1-Benzyl-2-acetamido-2-deoxy-α-D-galactopyranoside Blocks the Apical Biosynthetic Pathway in Polarized HT-29 Cells*

In previous work we reported that long term treatment of polarized HT-29 cells by 1-benzyl-2-acetamido-2-deoxy-α-d-galactopyranoside (GalNAcα-O-bn) induced undersialylation and intracellular distribution of apical glycoproteins such as dipeptidyl peptidase IV (DPP-IV), and we suggested therefore that sialylation could act as an apical targeting signal. In this work, the apical direct biosynthetic route was studied after transfection of polarized enterocyte-like HT-29 5M12 cloned cells with a murine cDNA coding for a soluble form of DPP-IV, which was secreted into the apical medium. A 24-h treatment of transfected cells by GalNAcα-O-bn markedly inhibited the apical secretion and the sialylation of this soluble murine DPP-IV, which became blocked inside the cell. A similar short GalNAcα-O-bn treatment also induced an intracellular distribution of both endogenous transmembrane DPP-IV and proteins involved in the regulation of the apical trafficking such as the apical t-SNARE syntaxin-3 and the raft-associated protein annexin XIIIb, whereas the basolateral t-SNARE syntaxin-4 kept its normal localization. These apical membrane proteins moved efficiently from trans-Golgi network to apical carrier vesicles but failed to be transported from carrier vesicles to the apical plasma membrane. Isolation of membrane microdomains showed that GalNAcα-O-bn induced the formation of abnormal lipid-rich microdomains in comparison to normal rafts, as shown by their lower buoyant density and their depletion in annexin XIIIb. In conclusion, GalNAcα-O-bn blocks the anterograde traffic to the apical surface of polarized HT-29 cells at the transport level or docking/fusion level of carrier vesicles.

In recent years, data of the literature have shown the role of glycosylation in the apical biosynthetic route in polarized epithelial cells. Association of proteins with glycosphingolipids has been proposed for the apical delivery (1,2). Glycosylphosphatidylinositol (GPI) 1 -anchored proteins and glycosphingolip-ids could be isolated from a Triton X-100-insoluble fraction (3). The dynamic clustering of glycosphingolipid-enriched microdomains in the TGN would constitute functional rafts for apical delivery of proteins (4). Thus, proteins of apical destination, such as GPI-anchored proteins, would be specifically recruited in these apical raft carriers. Besides, data obtained by mutation, deletion, or addition of glycosylation sites showed that Oor N-glycosylation could influence the targeting of glycoproteins toward the apical side of the cells (5)(6)(7)(8)(9)(10)(11)(12)(13)(14).
The putative role of N-and O-glycans was also shown through the use of drugs affecting their processing, i.e. tunicamycin, deoxymannojirimycin, swainsonine for N-glycans, and 1-benzyl-2-acetamido-2-deoxy-␣-D-galactopyranoside (GalNAc-␣-O-bn) for O-glycans (11, 14 -18). We previously observed that long term treatment of polarized goblet or enterocytic HT-29 cells by GalNAc␣-O-bn led (i) to a dramatic inhibition in the secretion of mucins, which are highly O-glycosylated; (ii) to a decrease in the apical membrane expression of brush border markers such as the transmembrane glycoprotein dipeptidylpeptidase-IV (DPP-IV), the GPI-anchored carcinoembryonic antigen, and the mucin-like glycoprotein MUC1; and (iii) to the abnormal presence of these glycoproteins inside the cells (19,20). In contrast, basolateral glycoproteins such as gp120 and gp525 kept a normal localization under GalNAc␣-O-bn treatment (19,21).
These data led us to analyze more deeply the interferences of exogenous GalNAc␣-O-bn with the intracellular processes of glycosylation in HT-29 cells. In a previous work, we showed that GalNAc␣-O-bn was highly converted into the benzyldisaccharide Gal␤1-3GalNAc␣-O-bn, which acts as a potent competitive inhibitor of ␣2,3-sialyltransferase ST3Gal I and involved in the terminal elongation of O-linked glycans (22). Thereafter, we reported that GalNAc␣-O-bn was extensively metabolized beyond Gal␤1-3GalNAc␣-O-bn, showing that the glycosylation of endogenous substrates by several glycosyltransferases could be inhibited, and in particular the sialylation of N-glycans by ␣2,3-sialyltransferase ST3Gal IV (21,23). In this way, the sialylation of N-and/or O-glycans was found inhibited on the endogenous glycoproteins DPP-IV, MUC1, and GPI-anchored carcinoembryonic antigen, and we suggested that undersialylation of glycoproteins may induce a defect in the direct apical targeting of glycoproteins (19 -21).
