INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Although pulmonary infection contributes to the morbidity of patients with cystic fibrosis (CF), the intense host inflammatory response largely accounts for the progressive, suppurative pulmonary disease in CF, which usually leads to death. Antiinflammatory agents have successfully been used to control airway inflammation and retard the pulmonary deterioration characteristic of CF (1, 2). Neutrophils are the prominent inflammatory cells in the lungs of patients with CF, including those with minimal pulmonary disease (3). The influx of neutrophils in response to Pseudomonas aeruginosa endobronchitis is mediated by a number of chemoattractants, such as interleukin (IL)-8 and leukotriene-B₄. The neutrophil granules contain neutrophil elastase (NE), a proteolytic enzyme that degrades the outer membrane of Pseudomonas, but unfortunately also degrades structural elastin, which weakens the CF airway and results in bronchiectasis and bronchomalacia. Local defenses, such as human secretory leukocyte protease inhibitor (SLPI), have evolved to protect the airway (4). However, these natural defenses may be overwhelmed or compromised as the lung disease characteristic of CF progresses, and under these circumstances it is important to neutralize inflammatory stimuli at the surface of the airway epithelium.

₁-Antitrypsin (₁-AT), a glycoprotein of 52 kD that is the most abundant of the antiproteases in human serum, is produced naturally by hepatocytes and mononuclear phagocytes, and diffuses into the alveoli and pulmonary interstitium, where it acts as the primary inhibitor of elastase released from neutrophils (5). ₁-AT can block the effect of NE on the respiratory epithelium, which should reduce IL-8 and macromolecule secretion, as well as prevent degradation of structural proteins of the airway wall. Although ₁-AT is increased in the airway surface fluid of patients with CF, the abundance of NE in their airways overwhelms this antiprotease in the lung, resulting in inadequate protection of the lower respiratory tract against the lytic effects of NE (6). Moreover, circulating ₁-AT diffuses primarily to the alveolar space and not the airways, the site at which neutralization of NE is required in CF.

It may be possible to efficiently deliver the ₁-AT to relatively inaccessible CF airways by targeting the respiratory epithelium via the polymeric immunoglobulin receptor (pIgR). Once synthesized, the pIgR is trafficked to the basolateral surface of epithelial cells, where it is specifically adapted for the internalization and nondegradative transfer of polymeric antibodies (7) (i.e., dimeric immunoglobulin A [dIgA] and pentameric immunoglobulin M [pIgM]). Once internalized, the receptor-ligand complex is transported across the cell to the apical surface, where the receptor is cleaved from the complex, releasing dIgA bound to the ectoplasmic domain of the receptor, or secretory component (SC), into the lumen of the airway. In humans, the receptor is expressed in airway epithelial cells that reach the luminal surface, and in cells of the submucosal glands, especially serous cells (8). The pIgR may be ideal for the transport of proteins to the airways, and could permit the delivery of the ₁-AT to the apical surface of the respiratory epithelium.

We have developed a unique strategy for circumventing these difficulties and protecting the respiratory epithelial cell surface directly, by capitalizing on the properties of the pIgR (9). We have shown that fusion proteins containing a single-chain Fv antibody directed against human SC, the extracellular portion of the pIgR, can transport ₁-AT from the basolateral to the apical surface of a monolayer consisting of Madin-Darby canine kidney (MDCK) cells transfected with the complementary DNA (cDNA) for the human pIgR (Figure 1). If this approach is successful in respiratory epithelia, it may provide the ability to deliver a therapeutic protein directly to the luminal surface of the epithelium. In this case the highest concentration of ₁-AT will be at the apical surface, where it can protect epithelial cells against stimulation by NE and protect the airway wall from degradation. We report here that such a fusion protein was effectively transported in a vectorial fashion in two model systems consisting of polarized 16HBEo human airway epithelial cells transfected with human pIgR cDNA and of human respiratory epithelial cells grown in primary culture at an air-liquid interface, which spontaneously express the receptor. Moreover, high levels of functional antiprotease were measured in the apical surface fluid of the primary respiratory cells, approximating the concentration needed to neutralize free NE in the CF airway.

