INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

THE PLEIOTROPIC EFFECTS of cystic fibrosis (CF) appear to result from the mislocalization or impaired activity of an apical membrane chloride channel of epithelial cells, the cystic fibrosis transmembrane conductance regulator (CFTR). CFTR may also modulate intracellular chloride conductances and thus affect the pH of organelles along the secretory and endocytic pathways (4-6). An elevated Golgi pH in CF cells could explain the observations that the glycosylation profiles of proteins, glycolipids, and mucins differ between CF and normal cells. In particular, it has been suggested that CF glycoconjugates are more heavily sulfated and fucosylated than normal and contain less sialic acid (6, 9, 17, 36). Because the pH optimum for sialyltransferase is considerably lower than that of fucosyltransferase or sulfotransferase [pH 5.8 vs. pH 6.8 for sulfotransferase and pH 7.0-8.5 for fucosyltransferase (7, 27, 35)], these observed differences are consistent with the predicted effects of increased intra-Golgi pH in CF cells. Furthermore, because many strains of bacteria have been shown to bind avidly to asialoconjugates, such as the glycolipid asialoGM1 (23), differences in sialylation between CF and normal cells may explain the observation that increased levels of bacteria bind to CF tissue and cells compared with normal cells (22, 31, 32). However, it should be noted that, in other studies, no difference in bacterial binding to cultures of CF and non-CF nasal polyps was observed (8, 12).

Several issues complicate the studies of glycosylation and bacterial binding in primary tissues. In general, it has been difficult to obtain large enough quantities of normal mucin for detailed structural and compositional studies (24, 29). Furthermore, not all of the mucus-secreting cells in the airway express CFTR (16). Another problem is that CF patients frequently have severe airway infection and inflammation. Inflammatory mediators and bacteria have been shown to enzymatically alter the composition of cell surface glycoconjugates and mucous secretions, as well as the amount of mucin secreted (10, 21, 25, 29, 31, 32). Although these observations support the idea that the surface glycosylation profiles of cells from CF patients differs from normal, the basis for these differences is not clear.

To avoid the problems of inflammation and bacterial contamination, we have used a quantitative lectin binding assay to compare cell surface glycosylation profiles of CF and rescued immortalized cell lines. For our studies, we chose the CF bronchial epithelial cell line IB3-1 and its clonally derived counterpart S9, which expresses functional CFTR. Although we measured significant and reproducible differences in terminal glycosylation profiles of these two closely matched cell lines, our data suggest that these differences are independent of the CF genotype of the cells.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Cell lines. IB3-1 cells, derived from CF bronchial epithelia and immortalized using adeno-12-SV-40 (38), and S9 cells, generated from IB3-1 by transfection with a recombinant adeno-associated viral vector encoding full-length, wild-type CFTR (15), were kindly provided by Dr. Pamela Zeitlin. Cells were maintained in F-12 medium supplemented with 10% fetal bovine serum and penicillin/streptomycin.

Detection of functional CFTR by 6-methoxy-N-(3-sulfopropyl)quinolinium. Adenosine 3',5'-cyclic monophosphate (cAMP)-dependent anion efflux was monitored by 6-methoxy-N-(3-sulfopropyl)quinolinium (SPQ; Molecular Probes, Eugene, OR) fluorescence changes in living cells (37). Briefly, subconfluent cells grown on 25-mm glass coverslips were loaded with SPQ (10 mM) by 12 min of exposure at 37°C to hypotonic NaI buffer (1:1 with water). Buffer composition (in mM) was 130 NaI, 4 KNO₃, 1 Mg(NO₃)₂, 1 Ca(NO₃)₂, 10 glucose, and 20 N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid (HEPES), pH 7.4. Cells were mounted in a perfusion chamber placed in a heating stage set to 37°C and were perfused with buffers throughout the experiment. Imaging was performed on a Nikon Diaphot 300 inverted microscope equipped with a ×40 oil immersion objective (Nikon CF fluor), image intensifier, and video camera. Excitation was at 330 nm, and image acquisition and analysis were performed by Metafluor software (Universal Imaging, West Chester, PA). The average fluorescence intensity of individual cells in a field was monitored every 15 s (8 frame average) throughout the assay, which ran for 14 min. Cells were perfused with isotonic NaI buffer for 2 min, nitrate buffer (NaNO₃ replaced NaI) for 4 min to assess the rate of iodide leakage/exchange from unstimulated cells, 4 min in nitrate buffer supplemented with 10 µM forskolin and 200 µM 3-isobutyl-1-methylxanthine (IBMX), and then 4 min in iodide buffer to requench intracellular SPQ. Functional CFTR was detected as an increase in the rate of dequenching of SPQ upon addition of forskolin-IBMX. Assays on adenoviral-infected cells were performed using a blinded technique.