Later on, the effect of GalNAc␣-O-bn was studied by others using other cell lines, and the induction of a shift from the apical membrane to the basolateral membrane was reported in the targeting of the brush border glycoproteins sucrase isomaltase and DPP-IV using MDCK and Caco-2 cell lines (11,18). However, the expression of peripheral glycan epitopes is a cell type-specific process that depends on the expression pattern of glycosyltransferases. Interestingly, we and others found that GalNAc␣-O-bn treatment of different cell lines in culture did not result in the identical alterations, regarding glycosylation and intracellular trafficking (21,(23)(24)(25)(26)(27), suggesting that the cell type specificity in the cellular responses to GalNAc␣-O-bn was connected to the cell type specificity in the modification of the glycosylation pattern.
GalNAc␣-O-bn appeared to be an interesting tool to study polarized HT-29 cells, regarding the fact that, in this cell type specifically, (i) the terminal sialylation was identified as the target of inhibition, and (ii) the normal apical localization of brush border glycoproteins was disrupted with intracellular localization. Thus, we investigated the effect of this inhibitor of glycosylation on de novo apical trafficking in enterocyte-like HT-29 cells.

MATERIALS AND METHODS
Cell Culture-HT-29 clone 5M12 cells (cloned from a HT-29 cell subpopulation resistant to methotrexate (Ref. 28)) were cultured as previously described (19). Cells were cultured in 25-cm 2 T flasks (Corning Glass Works, Corning, NY) for cell maintenance and on glass coverslips or 24.5-mm tissue culture-treated Transwell polyester membrane filters (0.4-m pore size) (Costar, Cambridge, MA) for confocal microscopy and on 6-well culture dishes or 24.5-mm tissue culturetreated Transwell polyester membrane filters for metabolic labeling. GalNAc␣-O-bn was used at a concentration of 2 mM in Dulbecco's modified Eagle's medium with fetal bovine serum during a short-term treatment (from 0 to 44 h) starting from confluence. In all experiments, GalNAc␣-O-bn had no effect on cell viability, as assessed by the absence of cells in suspension and trypan blue exclusion. For the analysis of cell culture media by two-dimensional gel electrophoresis, cells were cultured for 24 h in serum-free medium before collection.
Transfection of Soluble Mouse DPP-IV-To generate a secreted mouse DPP-IV (sDPP-IV), the full-length mouse DPP-IV cDNA (29) was digested with PvuII restriction enzyme to remove the first 105 nucleotides and was subcloned downstream of a sequence coding for a cleavable peptide signal into the eukaryotic expression vector pTEJ8 that contains the G418 resistance gene (18). HT-29 5M12 cells were transfected using LipofectAMINE according to the instructions from the manufacturer (Invitrogen). Resistant colonies growing in the presence of 400 g/ml G148 were isolated using cloning cylinders and screened for secreted DPP-IV activity in the medium. Clone 5M12 Cl2 was selected for further experiments. DPP-IV activity was measured by a kinetic method with the fluorogenic substrate H-Gly-Pro-AMC (Bachem) (21).
Two-dimensional Gel Electrophoresis and MALDI Time-of-flight (TOF) Mass Fingerprinting of Tryptic Peptides-HT-29 5M12 Cl2 cells culture media were concentrated and precipitated with trichloroacetic acid. Amounts used for two-dimensional electrophoresis were normalized to the same amount of cells. Individual 17-cm ReadyStrips IPG strips (Bio-Rad), pH 3-10 or 4 -7, were rehydrated with 400 l of urea buffer (7 M containing the sample. Isoelectric focusing was carried out using a Protean IEF Cell according to the instructions from the manufacturer (Bio-Rad). Prior to the second dimension, the IPG gel strips were equilibrated twice in 10 ml of equilibration solution (50 mM Tris-HCl buffer, pH 8.8, containing 6 M urea, 3.3% (v/v) glycerol, 1% (w/v) SDS) for 20 min, first in the presence of dithiothreitol (65 mM) and then in the presence of iodoacetamide (87 mM). Second dimension gels consisted in a 6-16% acrylamide gradient, cross-linked with piperazine diacrylamide. Gels were run using the Protean II xi Cell (Bio-Rad), at a constant current of 15 mA/gel overnight. The analytical two-dimensional gels were silver-stained and scanned (ImageMaster, Amersham Biosciences). The preparative two-dimensional gels were stained for 4 days in Coomassie Brilliant Blue G solution.