View larger version (34K): [in this window] [in a new window]   Figure 1.   Schematic diagram of transport of fusion protein from the systemic circulation to the epithelial surface. The fusion protein is bound to pIgR at the basolateral surface, trafficked to the apical membrane, and released into the airway lumen bound to SC, where the antiprotease component binds and neutralizes elastase.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Construction of Antihuman SC Single-Chain Fv

Balb/c mice (Jackson Laboratories, Bar Harbor, ME) were hyperimmunized with purified human SC, which was isolated from human colostrum. Splenocytes were harvested from seropositive mice, fused with the SP2/0 mouse myeloma cell line through the use of standard methods (10), and placed in selective media containing hypoxanthine and thymidine. Several subclones continued to produce antihuman SC antibodies, as detected with an enzyme-linked immunosorbent assay (ELISA). The antibodies were screened for their ability to bind secretory IgA, which reduced the likelihood that the single-chain Fv that we generated would compete with the natural ligand for binding site(s) on the pIgR. The variable light (V_L) and variable heavy (V_H) chain domains of the antihuman SC antibodies were then cloned from the hybridoma cell lines (11). The DNA sequences of the V_L and V_H domains were then spliced together, with separation from one another by an interdomain linker that encodes 14 amino acids (12, 13); the domains of the single-chain Fv were assembled in the order V_L- linker-V_H.

Expression of the Anti-SC Fv/₁-AT Fusion Protein

A chimeric gene, consisting of cDNA sequences encoding the full-length, antihuman SC Fv and human ₁-AT, was constructed as described previously (9). The coding sequence for the mouse immunoglobulin kappa leader was inserted to permit secretion of the fusion protein from transduced eukaryotic cells (14). The fidelity of the chimeric gene was verified by restriction-site analysis. Moreover, the nucleotide sequence of the chimeric gene was examined by dideoxy chain termination, and no rearrangements were noted. The fusion construct was subcloned into the expression plasmid pPac, containing the Drosophila actin 5C promoter for constitutive expression, and was then transfected into Drosophila Schneider Line 2 (S2) cells. Stable cell lines were produced after selection with hygromycin B, and were maintained through established techniques (9). Maximal secretion of the fusion protein (Mr = 78 kD) was 5 µg/ml/wk in flasks of confluent cultures. The fusion protein was stored at 4° C before use.

Cell Models

Cells of the 16HBEo bronchial cell line were transduced with cDNA encoding the human pIgR (15). Stably transfected cells were selected for neomycin resistance, and positive cells were sorted repeatedly to select those expressing the highest level of the human pIgR, using an EPICS-Elite ESP fluorescence activated cell sorter (Coulter, Inc., Hialeah, FL). Cells expressing high levels of the receptor were isolated and maintained in Dulbecco's modified Eagle's medium (DMEM) and Ham's F-12 medium (Gibco Laboratories, Grand Island, NY) supplemented with 10% heat-inactivated fetal calf serum, 100 U/ml penicillin, and 100 µg/ml streptomycin. For these experiments, nontransfected and receptor-bearing 16HBEo cells were grown on semipermeable filters (Becton Dickinson, Franklin Lakes, NJ). When grown in this manner, the transfected 16HBEo cells form tight junctions and appropriately traffic human pIgR. All media, sera, and antibiotics were purchased from Gibco Laboratories.

Airway epithelial cells in primary culture were also used in these investigations (16, 17). These cells were isolated from the nasopharynx, trachea, and bronchi of normal humans and were seeded onto collagen-coated, semipermeable Millicel-HA membranes (Millipore, Bedford, MA). The resultant respiratory cell monolayers were grown at an air-liquid interface, and maintained in DMEM and Ham's F12 medium (DME/F12) supplemented with 2% Ultroser (Bioseptra, Villeneuve, France) and antibiotics, as previously described (16). Measurements of transepithelial resistance showed that the cell monolayers were intact.

Transcytosis of Antihuman SC Fv/₁-AT Fusion Protein across Respiratory Epithelia

Purified fusion protein or human ₁-AT was added to culture medium on either the apical or basolateral side of confluent monolayers of respiratory epithelia grown on semipermeable filters. After incubation for 24 h at 37° C, the media or washing from the opposite compartment were collected and analyzed for ₁-AT and human SC, using immunoblot analyses or a sandwich-type ELISA.

