT expressing S2 cells were incubated in microtiter wells coated with purified human SC. Bound fusion protein was detected using an antibody against the A1AT domain. Conditioned media from nontransfected S2 cells were negative at all dilutions examined. (Lower panel) The association rate constants for purified A1AT, scFv-A1AT fusion protein, and fusion bound to human SC were measured as described in MATERIALS AND METHODS. Values of k obtained from four association reactions of A1AT (dashed line, open boxes), scFv-A1AT fusion (solid line, open circles), and the fusion bound to human SC (dotted line, open triangles) were plotted against A1AT or scFv-A1AT concentrations according to Equation 2 (see MATERIALS AND METHODS). Slopes of the plots were calculated by linear regression analyses assuming that k has constant percent error. The slopes of these linear plots are the association rates (kassoc), and the intercepts are the dissociation rates (kdissoc) of the inhibitors, which here have been fixed to zero.

Transcytosis of the scFv-A1AT Fusion Protein

Using MDCK cells stably transfected with the human pIgR (MDCK-pIgR) as a model polarized epithelium (15), we demonstrated that the antihuman SC scFv-A1AT fusion protein was transported specifically from the basolateral to apical surface and was released into the apical (lumenal) medium (Figure 3, upper panel). Considerably less of the fusion protein was transported in the apical to basolateral direction by receptor-expressing cells. Untransfected MDCK cells that do not express the receptor failed to specifically transport the fusion protein, and neither MDCK nor MDCK-pIgR cells transported free A1AT. A human airway epithelial cell line transfected with the human pIgR, 16HBEo-pIgR, transported the scFv-A1AT fusion protein with similar results (data not shown). In addition, basolateral to apical transcytosis of the fusion protein by receptor-expressing cells could be inhibited by greater than 90% by competition with a 20-fold molar excess of Fab fragments of the parental anti-pIgR antibody (Figure 3, middle panel), providing further evidence for specific, receptor-mediated transport. As predicted, the addition of as much as a 16-fold molar excess of dIgA, the natural ligand for the pIgR, did not inhibit transcytosis of the fusion (Figure 3, lower panel).

View larger version (18K): [in this window] [in a new window]   View larger version (24K): [in this window] [in a new window]   View larger version (40K): [in this window] [in a new window]   Figure 3.   Transcytosis of the scFv-A1AT fusion protein by pIgR-expressing epithelial cell monolayers. Data are plotted as means ± SEM for four filters. (Upper panel) Culture medium containing 20 nM scFv-A1AT fusion protein or purified A1AT was placed on the apical or basolateral side of pIgR-transfected ( filled bars) and untransfected (open bars) MDCK cell monolayers grown on permeable filters. After 24 h, the opposite medium was collected and assayed for A1AT by sandwich ELISA. A = apical, B = basolateral. Asterisk indicates that basolateral to apical transcytosis of scFv-A1AT fusion protein by MDCK-pIgR cells is greater than apical to basolateral transcytosis, and greater than transcytosis of free A1AT, P < 0.005. (Middle panel) Competition of transcytosis of scFv-A1AT by parental anti-pIgR IgG in MDCK-pIgR cells. In addition to the scFv-A1AT fusion protein, a 20-fold molar excess of Fab fragments of the parental monoclonal antibody from which the scFv was derived, or from an irrelevant mouse antibody, was added to the basolateral or apical compartment of MDCK-pIgR cell monolayers. After 24 h, the opposite medium was removed and assayed for A1AT by sandwich ELISA. Asterisk indicates that basolateral to apical transcytosis of the scFv-A1AT fusion protein was significantly inhibited by parental Fab fragments, P < 0.005. (Lower panel) Excess natural ligand did not affect transport of the fusion protein across receptor-bearing cells. Increasing molar excess of dIgA was added with the scFv-A1AT fusion protein to the basolateral medium of monolayers of MDCK-pIgR cells. At 24 h later, the apical medium was removed and assayed for A1AT by ELISA. Excess natural ligand did not inhibit transcytosis of the fusion protein.

Antiprotease Activity of the scFv-A1AT Fusion Protein after Transcytosis

To investigate whether the antiprotease domain of the fusion protein remained active following transcytosis and apical release, we immunoprecipitated the complexes with an antibody directed against SC, the extracellular domain of the receptor that remains bound to the ligand after secretion. As A1AT binds irreversibly to its substrate NE to form an inactivation complex (27), antiprotease activity can be demonstrated by a 29-kD molecular weight shift of the A1AT on a Western blot. When the fusion protein was recovered by immunoprecipitation from the apical medium after transcytosis, such a shift was detected after incubation with purified NE, indicating that the A1AT domain retained antiprotease activity after transcytosis (Figure 4).