Electrophoretically separated proteins were excised from the Coomassie Blue-stained gels. Gel pieces were reswollen in a trypsin solution containing 0.2 g of sequence-grade modified porcine trypsin (Promega, Madison, WI) at 37°C during 18 h. The protein fragments were extracted with 25 l of 50% acetonitrile, 0.1% trifluoroacetic acid for 30 min. The extracts were pooled, dried, and desalted with micro ZipTip C18 (Millipore) before the MALDI analysis. The peptide mixtures from the tryptic digests were crystallized in a matrix consisting of hydroxycinnaminic acid (Aldrich) prepared in 50% (v/v) acetonitrile, 0.1% (v/v) trifluoroacetic acid. Spectra were acquired using a PerSeptive Biosystems (Framingham, MA) Voyager-DE STR instrument and were internally calibrated using trypsin-autodigested peptides. The MALDI spectra were explored for database searches using the software Data Explorer (Applied Biosystems). The monoisotopic peptide masses obtained for each protein digest was submitted to Profound software to identify the proteins (prowl.rockfeller.edu/cgi-bin/Profound).
Quantitative analyses were performed by measuring the ratio apical fluorescence/inside fluorescence from confocal images. Fluorescence was quantified on a predefined surface for each section using the quantification program from Leica. The ratio apical fluorescence/inside fluorescence was calculated for the selected sections shown in the illustrations.
Vesicle Isolation from Perforated Cells-Isolation of TGN-derived vesicles was carried out according to the procedure described by Wandinger-Ness et al. (31). Briefly, HT-29 5M12 cells were seeded on 12 100-mm filters, which were cultured up to day 10. Six filters were treated by 2 mM GalNAc␣-O-bn for 18 h before the experiment. After the transfer of filters to 14-cm dishes, cells were perforated using a nitrocellulose acetate filter (HATF, 0.45-m pore) and carrier vesicles were released by incubation in the presence of an ATP-regenerating system for 1 h at 37°C. The carriers vesicles were then isolated from the incubation medium by flotation onto a 0.25 M sucrose cushion, pelleted, and purified on a discontinuous density gradient (1.5 M/1.2 M). The perforated cells were collected from the filters.
Ultrastructural Immunochemistry-Immunolabeling was carried out on isolated carrier vesicles of HT-29 5M12 cells according to the procedure described by Scheiffele et al. (32). Briefly, the vesicle preparation was laid on parlodion-coated nickel grids and fixed with 4% paraformaldehyde for 5 min. Single or double labeling was performed. The grids were incubated with a first primary antibody (polyclonal anti-syntaxin-3 (1/20) or polyclonal anti-annexin XIIIb (1/40)) for one night at 4°C in PBG buffer (PBS containing 0.5% bovine serum albumin and 0.2% gelatin), then with anti-rabbit Ig-coupled 18-nm gold particles (1/20) for 1 h at room temperature in PBG buffer. Subsequently, after a second fixation, grids were incubated with the second primary antibody (monoclonal anti-human DPP-IV antibody (1/20)) for 5 h at room temperature and then with anti-mouse Ig-coupled 12-nm gold particles. The grids were counterstained with 0.3% uranyl acetate and 1.8% methylcellulose. Controls grids were processed without the second primary antibodies and showed the absence of detection of 12-nm gold particles.
Metabolic Labeling and Immunoprecipitation of Secreted Mouse DPP-IV and Endogenous Human DPP-IV-For monitoring the secretion of transfected sDPP-IV, HT-29 5M12 Cl2 cells were cultured on filters; otherwise, HT-29 5M12 cells were cultured in 6-well plates. Cells were cultured in standard conditions until day 10. Subsequently, control cells were pulse-labeled for 30 min with 200 Ci/well of [ 35 S]methionine (ICN, Irvine, CA) in 1 ml of methionine-free medium, and then chased for the indicated periods of time with 1 ml of 0.01 M methionine in regular medium. A similar protocol was applied in the presence of 2 mM GalNAc␣-O-bn throughout the experiment. To follow the secretion of sDPP-IV, we collected the apical medium and we immunoprecipi-tated sDPP-IV with rat mAbs 1082 and 773. For analyzing the detergent extractability of cellular endogenous human DPP-IV, we used the procedure described by Alfalah et al. (14). HT-29 5M12 cells were rinsed in PBS, and cells were solubilized in 25 mM Tris-HCl, pH 8.0, 50 mM NaCl, containing 1% Triton X-100 and protease inhibitors for 2 h at 4°C. The detergent extracts were centrifuged at 100,000 ϫ g for 1 h at

FIG. 1. Secretion of recombinant soluble murine sDPP-IV by control and GalNAc␣-O-bn-treated HT-29 5M12 Cl2 cells.
