Benzyl-N-acetyl-α-d-galactosaminide Induces a Storage Disease-like Phenotype by Perturbing the Endocytic Pathway*

The sugar analogO-benzyl-N-acetyl-α-d-galactosaminide (BG) is an inhibitor of glycan chain elongation and inhibits α2,3-sialylation in mucus-secreting HT-29 cells. Long-term exposure of these cells to BG is associated with the accumulation of apical glycoproteins in cytoplasmic vesicles. The mechanisms involved therein and the nature of the vesicles have not been elucidated. In these cells, a massive amount of BG metabolites is synthesized. Because sialic acid is mainly distributed apically in epithelial cells, it has been proposed that the BG-induced undersialylation of apical membrane glycoproteins is responsible for their intracellular accumulation due to a defect in anterograde traffic and that sialic acid may constitute an apical targeting signal. In this work, we demonstrate that the intracellular accumulation of membrane glycoproteins does not result mainly from defects in anterograde traffic. By contrast, in BG-treated cells, endocytosed membrane proteins were retained intracellularly for longer periods of time than in control cells and colocalized with accumulated MUC1 and β1 integrin in Rab7/lysobisphosphatidic acid+ vesicles displaying features of late endosomes. The phenotype of BG-treated cells is reminiscent of that observed in lysosomal storage disorders. Sucrose induced a BG-like, lysosomal storage disease-like phenotype without affecting sialylation, indicating that undersialylation is not a requisite for the intracellular accumulation of membrane glycoproteins. Our findings strongly support the notion that the effects observed in BG-treated cells result from the accumulation of BG-derived metabolites and from defects in the endosomal pathway. We propose that abnormal subcellular distribution of membrane glycoproteins involved in cellular communication and/or signaling may also take place in lysosomal storage disorders and may contribute to their pathogenesis.


The sugar analog O-benzyl-N-acetyl-␣-D-galactosaminide (BG) is an inhibitor of glycan chain elongation and inhibits ␣2,3-sialylation in mucus-secreting HT-29 cells.
Long-term exposure of these cells to BG is associated with the accumulation of apical glycoproteins in cytoplasmic vesicles. The mechanisms involved therein and the nature of the vesicles have not been elucidated. In these cells, a massive amount of BG metabolites is synthesized. Because sialic acid is mainly distributed apically in epithelial cells, it has been proposed that the BG-induced undersialylation of apical membrane glycoproteins is responsible for their intracellular accumulation due to a defect in anterograde traffic and that sialic acid may constitute an apical targeting signal. In this work, we demonstrate that the intracellular accumulation of membrane glycoproteins does not result mainly from defects in anterograde traffic. By contrast, in BGtreated cells, endocytosed membrane proteins were retained intracellularly for longer periods of time than in control cells and colocalized with accumulated MUC1 and ␤ 1 integrin in Rab7/lysobisphosphatidic acid ؉ vesicles displaying features of late endosomes. The phenotype of BG-treated cells is reminiscent of that observed in lysosomal storage disorders. Sucrose induced a BGlike, lysosomal storage disease-like phenotype without affecting sialylation, indicating that undersialylation is not a requisite for the intracellular accumulation of membrane glycoproteins. Our findings strongly support the notion that the effects observed in BG-treated cells result from the accumulation of BG-derived metabolites and from defects in the endosomal pathway. We propose that abnormal subcellular distribution of membrane glycoproteins involved in cellular communication and/or signaling may also take place in lysosomal storage disorders and may contribute to their pathogenesis.
Glycosylation plays a fundamental role in various biological processes such as the folding, oligomerization, and biochemical maturation of proteins, and glycans play an important role in cell-cell and cell-matrix adhesion during development and tumor progression and in host-pathogen interactions, among others (1,2). The use of sugar analogs has been fundamental for the study of glycan function, mainly in the study of N-glycans, as no specific inhibitors of O-glycosylation are currently available (2).
Benzyl-N-acetyl-␣-D-galactosaminide (BG), 1 as well as other GalNAc analogs, was initially reported to inhibit O-glycosylation, presumably as a result of the inhibition of GalNAc-O-Ser/ Thr elongation (3). More recent studies of cells treated with BG or other GalNAc analogs have shown that their overall effects are likely due to complex modifications of glycan biosynthesis (4 -10). A detailed study of mucus-secreting HT-29 colon cancer cells treated chronically with BG revealed dramatic and pleiotropic effects, including a reduction in cell proliferation, an increase in cell size, and the accumulation of electron-lucid cytoplasmic vesicles containing predominantly undersialylated apical glycoproteins with N-and O-glycans (i.e. dipeptidyl peptidase IV and MUC1) (8). In HT-29 glycoproteins, sialic acid is mainly linked in an ␣2,3-configuration, and ␣2,3-sialyltransferase activity is profoundly inhibited upon treatment with BG. This is accompanied by a marked decrease in the sialylation of glycoproteins, detected with the ␣2,3-sialic acid-specific Maackia amurensis lectin (MAL), and an increase in the levels of glycoprotein-associated Gal-GalNAc precursor epitope detected with peanut agglutinin (PNA; Arachis hypogaea) (4,5,8). In these cells, changes in sialylation are a consequence of the synthesis of BG-derived metabolites, such as benzyl-Gal-NAc-Gal, that inhibit sialyltransferases, thus leading to the accumulation of the precursor carbohydrate chains of apical glycoproteins (5,6). Based on these findings, on the apical distribution of sialic acid in epithelial cells (11), and on the proposed role of N-and O-glycans in apical targeting (12,13), it was suggested that the altered subcellular distribution of apical glycoproteins and the accumulation of cytoplasmic vesicles in BG-treated cells might result from glycoprotein undersialylation and that sialic acid might play a role in apical targeting (8,9). We have shown that BG also perturbs the processing of lysosomal enzymes such as acid ␣-glucosidase (AAG) and cathepsin D, indicating broader effects of BG on intracellular traffic (14). However, the available evidence did not allow us to unravel several key issues related to the biochemical effects of this drug: 1) no definite causal relationship has been established between the effects on sialylation and the abnormal subcellular distribution of membrane glycoproteins, and 2) there are few data supporting the contribution of an anterograde (Golgi-to-plasma membrane) traffic defect versus endo-cytic pathway defects leading to the vesicular accumulation of membrane glycoproteins in cells treated with BG for prolonged periods of time. In this regard, the best evidence for the contribution of an anterograde traffic defect comes from the analysis of AAG maturation in BG-treated cells, which showed a blockade in a post-trans-Golgi network compartment (14).
