Limited Entry of Adenovirus Vectors into Well-Differentiated Airway Epithelium Is Responsible for Inefficient Gene Transfer

Virus preparations.

Human Ad5 vector (AdV), with deletions of the E1a, E1b, and E3 genes, contained the Escherichia coli lacZ gene under the control of the cytomegalovirus promoter (9). [³⁵S]methionine-radiolabeled AdVCMVlacZ (³⁵S-AdV) was prepared as described previously (16) with modification for 293 cells by harvesting at 48 h postinfection. The adenovirus particle-to-infectious-unit ratio was routinely 100:1 and was determined by plaque assay on either 293 or 911 cells (similar values were obtained for the two cell lines). Serial dilutions of virus in Dulbecco’s modified Eagle medium (DMEM-H) were incubated with 50% confluent cell cultures in six-well plates at a volume of 1 ml per well (fluid height, 0.1 cm). The cultures were incubated for 1 h at 37°C without agitation. The medium was removed and replaced with 3 ml of agar overlay (1× DMEM-H, 10% fetal bovine serum [FBS], and 1% SeaPlaque LGT agarose). The cultures were fed with 1 ml of overlay (containing 2% FBS) every 3 days until the plaques were counted at 6 days postinfection for 911 cells or 10 to 12 days postinfection for 293 cells. The specific activity of the radiolabeled vector was approximately 4 × 10⁻⁵ cpm per particle. To directly visualize AdV, Cy3 FluoroLink (Amersham Life Science Inc., Arlington Heights, Ill.) was conjugated to the AdV capsid coat by incubation of 10¹² particles of AdVlacZ as described previously to form CyAdV (23). This process reduced the AdV titer by less than 10-fold (23). CyAdV was characterized by assessment of fluorescent attachment and transgene expression with HeLa and CHO K1 cell lines in the absence and presence of competing purified fiber-knob protein. The Ad5 fiber-knob protein was produced by expressing pBEVαfibre (a gift from Robert Gerard, Katholieke Universtiet Leuven, Louvain, Belgium) in E. coli TG-1 cells, and purified fiber-knob protein was obtained exactly as described previously (19).

Cell culture.

Human tracheobronchial epithelial cells were derived from non-CF airway specimens and were cultured by procedures similar to those described by Gray et al. (15). Portions of the lower trachea and mainstem bronchi representing excess donor tissue were obtained at the time of lung transplantation under institutional review board-approved protocols. Epithelial cells were removed from the specimens by protease XIV digestion as described previously (34), and 10⁶ cells were plated per 100-mm tissue culture dish in modified LHC9 medium (22). The modifications were increased epidermal growth factor concentration to 25 ng/ml, adjustment of the retinoic acid concentration to 5× 10⁻⁸ M, and supplementation with 0.5 mg of bovine serum albumin per ml and 0.8% bovine pituitary extract. At approximately 75% confluence, the cells were harvested with trypsin and passage 1 cells were plated at a density of 3.33 × 10⁵ cells on Transwell-Col inserts (diameter, 24 mm; pore size, 0.4 μm; Corning-Costar, Cambridge, Mass.) in modified medium. The medium is similar to the supplemented LHC9, except that a 50:50 mixture of LHC Basal (Biofluids Inc., Rockville, Md.) and DMEM-H was used as the base, amphotericin and gentamicin were omitted, and the epidermal growth factor concentration was reduced to 0.5 ng/ml. After 4 to 6 days, the cells became confluent and were used as PD cultures. For the production of WD cultures, confluent cultures were maintained with an air/liquid interface for another 25 to 30 days.

Rat tracheal epithelial cells were isolated from pathogen-free male F344 rats (200 g), and 2 × 10⁵ cells were plated on permeable Transwell-Col matrix supports (as above) by the method of Kaartinen et al. (21). After 5 days of culture, the cells became confluent and were used as PD cultures. For the production of rat WD cultures, after the cells became confluent the apical surfaces of the cultures were given an air/liquid interface for at least 19 days.