A, cells were cultured on filters in standard conditions up to day 14. Subsequently, cells were cultured for 24 h in serum-free medium with or without GalNAc␣-O-bn. Apical and basolateral media were collected and analyzed by two-dimensional electrophoresis. In the apical medium of control cells, spots 1-5 were identified as sDPP-IV by MALDI-TOF mass spectrometry. The amount of sDPP-IV decreased under GalNAc␣-O-bn treatment, particularly the most acidic forms. sDPP-IV was not significantly secreted into the basolateral medium. B, pattern of apical secretion of sDPP-IV after pulse-chase metabolic labeling. HT-29 5M12 Cl2 cells were pulse-chase-labeled with [ 35 S]methionine in the absence (control) or presence (treated) of GalNAc␣-O-bn. The apical medium was collected and immunoprecipitated with anti-murine DPP-IV antibody, and immunoprecipitates were subjected to SDS-PAGE. C, lectin blot with MAA lectin of immunoprecipitated sDPP-IV from the apical medium of control and GalNAc␣-O-bn-treated cells. Note that sDPP-IV secreted by control cells reacts with MAA but not sDPP-IV secreted by GalNAc␣-O-bn-treated cells.
Isolation of Raft Microdomains-Rafts were isolated from series of 12 75-cm 2 culture flasks of HT-29 5M12 cells according to the procedure of Western Blotting-Samples were separated on a 5-30% SDS-PAGE (34), and the detection of proteins was carried out by luminescence using the ECL Western blotting system (Amersham Biosciences, Aylesbury, UK).

RESULTS
In this work, we studied the effect of GalNAc␣-O-bn on the direct apical biosynthetic route using a cloned population of HT-29 cells of enterocytic phenotype (HT-29 5M12) and short exposure times to GalNAc␣-O-bn. In addition to the study of endogenous transmembrane DPP-IV, which is endocytosed and recycled to the apical membrane, we studied a soluble form of murine DPP-IV, which was expected to be secreted in the medium under a polarized fashion after transfection in HT-29 5M12 cells. Indeed, after transfection of a soluble form of rat DPP-IV in MDCK cells, Weisz et al. (35) reported that this soluble form was predominantly secreted into the apical medium, indicating that the lumenal domain of DPP-IV was likely to contain the apical sorting information.
Inhibition in the Secretion of a Soluble Murine DPP-IV Form-HT-29 5M12 cells of enterocytic phenotype were stably transfected with an eukaryotic expression vector coding for a secreted mouse DPP-IV, and we selected the clone HT-29 5M12 Cl2 that secreted DPP-IV enzymatic activity in the medium.
We first investigated the secreted DPP-IV forms in the apical and basolateral medium using two-dimensional electrophoresis and protein identification by mass spectrometry of tryptic digests (Fig. 1A). Cells were cultured on filters in standard medium up to day 10 and then for 24 h in serum-free medium with or without GalNAc␣-O-bn. Apical and basolateral cell culture media were collected, concentrated, and precipitated with trichloroacetic acid. The amounts of cell culture media used for two-dimensional electrophoresis were normalized in reference to the same number of cells. In the apical medium of control cells, a train of 5 spots of high intensity was visualized between 80 and 90 kDa with a range of pI (isoelectric point) from 5.5 to 6 approximately. These spots were excised, digested with trypsin and analyzed for their peptidic map by MALDI-TOF. Results showed that these spots corresponded to murine DPP-IV (data not shown). The two-dimensional pattern of the basolateral medium showed that sDPP-IV was not significantly secreted into this medium. This shows that the recombinant sDPP-IV is predominantly secreted in the apical medium of HT-29 5M12 Cl2 cells and confirmed the existence of apical targeting signal(s) in the ectodomain of the molecule, and also indicated the involvement of a specific machinery for its transport toward the apical side. In the apical medium collected after GalNAc␣-O-bn treatment, we visualized in the area of sDPP-IV only three spots (spots 1, 2, and 3) with lower intensity. Spots 4 and 5, corresponding to the most acidic forms, were not detected, and spot 3, which accounted for the major spot, showed a marked decrease in comparison to spots 1 and 2. sDPP-IV was not recovered in the basolateral medium of treated HT-29 5M12 cells, although a slight increase was detected at the area of sDPP-IV. To further assess the intracellular retention of sDPP-IV, we evaluated the DPP-IV enzymatic activity in the cytosol of control and treated cells. Results showed a 1.6-fold increase in DPP-IV enzymatic activity after GalNAc␣-O-bn exposure. These data clearly showed that Gal-NAc␣-O-bn impaired the apical secretion of sDPP-IV, particularly that of the most acidic sDPP-IV forms, i.e. likely the most sialylated forms.