Our finding that BG treatment results in misprocessing of lysosomal enzymes (14) is reminiscent of the effects of sucrose on fibroblasts (15) and has led us to propose that BG could induce a reversible lysosomal storage disease-like phenotype in HT-29 cells (14). Lysosomal storage diseases are a diverse group of inherited conditions characterized by functional defects in endosomes/lysosomes leading to the accumulation of intracellular vesicles. They are caused by mutations in a wide variety of genes and are associated with the accumulation of cholesterol and/or sphingolipids in late endosomes/lysosomes (16). A misrouting of lactosylceramide and possibly other lipids appears to be characteristic of storage diseases, although the precise mechanisms involved are unknown (17).
The aims of this work have been 1) to analyze the contribution of abnormalities in endocytic traffic to the BG phenotype, 2) to determine whether a causal relationship exists between the undersialylation of membrane glycoproteins induced by BG and their intracellular accumulation, and 3) to provide additional support to the hypothesis that BG induces a lysosomal storage-like phenotype. Our findings indicate that in HT-29 cells, as well as in IMIM-PC-1 pancreatic cancer cells, the "BG phenotype" (i.e. dramatic accumulation of cytoplasmic vesicles containing membrane glycoproteins) 1) is due, to a large extent, to defects in the endocytic pathway; 2) does not depend on a defect in glycoprotein sialylation; 3) is associated with the accumulation of intracellular vesicles with variable phenotype, including vesicles containing late endosomal markers as well as endocytosed membrane proteins, as in storage diseases; and 4) is similar to that induced by prolonged treatment with sucrose, except that the latter does not perturb glycoprotein sialylation. Altogether, these results support the notion that defective sialylation is not required for the abnormal subcellular distribution of membrane glycoproteins induced by BG and that the latter may also take place in lysosomal storage diseases and contribute to their pathogenesis.

MATERIALS AND METHODS
Cell Culture-HT-29 colon cancer cells selected with 10 Ϫ6 M methotrexate (here designated as HT-29 M6 cells) were obtained from Drs. Alain Zweibaum and Thécla Lesuffleur (INSERM, Paris, France). IMIM-PC-1 pancreatic cancer cells were established in our laboratory (18). The A13 clone was obtained by seeding IMIM-PC-1 cells at 100 cells/plate in 10-cm diameter dishes and isolating single cell-derived populations with cloning cylinders. Human infantile sialic acid storage disease (ISSD) skin fibroblasts (GM5520) and control skin fibroblasts from the unaffected parents (GM5521) were kindly provided by Dr. Josef Glössl (University of Agricultural Sciences, Vienna, Austria) (15). HT-29 M6 and IMIM-PC-1 cells were seeded on plastic at 2 ϫ 10 4 and 5 ϫ 10 3 cells/cm 2 , respectively. All cells were maintained in Dulbecco's modified Eagle's medium (Invitrogen, Glasgow, UK) supplemented with 10% fetal bovine serum (Invitrogen) at 37°C in a 5% CO 2 atmosphere. Unless specified otherwise, when cells were cultured in the presence of BG (Sigma), this drug was added to the medium 24 h after seeding at a final concentration of 2 mM.
Transmission Electron Microscopy-Transmission electron microscopy was performed on cells grown on plastic as previously described (18). Samples embedded in Epon (Polysciences Inc., Warington, PA) were re-embedded to make sections perpendicular to the bottom of the flask. Ultrathin sections were visualized using a Phillips CM100 electron microscope.
Fluorescence Microscopy-For double immunofluorescence staining, cells were grown on coverslips and fixed with 4% paraformaldehyde for 10 min at room temperature, incubated with 50 mM NH 4 Cl for 30 min, and permeabilized with 0.5% Triton X-100 or 0.1% saponin in phosphate-buffered saline (PBS)/bovine serum albumin (BSA) for 30 min. In the case of reactions employing antibody 6C4, the permeabilization step was carried out using 0.05% saponin in PBS/BSA. Cells were then sequentially incubated with primary antibodies and fluorescein isothiocyanate (FITC)-conjugated goat anti-mouse Ig or TRITC-conjugated goat anti-rabbit Ig (Dako, Glostrup, Denmark); antibodies were added for 1 h, followed by three washes with PBS. Biotin-conjugated MAL (20 g/ml) was added in lectin buffer (50 mM Tris-HCl (pH 7.5), 15 mM KCl, 5 mM MgCl 2 , and 0.1% saponin) for 1 h, followed by FITC-streptavidin (5 g/ml). Fluorescence microscopy analysis was performed using a Leica TCS-SP2 confocal unit.