HeLa cells (American Type Culture Collection) were plated on Transwell-Col supports (as above), grown to confluence, and maintained in Eagle’s minimum essential medium supplemented with nonessential amino acids and 10% FBS. HeLa cells expressing the tetracycline-sensitive wild-type and K44a mutant form of dynamin were kind gifts from Sandra Schmid (Scripps Research Institute, La Jolla, Calif.) (7) and were initially maintained in DMEM–10% FBS–400 μg of gentamicin per ml–200 ng of puromycin per ml–1 μg of tetracycline per ml. To induce dynamin overexpression, the cells were cultured in the absence of tetracycline for 2 days before being exposed to AdV.

Expression, entry, and attachment studies.

Analyses of expression, entry, and attachment were performed within a single batch of cells, on the same day, with identical reagents. For the lacZ expression studies, the luminal surfaces of cultures were exposed to 10¹⁰ particles of AdV (multiplicity of infection, ∼100) at 37°C for 6 h, unbound virus was removed from the cells by three washes in ice-cold medium, and the cells were returned to 37°C before gene expression analyses 48 h after the initial exposure to AdV. β-Galactosidase (β-gal) enzyme activity was assessed as described previously (18).

Internalization of radiolabeled AdV was assessed as follows. After exposure of cultures to AdV (10¹⁰ particles) for 6 h at 37°C, the cultures were transferred to 4°C and washed three times in ice-cold medium. The cells were then rinsed with an acid-salt wash (0.2 N acetic acid, 0.5 M NaCl [pH 2.5]) at 4°C and exposed for 1 h to protease (0.25% pronase XIV with 0.0025% DNase in serum-free culture medium at 4°C) to remove extracellular bound vector. This method effectively removed more than 95% of the vector attached to the cell surface at 4°C, as determined by assessing removal-resistant counts after this treatment. After further washing in ice-cold medium, the cells were solubilized in 1% sodium dodecyl sulfate (SDS)–0.3 N NaOH and the counts were assessed by liquid scintillation counting. For attachment studies, cultures maintained at 4°C were exposed to ³⁵S-AdV (10¹⁰ particles) for 6 h. The cultures were then washed three times in ice-cold medium, and the cells were solubilized and subjected to liquid scintillation counting. For the studies with purified fiber-knob protein and RGD peptides, the cultures were preexposed to fiber-knob protein (10 μg/ml), RGD peptide (4.0 mg/ml; Gibco BRL, Bethesda, Md.) or cyclical RGD peptide (0.4 mg/ml; Immunodynamics, La Jolla, Calif.) for 2 h at 4°C before the addition of AdV for 6 h at 4°C. The cultures were then washed three times in ice-cold medium and either immediately solubilized as above for scintillation counting or maintained at 37°C for a further 48 h before being subjected to expression analyses.

For both species, only fully confluent cultures were exposed to AdV to ensure a standard surface area and to avoid nonspecific binding of AdV to the matrix support. The transepithelial resistance (R_t) values of the cultures at the time of AdV exposure were as follows: human PD and WD cultures, 1,076 ± 95 (n = 48) and 1,342 ± 53 (n = 36) Ω · cm², respectively; rat PD and WD cultures, 223 ± 5 and 2,653 ± 75 Ω · cm², respectively (n = 12 for both). Radioactive counts per minute and β-gal activity were standardized with respect to the nominal surface area of the culture surface, since we consider the apical surface area of cells exposed to vector to be the most appropriate denominator, allowing direct comparison to the epithelium in vivo. The β-gal activity was measured as microunits per square centimeter of epithelium, where 1 U of enzyme will hydrolyze 1 μmol of o-nitrophenyl-β-d-galactopyranoside per min at pH 7.3 and 37°C.

For attachment and expression studies with HeLa cell mutants, cells were plated onto six-well plates in tetracycline-deficient medium to induce wild-type or mutant dynamin expression. After 2 days in culture, the cells were cooled to 4°C and exposed to 10¹⁰ particles of AdVlacZ per ml for 2 h. After three washes with ice-cold medium the cells were prepared for liquid scintillation counting as above or placed at 37°C for 48 h until used for expression analyses. For each stage of the experiment, parallel cultures were used to determine cell numbers.