Enzyme-linked lectin assay. IB3-1 and S9 cells were seeded in 96-well plates (6 wells/experimental group) and were incubated at 37°C for 4 h in media. The cells were washed two times with phosphate-buffered saline (PBS) containing 1 mM CaCl₂ and 1 mM MgCl₂ at room temperature. Cells in some wells were lysed with 1% Triton X-100 in PBS for determination of total protein using the bicinchoninic acid assay (Pierce, Rockford, IL). The remaining cells were fixed with 3% paraformaldehyde in PBS for 15 min and then were washed with 0.2% gelatin (Sigma Chemical, St. Louis, MO) in PBS containing 10 mM glycine (PBS-G). Endogenous peroxidase activity was blocked before incubation with lectin-horseradish peroxidase (HRP) conjugates by treatment with 0.3% hydrogen peroxide in methanol for 15 min. The cells were then incubated with 2 µg/ml of HRP-conjugated peanut agglutinin (HRP-PNA) or HRP-conjugated wheat germ agglutinin (HRP-WGA; HRP-lectin conjugates were from EY Laboratories, San Mateo, CA) in PBS-G for 30 min and then washed with PBS-G. The bound lectin enzyme complex was visualized after light-protected incubation with 0.4 mg/ml of substrate o-phenylenediamine (Sigma) and hydrogen peroxide. The reaction was stopped by addition of 1 M H₂SO₄, and plates were read at 490 nm optical density using a microtiter plate spectrophotometer (Molecular Devices, Sunnyvale, CA). In some experiments, cells were pretreated with 0.05 U/ml of neuraminidase (Calbiochem, La Jolla, CA) at 37°C for 30 min before fixation.

Adenoviral infection. IB3-1 cells were mock infected or infected with replication-defective adenoviruses encoding either nuclear-localized -galactosidase (Ad--Gal) or wild-type CFTR (Ad-CFTR) driven by the cytomegalovirus (CMV) promoter [both viruses provided by Genzyme, Cambridge, MA (2, 28)] at a multiplicity of infection (MOI) of 25-50. Two days postinfection, cells were trypsinized and seeded into 96-well plates. PNA binding was measured using 2 µg/ml HRP-PNA as described above. The percentage of cells infected was monitored by histochemical staining of -Gal in cells infected with Ad--Gal as described previously (26).

Immunoprecipitation and phosphorylation. IB3-1 cells (uninfected or infected with Ad-CFTR) and S9 cells were solubilized in lysis buffer (1% Nonidet P-40 and 1 mM EDTA in 20 mM HEPES, pH 7) containing protease inhibitor cocktail (10 mM leupeptin, 1 mM pepstatin A, 2 mg/ml soybean trypsin inhibitor, 2 mg/ml aprotinin, 40 mg/ml phenylmethylsulfonyl fluoride, and 0.2 mM dithiothreitol) for 1 min at 4°C. The cell lysates were precleared by incubation with normal rabbit serum at 1:20 followed by precipitation using 25 µl of protein A-Sepharose 6MB (Pharmacia Biotechnologies, Piscataway, NJ). After centrifugation, supernatants were adjusted to 1× RIPA buffer [1% Triton X-100, 1% sodium deoxycholate, and 0.1% sodium dodecyl sulfate (SDS) in 50 mM tris(hydroxymethyl)aminomethane (Tris) · HCl, pH 7.5]. The samples were incubated with monoclonal antibodies against the COOH terminus and R domain of CFTR (Genzyme) for 90 min, and antibody-antigen complexes were isolated using protein A-Sepharose. After being washed sequentially with RIPA buffer followed by 50 mM Tris · HCl, pH 7.5, the Sepharose was resuspended in protein kinase A (PKA) buffer (10 mM MgCl₂ in Tris buffer) and was incubated with 5 units of cAMP-dependent PKA catalytic subunit (Promega, Madison, WI) and 10 µCi [-³²P]ATP (NEN, Boston, MA) at 30°C for 60 min. The immune complex was released from Sepharose by incubating in electrophoresis sample buffer [125 mM Tris · HCl, pH 6.8, 5% (wt/vol) SDS, and 25% (wt/vol) sucrose] containing 5% (vol/vol) 2-mercaptoethanol at 37°C for 5 min. The samples were run on a 6% SDS-polyacrylamide gel electrophoresis (PAGE), and the dried gel was exposed on X-ray film (X-OMAT AR; Eastman Kodak, Rochester, NY).

