INTRODUCTION

N-Linked glycosylation of proteins begins at the lumen of the rough endoplasmic reticulum, where a subset of Asn-X-Ser/Thr residues on newly synthesized proteins is subjected to addition with Glc₃Man₉GlcNAc₂. The oligosaccharides are then remodeled or processed as newly synthesized glycoproteins, and transported through the Golgi compartments toward the cell surface (1, 2). The oligosaccharide structures on the cell surface undergo significant changes during embryogenesis, differentiation, and malignant transformation (3, 4). Although there have been several investigations on oligosaccharides to determine the specific structure of a sugar chain, which is responsible for protein sorting, the authors only reported glycoproteins with or without N-glycan, without referring to a specific structure of oligosaccharides (5, 6). The introduction of N-glycosylation sites into the rat growth hormone leads to apical sorting and delivery of this secretory protein in epithelial cells, suggesting that N-glycans play roles as sorting signals (5). One striking role of N-glycans in the biosynthetic traffic after proper protein folding in the endoplasmic reticulum is the mannose 6-phosphate modification of lysosomal enzymes (7). However, how specific structures of N-glycans are responsible for the regulation of protein sorting has not yet been determined.

UDP-N-acetylglucosamine:-D-mannoside--1,4-N-acetylglucosaminyltransferase III (GnT-III)¹ is an enzyme that catalyzes the attachment of an GlcNAc residue to 1-4 mannose in the core region of N-glycans, as shown by Scheme I (8).

The product of GnT-III, designated as the bisecting GlcNAc structure, has been described in complex and hybrid oligosaccharides of various glycoproteins (9), and it was found that it inhibits further addition of other sugar chains by other glycosyltransferases such as N-acetylglucosaminyltransferases II, IV, and V, and -1,4-galactosyltransferase (2).

GnT-III activity was first observed in hen oviduct (8) and purified from rat kidney (10). Although the expression of GnT-III is very low in normal rat liver, it is increased during hepatocarcinogenesis in a rodent model (11) and in liver regeneration after partial hepatectomy (12). High GnT-III activity has been reported in many tumor cells, such as Novikoff ascites tumor cells (13), AH-66 hepatoma ascites cells (14), and Huh6 cells (15), and in both sera and livers of patients with liver cancer and cirrhosis (16). These previous reports suggested that GnT-III is related to malignant transformation in the liver. However, the mechanism by which this enzyme is induced in hepatocarcinogenesis and liver regeneration remains unclear and appears to be complicated. In order to determine whether or not bisecting GlcNAc residues are involved in the sorting of N-glycans, a hepatoma cell line, mRLN31 (M31), was treated with forskolin to enhance GnT-III at the transcriptional level.

Forskolin is widely used as an adenylyl cyclase activator to induce different kinds of proteins in many cell lines, through the stimulation of c-AMP (17-19), we found that forskolin is one of the agents that enhances GnT-III in hepatoma cells and normal hepatocytes. Furthermore, to determine the correlation between the oligosaccharide structure of a specific glycoprotein and its distribution, especially after bisecting GlcNAc residues' addition, we examined a number of glycoproteins that have different roles and expression sites, such as Lamp-1 and -glucuronidase as lysosomal proteins, -GTP as a plasma membrane protein, ceruloplasmin, and -fetoprotein as a secretory protein. Lamp-1 is expressed on the surface of many tumor cells (20, 21), although the majority of this molecule resides in lysosomes (22, 23).

We found that the enhanced GnT-III resulted in an increase of bisecting GlcNAc residues of glycoproteins in whole cell lysates but inhibited the glycoproteins sorting on the cell surface, such as that of Lamp-1 and -GTP, while it did not affect the secretion levels of secretory proteins such as ceruloplasmin and -fetoprotein, despite changing the oligosaccharide structures.