In a new experiment (Fig. 1B), we studied the secretion process of mouse sDPP-IV after metabolic labeling and immunoprecipitation of sDPP-IV from the apical medium using antimurine DPP-IV antibody. In control cells, sDPP-IV became clearly detected in the apical medium after a chase of 3 h. Thereafter, the secretion increased up to 24 h. In the presence of GalNAc␣-O-bn, there is barely no secretion of sDPP-IV. In addition, sDPP-IV appeared as two bands with lower M r in comparison to sDPP-IV secreted by control cells. This decrease in the apparent M r of sDPP-IV was assumed to result from the inhibition in the sialylation of sDPP-IV, in agreement with data previously obtained for human DPP-IV (19).
Finally, both the absence of the most acidic sDPP-IV forms and the decrease in the apparent M r of sDPP-IV suggested that GalNAc␣-O-bn inhibited the sialylation of soluble sDPP-IV. This hypothesis could be confirmed by the lectin blotting analysis of sDPP-IV with the Maackia amurensis lectin (MAA), which specifically recognized ␣2-3-linked sialic acid (Fig. 1C). A MAA reactivity was found for control cells but was not found for GalNAc␣-O-bn-treated cells, showing a defect in ␣2,3-sialylation for the sDPP-IV forms secreted by treated cells.

Alteration in the Distribution of Endogenous Human DPP-IV-We then analyzed the cellular distribution of the endogenous transmembrane DPP-IV in HT-29 5M12 cells after increasing exposures times to GalNAc␣-O-bn (from 6 to 44 h)
using quantitative confocal microscopy. Results (Fig. 2) showed that DPP-IV progressively lost its localization at the apical membrane and displayed an overall cellular localization. The quantitative measure of the ratio apical fluorescence/inside fluorescence showed that, after a 18-h exposure time, most DPP-IV was localized inside the cells. To confirm that DPP-IV only redistributed to the inside of the cells and not to the basolateral membrane, we further analyzed the localization of DPP-IV by double-labeling experiment using E-cadherin as a marker of the basolateral membrane (data not shown). These results were consistent with an intracellular trapping of endogenous transmembrane DDP-IV.
Specific Alteration in the Distribution Proteins Involved in Apical Regulation-We then hypothesized that GalNAc␣-O-bn treatment could specifically affect the vesicular trafficking machinery to the apical surface. Both trafficking pathways toward the apical and basolateral membranes of polarized cells are regulated by the SNARE machinery, but different SNARE complexes are involved in each targeting. The t-SNAREs syntaxin-3 and -4 were, respectively, localized at the apical and basolateral membranes of polarized MDCK and Caco-2 cells (36,37). Munc18-2/Munc18b is a Sec1 homologue, which interacts with the apical t-SNARE syntaxin-3 (38). Annexin XIIIb is an N-myristoylated protein, which was reported to participate in the apical transport via an association with the raft lipid microdomains (39,40). We thus investigated the distribution of these SNARE or SNARE partners in control and GalNAc␣-Obn-treated polarized HT-29 5M12 cells (Fig. 3). Syntaxin-3 was faintly detected at the apical side of control cells. After Gal-NAc␣-O-bn treatment, we observed a significant syntaxin-3 labeling inside the cells. Munc18-2/Munc18b and annexin XIIIb were clearly detected at the apical surface of control cells. After GalNAc␣-O-bn treatment, Munc18-2/Munc18b and annexin XIIIb became localized inside the cells. In contrast, syntaxin-4 was expressed at the basolateral membrane of HT-29 5M12 cells, and this localization was unchanged under Gal- NAc␣-O-bn treatment. Altogether, these data showed that Gal-NAc␣-O-bn also inhibited the trafficking of proteins involved in apical regulation.
Correct Transport of Apical Proteins from TGN to Carrier Vesicles-Apical delivery has been proposed to occur via the recruitment of apical proteins within rafts in TGN and the budding of raft carrier vesicles (4). Regarding published data on the role of N-and/or O-glycans on raft association, we searched to determine whether apical proteins were transferred into carrier vesicles under GalNAc␣-O-bn treatment. Thus, we have isolated TGN-derived carrier vesicles from perforated HT-29 5M12 cells and analyzed them at the ultrastructural level by immunogold labeling (Fig. 4A) and at the biochemical level by Western blotting (Fig. 4B). We examined endogenous DPP-IV, annexin XIIIb, and syntaxin-3 by single labeling. As syntaxin-3 and annexin XIIIb were previously localized in apical carrier vesicles in MDCK cells (40,41), two immunogold double labelings were performed: DPP-IV/syntaxin-3 (data not shown) and DPP-IV/annexin XIIIb. Results clearly showed the presence of these apical proteins in carrier vesicles in both control and GalNAc␣-O-bn-treated cells. These data indicated that GalNAc␣-O-bn treatment did not alter the pathway from TGN to the TGN-derived carrier vesicles.