Generation of hCAR-expressing cell lines.

To generate stable expression of hCAR in cell lines, hCAR cDNA was placed in a Moloney murine leukemia virus-based retroviral vector. The hCAR cDNA was a gift from Jeffrey Bergelson (Dana-Farber Cancer Institute, Boston, Mass.). Briefly, hCAR cDNA was cloned into the LxPin retroviral vector (24a) containing a poliovirus internal ribosome entry site and a neomycin resistance gene. The resulting plasmid was packaged by transient transfection of PA317 cells and pseudotyped with vesicular stomatitis virus glycoprotein G.

Functional analyses of the retroviral construct was performed by introducing hCAR into a cell line that exhibits reduced susceptibility to AdV transduction. Chinese hamster ovary (CHO K1; American Type Culture Collection) cells were infected with retroviral vectors containing hCAR-Neo (CAR-CHO) or Neo alone (CHO) as a control, and stably expressing cell lines were selected by standard methods (4). To test the functional expression of hCAR, cells were grown in 12-well culture dishes until confluent, ³⁵S-AdVlacZ was exposed to the cultures (10¹⁰ particles/ml for 2 h at 4°C) in the absence or presence of excess purified fiber-knob protein (10 μg/ml), and attachment and expression were assessed as above.

HeLa, CHO, and CAR-CHO cells were exposed to monoclonal antibody (MAb) RmcB (a hybridoma cell culture supernatant generated against hCAR [20], a gift from Jeffrey Bergelson) as follows. Cells grown on glass coverslips were cooled to 4°C and blocked with 3% bovine serum albumin. After incubation with RmcB, the cells were washed and exposed to fluorescein isothiocyanate-conjugated goat anti-mouse immunoglobulin G1 (IgG1) (Jackson ImmunoResearch Labs Inc., West Grove, Pa.) in the presence of CyAdV (10¹⁰ particles). The cells were washed in phosphate-buffered saline and fixed in paraformaldehyde (4%), mounted with Vectashield containing 4′,6-diamidino-2-phenylindole (DAPI) (Vector Laboratories Inc.), and viewed by conventional fluorescence microscopy. Controls consisted of untransfected CHO cells or CHO cells expressing neo alone. In addition, the absence of primary antibody or the use of irrelevant isotyped MAb (anti-bromodeoxyuridine; Boehringer-Mannheim Corp., Indianapolis, Ind.) were used as controls. Localization of hCAR in human airway cultures was performed as follows. Human PD and WD cultures were fixed and permeabilized in paraformaldehyde (4%) and Triton X-100 (0.2%), respectively. The cultures were then exposed to RmcB (or anti-bromodeoxyuridine as a control) and, after being washed, exposed to Texas Red-conjugated goat anti-mouse IgG1 (Jackson ImmunoResearch Labs Inc.). After being washed, the cultures were fixed in paraformaldehyde and viewed by confocal fluorescence microscopy, and XZ sections were generated.

Electron and confocal microscopy.

For the transmission electron microscopy studies, the luminal surfaces of human cultures were exposed to AdV (10¹⁰ particles) for 6 h at 4°C, washed as above, and fixed in 1.5% glutaraldehyde overnight before being processed for electron microscopy by standard methods (28). For experiments with fluorescent microspheres, the luminal surfaces of human cultures were exposed to 100-nm fluorescent microspheres (0.02% in medium; Molecular Probes) for 6 h at 37°C and then washed three times with medium. En face fluorescent images were taken with a conventional inverted fluorescence microscope (Leica DM IRB). Images perpendicular to the cell layer were captured by XZ sectioning with a confocal microscope (Leica TCS/4D), and images were generated with the Metamorph image analysis system (Universal Image Co.).

Immunoprecipitation and Western analyses.