Statistical analysis. Results were analyzed using the SigmaStat statistics program (Jandel Scientific, San Rafael, CA). Data derived from three or more groups were compared by one-way analysis of variance. Groups that differed significantly from the control were identified using Dunn's test. Individual means from three or more experiments were compared using the paired t-test. P < 0.05 was considered significant.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

To determine the effect of CFTR on cell surface glycosylation, it was necessary to confirm that the IB3-1 and S9 cells used in these studies expressed the appropriate chloride secretory phenotype. Functional CFTR activity was monitored using the halide-sensitive fluorophore SPQ as described in MATERIALS AND METHODS (Fig. 1). S9 cells showed an increase in the rate of SPQ dequenching upon stimulation with forskolin and IBMX, indicating that these cells express functional CFTR (Fig. 1A). In contrast, cAMP stimulation cocktail had no effect on fluorescence in IB3-1 cells (Fig. 1B).

View larger version (18K): [in this window] [in a new window]   Fig. 1.   cAMP-dependent anion permeability of IB3-1 and S9 measured using the halide-sensitive fluorophore 6-methoxy-N-(3-sulfopropyl)quinolinium (SPQ). Iodide efflux was measured in S9 (A) and IB3-1 (B) cells grown on glass coverslips as described in MATERIALS AND METHODS. At time (t) = 2 min, cells were switched to nitrate buffer; at t = 6 min, cAMP-stimulating cocktail was added; and at t = 10 min, cells were returned to iodide buffer. Arrows indicate the time at which buffers reached the cells. S9 cells demonstrate an increased rate of SPQ dequenching upon cAMP stimulation, suggesting that they express functional cystic fibrosis transmembrane conductance regulator (CFTR), whereas the rate of iodide leakage in stimulated IB3-1 cells is unchanged. Data are normalized to 100% at the point of maximum dequenching and to 0 at the 2-min point when buffers are switched to nitrate buffer. Tracings represent means ± SE from 12 cells (A) and 16 cells (B). Four S9 cells did not respond to forskolin (not shown), suggesting that a population of these cells may not express wild-type CFTR.

Enzyme-linked lectin assays have been widely used for detection and quantitation of terminal cell surface saccharides (3, 18). Within the linear range, this sensitive assay allows direct comparison of lectin binding to two cell lines, although the total number of lectin binding sites cannot be measured. To determine whether CFTR expression might influence the level of terminal sialylation in cells as previously suggested, we measured the binding of HRP-PNA to IB3-1 and S9 cells. This lectin binds preferentially to galactosyl(-1,3) N-acetylgalactosamine [Gal(-1,3)GalNAc] linkages found on O-linked oligosaccharides such as mucin-type oligosaccharides, as well as to the glycolipid asialoGM1. IB3-1 and S9 cells were seeded in 96-well plates, and HRP lectin binding was measured as described in MATERIALS AND METHODS. Lectin binding was linear both with respect to cell number per well and to the concentration of lectin added (Fig. 2). Interestingly, HRP-PNA binding to IB3-1 cells was consistently higher than to S9 cells, suggesting that IB3-1 cells had increased levels of terminal galactose residues, as would be expected if sialyltransferase activity was reduced. The results were identical whether normalized to cell number per well or total protein per well, suggesting that the two cell lines did not differ markedly in size (not shown). Inclusion of 50 mM galactose during the HRP-PNA incubation step completely abrogated binding to both cell lines, suggesting that the binding we observed was due to PNA interaction with cell surface glycoconjugates (not shown). Furthermore, this result was highly reproducible and was observed regardless of the passage number of the cells used (between passages 20 and 35).

