GalNAc-α-O-benzyl Inhibits NeuAcα2-3 Glycosylation and Blocks the Intracellular Transport of Apical Glycoproteins and Mucus in Differentiated HT-29 Cells

HT-29–differentiated subpopulations were derived from the original cell line (Fogh and Trempe, 1975) that was obtained from the late J. Fogh (Sloan-Kettering Memorial Cancer Center, Rye, NY). The cells adapted to MTX 10⁻⁶, 10⁻⁵, and 10⁻³ M (Lesuffleur et al., 1990, 1991) were used after several weekly passages of reversion to drug-free medium (10–40 passages) and are referred to as HT29-RevMTX10⁻⁶, RevMTX10⁻⁵ (mucus-secreting cells), and RevMTX10⁻³ (enterocytic cells). Cells were grown in DME (Life Technologies, Inc., Cergy-Pontoise, France) supplemented with 10% inactivated FBS for 30 min at 56°C (Boehringer Mannheim Biochemicals, Mannheim, Germany). GalNAc-α-O-benzyl (Sigma Chemical Co., St. Louis, MO) was solubilized in DME. All experiments and maintenance of the cells were done in 25- or 75-cm² T flasks (Corning Glass Works, Corning, NY), and in 24-well cell culture clusters (Costar, Cambridge, MA) at 37°C in a 10% CO₂ / 90% air atmosphere. Cells were seeded at 2 × 10⁴ cells/cm². The same conditions were applied to cells grown on tissue culture-treated Transwell polycarbonate membranes with a 24.5-mm diam and a 0.4-μm pore size (Costar). For maintenance purposes, cells were passaged weekly, using 0.025% trypsin in 0.53 mM EDTA in PBS Ca²⁺Mg²⁺-free (PBS⁻). The medium was changed daily in all culture conditions. For growth curves, cells grown in 24-well culture clusters were detached with trypsin and counted with a hemocytometer. Cell volume was determined in a hematocrit. Control Caco-2 cells were cultured as previously reported (Pinto et al., 1983) and analyzed between passages 70 and 80.

Mouse mAbs HBB 3/775/42 (Hauri et al., 1985) and G1/136 (Eilers et al., 1989) specific for human DPP-IV and a 120-kD basolateral glycoprotein respectively, were a gift of H.P. Hauri (Biocenter of the University of Basel, Basel, Switzerland). Rat mAb 4H3 against human DPP-IV (Gorvel et al., 1991) was obtained from S. Maroux (CNRS Unité de Recherche Associee 1820, Faculté des Sciences de Saint Jerôme, Marseille, France). Mouse mAb 517 (Le Bivic et al., 1988) against CEA was obtained from A. Le Bivic (Faculté des Sciences de Luminy, Marseille, France). Mouse mAb BC-2 (Xing et al., 1989), which recognizes a sequence in the tandem repeat of the MUC1 gene product was obtained from I. McKenzie (Austin Cancer Research Institute, Heidelberg, Victoria, Australia). Mouse mAb B72.3 against sialyl-Tn (Nuti et al., 1982) was obtained from K.O. Lloyd (Memorial Sloan-Kettering Cancer Center, New York). Mouse mAb TS2/16 against the integrin β1 subunit (Arroyo et al., 1992) was obtained from F. Sanchez-Madrid (Universidad Autónoma de Madrid, Madrid, Spain). Rabbit polyclonal Abs against porcine villin (Robine et al., 1985) and the tight junction protein ZO-1 (Willot et al., 1992) were obtained from D. Louvard (Institut Curie, Paris, France) and J.M. Anderson (Yale University, New Haven, CT), respectively. For the detection of gastric mucins we used the same rabbit polyclonal Ab L56/C as previously used for cloning the L31 mucin cDNA encoding the 3′ end of MUC5AC (Lesuffleur et al., 1995). Fluorescein-conjugated Maackia amurensis agglutinin (MAA) (Wang and Cummings, 1988), Sambucus nigra agglutinin (SNA) (Shibuya et al., 1987), and Peanut (Arachis hypogaea) agglutinin (PNA) (Lotan et al., 1975), which recognize the oligosaccharide species NeuAcα2-3Gal-R, NeuAcα2-6Gal, and Galβ1-3GalNAc-R, respectively, were from Vector Labs Inc. (Burlingame, CA).