Apical or basolateral membranes of human and rat PD and WD cultures were exposed to sulfosuccinimidobiotin (0.5 mg/ml; Pierce Chemical Co.) for 30 min at 4°C and then the cells were washed in ice-cold phosphate-buffered saline containing protease inhibitors (2 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, and 1 mM [each] leupeptin-pepstatin) and solubilized in lysis buffer (150 mM NaCl, 1% Nonidet P-40, 0.5% deoxycholate, 0.1% SDS, 50 mM Tris-HCl [pH 7.5] with protease inhibitors). Before immunoprecipitation of integrins, the cell lysates were precleared with normal rabbit serum and protein G beads (Pierce Chemical Co.). The α_vβ_3/5 integrins were immunoprecipitated by rabbit anti-human/rat α_vβ_3/5 integrin polyclonal antibody (838 [3], a kind gift of S. Albelda, University of Pennsylvania, Philadelphia) or a rabbit anti-human vitronectin receptor polyclonal antibody (AB1904; Chemicon International Inc.). Precipitates were subjected to SDS-polyacrylamide gel electrophoresis (14% Tris-glycine gel under reduced conditions) and transferred to polyvinylidene difluoride, and biotinylated membrane proteins were detected with streptavidin-conjugated peroxidase and visualized by enhanced chemiluminescence.

Inulin uptake studies.

To measure the nonspecific uptake capacity of cells, confluent PD human primary airway epithelial cells were generated on tissue culture plastic and WD cultures were generated as described above. The luminal surfaces of the cultures were exposed to freshly prepared [³H]inulin (35 μg/ml, 4 cpm/nl; Amersham Life Sciences Inc.) at either 4 or 37°C for 6 h. The cultures were then returned to 4°C, washed five times with ice-cold medium containing excess inulin (1 mg/ml), and solubilized for liquid scintillation counting. To measure cell uptake of [³H]inulin, cell-associated counts at 37°C were subtracted from those at 4°C as previously described (31). The volume of fluid uptake into the cells was calculated based on the known counts per minute of the applied solution.

Statistics.

Statistical analysis was performed by Student’s t test, and P < 0.05 was considered significant.

Analyses of gene transfer, vector entry, and attachment with models of respiratory epithelium. (i) Human cultures.

Modifications of a cell culture system reported by Gray et al. (15) generate from human tracheobronchial airway epithelial cells polarized cultures with PD and WD cellular phenotypes from the same patient (Fig. 1A). The morphology of WD cultures resembles the pseudostratified ciliated epithelium exhibited by human cartilaginous airway in vivo. Exposure of the apical surfaces of human WD and PD cultures to AdVlacZ resulted in significantly less β-gal expression in WD cultures than in PD cultures (Fig. 1B). These results demonstrate that human WD and PD airway epithelial cells are resistant and susceptible, respectively, to AdV-mediated gene transfer, recapitulating the phenomenon observed in vivo for human and rodent cartilaginous airway epithelia (18).

We have previously shown that the reduced efficiency of gene transfer to WD cells compared to PD cells is not due to a specific interaction of the transgene promoter with these cell types or to the proliferative status of the cells at the time of transduction (25). In the present study, we tested the hypothesis that early steps in the vector-cell interaction determine the gene transfer efficiency. Using radiolabeled AdV, we measured the penetration of AdV into cells and determined which step(s) is rate limiting for efficient AdV-mediated gene transfer to human airway epithelial cells.

To determine if WD and PD cultures internalized different amounts of AdV, both culture types were exposed to ³⁵S-AdV and the quantity of internalized AdV was determined by measuring the cell-associated counts that were resistant to removal from the external cell surface by acid and protease treatment. Figure 1C shows that the internalization of AdV into WD cultures was markedly reduced compared to the amount internalized into PD cultures. To investigate whether the differences in entry reflect the degree of attachment of AdV, the apical surfaces of human WD and PD cultures were exposed to ³⁵S-AdV at 4°C to measure cellular attachment of vector in the absence of internalization. These studies showed that the surface of WD cultures bound markedly less AdV than the PD cultures (Fig. 1D). Collectively, these results suggest that the reduction in gene transfer to WD cultures results from a reduced internalization of AdV into this cell type, which may be related to the absolute amount of AdV attached to the luminal surface of the cultures.