ELISAs

For quantitation of ₁-AT, washings or media from either the apical or basolateral compartment were incubated in microtiter plates coated with goat antihuman ₁-AT (INCSTAR, Stillwater, MN) as a capture antibody. The bound protein was detected by incubation with rabbit antihuman ₁-AT antibody (Sigma Immunochemicals, St. Louis, MO), followed by a horseradish peroxidase-conjugated secondary antibody. The plates were developed with a tetramethylbenzidine (TMB) peroxidase substrate system and absorbance was measured at 450 nm. Purified human ₁-AT was used for standardization. For competition assays, human SC was bound to microtiter plates and increasing concentrations of the antihuman SC antibody produced by the parental hybridoma cell line, an irrelevant monoclonal IgG₁, and dIgA were coincubated individually with the antihuman SC Fv/₁-AT fusion protein at molar ratios relative to the amount of fusion protein applied to the microtiter wells, ranging from no competitor to a 200-fold molar excess of competitor. The ELISA for the antiprotease portion of the fusion protein was then performed as described earlier, and the amount of bound ₁-AT was expressed as a percentage of fusion protein binding in the absence of competitors.

The respiratory epithelial cell monolayers were also analyzed for human SC with a sandwich-type ELISA. To capture human SC, apical washings and basolateral media were applied to wells of microtiter plates that had been coated with 5 µg/ml of a goat-derived, polyclonal antihuman SC antibody (Sigma Immunochemicals), and the plates were incubated overnight at 4° C. On the following day the wells were washed and then blocked. Bound human SC was detected by sequential incubations with antihuman SC antibody and horseradish peroxidase-conjugated antimouse IgG secondary antibody. The readings were compared with those for standard concentrations of purified human SC. Concentrations of proteins in the airway surface fluid were calculated on the basis of previously measured volumes (16).

Immunoblot Analysis of SC and Human ₁-AT

Proteins from washings or media collected from the apical and basolateral compartments were subjected separately to electrophoresis in sodium dodecylsulfate (SDS)/10% polyacrylamide gels, and were transferred onto nitrocellulose membrane filters using standard techniques. The blots were blocked and then sequentially incubated at room temperature (RT) with either rabbit-derived antihuman ₁-AT or antihuman SC antibody, followed by incubation with a horseradish peroxidase-conjugated secondary antibody. The membranes were washed repeatedly, and 10 ml of enhanced chemiluminescence detection solution (Amersham, Arlington Heights, IL) was applied for 1 min. The luminescence emitted from the filter was detected by exposure to photographic film.

Antiprotease Activity of the Transported Fusion Protein

To investigate the ability of the ₁-AT domain of the fusion protein to form an inactivated complex with human NE, 0.4 pmol of the purified protease (Calbiochem, La Jolla, CA) was incubated at RT for 30 min with an equivalent amount of fusion protein that was transported to the apical surface of respiratory epithelial cell monolayers. The protease-antiprotease complexes were subjected to electrophoresis in SDS/10% polyacrylamide gels, and were detected by Western blotting (9).

Statistical Analysis

Data are expressed as mean ± SEM. Statistical comparisons of the control with the various treatment groups were made with paired Student's t tests or single-factor analysis of variance (ANOVA). The relationship between transcytosis of the fusion protein and human SC release was examined by determining the Pearson's product moment correlation coefficient (r) and least squares linear regression analysis.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Transport of Fusion Proteins Across Transfected Bronchial Epithelial Cells that Overexpress the pIgR

We examined the transport of fusion proteins that target the pIgR across the respiratory cell models used in our study. The fusion protein, consisting of single-chain Fv protein directed against human SC and linked to human ₁-AT, was secreted from transfected Drosophila Schneider Line 2 cells and extracted through affinity column chromatography. It was recognized by an antihuman ₁-AT antibody, and the Fv domain of the fusion protein retained its antigen-binding activity, as demonstrated by its ability to recognize immobilized human SC in an ELISA (Figure 2). This binding could be blocked by the addition of increasing amounts of the parental hybridoma cell anti-SC antibody, but not by irrelevant antibodies or by dIgA (Figure 2), even at the level of receptor saturation. The lack of competition of the natural ligand for the pIgR was not unexpected, since the antibody used to generate the Fv protein recognizes secretory IgA as well as free SC.