To detect cholesterol, cells were grown on coverslips, fixed with 4% paraformaldehyde for 10 min at room temperature, and stained with filipin (Sigma) at 125 g/ml in PBS for 30 min. Fluorescence analysis was performed using conventional fluorescence microscopy.
Metabolic Labeling and Immunoprecipitation-Cells were maintained in methionine-free minimal essential medium containing 10% dialyzed fetal bovine serum for 30 min, pulse-labeled for 1 h with 50 Ci/ml [ 35 S]Met/Cys (Tran 35 S-label, ICN Biomedicals, Costa Mesa, CA) in the same medium, and chased with Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and 2 mM unlabeled methionine for 24 h. In the case of cells treated with BG, the culture medium was supplemented with the drug at the concentration indicated. Cells were washed three times with PBS and lysed with 50 mM Tris-HCl (pH 8), 1% Triton X-100, 62.5 mM EDTA, 2 mM Pefabloc, 1 g/ml aprotinin, and 1 g/ml leupeptin. Lysates were centrifuged at 10,000 ϫ g for 30 min, and supernatants were precleared with preimmune rabbit antiserum for 2 h and protein A-Sepharose (Roche Molecular Biochemicals) for 30 min. Antibodies were added to the precleared supernatants for 3-16 h in the presence of 0.2% SDS. When using mAbs, rabbit anti-mouse Ig (Dako) was added for 2 h prior to the incubation with protein A-Sepharose. Immunoprecipitates were washed three times with radioimmune precipitation assay (RIPA) buffer (10 mM Tris-HCl, 150 mM NaCl, 0.1% SDS, 1% deoxycholate, and 1% Nonidet P-40), three times with high salt buffer (10 mM Tris-HCl, 0.5 mM NaCl, 1 mM EDTA, 0.5% Nonidet P-40, and 0.1% SDS), and twice with PBS. All immunoprecipitation steps were carried out at 4°C. Immunoprecipitates were resuspended in sample buffer, resolved by SDS-PAGE, and developed by fluorography.
Quantitation of Protein Secretion-Cells were metabolically labeled as described above with 150 Ci/ml [ 35 S]Met/Cys for 1 h and chased for 24 h at 37°C; the medium was collected, and cells were lysed with RIPA buffer. Proteins from the medium and cell extracts were precipitated in 10% trichloroacetic acid for 1 h on ice. Samples were centrifuged at 13,000 rpm for 15 min; pellets were resuspended in 1 M NaOH; and labeled material was quantified in a scintillation counter. The proportion of secreted material was calculated as total secreted trichloroacetic acid-precipitable cpm/total cell lysate-associated trichloroacetic acidprecipitable cpm.
Endocytosis Assays-Cells were incubated on ice with 0.5 mg/ml sulfosuccinimidyl 2-(biotinamido)ethyl-1,3Ј-dithiopropionate in PBS, 1 mM MgCl 2 , and 1 mM CaCl 2 for 15 min to label membrane proteins; subsequently, cold Dulbecco's modified Eagle's medium was added for 15 min as a blocking reagent. Cells were incubated at 37°C for the indicated periods of time with Dulbecco's modified Eagle's medium supplemented with 10% prewarmed fetal bovine serum. Cells were placed on ice, and biotin bound to membrane proteins was stripped by incubating the cells twice in cold 60 mM glutathione in PBS for 20 min. Cells were lysed with RIPA buffer supplemented with protease inhibitors; lysates were centrifuged at 13,000 rpm for 30 min at 4°C; and supernatants were fractionated by SDS-PAGE and transferred to a nitrocellulose membrane. Biotinylated proteins were detected with peroxidase-conjugated streptavidin (Zymed Laboratories Inc., South San Francisco, CA). The proportion of biotinylated proteins was determined by densitometric analysis using GenQuant software (Amersham Biosciences).
Lectin Blotting-Cells were lysed in 50 mM Tris-HCl (pH 8.0), 62.5 mM EDTA, and 1% Triton X-100, supplemented with protease inhibitors, and lysates were centrifuged at 13,000 rpm for 30 min at 4°C. Protein extracts were fractionated by 6% SDS-PAGE and transferred to a nitrocellulose membrane. MAL blotting was carried out with reagents from the digoxigenin glycan differentiation kit (Roche Molecular Biochemicals) according to the manufacturer's instructions. Lectin reactivity of MUC1 molecules immunoprecipitated from IMIM-PC-1 cell lysates was examined as described (8).
Immunoblotting-Cells were lysed in RIPA buffer supplemented with protease inhibitors. Cell extracts (30 g) were fractionated by SDS-PAGE and transferred to a nitrocellulose membrane. Filters were blocked with 5% skim milk for 30 min, incubated with primary antibodies for 1 h, washed, and incubated with peroxidase-conjugated goat anti-rabbit or anti-mouse Ig antibodies. Proteins were detected by the enhanced chemiluminescence system (Amersham Biosciences) according to the manufacturer's instructions.

RESULTS
Treatment of HT-29 M6 cells with BG leads to the accumulation of undersialylated apical glycoproteins in cytoplasmic vesicles. The interpretation of these effects has relied upon the assumption that BG induces a blockade in the anterograde traffic, leading to the vesicular accumulation of apical glycoproteins en route to the plasma membrane (8). To obtain insight into the rate of overall anterograde protein traffic, HT-29 M6 cells were cultured on Transwells, and the impermeability of the cell layers was assessed using [ 14 C]mannitol. BG was or was not added to the medium of impermeable cell cultures for 7 days; cells were metabolically labeled with [ 35 S]Met/Cys; apical and basolateral media and labeled cells were collected; and total constitutive protein secretion and cell-associated proteins were quantified. The secreted/cell-associated label ratio was expressed as an index. There was no significant inhibition of total apical or basolateral protein secretion (Fig. 1A).