We confirmed this conclusion with a second human culture system, containing cellular islands that exhibit a WD phenotype in the center and a PD phenotype at the edges (24). These cultures were exposed to fluorescence-labeled AdV (CyAdV) at 37°C and viewed by fluorescence microscopy after being washed. At 6 h after exposure, CyAdV was routinely associated with PD cells at the periphery and only rarely associated with individual cells within the WD regions (Fig. 2A). This cellular distribution of CyAdV localization is paralleled by the cellular distribution of lacZ expression in the same cultures 24 h later, assayed by 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal) histochemistry (Fig. 2B). Therefore, by direct visualization of vector, AdV entry into PD cells far exceeds that into WD cells.

To investigate whether the differences in attachment and internalization with WD and PD culture types are specific to AdV, the uptake of fluorescent microspheres that approximate the size of the vector (100 nm) into PD and WD cultures was examined. Incubation of cultures with microspheres for 6 h at 37°C resulted in a large quantity of microspheres associated with PD but not WD cultures, as shown by conventional fluorescence microscopy (Fig. 2C and E, respectively). To determine the cellular localization of the microspheres, confocal microscopy-generated XZ sections revealed that they were both attached and internalized only into PD cultures (Fig. 2D). These data suggest that PD cultures can internalize nonspecifically attached particles and that the apical membrane of PD cultures can undergo nonspecific uptake processes. In contrast, microspheres did not enter WD cultures (Fig. 2F), due to reduced nonspecific attachment and possibly to reduced uptake into the cells.

To determine whether AdV attachment to PD and WD cultures is a specific fiber-knob-mediated interaction, transgene expression and AdV attachment were measured in the presence of excess purified fiber-knob protein. Experiments with HeLa cells showed that purified fiber-knob protein was capable of inhibiting transgene expression by more than 99% and AdV attachment by ∼75% (Fig. 3), indicating that specific fiber-knob binding to HeLa cells accounted for the subsequent transgene expression. Excess fiber-knob protein inhibited ∼70% of transgene expression in PD cultures and completely inhibited the low level of expression in WD cultures, suggesting that the predominant route of AdV entry into both culture types is via fiber-knob receptor-mediated endocytosis. In the PD but not the WD cultures, however, a significant portion of transgene expression was not blocked by excess fiber-knob protein. This non-fiber-knob-protein-mediated expression may be due to nonspecific uptake processes, as described above (Fig. 2). In contrast to transgene expression, AdV attachment to both PD and WD cultures was not inhibited by fiber-knob protein, suggesting that we cannot detect fiber receptor-specific attachment given the large component of “nonspecific” attachment in both culture types.

To investigate the mechanism of nonspecific attachment of AdV, we visually assessed the binding of AdV to PD and WD cultures by transmission electron microscopy (TEM). The majority of AdV attached to the human PD culture surface was associated with the abundant glycocalyx present on the microvilli (Fig. 4A), suggesting that the glycocalyx itself was binding AdV by a fiber-knob-independent mechanism. Assessment of the binding of AdV to human WD cultures by TEM (Fig. 4B) showed reduced levels of glycocalyx on the WD cultures and little AdV associated with the cell surface. Therefore, we speculate that the great nonspecific attachment of AdV to PD cultures reflects the relative mass of glycocalyx on these cells.