View larger version (15K): [in this window] [in a new window]   Fig. 2.   Enzyme-linked lectin quantitation of cell surface galactose residues on subconfluent IB3-1 and S9 cells using peanut agglutinin (PNA). Cells (1-3 × 104/well; 4 wells each/time point) were plated in 96-well dishes. After 4 h, cells were rinsed, fixed, blocked, and incubated with peroxidase-conjugated PNA (1 and 2 mg/ml). After they were washed, cells were incubated with o-phenylenediamine and peroxide, and the amount of lectin-enzyme complex bound to the cells was quantitated using a microtiter plate reader. O.D., optical density. Mean ± SD for each condition is plotted. We consistently find that IB3-1 cells bind more PNA than S9 cells (P < 0.05).

To confirm that the difference in HRP-PNA binding to IB3-1 and S9 cells was not an artifact due to differences in cell size or shape, we compared binding of HRP-WGA to these cells. This lectin binds to N-acetylglucosamine residues (GlcNAc) and has high affinity for the di-GlcNAc-containing core of N-linked oligosaccharides. Because the addition of the N-linked oligosaccharide core occurs cotranslationally in the endoplasmic reticulum, the level of WGA binding to cells should be independent of CFTR expression. As shown in Fig. 3A, binding of HRP-PNA to S9 cells was on average 46 ± 13% of binding compared with IB3-1 cells (considered as 100%). By contrast, HRP-WGA bound equally to IB3-1 and S9 cells (Fig. 3B).

View larger version (36K): [in this window] [in a new window]   Fig. 3.   IB3-1 and S9 cells express similar levels of horseradish peroxidase (HRP) conjugated-wheat germ agglutinin (WGA) cell surface binding sites. IB3-1 and S9 cells (2 × 104/well) were incubated with 2 mg/ml HRP-PNA (A) or HRP-WGA (B) and developed in the substrate buffer as described in MATERIALS AND METHODS. Absorbance readings were normalized against total protein recovered per well, and the IB3-1 values were normalized to 100%. Data from 4 (A) and 3 (B) separate experiments are expressed as means ± SE. Analysis using the paired t-test shows a significant difference in the binding of HRP-PNA binding to IB3-1 and S9 cells but no difference in HRP-WGA binding to the two cell lines.

We next wanted to determine whether the difference in PNA binding to IB3-1 and S9 cells reflected a difference in the level of exposed terminal galactose residues at the cell surface. To do this, we measured PNA binding to cells before and after pretreatment with neuraminidase. PNA binding to untreated S9 cells was 58% compared with IB3-1 (Fig. 4A) as observed previously. Upon neuraminidase treatment, binding to both cell lines increased dramatically (Fig. 4B), and no significant difference between the two cell lines was detected. These data indicate that the percentage of exposed galactose residues in IB3-1 cells is significantly greater than in S9 and support the idea that sialylation in CF cell lines is impaired.

View larger version (34K): [in this window] [in a new window]   Fig. 4.   PNA lectin binding to neuraminidase-treated IB3-1 and S9 cells is similar. Cells were plated, mock treated (A) or treated with 0.05 U/ml neuraminidase at 37°C for 30 min (B), and then processed as described in MATERIALS AND METHODS. Neuraminidase-treated samples were developed in substrate buffer for 3 min. Untreated samples were developed for 20 min. Difference in PNA binding to IB3-1 compared with S9 cells is statistically significant [means ± SE from 3 experiments (P = 0.01)], whereas PNA binding to the neuraminidase-treated cell lines is not (P = 0.07). Results shown are from a representative experiment that was performed 3 times with similar results.

To determine whether the difference in HRP-PNA binding to IB3-1 and S9 cells was CFTR dependent, we asked whether heterologous expression of wild-type CFTR in IB3-1 cells would reduce the number of exposed galactose residues at the cell surface. For these studies, we infected cells with a recombinant replication-defective adenovirus encoding CFTR. We tested the efficiency of infection of these cells using a similar adenovirus encoding -Gal. Cells were infected with Ad--Gal at an MOI of 25, and -Gal activity was detected colorimetrically after 2 days. Whereas uninfected cells showed no staining (not shown), >90% of the virally infected cells stained positively for -Gal activity (Fig. 5). Therefore, under these conditions, we were able to infect the majority of cells.