Indirect immunofluorescence was performed on cryostat sections of cell layer rolls as reported (Lesuffleur et al., 1990). Briefly, late (day 21) cultures of cells grown in 25-cm² T flasks were rinsed with Ca²⁺Mg²⁺-free PBS, the T flask was cut up with a soldering iron, and then the cell layer gently was scraped with a rubber policeman and poured in a bath of liquid nitrogen. The resulting frozen cell pellet was either stored in liquid nitrogen for further analysis, or immediately processed for cryostat sections. This method has the double advantage of visualizing a large part of the cell layer and allowing the concomitant detection of apical, basolateral, and intracellular proteins on the same section. Double immunofluorescence was performed on sections postfixed with 3.7% paraformaldehyde in PBS⁻ for 10 min at room temperature using secondary antibodies fluorescein-coupled sheep anti–mouse or anti–rabbit Ig (Institut Pasteur Production, Marne-la-Coquette, France) or rhodamine-coupled sheep antiglobulins (Boehringer Mannheim Biochemicals). Desialylation was performed by incubation for 16 h at 37°C of paraformaldehyde-fixed cryostat sections with sialidase from Clostridium perfringens (Sigma Chemical Co.) (50 mU/ml in 50 mM citrate buffer, pH 6.0, 0.9% NaCl, 0.1% CaCl₂). For confocal microscopy analysis, cells grown on glass coverslips were fixed with 4% paraformaldehyde for 10 min, incubated with 50 mM NH₄Cl for 30 min, and then permeabilized with 0.1% saponin in 1% BSA/ PBS for 30 min. To detect β1 integrin, mAb TS2/16 (0.5 μg/ml in 0.1% saponin in 1% BSA/PBS) was added for 1 h, followed by FITC-conjugated goat anti–mouse; MAA (Vector Labs, Inc.) (20 μg/ml in 0.1% saponin in Tris-HCl, pH 7.5, 15 mM KCl, 5 mM MgCl₂) was added for 1 h, followed by streptavidin-rhodamine (Pierce Chemical Co., Rockford, IL). Confocal microscopy analysis was performed using a Leica instrument (model TCS 4D; St. Gallen., Switzerland). Histological staining with alcian blue, pH 2.5, and nuclear red was done on cryostat sections postfixed in absolute ethanol for 10 min at room temperature.

Classical transmission EM was performed as previously reported (Lesuffleur et al., 1990, 1991) on cells grown in 25-cm² plastic flasks. Samples embedded in Epon (Polysciences, Inc., Warington, PA) were reembedded to make sections perpendicular to the bottom of the flask. Thin sections were stained with toluidine blue. Ultrastructural immunochemistry was performed as previously described (Hennebicq-Reig et al., 1996). After rinsing three times in PBS, cells cultured in 25-cm² flasks were fixed in phosphate buffer containing 4% paraformaldehyde and 0.05% glutaraldehyde. The cell layer was scraped with a rubber policeman, the cell pellet was infiltrated with phosphate buffer containing 2.3 M sucrose and 20% polyvinylpyrolidone, and then frozen in liquid nitrogen. Ultrathin cryosections were successively incubated with PBS containing 10% FBS, mouse mAb HBB 3/775/42 (DPP-IV), rabbit anti–mouse Ig antibody, and 8-nm gold-conjugated protein A. All antibodies and gold-conjugated protein A were diluted in PBS containing 10% FBS. The grids were finally counterstained with methylcellulose uranyl acetate and observed using an electron microscope (model 902; Carl Zeiss, Inc., Thornwood, NY). The same procedure was applied for mAb 517 (CEA) and mAb BC-2 (MUC1).

For detection of DPP-IV and villin mRNAs, total RNA was isolated from the cells 16 h after medium change by lysis with guanidium isothiocyanate and centrifugation through a CsCl gradient (Chirgwin et al., 1979). Samples of total RNA, denatured in 1 M glyoxal (Thomas, 1980), were fractionated by electrophoresis through 1% agarose gels and then transferred to nylon (model Hybond N; Amersham Corp., Amersham, UK) in the presence of 20× SSC. Filters were incubated overnight at 42°C in prehybridization buffer containing 50% formamide, 5× SSC, 10× Denhardt's solution, 50 mM sodium phosphate, pH 6.5, and 250 μg/ml sonicated and denatured salmon sperm DNA. Filters were then hybridized with the ³²P-labeled probe for 20 h at 42°C in prehybridization buffer containing 10% dextran sulfate (Thomas, 1980). Blots were washed twice with 2× SSC, 0.1% SDS at room temperature, once with 0.1× SSC, 0.1% SDS at 50°C, and once, using the same solution, at 65°C for 15 min. Blots were then processed for autoradiography. To normalize for RNA, filters were dehybridized and stained with methylene blue. Methylene blue staining was preferred to hybridization with actin or glyceraldehyde 3-phosphate dehydrogenase since it was found that the levels of these transcripts differ in dividing and postconfluent cells. DPP-IV was detected with cDNA DPI-101 (Darmoul et al., 1990) and villin with cDNA V19 (Pringault et al., 1986), obtained from D. Louvard. The probe for human ST3Gal I (Kitagawa and Paulson, 1994) was a 537-bp PCR-amplified fragment from HepG2 cells cDNA corresponding to the coding region from nucleotide 361 to 898 (Recchi et al., 1998).