The recent identification of hCAR as a putative specific AdV attachment receptor (1, 29) prompted investigation of the localization of this receptor in human airway epithelial cultures. Retrovirus-mediated overexpression of hCAR cDNA in CHO cells (normally resistant to AdV-mediated gene transfer) leads to increased specific AdV attachment and increased specific AdV-mediated gene transfer to levels above that observed for HeLa cells (Fig. 5A). Immunofluorescence detection of hCAR was performed with MAb RmcB on HeLa, CHO, and CAR-CHO cells. RmcB detected hCAR on HeLa and CAR-CHO cells but not on control CHO cells (Fig. 5B, panels I, III, and II, respectively). In addition, coincubation of the cells with RmcB and CyAdV showed a similar distribution for the receptor and ligand, respectively, indicating that increased hCAR expression led to increased AdV attachment. These results show that hCAR mediates AdV attachment and AdV-mediated gene transfer in these cell lines and that RmcB can detect the presence of hCAR.

To determine whether hCAR is expressed in human airway epithelia, human PD and WD cultures were permeabilized and probed for hCAR expression with RmcB. In permeabilized human PD cultures, hCAR immunoreactivity was detected on the surfaces of all of the epithelial cells (Fig. 6A, panel I) whereas in permeabilized WD cultures hCAR was restricted to the basolateral membranes of the columnar cells (panel II). Control cultures probed with an irrelevant MAb showed little immunoreactivity above autofluorescence (panels III and IV). These results parallel the earlier findings of the AdV attachment and expression profiles for these culture types.

The basolateral distribution of both hCAR and α_vβ_3/5 integrins (see below and reference 14) in WD cells suggests that these cultures may be more susceptible to AdV gene transfer if access to the basolateral membrane was feasible. To test this notion, AdV was applied to either the basolateral or apical surface of human WD cultures. These studies showed that the gene transfer efficiency is enhanced when the basolateral surface rather than the apical surface is exposed to AdV (Fig. 6B).

(ii) Rat cultures.

Preliminary experiments suggested that PD airway epithelial cells derived from the rat tracheal epithelium exhibited less nonspecific AdV attachment than human PD cells (26). We therefore compared this well-characterized system (Fig. 7A) with our human studies. In agreement with the human model and in vivo studies, WD cultures were resistant and PD cultures were susceptible to AdV-mediated gene transfer (Fig. 7B). In contrast to human airway epithelia, the attachment of AdV to PD and WD cultures was similar (Fig. 7D). Because of the lower total attachment, specific fiber-knob inhibitable attachment could be detected and accounted for approximately 50% of the total attachment (rat PD, 648 ± 125 and 320 ± 73 cpm per cm² in the absence and presence, respectively, of fiber-knob protein [n = 9 for each]; rat WD, 449 ± 49 and 224 ± 60 cpm per cm² in the absence and presence, respectively, of fiber-knob protein [n = 9 for each]). Thus, the rat model allows a comparison of AdV internalization when specific and nonspecific AdV attachment was similar on both culture types. This comparison shows that the amount of AdV internalized into WD cells is far smaller than that internalized in PD cells, which correlates with the gene transfer efficiencies (Fig. 7C). These data indicate that the rate-limiting step for efficient gene transfer to the rat WD cultures is the internalization of AdV.

Mechanisms of internalization. (i) α_vβ_3/5 integrin localization and interactions with AdV.

AdV interactions with fiber-knob receptors and/or α_vβ_3/5 integrins are thought to initiate endosomogenesis via a mechanism involving receptor clustering. To test the hypothesis that reduced gene transfer in WD cultures compared to PD cultures reflects the absence of α_vβ_3/5 integrin “initiation” of coated-pit endosomogenesis, we measured the distribution of α_vβ_3/5 integrins in the apical and basolateral regions of PD and WD cultures. For human cultures (Fig. 8A, panel i) and rat cultures (data not shown), the α_vβ_3/5 integrins are distributed predominantly in the basolateral membranes of both PD and WD cultures. The α_vβ_3/5 integrins are absent from the WD apical membranes of both species and were expressed at low levels in the apical membranes of the PD cultures, a finding in agreement with that of Goldman and Wilson (14).