View larger version (53K): [in this window] [in a new window]   Figure 4.   Ability of the scFv-A1AT fusion protein to form an irreversible complex with NE. The scFv-A1AT fusion proteins bound to human SC or transcytosed across receptor-expressing cells were immunoprecipitated from culture medium using an antihuman SC antibody. Purified human A1AT was added to mock immunoprecipitations as control. Purified human NE was added to the immunoprecipitated protein to form an inactivation complex, which was detected by Western blot as a 29-kD shift in molecular weight. Lane 1: conditioned medium; lane 2: conditioned medium with scFv-A1AT bound to SC; lane 3: conditioned medium with scFv-A1AT bound to SC after incubation with NE; lane 4: conditioned medium containing transcytosed scFv-A1AT; lane 5: conditioned medium containing transcytosed scFv-A1AT, after incubation with NE; lane 6: purified A1AT control; lane 7: purified A1AT after incubation with NE.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

In CF, the intense host response to infection with P. aeruginosa and other airway pathogens contributes strongly to the decline in pulmonary function, ultimately leading to death. Neutrophils are the major inflammatory cell in the airways of patients with CF (3), and large amounts of NE are released in an attempt to clear the bacteria. NE directly causes structural damage to the lung, and perpetuates the inflammatory response by triggering the release of proinflammatory mediators from the respiratory epithelium (6). For these reasons, antiproteases capable of inhibiting NE are an attractive therapeutic strategy for CF. To be effective, antiproteases must be delivered in such a way that they are present in sufficient quantity at the site of the major NE burden for CFthe surface of the airway epithelium. Neither systemic nor aerosol delivery results in adequate protection at this critical site.

We proposed to circumvent these difficulties and deliver A1AT directly to the respiratory epithelial surface by capitalizing on the properties of the pIgR. In the airway, this receptor is expressed in the submucosal glands and in the epithelium of the trachea, bronchi, and bronchioles (14, 28, 29). In addition to its ideal expression pattern, the pIgR has a number of other characteristics that make it a suitable target for specific delivery to the airway. The normal biologic function of the receptor is to transport dIgA and pentameric IgM, as well as their immune complexes, from the basolateral (blood) side of the epithelium into the secretions (15). Thus, the pIgR is specifically adapted for the uptake and nondegradative transfer of large molecules. Second, because the receptor is expressed on the basolateral surface of the epithelium, the bloodstream is the chosen route of delivery, avoiding the complications of airway delivery to a diseased lung. The therapeutic protein could bypass the intense inflammatory reaction and obstruction of the bronchi and bronchioles that is characteristic of the lung disease in CF. Moreover, it is possible that delivery of the fusion protein will be enhanced to the more inflamed airways because blood flow via the bronchial arteries will be greater to these areas. Consequently, relatively greater amounts of the fusion will be shunted to these regions, the areas that require the antiprotease the most. Although the natural pIgR ligands are released locally by plasma cells in the lamina propria (30), studies in rodents have shown that intravenously administered anti-pIgR antibodies are also rapidly cleared from the circulation and transported across receptor-expressing tissues (T. Ferkol, unpublished observations). The targeting of proteins by the pIgR in humans could potentially provide an additional level of safety for the patient, because the fusion protein not delivered to the lung will be transported to the intestinal lumen, either through the enterocyte or in bile, where it will be excreted. Indeed, the trafficking pattern of the antihuman SC scFv-based fusion protein could also potentially be exploited to deliver proteins to other epithelia.

To target the pIgR for delivery to the airway, we produced a recombinant fusion protein consisting of an antihuman pIgR scFv antibody linked to human A1AT. scFv fragments, the smallest units capable of antigen binding, are comprised of the amino-terminal variable domains of the heavy and light Ig chains, linked together and stabilized by a flexible peptide linker (25). There is considerable experience with expressing fusion proteins containing scFv antibodies and retaining function of both components. Indeed, fusion proteins have been effective and versatile. Recombinant scFvs, directed against cell-surface antigens such as the IL-2 receptor, have been used to localize fusion proteins to target cells, and investigators have used such fusions to deliver a therapeutic payload (e.g., Pseudomonas exotoxin) to cells in vitro and in vivo that express the appropriate receptor (31, 32). Because most of the species-specific sequences have been removed, scFv antibodies are likely to be less immunogenic than intact antibodies. In addition, it is possible to completely "humanize" remaining framework regions to reduce even further the immunogenicity of scFv therapeutics (33).

We demonstrated that the recombinant, anti-pIgR scFv-A1AT fusion protein retains bifunctional activity. The scFv domain maintains its binding activity for the pIgR, and A1AT preserves its full ability to bind to and inhibit NE, so the critical structural features of both portions of the fusion protein are apparently not compromised. The fusion protein is transported specifically from the basolateral surface of an epithelial cell monolayer to the apical surface. After transcytosis, the fusion protein is released from the cell bound to SC and retains the ability to form an inactivation complex with NE. The kinetics of elastase inhibition by the fusion protein alone are similar to those of the fusion bound to the extracellular portion of the pIgR. It is crucial that the binding domain of the fusion inhibits elastase with kinetics similar to those of the native antiprotease, because it will have to compete with an excess of natural substrates in the lung, such as elastin, for elastase binding sites (34).

In conclusion, this approach provides us with the ability to deliver a therapeutic antiprotease preferentially to the apical surface of the respiratory epithelium, under the mucus blanket, where it will be in highest concentration at the site likely to be critical for preventing structural damage and interrupting the inflammatory response in the CF lung. If this approach is successful in vivo, it could serve as a prototype for other therapeutic proteins, such as antiinflammatory cytokines and peptide antibiotics, that would be fused to an anti-pIgR scFv for delivery to the relatively inaccessible, receptor-expressing lumenal surfaces of the epithelium.