To obtain additional evidence that BG has no significant effect on total protein secretion, we performed similar experiments using IMIM-PC-1 pancreatic cancer cells. This cell line was selected from a panel of lines in which the effects of prolonged BG treatment had been tested. 2 Upon treatment with BG, IMIM-PC-1 cells displayed a phenotype that is similar to that of BG-treated HT-29 M6 cells and is characterized by reduced cell proliferation, increased cell volume, reduced ␣2,3glycoprotein sialylation, and accumulation of electron-lucid cytoplasmic vesicles (Fig. 2, A-D). Unlike HT-29 M6 cells, BGtreated IMIM-PC-1 cells displayed the intracellular accumulation of both apical (i.e. MUC1) and basolateral (i.e. ␤ 1 integrin) glycoproteins (Fig. 2E). As shown for HT-29 M6 cells, BG did not significantly decrease total protein secretion in IMIM-PC-1 cells (Fig. 1B). Therefore, the BG-induced cytoplasmic accumulation of membrane glycoproteins is not primarily due to alterations in anterograde transport.
Endocytosed Membrane Glycoproteins Accumulate in Cytoplasmic Vesicles in BG-treated Cells-To determine whether intracellular glycoprotein accumulation results from altered endocytosis/recycling of plasma membrane glycoproteins, IMIM-PC-1 cells were used because they are flatter and allow a better morphological resolution than HT-29 M6 cells and because they express lower levels of membrane-associated mucins that impair cell-surface biotinylation. IMIM-PC-1 membrane proteins were biotinylated at 4°C and allowed to internalize at 37°C for various periods of time. Biotin that remained bound to plasma membrane glycoproteins that had not been endocytosed was dissociated using glutathione, so only the internalized glycoproteins originating from the plasma membrane could bind streptavidin. Endocytosis of biotinylated membrane proteins was analyzed and quantified by Western blotting with peroxidase-conjugated streptavidin (Fig. 3A). In addition, the subcellular distribution of biotinylated glycoproteins was examined using FITC-streptavidin by confocal fluorescence microscopy (Fig. 3B). In control and BG-treated cells, labeled proteins were distributed mainly at the plasma membrane 10 min after shifting the temperature to 37°C, and intracellular labeling increased progressively during the first 2 F. Ulloa and F. X. Real, unpublished data. Cells were labeled with [ 35 S]Met/Cys for 1 h; apical and basolateral media were collected; and cells were lysed. Proteins present in the culture medium and cell lysates were precipitated in trichloroacetic acid, and radioactivity was determined using a scintillation counter. B, IMIM-PC-1 cells cultured on plastic were processed as described for A, except that they were cultured on plastic, as they are not polarized and do not form impermeable monolayers.
2 h. Subsequently, a decrease in cytoplasmic labeling was observed only in control cells. In BG-treated cells, intracellular labeling progressively increased during the first 30 min after the temperature shift, and an intracellular accumulation of labeled proteins was demonstrated for up to 24 h. Quantitative analysis showed that endocytosis occurred at a relatively slower rate in BG-treated cells during the first 2 h, whereas intracellular accumulation of biotinylated glycoproteins occurred to a greater extent than in control cells (Fig. 3, A and B). At 24 h, BG-treated cells showed a 5.5-fold higher accumulation of membrane proteins compared with control cells. These results indicate that the phenotype of BG-treated cells involves the intracellular accumulation of endocytosed membrane glycoproteins.
Endocytosed Membrane Glycoproteins Accumulate in Vesicles Containing MUC1 and ␤ 1 Integrin-To determine whether compartments in which MUC1 and ␤ 1 integrin accumulate also contain endocytosed biotinylated membrane proteins, cells were cultured for 8 days in the presence of BG, surface-biotinylated at 4°C, and incubated at 37°C for various periods of time. As described above, in BG-treated IMIM-PC-1 cells, MUC1 and ␤ 1 integrin colocalized and accumulated in cytoplasmic vesicles (Fig. 4). A progressive increase in the amount of biotinylated membrane glycoproteins was observed in MUC1and ␤ 1 integrin-containing vesicles starting 30 min after initiation of internalization for up to 24 h (data not shown). At the 4-h time point, numerous vesicles showed double labeling of endocytosed membrane proteins with either MUC1 or ␤ 1 integrin, whereas others contained exclusively one type of protein marker (Fig. 4). These findings indicate that the cytoplasmic vesicles that accumulate in BG-treated cells indeed receive endocytosed membrane material. MUC1 and ␤ 1 integrin have been shown to be recycled to the plasma membrane (28,29), strongly suggesting that the cytoplasmic accumulation of these proteins is secondary to defective recycling.
Cytoplasmic Vesicles Exhibit Markers Typical of Late Endosomes-To obtain insight into the nature of the cytoplasmic vesicles, double immunolabeling for MUC1 or ␤ 1 integrin and markers corresponding to the Golgi complex (GRASP65 and TGN46) (30), early endosomes (EEA1 and Rab5) (31,32), and late endosomes (LBPA and Rab7) (20, 34) was performed. Results are shown for ␤ 1 integrin colocalization experiments, and similar findings were obtained with anti-MUC1 antibodies. Cells were cultured for 8 days in the presence or absence of BG,

FIG. 3. BG treatment induces marked changes in the internalization rate of cell-surface glycoproteins in IMIM-PC-1 cells.