Whereas these data on integrin distribution correlate with the high and low levels of gene transfer to the respective culture types, attempts to demonstrate the functional importance of integrins in AdV-mediated gene transfer were unsuccessful. In our models of airway epithelial cells, neither cyclical RGD (cRGD) peptide (0.4 mg/ml) nor RGD peptide (4 mg/ml) significantly altered the level of transgene expression in human or rat PD cultures, respectively (Fig. 8A, panel ii). AdV attachment to human or rat cultures was also not altered by cRGD or RGD peptides, respectively (results not shown). These data suggest that mediation of endosomogenesis and endosomolysis leading to internalization of AdV into PD cultures may occur by mechanisms other than α_vβ_3/5 integrin interactions, suggesting that the absence of α_vβ_3/5 integrins from the apical membrane of WD cells may not entirely account for the resistance of these cells to gene transfer.

(ii) Dynamin-mutant cells deficient in receptor-mediated endocytosis.

The data from rat WD cultures emphasize the importance of entry across the apical membrane as a limiting variable for gene transfer. Whereas there are multiple modes of entry across the cellular membrane (see below), morphological studies have identified coated-pit vesicles as an important path for AdV entry (11, 30). We initiated studies to functionally test the importance of this path for AdV-mediated gene transfer with HeLa cells and HeLa cell mutants that have reduced coated-pit-mediated endocytosis. Dynamin is responsible for “pinching off” endocytotic invaginations formed during receptor-mediated endocytosis (7). While overexpression of wild-type (Wt) dynamin in HeLa cells does not affect endocytotic processes (7), overexpression of the K44a dynamin mutant selectively and reproducibly reduces receptor-mediated endocytosis. Overexpression of either Wt or K44a dynamin did not affect AdV attachment to the cells (Fig. 8B, panel i), but β-gal expression was significantly reduced in K44a dynamin-expressing cells compared to Wt dynamin-expressing cells (panel ii). These findings show that the predominant route of entry of AdV into HeLa cells occurs via receptor-mediated endocytosis and that cells expressing identical amounts of AdV attachment/internalization receptors can display a reduced gene transfer efficiency reflecting a reduced plasma membrane uptake pathway process.

Relationship between attachment and expression in PD and WD cultures.

Increased efficiency of AdV-mediated gene transfer to cells in vitro has been achieved by the manipulation of the AdV capsid coat to increase cellular attachment (10, 33). To determine if increased attachment of AdV to human PD and WD cultures leads to enhanced gene transfer efficiency, increasing concentrations of AdV were applied to both culture types. As shown in Fig. 9A, panel i, exposure of PD cultures to a 10-fold increase in AdV concentration resulted in significant enhancement of both the amount of AdV attached to the cultures and transgene expression. These data suggest that in this culture type, similar in morphology to many cell types grown in vitro, the absolute amount of AdV attachment directly correlates with the subsequent level of gene transfer. For WD cultures (panel ii) exposed to AdV concentrations similar to or up to 1,000-fold higher than those to which PD cultures were exposed, the level of gene transfer was not enhanced even though the absolute amount of AdV attachment was increased to levels associated with significant gene transfer in the PD cultures. These data clearly show that increased AdV attachment to WD cultures failed to overcome the inefficient gene transfer.

The direct relationship between the quantity of attached vector and expression in PD cells suggests that once attachment has occurred, entry can be achieved by constitutive nonspecific pathways, e.g., pinocytosis and phagocytosis. In contrast, although vector can nonspecifically attach to WD cells, the absence of expression suggests that few or no constitutive or nonspecific entry pathways exist in the apical surface of this cell type. To test the capacity for constitutive (unstimulated) uptake of luminal solutions by airway epithelia in different states of differentiation, we measured the uptake of a fluid-phase marker and correlated this parameter with gene transfer efficiency. Human cells grown on tissue culture plastic (i.e., PD cells) internalized measurable amounts of [³H]inulin, whereas WD cells did not (Fig. 9B, panel i). These differences were paralleled by the differences in gene transfer (panel ii). These findings suggest that a relative reduction in the constitutive activity of the aggregate apical membrane entry pathways in WD cells contributes to the resistance of these cells to gene transfer.