View larger version (135K): [in this window] [in a new window]   View larger version (187K): [in this window] [in a new window]   Fig. 5.   Recombinant adenoviruses efficiently infect IB3-1 cells. IB3-1 cells were mock infected or infected with nuclear-localized -galactosidase (Ad--Gal), a recombinant adenovirus encoding -Gal, at a multiplicity of infection (MOI) of 25. At 2 days postinfection, cells were stained with X-Gal to determine the infection efficiency. Greater than 90% of the cells express -Gal (inverted phase photomicrograph, magnification ×100).

To confirm that infection with Ad-CFTR resulted in the production of full-length functional protein, we immunoprecipitated protein from uninfected and virally infected cells (Fig. 6A). Cells were infected with Ad-CFTR at an MOI of 25 and were incubated for 2 days. Because CFTR expression is driven by the butyrate-inducible CMV promoter in this virus, we also tested the effect of overnight induction with 1 mM butyrate on CFTR production. Cells were lysed and CFTR was immunoprecipitated using commercially available monoclonal antibodies directed against the COOH terminus and regulatory domain of the protein. CFTR was detected on SDS-PAGE gels after in vitro phosphorylation with PKA. As shown in Fig. 6A, lane 1, no CFTR could be detected in uninfected IB3-1 cells. Upon infection with Ad-CFTR, radiolabeled bands consistent with the known electrophoretic mobilities of the core glycosylated (Fig. 6A, band B, ~130 kDa) and mature (Fig. 6A, band C, ~180 kDa) forms of CFTR were observed (Fig. 6A, lane 2). Induction of CFTR expression in Ad-CFTR-infected IB3-1 cells with sodium butyrate dramatically increased the amount of material in both bands (Fig. 6A, lane 3; for clarity, the exposure shown is four times shorter than the other lanes). S9 cells expressed a low but detectable level of both forms (Fig. 6A, lane 4).

To confirm that the CFTR expressed in these cells was functional, we measured cAMP-stimulated halide efflux in virally infected cells using the SPQ assay (Fig. 6B). In contrast to uninfected IB3-1 cells or cells infected with Ad--Gal, cells infected with Ad-CFTR and induced with sodium butyrate demonstrated a dramatic cAMP-dependent increase in the rate of halide efflux, indicative of functional CFTR expression. Together, these data demonstrate that we were able to express functional CFTR in a large proportion of the cells.

View larger version (29K): [in this window] [in a new window]   Fig. 6.   IB3-1 cells express functional CFTR after infection with recombinant adenovirus. A: subconfluent dishes of IB3-1 cells were infected with wild-type CFTR (Ad-CFTR) at an MOI of 25. Before further analysis at 2 days postinfection, 1 dish was induced overnight with 1 mM sodium butyrate. CFTR proteins were immunoprecipitated using monoclonal antibodies against the COOH terminus and R domain of the protein. Antibody-antigen complexes were collected using protein A-Sepharose and were phosphorylated with protein kinase A catalytic subunit and [-32P]ATP. Proteins were separated by SDS-polyacrylamide gel electrophoresis, and radiolabeled bands were detected by exposure to X-ray film. Lane 1: uninfected IB3-1 cells; lane 2: IB3-1 cells infected with Ad-CFTR; lane 3: butyrate-induced Ad-CFTR-infected IB3-1 cells; lane 4: S9 cells. Positions of expected migration of the core glycosylated form (band B) and mature form of CFTR (band C) are noted. Lanes 1, 2, and 4: 16-h exposure; lane 3: 4- h exposure. B: IB3-1 cells grown on glass coverslips were mock infected (IB3) or infected with Ad-CFTR (IB3-Ad-CFTR) or Ad--Gal (IB3-Ad--Gal) at an MOI of 50 and induced with sodium butyrate as described above. Functional CFTR activity of uninfected and infected cells was monitored using the SPQ assay as described in MATERIALS AND METHODS. Arrows indicate the time at which buffers reached the cells. Tracings represent means ± SE from 19, 37, and 14 cells (IB3, IB3-Ad--Gal, and IB3-Ad-CFTR, respectively) and are normalized to 100% at the point of maximum dequenching of IB3-Ad-CFTR and to 0% at the 2-min point when buffers were switched to nitrate buffer. All of the IB3-Ad-CFTR cells responded to forskolin, suggesting that they expressed functional CFTR.