MATERIALS AND METHODS

Rat hepatoma cell lines mRLN-31 (M31) and AH66 and human hepatoma cell lines Huh6 and Huh7 were provided by the Japanese Cancer Resources Bank (Tokyo, Japan). A human hepatoma cell line, Hep3B, was provided by the ATCC (American Tissue Cell Culture) collection. Huh6 and Huh7 cells were grown in RPMI 1640 medium (Nikken Bio Medical Laboratory, Kyoto, Japan) supplemented with 10% fetal bovine serum (Life Technologies, Inc.), 50 units/ml penicillin sulfate (Life Technologies, Inc.), and 100 µg/ml kanamycin sulfate (Wako Chemical Industries Ltd., Japan) at 37 °C under 5% CO₂, 95% air. The other hepatoma cell lines were grown in Dulbecco's modified essential medium supplemented with the same agents. For primary cultures, rat and mouse hepatocytes were isolated from 8-week-old male Sprague-Dawley rats and BDF1 mice by the two-step collagenase perfusion method described by Seglen et al. (24), and then the cells were cultured in Eagle's medium containing 10⁴ mM insulin, 10⁵ mM dexamethasone, and 5% fetal bovine serum. Hepatoma cells and hepatocytes at confluency conditions were treated with various concentrations of forskolin (Sigma).

About 5-10 × 10⁶ cells were used to assay GnT-III. After treatment with 10 µM forskolin, the cells were washed with phosphate-buffered saline (PBS), pH 7.4, twice and then centrifuged at 1500 × g for 10 min. The precipitated cells were resuspended in 0.1-0.2 ml of PBS, and then sonicated. The sonicated cell lysates were assayed for GnT-III. A fluorescence-labeled, pyridylaminated biantennary sugar chain was used as a substrate for the enzyme activity assay (25). The reaction mixtures were analyzed by high performance liquid chromatography on a TSK-gel ODS-80Tm column (4.6 × 150 mm; Tosoh, Tokyo, Japan). The details of the standard assay were described previously (26-28). Protein concentrations were determined with a bicinchoninic acid kit (Pierce) using bovine serum albumin as a standard.

Total RNA was prepared from cells according to the method reported previously (29). 20 µg of RNA was electrophoresed on a 1% agarose gel containing 2.2 M formaldehyde and then transferred to a Zeta-Probe membrane (Bio-Rad) by capillary action (30). The membrane filter was hybridized with ³²P-labeled GnT-III cDNA (10) or ³²P-labeled -actin at 42 °C in a hybridization buffer (30). The filter was washed at 55 °C with 2 × standard saline citrate and 0.1% sodium dodecyl sulfate for 30 min, and then with 0.2 × standard saline citrate and 0.1% sodium dodecyl sulfate for 30 min twice, and finally exposed to x-ray film ((Eastman Kodak Corp.) with an intensifying screen at 80 °C for 3 days.

8 µg of protein from total cell lysates was electrophoresed on a 12% SDS-acrylamide gel (31). After SDS-PAGE, the gel was blotted onto a nitrocellulose membrane (Schleicher & Schuell, Dassel, Germany) using instructions TE70 and TE77 Semiphor semi-dry units (Hoefer Scientific Instruments, San Francisco, CA). The membrane was prehybridized with PBS including 3% bovine serum albumin for 3 h, and then hybridized with 2 µg/ml biotinylated E-PHA, L-PHA, ConA, and SSA (Seikagaku Industrial Corp., Tokyo, Japan) for 2 h. After washing four times each for 10 min with PBS containing 0.05% Tween 20, the membrane was incubated with a 1/2000 dilution of horseradish peroxidase avidin D (Vector Industrial Corp., CA) for 1 h, and then washed four times each for 10 min with PBS containing 0.05% Tween 20. Staining was performed with a Western blot detection reagent, ECL (Amersham Life Science), for 1 min at room temperature. The membranes were exposed to a Kodak scientific imaging film.

M31 cells treated with various concentrations of forskolin for 12 h were removed from 10-cm culture dishes using PBS containing 0.2% EDTA. They were then centrifuged at 1500 × g for 5 min, and the precipitate was resuspended in 100 µl of PBS. Fluorescein isothiocyanate-labeled E-PHA or L-PHA (as a control) (Seikagaku Corp., Japan) was added to a final concentration of 5 µg/ml. After incubation for 20 min at room temperature, the cells were pelleted and washed three times with cold PBS, followed by sorting with a fluorescence-activated cell sorter (FACSORT, Becton Dickinson). Background was eliminated by establishing gates to monitor live cells only, but not cell debris. Unstained cells were served as controls. A fluorescence histogram and mean fluorescence were determined and analyzed from these data using the Macintosh Cell Quest computer program (FACSORT). The binding capacity as to E-PHA or as to L-PHA was evaluated as the difference between the mean fluorescence of stained cells and the mean autofluorescence of the cells.