DPP-IV activity was measured in the cell homogenates as previously reported (Lesuffleur et al., 1990), according to the method of Nagatsu et al. (1976) using 1.5 mM glycyl-l-proline-4-nitroanilide as substrate. Results are expressed as mU/mg of protein. One unit is defined as the activity that hydrolyzes 1 mmol of substrate/min at 37°C. Proteins were measured with the BCA protein assay reagent (Pierce Chemical Co.). Galβl-3GalNAc α2,3-sialyltransferase activity was measured in cell homogenates prepared by lysing the cells at 0°C with 10 mM sodium cacodylate buffer, pH 6.5, containing 1% Triton X-100 (Sigma Chemical Co.), 20% glycerol, 0.5 mM dithiothreitol, and 5 mM MnCl₂ (1 ml per 2.5 × 10⁷ cells). After 10 min of incubation under continuous stirring, cell homogenates were centrifuged at 10,000 g for 15 min and the supernatants were used for enzymatic assay. Protein concentration was determined according to Peterson (1977) using BSA as standard. Cell homogenates (40 μg of protein) were brought to a final volume of 120 μl with 0.1 M sodium cacodylate buffer, pH 6.5, 1% Triton X-100, 0.1 M galactose (as inhibitor of β-galactosidase), 1 mM 2,3-dehydro-2-deoxy-Neu5Ac (as inhibitor of sialidases), 52.9 μM CMP-[¹⁴C]- Neu5Ac (0.58 GBq/mmol; 3.68 kBq/120 μl) (Amersham Corp.), containing 1 mM of Galβl-3GalNAcα-O-pNp (Sigma Chemical Co.) and incubated for 1 h at 37°C. The reactions were stopped by adding 1 vol of ethanol. Samples were centrifuged at 3,000 g for 5 min and then supernatants were directly developed by descending paper chromatography with ethyl acetate/pyridine/water (10:4:3 by vol) (Delannoy et al., 1993). Assays were performed in duplicate. The rates of reactions were linear with time, at least for 1 h. The incorporation of [¹⁴C]-Neu5Ac was determined by subtraction of the radioactivity measured in the absence of exogenous acceptors and results are expressed as average values in nmol of Neu5Ac transferred per milligram of protein and per hour.

Cells were homogenized by sonication in Tris/Mannitol buffer. Immunoprecipitation of DPP-IV, CEA, and MUC1 was performed as in Hauri et al. (1985), using mAbs 3/775/42, 517, and BC-2 previously coated on protein A–Sepharose beads (Pharmacia Fine Chemicals, Uppsala, Sweden). SDS-PAGE was performed under reducing conditions on 4–20% gradient polyacrylamide gels (Laemmli, 1970) either with 50 μg of total cellular protein per lane or with DPP-IV, CEA, or MUC1 immunoprecipitates. After electrophoresis, proteins were transferred to a nitrocellulose membrane (model BioTrace NT; Gelman Sciences Inc., Ann Arbor, MI) as described in Vaessen et al. (1981). The membranes were then treated for 2 h with polyvinylpyrolidone K-30 (2% in TBS). Immunodetection of DPP-IV, CEA, and MUC-1 was performed with mAbs 4H3, 517, and BC-2, respectively, using secondary antibodies peroxidase-coupled anti-rat or anti-mouse Ig accordingly (Biosys, Compiègne, France). For glycan detection, membranes were incubated with digoxigenin-labeled lectins from Boehringer Mannheim Biochemicals at concentrations of 5 μg/ml in TBS for MAA and SNA, and 2 μg/ml in TBS for PNA-digoxigenin. Then, the nitrocellulose membranes were incubated for 1 h with alkaline phosphatase-labeled antidigoxigenin Fab fragments (1 μg/ml in TBS) (Boehringer Mannheim Biochemicals). After washing, labeled glycoproteins were revealed by 4-nitro blue tetrazolium chloride 5-bromo-4-chloro-3-indolyl-phosphate staining. Desialylation was performed as for cryostat sections before incubation with the digoxigenin-labeled lectins. Digestion with endoglycosidase H (Boehringer Mannheim Biochemicals) was carried out in 0.1 M phosphate buffer, pH 5.5, containing 0.02% SDS, 1% Triton X-100, 1% β-mercaptoethanol, 1 mM phenylmethylsulfonylfluoride, 10 μg/ml leupeptin, 10 μg/ml pepstatin, using 10 mU of endoglycosidase H overnight at 37°C. Digestion with endoglycosidase F/N-glycosidase F (Boehringer Mannheim Biochemicals) was carried out in 0.1 M phosphate buffer, pH 7.0, containing 0.05% SDS, 1% Triton X-100, 1% β-mercaptoethanol, 1 mM phenylmethylsulfonylfluoride, 10 μg/ml leupeptin, and 10 μg/ml pepstatin, using 0.5 U of endoglycosidase F/N-glycosidase F overnight at 37°C. Controls were incubated in the same buffers overnight at 37°C without glycosidase.