View larger version (15K): [in this window] [in a new window]   Figure 2.   Inhibition of fusion protein binding to human SC by ligands. Human SC was bound to microtiter plates and the antihuman SC-based fusion protein was coincubated with increasing concentrations of the ligands, parental hybridoma antibody, and dIgA, as well as with an irrelevant monoclonal IgG1. ELISAs were performed to quantitate the binding of the fusion protein to human SC. The percentage of fusion protein bound to the receptor in the presence of the potential competitors was then calculated. The data shown are from two different experiments (n = 3 to 6 replicates). Increasing amounts of the parental hybridoma antibody virtually eliminated the binding of fusion protein to the receptor, whereas the irrelevant IgG1 and the natural ligand dIgA had no effect on the ability of the fusion protein to recognize human SC.

Cells of the human bronchial epithelial cell line 16HBEo were transfected with the cDNA encoding the human pIgR. Human SC was asymmetrically released at the apical surface of epithelial cell monolayers, and significantly less receptor was secreted into the basolateral medium (approximately 20% of the quantity released apically). In addition, dIgA, the natural ligand of SC, and monoclonal antihuman SC antibodies, were transported vectorially from the basolateral to the apical membrane, where they were released into the medium. With this system as a model polarized respiratory epithelium, the fusion protein was transported through receptor-bearing 16HBEo cells from the basolateral to the apical surface, where it was released into the luminal medium (Figure 3). These receptor-expressing cell trafficked less fusion protein across themselves in the opposite (i.e., apical-to-basolateral) direction. Since the Fv domain remains bound to SC after it has reached the apical surface, the fusion protein cannot be transported back across the epithelium. Indeed, the addition of human SC to the basolateral medium blocks transcytosis across receptor-bearing cells (data not shown). Free, unmodified ₁-AT did not penetrate the cell layer in either direction (Figure 3), and the fusion protein did not penetrate nontransfected 16HBEo cells (data not shown).

View larger version (18K): [in this window] [in a new window]   Figure 3.   Transcytosis of the fusion protein through monolayers of 16HBEo cells that express pIgR. The antihuman SC Fv/1-AT fusion protein or purified 1-AT (26 pmol/well) was added to media in the apical (A) or basolateral (B) compartments of receptor-bearing 16HBEo cell monolayers grown on permeable filters. After 24 h the medium from the opposite compartment was collected and assayed for 1-AT with a sandwich-type ELISA (n = 4) in each treatment group. The basolateral-to-apical transcytosis (solid columns) of the scFv/1-AT fusion protein by pIgR-expressing cells was significantly greater than its apical-to-basolateral (open columns) transcytosis (4.4 ± 0.7 pmol/filter/d and 2.0 ± 0.2 pmol/filter/d, respectively). Free 1-AT was not effectively transported across the monolayer in either direction (0.2 ± 0.0 pmol/filter/d and 0.2 ± 0.1 pmol/filter/d, respectively).

Transcytosis of Fusion Protein across Primary Cultures of Respiratory Epithelial Cells

We examined the transcytosis of intact antihuman SC antibody or the natural ligand (dIgA) of the pIgR across monolayers of airway epithelial cells grown in primary culture at an air-liquid interface. Cells isolated from human nasopharynx, trachea, or bronchi and seeded onto collagen-coated porous filters produced human pIgR in culture, on the basis of immunostaining for the receptor and asymmetric release of SC into apical washings (Figure 4). These cells, when grown in this manner, spontaneously expressed the pIgR. Consistently, both dIgA and antihuman SC antibody were transported in the basolateral-to-apical direction across monolayers of these epithelial cells (Figure 5a). Irrelevant monoclonal antibodies, however, did not penetrate the monolayer. Thus, transport across respiratory epithelium was specifically mediated through the pIgR. The natural ligand applied to the basolateral surface of the epithelial monolayer was trafficked in a dose-dependent manner (Figure 5b), and little transport of the natural and monoclonal antibody ligands occurred in the opposite direction. In this particular model, the respiratory epithelial cells were grown at an air-liquid interface, and only an extremely thin layer of fluid covered the monolayer, as occurs in vivo. The transcytosed ligands were concentrated atop the apical surface of the epithelial cells (Figure 5c).