We further tried to obtain quantitative data on the relative distribution of these apical proteins in the carrier vesicles and at the plasma membrane. The isolated carrier vesicles were analyzed by immunoblotting in comparison to the membrane fraction isolated from the corresponding ghosts of perforated cells. Samples were normalized to a similar amount of ghost membrane fractions. We analyzed the relative distribution of the raft-associated protein annexin XIIIb (Fig. 4B). No significant change was found as for the level of annexin XIIIb in the carrier vesicles of control and GalNAc␣-O-bn-treated cells in contrast to a significant decrease of annexin XIIIb in the plasma membrane. These data showed that apical proteins were correctly transported from TGN to carrier vesicles, and that the defect in the anterograde traffic instead affected the transport and/or the docking/fusion of carrier vesicles to the apical plasma membrane.
Raft Association of DPP-IV along Its Biosynthetic Pathway-Raft association of endogenous DPP-IV was then followed along its biosynthetic pathway in control and GalNAc␣-O-bn-treated HT-29 5M12 cells after pulse-chase metabolic labeling (Fig. 5). In control cells, results showed that DPP-IV displayed a detergent insolubility after 4 h of chase. At this time, the maturation of the protein precursor into the mature glycosylated form of higher molecular weight appeared nearly complete and only the mature form was detergent-insoluble. Subsequently, the relative proportion of detergent-insoluble DPP-IV form decreased, and the glycoprotein reappeared under a detergentsoluble form after 6 and 24 h of chase. After 24 h of chase, a second shift in the apparent M r of DPP-IV was visualized, suggesting a late maturation by post-translational processing. In GalNAc␣-O-bn-treated cells, DPP-IV also showed the appearance of a detergent-insoluble form after 4 h of chase, which decreased thereafter. However, after 6 h of chase, the detergent-soluble form appeared only at a very low level, and became nearly undetectable after the chase of 24 h, showing that DPP-IV followed a degradative process after detergent insolubility in GalNAc␣-O-bn-treated cells. We have checked this pattern of biosynthesis and trafficking of DPP-IV in two other immunoprecipitation experiments. These data indicated that DPP-IV was able to associate transiently with lipid microdomains in GalNAc␣-O-bn-treated cells, but that the subsequent reversion of DPP-IV in a non-raft membrane fraction did not show a similar steady state to that in control cells.
Association of Apical Proteins with Rafts-To further analyze raft association, we used a slightly modified procedure of membrane fractionation using detergent insolubility and ultracentrifugation on discontinuous sucrose gradients as described by Fiedler et al. (33). The procedure was applied on control and GalNAc␣-O-bn-treated HT-29 5M12 cells. Twelve fractions were collected all over the gradient (Fig. 6A). Detergent-insoluble raft microdomains described to be present at the interface 0.15 M/1.1 M corresponded to the collected fraction 3, whereas detergent-soluble membranes was collected along fractions 4 -12. In control cells, a high amount of proteins were found in fractions 4 -12, but a peak of proteins was clearly visualized in fraction 3. In GalNAc␣-O-bn-treated cells, the peak of detergent-insoluble material shifted to fraction 2 of lower buoyant density.
The equilibrium distribution of DPP-IV and annexin XIIIb along this gradient was probed by Western blot after loading each lane with half of the entire volume of each fraction (Fig.  6B). Films were scanned, and the insoluble (fraction 2-3)/ soluble (fractions 4 -12) ratio was calculated. In control cells, DPP-IV was mainly present in detergent-soluble membrane fractions (ratio 0.09). We also observed an heterogeneity in the apparent M r of DPP-IV, as DPP-IV in fraction 4 showed a lower migration in comparison to other fractions, an observation connected to the pattern obtained by pulse-chase metabolic labeling. GalNAc␣-O-bn induced an enrichment of DPP-IV in detergent-insoluble fraction 2 (ratio 1.15). For annexin XIIIb, the distribution in control cells also showed a predominance in detergent-soluble membrane fractions (ratio 0.17), but in contrast to DPP-IV, GalNAc␣-O-bn markedly decreased the proportion of annexin XIIIb in the detergent-insoluble fractions (ratio 0.02).