Cells were cultured in six-well plates with or without the presence of 2 mM GalNAc-α-O-benzyl until day 11. Subsequently, untreated cells were pulse labeled for 15 min with 200 μCi/well of [³⁵S]-methionine (Amersham Corp.) in 1 ml of methionine-free medium, and then chased for the indicated periods of time with 1 ml 0.01 M methionine in regular medium. The same protocol was applied to treated cells, except for the presence of 2 mM GalNAc-α-O-benzyl throughout the experiment. Cells were rinsed in PBS and lysed in 1 ml of RIPA buffer (0.001 M Tris-HCl, pH 8.0, 0.01 M NaCl, 0.1% SDS, 1% Triton X-100, 0.5% sodium deoxycholate, 1% phenylmethylsulfonylfluoride, 0.001 M sodium ethylene diamine tetraacetate). Aliquots of 50 μg of proteins from cell lysates were incubated with mAb 4H3 overnight at 4°C. Immunocomplexes were collected on protein G–Sepharose 4B (Sigma Chemical Co.), eluted in SDS sample buffer (0.2 M Tris-HCl buffer, pH 6.8, containing 2% SDS and 30% glycerol) at 60°C for 5 min, and then analyzed on 5–30% SDS–polyacrylamide gels. For autoradiography, gels were fixed in 40% ethanol, 10% glycerol, 10% acetic acid (by vol), soaked in Amplify (Amersham Corp.) for 20 min, dried on Whatman paper, and then exposed to Cronex 4 NIF film (Dupont, Les Ulis, France).

To assay for a dose-dependent effect of GalNAc-α-O-benzyl, mucus-secreting HT-29 cells (RevMTX10⁻⁶) were treated, from the day of seeding on, with different concentrations of the drug in the 0.1–2 mM range. Regardless of the concentration used, no effect on cell viability was observed, as assessed by the absence of cells in suspension and trypan blue exclusion. As shown in Fig. 1, it has no effect on the doubling time of the cells in the first days in culture, but results in a dose-dependent lower cell density in the stationary phase. At the highest concentration (2 mM) the cells stop growing before reaching total confluence, with the cell layer occupying 75–80% of the surface of the flask (Fig. 1). This effect is associated with a dramatic swelling of the cells (Figs. 1 and 2), with their cytoplasm appearing filled with a honeycomb-like accumulation of vesicles of various sizes during transmission electron microscopy (see Fig. 4,c). Concomitant with these changes, the mucus gel is totally absent from the cells treated with the highest concentrations, even when the cells are maintained for a longer period (<30 d). The decrease in mucus content was further demonstrated by the analysis of sections of the cell layer (Fig. 2). At the highest concentration (2 mM), there was a total absence of alcian blue–stained material (Fig. 2,e). In addition, the dense immunofluorescent staining of apical mucus droplets observed in control cells with antigastric mucus antibodies was no longer observed in cells treated with 2 mM, having been replaced by a diffuse staining of the cytoplasm (Fig. 2 f).

Because the most clear-cut effect of GalNAc-α-O-benzyl was observed at 2 mM, all further experiments were done at this concentration. In control postconfluent HT29-RevMTX10⁻⁶ cells, DPP-IV, MUC1, and CEA are exclusively associated with the apical brush border of the cells, as previously reported (Lesuffleur et al., 1993) and shown in Fig. 3, g, i, and k. In contrast, in GalNAc-α-O-benzyl– treated cells, DPP-IV, MUC1, and CEA are present in the totality of the cytoplasm (Fig. 3, h, j, and l); this occurs without modification of the morphological polarity of the cells, substantiated by the apical expression of villin and ZO1 (Fig. 3, a–d). Interestingly, and in contrast to what observed for apical glycoproteins, the treatment had no effect on the basolateral expression of GP120 (Fig. 3, e and f). The cytoplasmic accumulation of apical glycoproteins in treated cells was further confirmed using immunoelectron microscopy: unlike in control cells where they are restricted to the apical brush border, DPP-IV, MUC1, and CEA are localized in the numerous vesicles that fill the cytoplasm (Fig. 4). Using DPP-IV as a marker of the effect, we further found that this altered distribution is associated with an increased level of expression of the enzyme at both mRNA (Fig. 5) and protein level as substantiated by higher enzyme activities and higher protein content, with a lower apparent molecular mass, however, as shown by Western blot (Fig. 5). The same results were obtained with mucus-secreting HT29-RevMTX 10⁻⁵ cells and enterocytic HT29-RevMTX10⁻³ cells (data not shown). The effect of GalNAc-α-O-benzyl on apical glycoproteins is not dependent on the support the cells are cultured on, as exemplified by the swelling of the cells and the intracytoplasmic accumulation of DPP-IV also observed in treated filter-grown cells (data not shown).