HRP-PNA binding to IB3-1, S9, and virally infected IB3-1 cells was determined using the enzyme-linked lectin assay. The results (Fig. 7) show that CFTR expression has no effect on lectin binding to IB3-1 cells. HRP-PNA binding to IB3-1 cells was unaffected by infection with Ad-CFTR regardless of whether CFTR expression was induced with sodium butyrate. Neither adenoviral infection with Ad--Gal nor sodium butyrate treatment of uninfected IB3-1 cells altered PNA binding significantly. In contrast, HRP-PNA binding to S9 cells was clearly different compared with IB3-1 cells (P < 0.0001). Thus it appears that the difference between HRP-PNA binding to IB3-1 and S9 cells is CFTR independent.

View larger version (68K): [in this window] [in a new window]   Fig. 7.   Infection with Ad-CFTR does not alter PNA binding to IB3-1 cells. IB3-1 cells were mock infected or infected with Ad-CFTR or Ad--Gal at an MOI of 25-50. At 2 days postinfection, cells were reseeded into 96-well plates, and HRP-PNA binding was measured after 4 h. Some dishes were treated with 1 mM sodium butyrate for 24 h before reseeding. Results were normalized against HRP-PNA binding to uninfected IB3-1 cells (considered as 100%) in each experiment and are presented as means ± SE. 1: Uninfected IB3-1 cells treated with sodium butyrate (n = 3); 2: IB3-1 cells infected with Ad-CFTR (n = 4); 3: IB3-1 cells infected with Ad-CFTR and induced with sodium butyrate (n = 4); 4: IB3 cells infected with Ad--Gal (n = 4); and 5: S9 cells (n = 9). Whereas HRP-PNA binding to S9 cells is statistically different from IB3 (P < 0.0001), there is no statistical difference between induced and/or infected IB3-1 cells compared with control (P > 0.1).

To address the possibility that the 2-day infection period was not long enough to remodel cell surface glycoconjugates sufficiently for measurement, we synthetically altered intra-Golgi pH by incubating cells for 2 days in 10 mM ammonium chloride and measured the effect on HRP-PNA binding (Fig. 8). This treatment, which elevates the pH of all acidic intracellular compartments (11), was not toxic to cells over this time period. PNA binding to both cell lines increased measurably in ammonium chloride-treated cells relative to control untreated cells, suggesting that this treatment interfered with sialylation as expected. Interestingly, the difference in PNA binding was preserved between ammonium chloride-treated IB3-1 and S9 cells, perhaps suggesting that the difference in terminal glycosylation between these two cell lines is pH independent.

View larger version (56K): [in this window] [in a new window]   Fig. 8.   Ammonium chloride treatment increases HRP-PNA binding to IB3-1 and S9 cells. IB3-1 and S9 cells were incubated with (+) or without () 10 mM ammonium chloride for 2 days before quantitation of HRP-PNA binding. Even after pretreatment with ammonium chloride, PNA binding to IB3-1 is significantly different from binding to S9 cells (means ± SD; P < 0.05). This experiment was performed 3 times with similar results.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

We have compared the glycosylation profile of a matched CF and rescued epithelial cell line using a quantitative enzyme-linked lectin assay. This assay was linear with respect to cell number (or protein concentration) and concentration of HRP-conjugated lectin added. Using this assay, we demonstrated a significant and highly reproducible difference in the amount of terminal galactose present on CF versus normal cells. This lectin recognizes Gal(-1,3)GalNAc linkages preferentially and thus is most specific for O-linked oligosaccharides and glycolipids, particularly asialoGM1. When cells were pretreated with neuraminidase, PNA binding to both cell lines increased dramatically, and no significant difference was observed in the level of total galactose residues on IB3-1 compared with S9 cells. Furthermore, binding of WGA lectin was identical between IB3-1 and S9 cells. Together, our results show that IB3-1 cells have an increased level of terminal galactose residues compared with S9 cells but otherwise are similar in surface glycoconjugate composition. These data validate the use of the enzyme-linked lectin assay as a quantitative method to compare cell surface glycosylation profiles between related cell lines.

Because our results supported the hypothesis that CF cells may have increased intraorganellar pH, we investigated whether restoring CFTR function to the IB3-1 cells would also rescue the glycosylation phenotype. Infection with recombinant, replication-defective adenovirus encoding CFTR for 2 days resulted in >90% transduction efficiency. We were able to qualitatively measure CFTR in the infected cells using immunoprecipitation followed by in vitro phosphorylation. Furthermore, the CFTR expressed in these cells was functional as assessed by the halide-sensitive fluorophore SPQ. However, expression of CFTR did not alter the glycosylation profile of IB3-1 cells, suggesting that the decreased sialylation that we observed in this cell line was CFTR independent.