A, cells were cultured for 8 days in the absence or presence of 2 mM BG, and membrane proteins were labeled with sulfosuccinimidyl 2-(biotinamido)ethyl-1,3Ј-dithiopropionate at 4°C. Subsequently, membrane internalization was allowed for various periods of time by shifting cultures to 37°C, and plasma membrane-associated biotin was released by adding 60 mM glutathione in PBS. Cells were lysed with RIPA buffer, and lysates were fractionated by SDS-PAGE. Biotinylated proteins were detected using peroxidase-conjugated streptavidin, and the signal was quantitated by densitometry. B, cells were cultured as described for A, and the distribution of sulfobiotin-labeled proteins was analyzed using FITC-streptavidin after incubating cells for various periods of time at 37°C. The two panels corresponding to the 10-min incubation time show the labeling pattern observed in different cells. Scale bar ϭ 10 m.
fixed, and permeabilized. Each of these markers displayed the expected distribution in control cells (Fig. 5A, upper panels). Immunofluorescence microscopy analysis of BG-treated cells showed no colocalization of MUC1 or ␤ 1 integrin with either Golgi or early endosome markers (Fig. 5). The MAL proteolipid, a proteolipid that is present in apical carrier vesicles, did not colocalize with intracellularly accumulated MUC1 or ␤ 1 integrin, and MAL proteolipid expression levels were very low in both control and BG-treated cells (data not shown). By contrast, considerable co-labeling of late endosomal markers and MUC1 or ␤ 1 integrin was demonstrated, although ␤ 1 integrin ϩ / Rab7 Ϫ and ␤ 1 integrin Ϫ /Rab7 ϩ vesicles were also present in treated cells, emphasizing the heterogeneous nature of the cytoplasmic vesicles (Fig. 5A). The transferrin receptor, a basolateral protein subject to extensive recycling, also colocalized with MUC1 or ␤ 1 integrin in some vesicles (Fig. 5A). Furthermore, BG-treated cells showed a marked increase in the total amount of Rab7 ϩ and LBPA ϩ vesicles and a slight increase in the amount of Rab5 ϩ vesicles. Western blot analysis confirmed increased levels of Rab7 in BG-treated cells (Fig. 5B). Because late endosomes are involved in the regulation of cholesterol transport (34), we also examined cholesterol distribution in cells using filipin (35). Fig. 5C shows an increase in filipin labeling upon treatment of IMIM-PC-1 cells with BG. Similar results were obtained using HT-29 M6 cells. Altogether, these results indicate that a large proportion of the cytoplasmic vesicles that accumulate in BG-treated cells share molecular features with late endosomes.

BG-treated Cells Display Ultrastructural and Biochemical Properties Reminiscent of Those of Sialic Acid Storage Diseases-
We have previously shown that HT-29 M6 cells treated with BG display an abnormal processing of lysosomal enzymes, viz. AAG and cathepsin D (14). An abnormal processing of cathepsin B has also been demonstrated in fibroblasts from patients with ISSD (15), a recessive syndrome caused by mutations in the lysosomal sialic acid transporter (36,37) leading to an accumulation of sialic acid in lysosomes (38). Therefore, we have proposed that similar mechanisms may contribute to the abnormal processing of the lysosomal enzymes observed in BG-treated HT-29 M6 cells and in ISSD cells, as the former synthesize massive amounts of sialylated BG-derived metabolites.
To further analyze this similarity, we examined AAG processing in fibroblasts from one patient with ISSD and from the patient's unaffected parents. AAG is synthesized as a precursor with a molecular mass of ϳ110 kDa, which is subsequently processed to intermediate and mature forms, which are the major species detected after 1 h of pulse and 24 h of chase (19). In normal fibroblasts, only the precursor AAG form was detected after 1 h of labeling; after 4 h of chase, only the intermediate AAG form was detected; and after 24 h of chase, both the intermediate and mature AAG forms were detected (Fig. 6) (19). We have previously reported a delay in AAG processing in untreated HT-29 M6 cells, so the three forms are detected after 24 h of chase (39). By contrast, in BG-treated HT-29 M6 cells labeled for 1 h and chased for 24 h, anti-AAG antibodies immunoprecipitated molecules with a mobility that was intermediate to that of the precursor and intermediate forms of AAG (AAG BG ), the abnormal AAG form found in cells cultured in the presence of BG (14). This molecular species has not been described in normal cells. ISSD fibroblasts showed defective maturation kinetics of AAG and the transient accumulation of an abnormally processed AAG species, designated AAG ISSD , with a mobility similar to that of AAG BG (Fig. 6). This species has not been reported in fibroblasts from normal individuals and was not found in fibroblasts from the unaffected parents (Fig. 6). Therefore, similar defects in lysosomal protein processing are present in BG-treated cells and in ISSD cells.
BG and Sucrose Induce the Accumulation of Membrane Glycoprotein-containing Cytoplasmic Vesicles-The results described above and those from the literature suggest that high concentrations of sialic acid or sialic acid-containing metabolites in endosomal/lysosomal compartments are associated with an accumulation of cytoplasmic vesicles. In BG-treated HT-29 cells, BG-derived metabolites accumulate up to 1 mg of hexose/mg of protein (40). To examine whether the acquisition of this phenotype is exclusive to this compound and whether reduced sialylation is a prerequisite, the effects of sucrose were studied. In normal fibroblasts, sucrose (50 -100 mM) induces a lysosomal storage-like phenotype characterized by the intracellular accumulation of electron-lucid vesicles, designated sucrosomes (41,42), as well as defects in cathepsin B processing (15).