M31 cells treated with 10 µM forskolin for 12 h were harvested with PBS containing 0.2% EDTA, washed twice with PBS, and then resuspended at a density of 1 × 10⁴ cells/ml in PBS. After permeabilization with or without 0.1% Triton X-100 treatment, the cells were stained with FITC-conjugated E-PHA (10 µg/ml) at 4 °C for 30 min. After washing twice with PBS, an aliquot of the cell suspension was placed on a slide and mounted in 10% glycerol in PBS. The fluorescence was viewed with an epifluorescence microscope (Olympus, Provis, Tokyo, Japan), and photographed with a 20-s exposure using a 200 × objective scale.

M31 cells treated with 10 µM forskolin for 12 h were washed with PBS, and then lysed in lysis buffer (6.7 mM potassium phosphate buffer, pH 7.4, 150 mM NaCl, 1% Nonidet P-40, 10 µM p-amidinophenylmethylsulfonyl fluoride, and 100 µM aprotinin). The glycoproteins were sequentially immunoprecipitated from the cell lysates by the addition of antibodies, as follows: Lamp-1 by a rat anti-mouse Lamp-1 monoclonal antibody (ID4B) (20, 32), which was a generous gift from Dr. J. T. August; -GTP by the addition the of a goat anti-rat -GTP polyclonal antibody (33, 34); -glucuronidase by an anti-rabbit -glucuronidase antibody (35); -fetoprotein by a horse anti-rat -fetoprotein antibody (36); and ceruloplasmin by a rabbit anti-human ceruloplasmin antibody (Wako Chemical Industries, Ltd., Japan).

Aliquots of samples were subjected to 12% SDS-PAGE according to the method reported by Laemmli (31) and then transferred to nitrocellulose membranes. The membrane filters were probed with 2 µg/ml biotinylated E-PHA, L-PHA, or an antibody. Detection was performed with each specific antibody and an ECL kit as described above.

Cell Surface Biotinylation and Immunoprecipitation of Lamp-1 and -GTP

To determine the expression levels of Lamp-1 and -GTP on the cell surface, cell surface proteins were biotinylated according to the method reported by Goishi et al. (37), and then immunoprecipitated with the anti-Lamp-1 antibody or anti--GTP antibody as described above. M31 cells treated with forskolin were plated on 10-cm dishes at the density of 3 × 10⁶ cells/dish, incubated for 0, 12, and 24 h, and then washed three times with ice-cold Hanks' buffer. Three ml of labeling buffer comprising 0.1 mg/ml sulfo-NHS-biotin (Pierce) in 50 mM HEPES, pH 7.5, and 0.15 M NaCl was added, followed by incubation for 15 min on ice with gentle shaking. Excess reagent was quenched and removed by washing with ice-cold Dulbecco's modified Eagle's medium, 10% fetal bovine serum. Cells were scraped off with a silicon rubber policeman and collected in 500 µl of Dulbecco's modified Eagle's medium. The biotinylated cells were lysed with a lysis buffer, and then immunoprecipitated as described above. Aliquots of the samples were subjected to 12% SDS-PAGE and then blotted on nitrocellulose membranes. After prehybridization, the filters were incubated with 1/2000 diluted horseradish peroxidase avidin D, and then stained with a Western blot detection reagent, ECL.

The isolation of subcellular organelles of M31 cells was performed according to OptiPrep^TM standard protocols (Nycomed Pharma, Oslo, Norway; Ref. 70) after treatment with or without 10 µM forskolin for 12 h. Solutions for the formation of gradients were prepared according to Tables 2 and 3 of the protocols (70). The fractionation of the major organelles of M31 cells was performed in self-generated gradients. Enzymes are commonly used as markers for the detection of organelles. 8 µg of protein from each fractionated sample was used for E-PHA blot analysis.

RESULTS

Initially, GnT-III activity and mRNA expression were determined in various hepatoma cells and hepatocytes after treatment with 10 µM forskolin for 12-24 h (Table I). Enhancement of GnT-III activity by forskolin was observed in Huh6 and M31 cells, and normal hepatocytes, but not in Huh7 and Hep3B cells. The highest enhancement level was observed in M31. The effect of forskolin on GnT-III enhancement was studied in M31 cells. When M31 and Huh7 cells were treated with various concentrations of forskolin (0-50 µM) for 12 h, the highest enhancement of GnT-III mRNA was observed in M31 cells on 10 µM forskolin treatment (Figs. 1 and 3), but not in Huh7 cells (Fig. 1). When M31 cells were treated with this concentration of forskolin for different time intervals (Fig. 2), GnT-III activity reached the maximum level at 12 h after treatment and then reached the initial level at 24 h after treatment. Northern blot analysis showed dramatic induction of GnT-III mRNA at 12 and 18 h after treatment (Fig. 3). These results suggested that forskolin enhanced the transcriptional level of GnT-III in M31 cells.