View larger version (45K): [in this window] [in a new window]   Figure 4.   Expression and release of human SC from polarized respiratory epithelial cells. Proteins secreted into the apical (A) or basolateral (B) compartments of cultured tracheal (Tr), bronchial (Br), or nasopharyngeal (N) epithelial cells that were grown at the air-liquid interface were separated by electrophoresis in SDS polyacrylamide gels and transferred, and human SC was identified by Western blot hybridization. Human SC (hSC) was asymmetrically released from all three respiratory epithelia, with greater concentrations of the receptor present in the apical surface fluid. Molecular weight standards (kD) are shown in the right margin.

View larger version (17K): [in this window] [in a new window]   View larger version (15K): [in this window] [in a new window]   View larger version (18K): [in this window] [in a new window]   Figure 5.   Transport of ligands for the pIgR across polarized respiratory epithelial cells in primary culture. (A) Antihuman SC IgG or dIgA (20 pmol/well) were applied to the basolateral surface of a bronchial epithelial cell monolayer, and apical washings were collected 24 h later. An irrelevant murine IgG1 (IgG) was also added to the basolateral compartment to examine the specificity of ligand transport and relative permeability of the cell monolayer. Cells that received no treatment (NT) were used as additional controls. The amounts of Ig transported into the apical surface fluid were then measured with an ELISA (n = 4) in each treatment group. The antihuman SC antibody and dIgA were effectively transported in the basolateral-to-apical direction across the respiratory epithelial cells, whereas virtually none of the irrelevant antibody penetrated the monolayer. (B) Increasing amounts of dIgA (0, 0.2, 2, and 20 pmol/well) were added to the basolateral medium of monolayers of bronchial epithelial cells, the apical surface was washed, and epithelial lining fluid was analyzed for the amount of transported dIgA (solid columns) and human SC (open columns) with an ELISA (n = 4) in each group. The amount of dIgA transported across the respiratory epithelial cells in a basolateral-to-apical direction rose in a linear manner with increasing concentrations of ligand added to the basolateral compartment (F = 13.2, p < 0.004). (C ) Twenty-four hours after the application of antihuman SC IgG antibody (26 pmol/well) and dIgA (21 pmol/well) to either the basolateral or apical surface of human bronchial epithelial cells, media and washings were collected from the apical (solid columns) or basolateral (open columns) compartments, respectively, and the amounts of human SC, antihuman SC antibody, and dIgA were measured with ELISAs. The actual concentrations of free SC and ligands in the epithelial lining fluid were then calculated on the basis of previously established volumes of apical surface fluid measured in this cell model.

The antihuman SC Fv/₁-AT fusion protein was transported in the basolateral-to-apical direction across monolayers of airway epithelial cells grown in primary culture. Significantly less fusion protein was transcytosed across these cells in the opposite direction (p = 0.002). Indeed, more than 25% of the fusion protein applied to the basolateral surface of monolayers penetrated the respiratory epithelia. The amount of fusion protein transcytosed correlated (r = 0.75) with the amount of SC that was released (Figure 6a). Linear regression analysis showed that the transcytosis of fusion protein was related to the amount of SC detected at the apical surface (F = 13.2; p = 0.004). In addition, because the fusion protein was transported vectorially to the apical surface and deposited into the small volume of epithelial lining fluid, the antiprotease was concentrated at the immediate apical surface of the epithelium (Figure 6b). Uptake of the fusion protein was mediated by the specific interaction of the anti-SC antibody with the human pIgR, and the fusion protein was transported to the apical surface of cells in vitro, resulting in high surface concentrations of the Fv-based ligands. Moreover, basolateral-to-apical transcytosis of the fusion protein by the respiratory epithelial cells use in the study could effectively be inhibited by the addition of excess anti-pIgR antibody from parental hybridoma cells, providing further evidence for specific, receptor-mediated transport. In contrast, the addition of excess dIgA, the natural ligand for the pIgR, did not inhibit transcytosis of the fusion protein (Figure 7).