Switching back the cells to drug-free medium after 20 d of treatment results in a rapid reversal of the phenotype described above: within 24 h the volume of the cells decreases, and in the following days apical glycoproteins redistribute to the apical surface (Fig. 6,e). At the same time, mucus secretion resumes, judged by the occurrence of a visible gel on the surface of the cell layer demonstrated by alcian blue staining (data not shown) and immunofluorescence reactivity to antimucus antibodies of cell layer sections (Fig. 6 f).

To characterize the oligosaccharide species associated with mucins and other glycoproteins from HT-29–differentiated cells, we used the lectin MAA, which reacts with NeuAcα2-3Gal-R terminal sequence and PNA, which reacts with the O-linked T antigen (Galβ1-3GalNAcα1-O-Ser/Thr). As shown in Fig. 7 via double immunofluorescence, mucus from control HT29-RevMTX10⁻⁶ cells is reactive with MAA (Fig. 7, top panel, a–c), whereas only a small proportion of the mucus reacts with PNA (Fig. 7, top panel, d–f). After treatment with sialidase from Clostridium perfringens, all the mucus is reactive with PNA (data not shown), confirming the fact that a large proportion of the mucus expresses the NeuAcα2-3Galβ1-3GalNAc sequence. Because mucus droplets are concentrated in the apical cytoplasm and do not allow to distinguish the reactivity of brush border– associated glycoproteins to lectins in mucus-secreting cells, we used enterocytic HT29-RevMTX10⁻³ cells to further characterize the reactivity of these glycoproteins to lectins. As shown on cryostat sections from postconfluent cells (Fig. 7, middle panel), MAA shows a strong apical reactivity (Fig. 7, middle panel, a), whereas PNA is unreactive (Fig. 7, middle panel, e). After sialidase treatment, there is no longer any reactivity to MAA (Fig. 7, middle panel, c), whereas PNA shows a strong apical staining (Fig. 7, middle panel, g), thus testifying the presence of a large amount of sialyl-T antigen linked to brush border–associated glycoproteins. In contrast, no basolateral labeling was observed with MAA by confocal microscopy (Fig. 7, bottom panel), thus indicating that sialyl-T antigen is not associated with basolateral glycoproteins in differentiated HT-29 cells. The association of NeuAcα2-3Galβ1-3GalNAcα1-O-Ser/Thr with glycoproteins in differentiated HT-29 cells was confirmed by Western blot analysis of cell homogenates from the different HT-29 cell populations (Fig. 8) which suggest, based on the reactivity to MAA and PNA before and after treatment with sialidase, that NeuAcα2-3Galβ1-3GalNAcα1-O-Ser/ Thr sequence is associated not only to mucins, but also to a number of glycoproteins with a molecular mass in the 80– 400 kD range. Among the glycoproteins that react with MAA are DPP-IV, MUC1, and CEA as shown by immunoblot analysis with MAA of these immunoprecipitated proteins (see Fig. 10). NeuAcα2-3 glycosylation of T antigen is further supported by the observation that ST3Gal I is expressed at all stages of the culture in mucus-secreting as well as in enterocytic HT-29 cells, as substantiated by analysis of ST3Gal I mRNA level and enzyme activity (Fig. 8). Interestingly, no reactivity was observed by immunoblotting cell homogenates with mAb B72.3 against sialyl-Tn (data not shown) or SNA that recognizes the terminal oligosaccharide species NeuAcα2-6Gal (Fig. 9).

Western blot analysis of the reactivity to lectins of cell homogenates from postconfluent HT29-RevMTX10⁻⁶ cells treated with various concentrations of GalNAc-α-O-benzyl shows a dose-dependent decrease of MAA-reacting glycoproteins associated with a dose-dependent increase in PNA-reacting glycoproteins, with a maximum effect observed at a drug concentration of 2 mM (data not shown). This was further confirmed by the observation, using immunofluorescence, that regardless of their phenotype, the cytoplasm of cells treated with 2 mM GalNAc-α-O-benzyl is heavily stained with PNA, in contrast to control cells (data not shown). Western blot analysis of immunoprecipitated DPP-IV, MUC1, and CEA with MAA and PNA shows that MAA reactivity of these glycoproteins is reduced in cells treated with GalNAc-α-O-benzyl. Concomitant to these changes, a reactivity to PNA was observed for MUC1 and DPP-IV, but not for CEA (Fig. 10). The changes in glycosylation are reversible upon removal of the drug; Western blot analysis of the reactivity to lectins of cell homogenates from cells reverted to drug-free medium shows a reappearance of MAA reactivity and the concomitant disappearance of PNA reactivity of a number of glycoproteins in the 80–400 kD range (data not shown). These include MUC1, DPP-IV, and CEA as shown by Western blot analysis with lectins of the immunoprecipitates of these glycoproteins (Fig. 10).