IMIM-PC-1 cells were cultured for 8 days in medium containing 50 or 100 mM sucrose, and the morphology and subcellular distribution of MUC1, ␤ 1 integrin, and sialic acid were analyzed. At the phase-contrast level, sucrose-treated cells showed morphological changes similar to those observed in BG-treated cells (Fig. 7A). In addition, confocal fluorescence microscopy demonstrated the intracellular accumulation of MUC1 and ␤ 1 integrin in vesicles (Fig. 7, B and C), especially at 100 mM. Lectin cytochemistry showed that, unlike BG, sucrose had no effects on glycoprotein sialylation. Reactivity with MAL and PNA was unchanged (Fig. 7D), and vesicles in which MUC1 accumulated showed strong reactivity with MAL (Fig. 7E).
Altogether, these findings indicate that BG and sucrose induce a similar storage disease-like phenotype characterized by the intracellular accumulation of vesicles containing membrane glycoproteins. The effects of BG, but not those of sucrose, are associated with reduced sialylation, as determined by MAL reactivity of cellular glycoproteins. These findings demonstrate that a blockade in sialylation is not required for the accumulation of membrane glycoproteins in cytoplasmic vesicles. DISCUSSION BG has been extensively used as an inhibitor of O-glycosylation in a variety of cell types (4, 5-7, 10, 43-45). Because no other drugs that effectively inhibit this process have been identified, it is important to unravel the mechanisms through which BG acts. Recent work has shown that, in fact, the effects of BG are much more complex than initially thought. (i) Acute (24 h) as well as prolonged (several days) exposure to BG is mainly associated with the inhibition of sialylation rather than with a blockade in O-glycan synthesis; (ii) both N-and Oglycans are affected by the drug; and (iii) in some cells, BG is used as a substrate by glycosyltransferases, and BG-derived metabolites are synthesized in large amounts, leading to their accumulation in cells and to their secretion to the culture medium (40). On the basis of these findings, it has been proposed that the accumulation of cytoplasmic vesicles containing membrane glycoproteins is due to defective sialylation of these proteins, leading to a blockade in their membrane targeting, and that sialic acid might constitute a membrane targeting signal (8).
Anterograde Traffic Versus Endocytosis/Recycling-In this work, we provide extensive evidence against the hypothesis FIG. 5. BG induces the predominant accumulation of glycoproteins in vesicles displaying markers of late endosomes. A, late endosomal markers colocalize with ␤ 1 integrin in BG-treated cells. IMIM-PC-1 cells were cultured for 8 days in the absence (CONTROL) or presence of 2 mM BG, fixed with 4% paraformaldehyde, permeabilized with 0.5% Triton X-100 in PBS and 1% BSA (or 0.05% saponin in PBS and 1% BSA in the case of LBPA), and incubated with antibodies detecting ␤ 1 integrin and the indicated molecules. The Golgi marker GRASP65 and the early endosomal marker EEA1 did not colocalize with intracellularly accumulated membrane glycoproteins. The transferrin receptor (TfR; a basolateral protein that recycles efficiently), Rab7, and LBPA showed a partial colocalization with ␤ 1 integrin in intracellular vesicles. Scale bar ϭ 12 m (main panels) and 5 m (insets). B, BG-treated cells display an increase in the levels of Rab7 (a late endosomal marker) and a decrease in the levels of EEA1. The levels of p115 (a cis-Golgi marker) were unaffected. Lysates of IMIM-PC-1 cells cultured for 8 days in the absence (Ϫ) or presence (ϩ) of 2 mM BG were fractionated by SDS-PAGE, and total protein levels were analyzed by Western blotting. Ponceau red staining is shown to compare the total amounts of proteins loaded. C, BG induces an increase in the levels of cholesterol and changes in its distribution. Cells were cultured for 8 days in the absence (CONTROL) or presence of BG and incubated with filipin, a molecule that emits fluorescence only when bound to cholesterol.
FIG. 6. Alterations in lysosomal AAG processing in fibroblasts from patients with ISSD are similar to those observed in HT-29 M6 cells treated with BG. Fibroblasts from a patient with ISSD and from the healthy progenitors were metabolically labeled with [ 35 S]Met/ Cys for 1 h, and cell lysates were obtained after chase at the indicated time points. Immunoprecipitates obtained with anti-AAG antibodies were analyzed by SDS-PAGE and developed by autoradiography. The results obtained with fibroblasts from both progenitors were indistinguishable from those obtained with normal fibroblasts. The electrophoretic mobility of normal precursor (p), intermediate (i), and mature (m) AAG forms is indicated. The mobility of the abnormally processed AAG form found in ISSD cells, AAG ISSD , is indicated. described above. First, we have shown that global anterograde protein traffic was not significantly diminished in two cellular models in which BG induced a similar phenotype, HT-29 M6 and IMIM-PC-1 cells. It has previously been shown that BG induces reduced mucin secretion in mucus-secreting HT-29 cells, a finding that led to the proposal of a blockade in anterograde traffic (5,6,9). Nevertheless, mucin secretion requires extensive post-translational modifications as well as appropriate maturation of mucus droplets, and it is likely that defects in these processes associated with BG treatment account for the observed findings (9). By contrast, the majority of membrane glycoproteins do not share such extensive glycosylation and thus appear not to be affected by BG in a similar fashion. To provide an alternative explanation for their intracellular accumulation, we examined the kinetics of endocytosis of membrane glycoproteins and the presence of endocytosed material in the cytoplasmic vesicles. As shown above, we found that, in BG-treated cells, membrane proteins were endocytosed slowly and remained intracellular for longer periods of time, possibly due to defective recycling to the plasma membrane, leading to their intracellular accumulation. In addition, we found that vesicles that accumulated MUC1 and ␤ 1 integrin membrane glycoproteins also contained endocytosed plasma membrane glycoproteins. Overall, these findings strongly support a predominant role of abnormal endocytosis, rather than altered anterograde traffic, in the generation of the phenotype induced by BG.