View larger version (12K): [in this window] [in a new window]   View larger version (20K): [in this window] [in a new window]   Figure 6.   Transport of fusion protein across respiratory epithelial cells in primary culture at an air-liquid interface. (A) The relationship between transcytosis of the fusion protein and human SC release was examined. Twenty-four hours after addition of the fusion protein (18 pmol/well) to the basolateral surface of epithelial cell monolayers, apical washings were collected and the amount of human SC or immunoreactive 1-AT transported was measured with an ELISA. The amount of fusion protein transported correlated with the amount of SC released (r = 0.75). (B) The antihuman SC Fv/1-AT fusion protein was effectively transported in the basolateral-to-apical direction (solid columns; n = 12), with high concentrations of both the human SC and fusion protein present in the epithelial surface fluid (5.4 ± 0.6 µM and 3.0 ± 0.3 µM, respectively). Considerably lower concentrations of receptor and fusion protein were measured in the basolateral compartment (open columns; n = 11) at 24 h after application of the protein to the apical surface (p = 0.002).

View larger version (28K): [in this window] [in a new window]   Figure 7.   Competition for transcytosis of fusion protein across respiratory epithelia in primary culture by other ligands. The antihuman SC Fv/1-AT fusion protein (18 pmol/well) was added to the basolateral compartment of respiratory epithelial cell monolayers with either 7.5 µg of the parental hybridoma bivalent monoclonal antibody (100 pmol ligand/well), from which the single chain Fv was derived, or with 30 µg dIgA (100 pmol/well). The apical surface was washed on the following day, and the epithelial lining fluid was collected and analyzed for 1-AT with a sandwich-type ELISA (n = 12 in each treatment group). The basolateral-to-apical transcytosis of the fusion protein-type was significantly inhibited by the parental hybridoma antibody (p < 0.001), but excess dIgA did not affect transport of the fusion protein across the respiratory epithelium.

We have previously shown that the antihuman SC Fv/₁-AT fusion protein became bound to and inhibited NE with an association rate constant similar to that of plasma ₁-AT (9). Similarly, the antiprotease portion of the fusion protein remained active after transcytosis and apical release from respiratory epithelium. Since ₁-AT irreversibly binds to its substrate to form an inactivation complex, antiprotease activity can be shown by an increase in size of the fusion protein by 29 kD, the molecular weight of elastase (18). Such an increase was found when the transported fusion protein from the apical medium was incubated with equimolar amounts of NE, thus demonstrating that protease-antiprotease complexes were formed and that the ₁-AT domain of the fusion protein retained antiprotease activity after transcytosis and binding to human SC (Figure 8).

View larger version (38K): [in this window] [in a new window]   Figure 8.   Binding of the antiprotease component of transcytosed fusion protein to NE. Purified antihuman SC Fv/1-AT fusion protein and fusion protein from the apical media after transcytosis were incubated with and without NE at 22° C for 30 min, and the resultant protease-antiprotease complexes were separated by electrophoresis in SDS polyacrylamide gel under nonreducing conditions, followed by Western blot analysis. Lane 1: purified human 1-AT (0.6 pmol); lane 2: purified human 1-AT after incubation with elastase (1.7 pmol); lane 3: fusion protein (0.6 pmol 1-AT content); lane 4: fusion protein after incubation with elastase; lane 5: fusion protein after transcytosis; lane 6: fusion protein after transcytosis, following incubation with elastase. The fusion protein became bound to NE, producing an inactivation complex that was detected by Western blotting as a 29-kD shift in molecular weight.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We have previously shown that the recombinant fusion protein consisting of a single-chain Fv directed against the pIgR and linked to human ₁-AT, maintains full elastase-inhibitory activity and can be specifically transported across a model system of MDCK cells transfected with human pIgR cDNA (19). In the present study, we extended this observation to respiratory epithelial cells. Two systems were studied: polarized human bronchial epithelial cells transfected with human pIgR cDNA, and human respiratory epithelial cells that spontaneously express pIgR. The fusion protein, bound to SC, is released from the cell and can form an inactivation complex with NE. This approach therefore provides the ability to deliver a therapeutic antiprotease preferentially to the apical surface of the respiratory epithelium, beneath the mucus blanket, where it will reach its highest concentration at the site likely to be critical for preventing structural damage and interrupting the inflammatory response in CF (Figure 9). Indeed, the antiprotease concentrations that we measured in the epithelial lining fluid of respiratory epithelial cells grown at an air-liquid interface approximate the level of free NE measured in bronchoalveolar lavage fluid from the CF lung.