To further characterize at which level the block induced by GalNAc-α-O-benzyl occurs, immunoprecipitates of apical glycoproteins from control and treated cells were analyzed for their sensitivity to endoglycosidase H and endoglycosidase F treatment. As exemplified for DPP-IV and CEA (Fig. 11,A), the same results were observed in treated and control cells with both proteins being endoglycosidase H–resistant and endoglycosidase F–sensitive, suggesting that the block occurs after the cis-Golgi. This was further confirmed by analysis of the biosynthesis of DPP-IV that shows the processing of the enzyme is similar in treated and in control cells (Fig. 11 B).

The present results suggest that in HT-29 cells, α2,3-sialylation plays a crucial role in intracellular transport of brush border membrane–associated glycoproteins and in mucus secretion. They rely on the exceptional conjunction of a number of factors: (a) the availability of polarized cells, isolated from the HT-29 cell line, expressing either an enterocytic or a mucus-secreting differentiated phenotype; (b) the neoplastic nature of these cells which, as such, show a modified pattern of protein glycosylation as compared with their normal counterpart with shorter oligosaccharide side chains (for review see Lesuffleur et al., 1994); (c) the fact that the main sialyltransferase activities expressed in HT-29 cells catalyze the transfer of sialic acid to the 3-position of Gal in the Galβ1-3GalNAc disaccharide sequence (Dall'Olio et al., 1993; Majuri et al., 1995; Delannoy et al., 1996) in contrast to most colon cancers (Sata et al., 1991) or cell lines, including the enterocytic cell line Caco-2, which mainly express the Galβ1-4GlcNAc α2,6-sialyltransferase, ST6Gal I as reported by others (Dall'Olio et al., 1992, 1996), and confirmed here by the reactivity to SNA of Caco-2 glycoproteins; (d) the availability of an O-glycosylation inhibitor, GalNAc-α-O-benzyl (Kuan et al., 1989; Huang et al., 1992; Byrd et al., 1995; DiIulio and Bhavanandan, 1995), which enters the cells and is metabolized into a compound that acts as a competitive inhibitor of Galβ1-3GalNAc α-2,3-sialyltransferases (Huet et al., 1995; Delannoy et al., 1996); and (e) the fact that, in differentiated HT-29 cells, NeuAcα2-3Galβ1-3GalNAc-R is the main oligosaccharide species associated not only with mucins, as previously reported (Capon et al., 1992; Lesuffleur et al., 1993; Huet et al., 1995), but also, as shown here, with a number of glycoproteins of the brush border concomitantly expressed in these cells. The association of NeuAcα2-3Galβ1-3GalNAc-R to apical glycoproteins relies exclusively on their reactivity to MAA, and not to a biochemical characterization that would require a huge amount of cells to be performed. However, the reliability of MAA characterization is validated by a recent structural characterization of carbohydrate chains of the mucus from HT29-RevMTX10⁻⁵ mucus-secreting cells (that can be easily performed since large quantities of mucus can be harvested daily) which has confirmed that NeuAcα2-3Galβ1-3GalNAc-R is the main oligosaccharide species associated with the mucus of these cells (Hennebicq-Reig, S., T. Lesuffleur, C. Capon, C. de Bolos, I. Kim, O. Moreau, C. Richet, B. Hémon, M.A. Recchi, E. Maës, J.P. Aubert, F.X. Real, A. Zweibaum, P. Delannoy, P. Degand, and G. Huet, manuscript submitted for publication). Finally, it must be noted, as demonstrated by confocal microscopy, that no MAA reactivity could be detected on the basolateral membrane of the cells.

Based on both the sugar specificity of MAA and PNA lectins and the results of sialidase treatment, the present data show that in addition to mucins (i.e., in HT29-RevMTX10⁻⁶ and 10⁻⁵), apical O-glycosylproteins such as MUC1 express NeuAcα2-3Galβ1-3GalNAc-R terminal sequences. This being established, four main observations can be drawn from the experiments performed with GalNAc-α-O-benzyl. First, alteration of α2,3-sialylation of mucins and of apical O-glycosylproteins is accompanied by their accumulation into intracytoplasmic vesicles. This accumulation, which is most likely responsible for the dramatic swelling of the cells, is not restricted to O-glycosylproteins, but is also observed for N-glycosylproteins such as DPP-IV and CEA. Regarding CEA, the absence of PNA binding in treated cells may be explained by a weak expression of the Galβ1-3GalNAc sequence, not in contradiction with the fact that the binding of MAA is decreased. Second, the block induced by GalNAc-α-O-benzyl occurs after the cis-Golgi as substantiated by endoglycosidase H resistance of apical glycoproteins and normal processing of DPP-IV. Third, these effects are reversible upon removal of the drug, resulting in a concomitant restoration of α2,3-sialylation and resumption of the secretion of mucins and of the apical delivery of brush border–associated glycoproteins. Fourth, these effects occur without any modification in the morphological polarity of the cells, as shown by the normal polarized expression of villin and ZO1, and without modification of distribution of the basolateral glycoprotein GP120.