Controls were then used to check the specificity of our data (Fig. 6C). The distribution of the raft marker protein flotillin-1/Reggie-2 was examined as a positive control. Flotillin-1 was found in raft fraction 3 of control cells but also in non-raft fractions 4 -12. In GalNAc␣-O-bn-treated cells, flotillin-1/Reggie-2 shifted to the fraction 2. The glycoprotein gp525, present at the basolateral membrane and not affected in its trafficking by GalNAc␣-O-bn treatment (21), was used as a negative control. As shown in Fig. 6C, the distribution of this glycoprotein was exclusively localized in non-raft fractions and appeared unchanged by GalNAc␣-O-bn treatment. Altogether, these data showed that GalNAc␣-O-bn treatment modified the composition of lipid microdomains. DISCUSSION In our previous work, we reported that long term treatment of polarized HT-29 cells by the inhibitor of glycosylation Gal-NAc␣-O-bn induced an abnormal intracellular localization of apical membrane glycoproteins. In this cell line, GalNAc␣-O-bn was found to inhibit primarily the terminal sialylation by the sialyltransferases ST3Gal I and ST3Gal IV, and we suggested that sialylation of apical glycoproteins may act as a targeting signal for apical delivery (19,22,23). The goal of this work was to examine the anterograde traffic in cloned HT-29 5M12 cells material of GalNAc␣-O-bn-treated cells shifted in fraction 2 of lower buoyant density. B, probing with antibodies against DPP-IV and annexin XIIIb (half of the entire volume of each fraction was used). The gels were scanned, and the detergent-insoluble (fractions 2 and 3)/detergent-soluble (fractions 4 -12) ratio was calculated and is shown on the right. C, probing with a positive control, i.e. the ubiquitous raft marker flotillin, and with a negative control, i.e. gp525, a basolateral glycoprotein not affected by GalNAc␣-O-bn treatment.
under GalNAc␣-O-bn exposure. Short exposure times were used to avoid the expression of a storage phenotype linked to the accumulation of GalNAc␣-O-bn metabolites (23,27,42).
It has been previously shown that rat and murine soluble recombinant DPP-IV were secreted predominantly into the apical medium of MDCK cells, indicating that the ectodomain of DPP-IV contained sorting signal(s) for apical targeting (18,35). To study the direct apical route in HT-29 5M12 cells, we transfected an expression vector coding for a secreted mouse DPP-IV. Murine sDPP-IV accounted for a major protein secreted into the apical medium of transfected HT-29 5M12 Cl2 cells, whereas it was barely detected in the basolateral medium. Interestingly, GalNAc␣-O-bn markedly decreased the apical secretion of this glycoprotein, and sDPP-IV did not shift to the basolateral medium but accumulated inside the cells. Altogether, these data showed that GalNAc␣-O-bn blocked the apical anterograde traffic in HT-29 5M12 Cl2 cells. In addition, we observed that GalNAc␣-O-bn also induced undersialylation of sDPP-IV. Recently, Ulloa and Real (43) tried to rule out the hypothesis of a defect in the apical direct traffic caused by undersialylation, because the storage phenotype induced by prolonged GalNAc␣-O-bn treatment led to a defect in the recycling of endocytosed membrane glycoproteins which became accumulated in late endosomes. However, our analysis of various cell lines treated by GalNAc␣-O-bn clearly showed that the storage phenotype and the intracellular retention of apical glycoproteins are two independent events (27). In this work overall, the finding of the intracellular retention of a DPP-IV form, which does not follow the endocytosis/recycling process but is exocytosed into the apical medium allowed to assert clearly the occurrence of a block on the apical anterograde traffic in GalNAc␣-O-bn-treated HT-29 5M12 Cl2 cells.
We had shown that the block occurred beyond the cis-Golgi, as substantiated by endoglycosidase treatment and processing of endogenous DPP-IV (19). Thus, we investigated the distribution of different proteins known to be involved in the regulation of the vesicular transport. Proteins of SNARE complex are known to play a role in the regulation of the direct apical biosynthetic route: t-SNARE syntaxin-3 and v-SNARE Ti-VAMP (36,37,44,45). In addition, the sec1-related protein Munc18-2 modulates the interaction of syntaxin-3 with other SNARE partners, and controls the vesicular apical transport (38). Annexin XIIIb has been also previously described to play a role in apical delivery through an association with rafts (40). Syntaxin-3, Munc18-2, and annexin XIIIb were localized at the apical side of control HT-29 cells, but shifted to an intracellular localization in GalNAc␣-O-bn-treated HT-29 5M12 cells. Therefore, GalNAc␣-O-bn also affected proteins regulating the apical trafficking such as proteins of the apical SNARE fusion machinery and raft-associated proteins, showing finally a general block in the vesicular transport toward the apex. The effect affected specifically the apical route, because the basolateral t-SNARE syntaxin-4 kept its normal basolateral localization.