View larger version (24K): [in this window] [in a new window]   Figure 9.   Alternative delivery systems for inhibitors of NE to the CF airway. Left: systemic administration of 1-AT delivers this antiprotease in greatest concentration to the alveoli and interstitium. Center: delivery of 1-AT by nebulization is inefficient, uneven, and deposits the drug atop the mucus blanket rather than at the apical surface of the cell, which is the critical site. Right: antihuman SC- based fusion protein penetrates the bronchial epithelium and accumulates in the airway surface fluid.

Our results with both cell models used in our study indicate that the trafficking machinery in airway epithelial cells, as in MDCK cells, will support the basolateral-to-apical transfer of fusion protein that retains its NE-inhibitory function. The results of studies done with respiratory epithelial cells grown at an air-liquid interface, in which the pIgR is spontaneously expressed, illustrate another advantage of this therapeutic strategy. These cultures mimic the situation in native epithelium, in the sense that the volume of fluid covering the apical surface is quite small. Most measurements have estimated that this layer is only 10 to 20 µm deep in the conducting airways (20, 21). Efficient, one-way delivery of therapeutic protein into this small volume results in high concentrations of the protein at the apical surface. In our experiments, as much as 25% of the transcytosed pIgR carried the fusion protein, and the concentration of the fusion protein was approximately 3 µM, a level exceeding that of uninhibited elastase in the airway epithelial surface fluid of most children with CF (22, 23), as well as of adults with mild pulmonary disease (3). We have shown that the binding and transport of the antihuman SC Fv/₁-AT fusion protein was not inhibited even with a marked excess of dIgA, because the single-chain Fv probably binds to a site distinct from that of the natural ligand, which most likely interacts with domain I of the pIgR (24). Thus, this system is capable of concentrating the therapeutic protein in the thin film of airway surface fluid to a level expected to be therapeutic in the CF airway for many patients.

Airway epithelium is a barrier. Epithelia are generally impermeable to macromolecules, and such agents are prevented from diffusing across these tissues by the tight junctions that connect adjacent epithelial cells (25). Yet the pIgR, specifically adapted for the transcytosis of polymeric antibodies across epithelia, could permit drugs to penetrate these tissues. Once internalized, the receptor-ligand complex is transported across the cell to the apical surface, where the receptor is cleaved, releasing dIgA bound to the ectoplasmic domain of the receptor, or SC, into the airway lumen. This extracellular ligand-binding domain of the pIgR contains five homologous repeating immunoglobulinlike domains of 100 to 110 amino acids each (26), and is highly conserved across species (27). The natural ligands of the pIgR need not be present for receptor endocytosis, and as discussed earlier, the natural ligand pIgR does not inhibit transport of the fusion protein via the pIgR, which is critical for purposes of drug delivery. Several other potential barriers between the vascular compartment and the epithelial surface must also be traversed for the successful delivery of drugs to the airway. The endothelial cell, interstitium, and epithelial basement membrane must be penetrated for the fusion protein to reach the pIgR on the basolateral surface of the epithelium. Clearly though, the bulky, natural ligands of the receptor can get to this target.