The phenotype of BG-treated cells might be explained by a defect in vesicular fusion events related to late endosomes. Proteins of the SNARE (soluble N-ethylmaleimide-sensitive factor attachment protein receptors) family, Rab proteins (regulators of vesicular traffic), and membrane lipids are candidates to play a role. Dominant-negative Rab7 proteins are able to disrupt lysosome biogenesis (46), suggesting that the overexpression of Rab7 present in BG-treated cells might be involved in the accumulation of cytoplasmic vesicles. The increased levels of LBPA and cholesterol present in BG-treated cells might directly result in changes in the biophysical properties of membranes, including the formation and distribution of rafts; and a role for LBPA in the regulation of the fusogenic properties of late endosomes has been proposed (47). In addition, the accumulation of cholesterol in the degradative compartments of cells treated with U18666A leads to changes in the membrane-to-cytosol recycling of Rab7 (48). Causal Role of Undersialylation in the BG Phenotype-A second tenet of the models proposed to account for the effects of BG has been that sialic acid might act as an apical targeting signal and that undersialylation is responsible for the accumulation of glycoproteins in cytoplasmic vesicles (8). Our data also rule out this hypothesis. First, sucrose treatment of IMIM-PC-1 cells did not affect glycoprotein sialylation, yet caused morphological changes, including the accumulation of MUC1 and ␤ 1 integrin in cytoplasmic vesicles, in a fashion similar to that observed with BG. Second, because IMIM-PC-1 cells display some heterogeneity in their response to BG, we derived clones by limiting dilution and analyzed their sensitivity to BG treatment. One of the clones analyzed, designated A13, displays a polarized phenotype and contains glycoproteins that are mainly sialylated in the ␣2,3-position, as do the parental IMIM-PC-1 population and HT-29 cells. In A13 cells, BG induced a profound reduction of overall ␣2,3-sialylation and MUC1-associated ␣2,3-sialic acid, without concomitant effects on ␣2,6-sialylation, and a marked accumulation of cytoplasmic vesicles. Yet, MUC1 did not accumulate intracellularly and displayed an apical distribution, as in control cells (Fig. 8). Therefore, although changes in sialylation may be associated with changes in the targeting (43) and/or subcellular distribution (8, 10) of specific membrane glycoproteins, sialic acid is not required for the appropriate apical targeting of MUC1. Furthermore, global constitutive apical and basolateral secretion proceeded unaltered in cells showing profound undersialylation.
Our findings are in agreement with data from other laboratories. Chinese hamster ovary Lec2 mutants contain inactive CMP-Neu5Ac transporters and show defective glycoprotein sialylation (49), yet no changes in the distribution of membrane glycoproteins have been described. Similarly, ricin-resistant Madin-Darby canine kidney cells show reduced sialylation of membrane glycoproteins, yet lack a vesiculated cytoplasm (50), again suggesting that sialylation defects are not the primary cause of the misdistribution of membrane glycoproteins.
In view of these findings, how can the effects of BG on HT-29 M6 and IMIM-PC-1 cells be explained? Delannoy et al. (6) initially reported that BG is metabolized to benzyl-GalNAc-Gal, and Zanetta et al. (40) have subsequently shown that HT-29 cells synthesize massive amounts of BG-derived metabolites. Preliminary data indicate that IMIM-PC-1 cells, but not other cells upon which BG has little effect, also produced large amounts of BG-derived metabolites. 3 Therefore, such metabolites are candidates to play a role in the generation of the BG phenotype. If so, one could predict that the levels of metabolites and the effects of BG would depend on at least three factors: BG concentration used, duration of treatment, and catabolism/secretion of the BG-derived metabolites. Although a quantitative assessment of these variables is not yet available, several observations support this notion in a series of experiments shown in Fig. 9, in which dipeptidyl peptidase IV and AAG processing was studied. HT-29 M6 cells were pretreated for various periods of time with 2 mM BG, pulse-labeled with [ 35 S]Met/Cys, chased for 24 h in medium containing 2 mM BG, and used in immunoprecipitation assays. A 30-min preincubation with BG was sufficient to inhibit dipeptidyl peptidase IV sialylation, identified by changes in electrophoretic mobility previously characterized as due to lack of sialic acid (8,14). Under these conditions, AAG processing was unaffected (Fig. 9A). By contrast, pretreatment of cells for 4 -10 days with BG led to predominant processing of AAG to the AAG BG aberrant species. In a similar series of experiments, cells were pretreated for 24 h with increasing concentrations of BG (0.05-2 mM), pulse-labeled, and chased in BG-containing medium at the appropriate concentration, and dipeptidyl peptidase IV and AAG immunoprecipitates were analyzed by SDS-PAGE. In these experiments, 0.5 mM BG led to inhibition of dipeptidyl peptidase IV sialylation, but had no effect on AAG processing; a partial effect on the latter was observed only at 2 mM (Fig. 9B). Overall, these experiments indicate that higher BG concentrations or longer exposure times are required for BG to affect AAG processing than to perturb dipeptidyl peptidase IV sialylation and that the effects of BG on the processing of various types of molecules take place through different mechanisms. Our findings are in agreement with the results reported by Ait-Slimane et al. (43) in Caco-2 cells, upon which BG induced changes in glycoprotein sialylation, but did not lead to the accumulation of intracellular vesicles or to the inhibition of AAG processing, even upon long-term exposure to the drug. 4 Nature of the Intracellular Vesicles-Our data indicate that 3 P. Delannoy, G. Huet, F. X. Real, F. Ulloa, and J. P. Zanetta, unpublished data. 4 C. Francí and F. X. Real, unpublished data. the vesicles accumulating in BG-treated cells share molecular features with late endosomes: they receive internalized membrane material, and MUC1 and ␤ 1 integrin colocalize in vesicles with the late endosomal markers Rab7 and LBPA, but not with early endosomal markers such as EEA1 and Rab5. The acidic nature of the vesicles that accumulate in HT-29 M6 cells, determined using the pH-sensitive marker Lysosensor Green DND-189 (Molecular Probes, Inc., Leiden, The Netherlands) (data not shown), further supports their endosomal nature. Nevertheless, it should be noted that the vesicles that accumulate in both HT-29 and IMIM-PC-1 cells display a considerable degree of molecular heterogeneity, a feature that is also characteristic of late endosomes (47). The working model proposed on the basis of the findings reported above is summarized in Fig. 10.