Proteins of the apical membrane have been proposed to be transported via apical carrier vesicles (4). Furthermore, such apical vesicles are also known to carry proteins involved in the regulation of apical trafficking (41). Analysis of carrier vesicles isolated from perforated HT-29 5M12 cells showed the presence of DPP-IV, syntaxin-3, and annexin XIIIb in carrier vesicles of both control and GalNAc␣-O-bn-treated cells, indicating that GalNAc␣-O-bn did not prevent the trafficking pathway from the TGN to the apical carrier vesicles, but rather inhibited the transport from the carrier vesicles up to the apical plasma membrane. Thus, undersialylated DPP-IV is still packaged into vesicles containing apical directed SNAREs. The correct transport from TGN to carrier vesicles was in accordance with the occurrence of an association of DPP-IV with detergent-insoluble microdomains in GalNAc␣-O-bn-treated cells as in control cells. This is in agreement with the fact that rafts have been described to constitute a dynamic platform for budding of raft carrier vesicles, which would transport the apical cargo up to the apical plasma membrane (4). It is not surprising that detergent insolubility of DPP-IV in control and GalNAc␣-O-bntreated HT-29 5M12 cells occurred after processing by posttranslational glycosylation, because N-or O-glycans were reported to favor raft association and apical delivery of membrane proteins in MDCK and Caco-2 cells (12,14). In the literature, human DPP-IV has been described as preferentially non-raft or raft-associated in MDCK or Caco-2 cells (14,18), and our results using HT-29 5M12 cells suggest that DPP-IV is raft-associated along a transient period of its biosynthetic pathway, i.e. in the Golgi apparatus and carrier vesicles, and subsequently stabilized in non-raft domains. Despite detergent insolubility and recruitment in budding carrier vesicles, DPP-IV failed to be transported to the apical membrane in GalNAc␣-O-bn-treated cells and appeared to follow in turn a degradation process, showing that carrier vesicles could not be normally processed for apical delivery, i.e. for transport along the microtubule and/or actin networks or for docking/fusion with the apical plasma membrane.
In this regard, the fractionation of a detergent extract of a membrane fraction on sucrose gradient showed an impact of GalNAc␣-O-bn on rafts. In control HT-29 5M12 cells, DPP-IV, annexin XIIIb, and an ubiquitous marker of raft domains, i.e. flotillin-1/Reggie-2 (46,47) were found in the typical raft fraction and also in non-raft membrane fractions. In GalNAc␣-Obn-treated HT-29 5M12 cells, (i) these raft-associated proteins shifted to a detergent-insoluble fraction of lower buoyant density, characterized by a different lipid composition in comparison to rafts of control cells (the total amount of glycolipids is decreased by 50% by GalNAc␣-O-bn treatment, and their composition is changed, the major feature being the complete disappearance of GM1 ganglioside) 2 and (ii) DPP-IV appeared 10-fold enriched in the lipid-rich microdomains, whereas annexin XIIIb appeared 10-fold depleted. The enrichment of DPP-IV in a detergent-insoluble fraction was in accordance with the pattern observed by biosynthetic labeling, showing that DPP-IV failed to reverse in non-raft membranes but was in turn degraded. In contrast, annexin XIIIb showed an inability to associate with the lipid microdomains of GalNAc␣-O-bntreated cells, and such finding might be connected to the abnormal apical delivery, as annexin XIIIb has been proposed to play a function in apical delivery either as an adaptor for the binding of apical carrier vesicles to the microtubule motors or as a factor involved in the docking and fusion machinery. Until now, very few data have been available concerning the mechanisms involved in the transport of apical carrier vesicles and in the docking/fusion process. The association of annexin XIIIb with a protein of the kinesin superfamily (KIF), the microtubule minus end-directed motor KIFC3, in Triton-insoluble membranes in MDCK cells was recently reported (48). However, annexin XIIIb is not directly associated with KIFC3 but involves other unknown intermediate partners (48). Our data can be connected to those obtained with the unmyristoylated recombinant form of annexin XIIIb, which inhibited the apical delivery without affecting the release of apical carriers (40,49,50). However, unmyristoylated annexin XIIIb was not bound to the membrane but exclusively present in the cytosolic fraction (50). Beyond the requirement of the myristoyl moiety for mem-brane binding (50), our data suggest that in HT-29 5M12 cells, the association of membrane-bound annexin XIIIb with lipid microdomains involves its interaction with sialylated epitopes, as we showed that GalNAc␣-O-bn inhibits sialylation of glycoproteins and likely glycolipids by the two sialyltransferases ST3Gal I and ST3Gal IV (22,23).
In conclusion, our data bring a new insight on the effect of GalNAc␣-O-bn in the apical trafficking in polarized HT-29 cells. The fact that undersialylated DPP-IV is incorporated into apical transport vesicles indicates that the sialylation is not essential as an apical targeting signal. The defect is in transport or docking/fusion of apical carrier vesicles rather than impaired packaging of undersialylated DPP-IV. In this context, our data show the role of glycosylation in the organization of raft microdomains and their functions in the transport of carrier vesicles to the apical membrane and/or in the docking/ fusion machinery.