Recombinant fusion proteins that target the pIgR may have additional advantages for delivery of therapeutic components to the lungs of patients with CF. The pIgR is expressed early in fetal life, and SC has been found in tracheal aspirates from premature neonates (28). Patients with CF express the receptor at normal or, in some instances, increased levels (T. Ferkol, S. Saeed, and M. Konstan, unpublished observation). In fact, inflammatory mediators released during pulmonary inflammation may augment the expression of this receptor in the airway. Various cytokines have been shown to increase SC release from epithelial cells in vitro (29, 30). Consequently, more inflamed airway surfaces may be better accessed by this strategy, since proinflammatory cytokines upregulate the expression of pIgR, making more receptors available to ferry the fusion protein to the apical epithelial surface in inflamed areas. In addition, inflamed areas have increased blood flow, increasing the delivery of blood-borne therapeutic agents. The natural ligand dIgA also appears to stimulate transcytosis through the pIgR (31).

SC itself in bronchial secretions may be protective. The natural ligand in secretions is often exposed to a harsh environment, but the binding of SC to dIgA shields this antibody from proteolytic degradation, thus enhancing its biologic effectiveness (32). It is possible that SC affords other proteins similar protection from proteases. Additionally, the use of a single-chain Fv protein has considerable appeal, in that a ligand for the pIgR is reduced to its essence. Since most of the species-specific constant regions of the receptor have been excluded from the construct, and the remainder can be altered if necessary, this component of the recombinant bifunctional protein should be poorly immunogenic.

Delivery of the fusion protein systemically has advantages over the administration of ₁-AT itself. Delivery of ₁-AT by the intravenous route depends on its leakage into the alveolar space from the pulmonary circulation and carriage to the airways via the mucociliary escalator, yielding a low concentration of ₁-AT. Because pulmonary capillary blood flow will be diverted away from poorly ventilated regions of the lung, this leakage is likely to be smaller in areas of significant airway obstruction, which are precisely where therapeutic agents are needed most in the CF lung. Indeed, concentrations of ₁-AT in the airway surface fluid after its intravenous administration were too low to be therapeutic in CF (6).

Human ₁-AT can be delivered directly to the lung by inhalation, but this route is especially problematic in patients with CF (33). In diseased lungs, ventilation is preferentially distributed to gas-exchange units with the lowest resistance to airflow, which diverts inhaled agents away from the diseased regions of lung that require them the most. The distribution of inhaled agents in the lungs of patients with CF is inhomogeneous because of mucus obstruction of the airways and atelectasis, and it is likely that little of a nebulized antiprotease would be deposited in the most diseased regions of the lung. Moreover, aerosols deposit atop the mucus blanket, and much of a nebulized ₁-AT will be neutralized by NE in inflamed airway secretions before it penetrates to the epithelial surface. Aerosol administration of plasma-derived ₁-AT to patients with CF produced highly variable concentrations of this antiprotease in the lung, although active NE in bronchoalveolar lavage samples was partly suppressed, and the antiprotease also reversed the inhibitory effects of NE on opsonophagocytosis of Pseudomonas aeruginosa by neutrophils (34). The difficulties associated with both systemic and aerosol administration of antiproteases underscore the importance of considering an alternative method of delivering these substances to the CF airway. Delivering ₁-AT beneath the mucus blanket in the airways clearly has the advantage of allowing it to act preferentially at the cell surface to block the noxious effects of NE.

Our data indicate that even for therapeutic agents required in considerable quantity, such as antiproteases in CF, ferrying of a fusion protein containing such an agent, by taking advantage of essentially unidirectional trafficking and the remarkably small volume into which it is delivered, may yield therapeutic levels of active agent at the cell surface. The therapeutic payloads are not limited to antiproteases. Indeed, a system such as that described in this report could preferentially deliver other functional proteins to the respiratory epithelial surface, especially agents that are required at lower concentrations. For instance, antiinflammatory agents could be transported to the epithelial surface in this manner to temper the intense inflammatory response induced by endobronchial infection in the CF lung. It may also be possible to control airway inflammation in CF by preventing the interaction between bacteria and epithelial cells that incites this inflammation. It may be crucial that protection occurs right at the epithelial surface, and that antimicrobial agents could be deposited in the periciliary space with a bifunctional protein such as that described here. Thus, the fusion protein reported here is a prototype of a much broader class of agents, of which the therapeutic index is improved by targeting them preferentially to the epithelial surface. This approach is generic.