How can the fact be explained that GalNAc-α-O-benzyl, an O-glycosylation inhibitor (Kuan et al., 1989; Huang et al., 1992; Byrd et al., 1995) which, in HT-29, is metabolized into a compound, acts as a competitive inhibitor of Galβ1-3GalNAc α2,3-sialyltransferases (Huet et al., 1995; Delannoy et al., 1996) may also inhibit the sialylation of N-glycosylproteins? Recent progress in the molecular cloning of sialyltransferases (for review see Harduin-Lepers et al., 1995; Tsuji, 1996; Tsuji et al., 1996) indicates that three different enzymes (ST3Gal I, ST3Gal II, and ST3Gal IV), encoded by different genes, located on separate chromosomes (Chang et al., 1995) are able to transfer sialic acid residues in the 3-position of Gal onto the disaccharidic Galβ1-3GalNAc-R sequence. The substrate specificity of ST3Gal I and ST3Gal II is strictly restricted to this particular disaccharidic sequence (Kojima et al., 1994) but ST3Gal IV can use both Galβ1-3GalNAc and Galβ1-4GlcNAc as acceptor substrates (Kitagawa and Paulson, 1994). The competitive inhibition of ST3Gal IV by Galβ1-3GalNAc-α-O-benzyl could explain the decrease of α2,3-sialylation of N-glycosylproteins in HT-29 cells. On the other hand, the presence of a high concentration of the competitive substrate (i.e., Galβ1-3GalNAc-α-O-benzyl) may also decrease the concentration of CMP-NeuAc in the Golgi lumen and therefore can compete with the other sialyltransferases expressed in HT-29 cells via the donor substrate.

Whatever the mechanism leading to the decrease of α2,3-sialylation, these results show that lack of terminal NeuAcα2-3Gal-R glycosylation is associated with a blockade of apical targeting of brush border membrane–associated glycoproteins and mucus secretion in HT-29 cells. They further suggest that α2,3-sialylation could play a role in regulating the intracellular traffic of these proteins, and in some way, support the view by Fiedler and Simons (1994) that “it may be that oligosaccharide side chains play a more important role in biosynthetic traffic than hitherto recognized”. Even if the role of glycans in the intracellular targeting of newly synthesized lysosomal enzymes via the mannose-6–phosphate receptor pathways was clearly demonstrated in human colon adenocarcinoma cell lines (Braulke et al., 1992), a role for glycans in the sorting machinery of membrane proteins in polarized cells has long been excluded on the basis of experiments using inhibitors of N-glycosylation such as tunicamycin, which blocks the transfer of Glc₃Man₉GlcNAc₂ from dolichol to Asn, thus resulting in the accumulation of unprocessed proteins in the rough ER (Green et al., 1981), or castanospermine, or 1-deoxymannojirimycin which block processing before trimming of the oligomannose chains (Duronio et al., 1988). Very few recent observations suggest, however, a possible role for terminal glycans in this regulation. Growth hormone, which is nonglycosylated and secreted from both sides of MDCK cell layers, is secreted from the apical side when N-glycosylated (Scheiffele et al., 1995). The sialoglycoprotein gp114 that is expressed on the apical membrane of MDCK cells is misglycosylated and predominantly basolateral in the MDCK mutant MDCKII-RCA^r (Le Bivic et al., 1993). The O-glycosylated stalk domain is required for apical sorting of neurotrophin receptors in polarized MDCK cells (Yeaman et al., 1997). In addition to these particularities that concern glycoproteins, it has been shown that among the increasing number of vesicular proteins presumably involved in protein sorting (Rothman, 1994), one of them, VIP36 (Fiedler et al., 1994), presents some homology with leguminous lectins (Fiedler and Simons, 1994) and binds GalNAc (Fiedler and Simons, 1996). Therefore, it is conceivable that terminal NeuAcα2-3Gal glycan sequences as signals, and lectins as receptors for these signals, could be involved in the sorting machinery of glycoproteins in polarized HT-29 cells. Because the glycosylation of both lectin-like resident Golgi proteins and in transit glycoproteins may be affected by GalNAc-α- O-benzyl, the precise level at which the sorting machinery is affected needs to be elucidated.