The accumulation of intracellular vesicles observed in some cells upon exposure to 2 mM BG appears to be closely related to changes in lysosomal enzyme processing and to a storage disease phenotype. Here, we also provide indirect evidence supporting the notion that the accumulation of BG-derived metabolites observed in HT-29 and IMIM-PC-1 cells and the accumulation of sialic acid described in cells from patients with ISSD lead to a similar morphological and molecular phenotype, as determined by the processing of lysosomal enzymes and the increased levels of cholesterol in cells.
The similarities between the cellular and molecular phenotypes of ISSD and BG-treated cells, together with the fact that a dramatic accumulation of plasma membrane glycoproteins takes place in the latter, provide support for a hypothesis that has recently been proposed, but for which there is so far no evidence: changes in the subcellular distribution of plasma proteins might take place in cells from patients with lysosomal FIG. 9. Effects of BG on dipeptidyl peptidase IV and AAG processing in HT-29 M6 cells occur at differential concentrations/times of exposure. A, cells cultured for 10 days in control medium were maintained for various periods of time (30 min to 10 days (d)) in medium containing 2 mM BG prior to metabolic labeling. Subsequently, cells were pulse-labeled with [ 35 S]Met/Cys for 1 h and chased for 24 h in medium lacking (Ϫ) or containing (ϩ) BG. Immunoprecipitates obtained with specific antibodies were fractionated by SDS-PAGE and revealed by fluorography. Pretreatment of cells with 2 mM BG for 30 min (plus the 24-h chase) was sufficient to reduce dipeptidyl peptidase IV sialylation. By contrast, at least a 4-day pretreatment with BG was necessary to induce changes in AAG processing. B, cells cultured for 10 days in control medium were maintained in medium containing increasing concentrations of BG for 24 h, pulse-labeled with [ 35 S]Met/Cys for 1 h, and chased for 24 h. Immunoprecipitates were isolated and analyzed as described for A. An effect on the migration of dipeptidyl peptidase IV (DPP-IV), likely to represent sialylation, was observed at 0.5 mM BG, whereas partial effects on AAG processing occurred only at 2 mM. The migration of the precursor (p), intermediate (i), and mature (m) forms of AAG is indicated. The migration of the abnormally processed AAG from BG-treated cells, AAG BG , is indicated.
FIG. 10. A working model to account for the effects of BG. A, BG enters cells (step 1) and is metabolized to benzyl-GalNAc-Gal by a galactosyltransferase (step 2), presumably in the Golgi complex. There, this molecule is used as an acceptor for a variety of glycosyltransferases (i.e. ST3 Gal␤1-3GalNAc ␣2,3-sialyltransferase I), leading to the synthesis of complex metabolites such as benzyl-GalNAc-Gal-Neu5Ac (step 3), which compete for terminal processing of glycans in glycoproteins (step 4). The BG-derived metabolites may be transported to a variety of subcellular compartments, including those distal from the trans-Golgi network (TGN) such as endosomes (step 5). B, the massive amounts of metabolites produced in certain cells (i.e. HT-29 and IMIM-PC-1) accumulate in late endosomes/lysosomes, resulting in the swelling of degradative compartments (step 6), possibly through osmotic effects, leading to changes in pH and function, altered maturation and transport (step 7), and the resulting formation of BG-associated cytoplasmic vesicles with mixed characteristics of late endosomes and lysosomes (step 8). Such vesicles are enriched in late endosomal markers such as LBPA and Rab7, contain cholesterol, and may display altered fusion properties. C, membrane glycoproteins are appropriately targeted, but regulated secretion may be partially blocked (step 9), e.g. in mucus-producing cells. Membrane glycoproteins are endocytosed, reach early endosomes, and are then targeted to BG vesicles (step 10), where they accumulate. Apical and basolateral glycoproteins appear to be differentially affected in various cells. storage diseases. MUC1, ␤ 1 integrin, and CD44 are some of the membrane glycoproteins that have been shown to display a subcellular misdistribution in BG-treated cells. All these proteins participate in cellular communication processes, including cell-cell and cell-matrix adhesion. It is conceivable that altered plasma membrane expression of these as well as other not yet studied membrane proteins as a result of the alterations in traffic that are typical of lysosomal storage diseases might secondarily contribute to the pathogenesis of these disorders. Simons and Gruenberg (16,33) have proposed that the amount of raft lipids tolerated by raft endosomes is limited and that the accumulation of one type of raft lipids, cholesterol or sphingolipids, would cause the accumulation of the other type of lipid, eventually leading to a traffic jam that contributes to the cellular and tissue phenotype associated with these conditions. A better understanding of these molecular events might provide valuable information for the treatment of lysosomal storage disorders. In this regard, the reversible nature of the effects of BG on cells might be of help in identifying pharmacological strategies for the systemic therapy of these disorders.