Given the large difference in PNA binding that we observed between IB3-1 and S9 cells and the estimated 90% adenovirus infection efficiency, we should have easily been able to detect a change in PNA binding if CFTR expression altered oligosaccharide processing in IB3-1 cells. One argument against our interpretation is that glycosylation turnover is too slow to detect differences by 2 days postinfection. Several studies have examined the half-life of membrane proteins in other cultured cell lines (1, 14, 19). Although the results of these studies suggest that the rate of protein turnover is somewhat dependent on cell type and cell growth conditions, protein turnover in rapidly dividing cells was generally found to follow diphasic kinetics, with between 25 and 35% of protein turning over rapidly (half-life of between 1 and 10 h) and the remainder being degraded more slowly (half-time between 1 and 5 days). These measurements do not take into account the synthesis of new membrane proteins required for continuing cell division and therefore represent an underestimate of the rate of protein remodeling in dividing cells. Under our growth conditions, IB3-1 cells divide approximately every 24 h, even after adenoviral infection. Therefore, we estimate that at least 75% of the total protein is newly synthesized within 48 h of infection. Because mRNA from recombinant adenoviruses can be detected within 6 h of infection (20), the majority of glycoconjugates synthesized in IB3-1 cells during the 2 days of postinfection incubation will have been synthesized in cells expressing wild-type CFTR. Furthermore, treatment with ammonium chloride for 48 h increased PNA binding to cells by ~30%, thus confirming that significant remodeling of the cell surface carbohydrate profile occurred during this period. We attempted to extend the length of time in culture after adenoviral infection; however, because the cells continue to proliferate and the adenovirus-mediated gene transfer is episomal (20), the percentage of transduced cells declined dramatically with time in culture. After 4 days, only 55% of cells infected with Ad--Gal showed detectable -Gal staining. At this time point, there was no difference in PNA binding between Ad-CFTR-infected and uninfected IB3-1 cells (not shown). In addition, we tried to express endogenous functional CFTR at the cell surface of IB3-1 cells using other nonviral approaches, including incubation for 2 days with 0.5 mM 4-phenylbutyrate (30), incubation at reduced temperature (27°C; see Ref. 13), and treatment with 0.1 mg/ml geneticin, which can promote read through from the W1282X allele present in IB3-1 cells (M. Howard and M. Gondor, personal communication). However, in our hands, none of these treatments produced detectable functional CFTR at the cell surface, as monitored by SPQ.

The increased level of terminal galactosylation in IB3-1 cells could indicate an increased number of bacterial binding sites on these cells compared with S9 cells. Interestingly, Imundo et al. (22) recently reported that IB3-1 cells bind two times as many Pseudomonas aeruginosa bacteria as another subclone (C38) that expresses functional CFTR. This observation may suggest that this subclone also expresses lower levels of sialylated glycoconjugates than IB3-1 cells.

There are several possible reasons for the differences in terminal galactosylation that we observed between IB3-1 and S9 cells. The most likely explanation is that the glycosylation pattern of the clonally derived S9 cells reflects the profile of a subpopulation of the original IB3-1 population from which it was generated. IB3-1 cells are polyploid and contain between 80 and 90 chromosomes. Thus any subclone of this heterogeneous population may express different levels or isoforms of any of the myriad cellular components that affect glycosylation, such as glycosyltransferases or sugar transporters. Because ammonium chloride did not equalize PNA binding between IB3-1 and S9 cells, we suspect that the difference in glycosylation between the two cell lines is pH independent. The difference in glycosylation is not likely to be due to differences in cell passage number, since we observed no difference in glycosylation of either line with increased passage and since the levels of WGA binding to the cells remained identical. Furthermore, in a detailed study examining oligosaccharide composition and structure, Swiedler et al. (33, 34) found virtually no change in the glycosylation profile of cells even after up to 1 yr of continuous passage.

Our data suggest that differences in glycosylation observed between CF and control cell lines must be interpreted with caution, as such differences are not necessarily CFTR dependent. To address the role of CFTR in glycosylation, it will be necessary to compare CF and corrected primary tissue from a single individual. The enzyme-linked lectin assay described here appears to be sensitive and reproducible enough to be used for such studies.