Whether or not α2,3-sialylation is exclusively involved in the apical sorting of glycoproteins in HT-29 cells cannot be concluded from the present work since it was only focused on brush border membrane glycoproteins and mucins. From the results obtained with Western blot, it is clear that the effects of GalNAc-α-O-benzyl are shared by a number of other proteins that remain to be characterized as to their nature and localization. Nevertheless, the present data show that a shift in the predominant glycosylation pattern of glycoproteins from NeuAcα2-3Galβ1-3GalNAc-R to Galβ1-3GalNAc-R results in what appears as a glycoprotein traffic jam that dramatically alters their normal delivery and also leads to cell hypertrophy.

Analysis of whether the present results uncover a more general involvement of terminal glycans in the intracellular protein transport machinery needs to consider two main evidences. First, there is the high polymorphism of terminal glycans which differ from one individual to another. Second, and with regard to a putative role of animal lectins as receptors for the glycan signal (Fiedler and Simons, 1994), there is the strict oligosaccharide specificity of lectins. This implies that if the results obtained with HT-29 cells rely on a general mechanism, each cellular system, whether from human or animal origin, is unique and only representative of the genetic background of the individual it originates from. With regard to normal or malignant intestinal cells it must be noted that there are only very few indications in which oligosaccharide species are associated with apical glycoproteins. The only available data concern the observation that, in the normal adult intestine, ABH blood group antigens are the main terminal glycans associated with the intestinal brush border hydrolases in humans (Triadou et al., 1983; Green et al., 1988) as well as in rabbits (Gorvel et al., 1982). With regard to differentiated colon cancer cells, the results reported here are the first observation showing that in HT-29 cells, the main terminal oligosaccharide species associated with apical glycoproteins is NeuAcα2-3Gal. Preliminary results obtained in the laboratory indicate that, as in rat developing enterocytes (Roth, 1993), the apical membrane of intestinal enterocytes from 10–20 gestational wk fetuses is strongly reactive with MAA; in contrast, epithelial cells in the adult human small intestine lack MAA reactivity (our unpublished results), differing from the situation found in the adult colon which has been shown to express an apical reactivity to MAA (Sata et al., 1991), but which is devoid of brush border–associated hydrolases. Therefore, the type of glycosylation demonstrated in HT-29–differentiated cells likely represents one more feature of the fetal type of differentiation which is known to be associated with the malignant phenotype (Zweibaum et al., 1991).

With regard to the oligosaccharidic individual variability it is clear that the results obtained with HT-29 cells cannot be extrapolated as such to other cellular systems. For example, in Caco-2 cells, in which the main expressed sialyltransferase is ST6Gal I (Dall'Olio et al., 1992, 1996) GalNAc-α-O-benzyl has no effect, even at much higher concentrations (10 mM), on the morphology of the cells and the apical polarity of brush border–associated glycoproteins (our unpublished results). This absence of effect is obviously consistent with the fact that in Caco-2 cells, apical glycoproteins such as sucrase-isomaltase or DPP-IV most likely express NeuAcα2-6, as suggested from the SNA reactivity of Caco-2 glycoproteins and that GalNAc-α-O-benzyl has no effect on ST6Gal I.

Finally the present results draw attention to the necessity of characterizing, in each experimental system, which oligosaccharide species is associated with the apical glycoproteins. This is a prerequisite for any further analysis of the role of terminal oligosaccharides in the intracellular traffic of apical glycoproteins. It also implies that new strategies should be developed to specifically block the terminal glycosylation of apical glycoproteins. One challenge is to analyze whether, for example, ABH blood group antigens are involved in the intracellular traffic of intestinal brush border hydrolases in humans.

We thank V. van Miegem (CNRS UMR III [Lille]), B. Hémon, and O. Moreau (both from INSERM U377 [Lille]) for excellent technical assistance. We are grateful to the Scientific and Medical Services of the University of Barcelona (Barcelona, Spain) for confocal microscopy analysis.

This work was supported in part by INSERM Consejo Superior de Investigaciones Científicas (CSIC) Cooperation Agreement, and grants from Fondo de Investigacion Sanitaria (94/1128), Generalitat de Catalunya (GRQ 93-01), Comisión Interministerial de Ciencia y Tecnología (CICYT) (SAF 97-0085), and Université Paris XI.

CEA

carcinoembryonic antigen

DPP-IV

dipeptidylpeptidase-IV

GalNAc-α-O-benzyl

benzyl-2-acetamido-2-deoxy-α-d-galactopyranoside

MAA

Maackia amurensis agglutinin

MTX

methotrexate

PNA

Arachis hypogaea agglutinin

SNA

Sambucus nigra agglutinin. The abbreviations for sialyltransferases are according to the new systematic nomenclature proposed by Tsuji et al. (1996)