Table I. Changes of GnT-III activities in hepatoma cells and hepatocytes treated with forskolin Various hepatoma cells and hepatocytes were treated with (+) or without () 10 µM forskolin for 12. All data represent the means ± S.D. of three experiments. Enzyme activities are described as pmol/h/mg protein. ND; not detected. Cell type GnT-III activity   Forskolin + Forskolin Human hepatoma cells   HuH6 57.2  ± 5.5 117  ± 7   HuH7 ND ND   Hep3B ND ND Rat hepatoma cells (M31) 51  ± 8.5 463  ± 19.7 Rat hepatocytes 10.9  ± 1.5 35  ± 5 Mouse hepatocytes 7.3  ± 2.1 26  ± 3.6

According to our previous work, enhanced expression of GnT-III resulted in an increase in the bisecting GlcNAc structure on N-glycans. In order to investigate this change in glycosylation, lectin blot analyses were performed using E-PHA, L-PHA, ConA, and SSA (Fig. 4). E-PHA blotting showed that the intensity of the bands between 46 and 97 kDa was enhanced at 12, 18, and 24 h after forskolin treatment. The binding with E-PHA thus appeared to be correlated with the increased GnT-III activity, implying the elevated GnT-III activity catalyzed the addition of the bisecting GlcNAc structure to the glycoproteins on whole cells that had been treated with forskolin. Lectin blotting of L-PHA, which binds with 1-6 branch (a product of GnT-V), showed that the binding capacity as to L-PHA decreased by 12-h forskolin-enhanced GnT-III activity, since actions of GnT-III and GnT-V are competitive. Lectin blotting of ConA, which binds with high affinity to high mannose-type, biantennary, complex, or hybrid type of asparagine-linked oligosaccharides but does not bind to bisecting structures (38), showed that some bands of approximately 30-45 kDa had disappeared by 12 h after forskolin treatment. SSA binds to Sia2-6Gal/GalNAc tightly but weakly to Sia2-3Gal/GalNAc (39). The SSA blot showed the disappearance of many bands, especially between 97 and 46 kDa, at 12 h after forskolin treatment. These patterns on lectin blot analysis were consistent with the increase of bisecting GlcNAc structures in many glycoproteins in whole cells on up-regulation of GnT-III.

To determine whether or not the increased GnT-III activity altered oligosaccharide structures on cell surface glycoproteins, FACS analysis was performed using FITC-conjugated E-PHA and L-PHA (as a control) (Fig. 5). Although GnT-III activity was highly enhanced on forskolin treatment, the binding capacity as to E-PHA decreased in a forskolin dose-dependent manner, in contrast, the binding capacity as to L-PHA increased. This finding suggested that the addition of bisecting GlcNAc to N-glycans inhibits the sorting of glycoproteins bearing bisecting-GlcNAc structures to the cell surface, but not N-glycans bearing 1-6 branch. Interestingly, these changes were observed at 12 h after forskolin treatment but not after 24 h (data not shown).

The results of lectin blot and FACS analyses suggest that the addition of bisecting structures to N-glycans inhibits the normal sorting of glycoproteins to the cell surface. To confirm this, an immunofluorescence microscopic study involving FITC-conjugated E-PHA was performed (Fig. 6). Broad staining was observed for untreated control cells. In contrast, high fluorescence was observed only intracellularly for forskolin-treated cells, while the fluorescence was much fainter on the cell surface than that of control cells. This finding supports our hypothesis that the addition of bisecting GlcNAc to N-glycans inhibits the sorting of glycoproteins bearing bisecting-GlcNAc structures to the cell surface.

To search for a correlation between the oligosaccharide structure of a specific glycoprotein and its distribution, we investigated the oligosaccharide structures and distributions of different glycoproteins, such as Lamp-1, which possesses many N-glycans of different structures and exists in the lysosome and on the cell surface; -glucuronidase (lysosomal enzyme); -GTP (plasma membrane); and secretory glycoproteins such as ceruloplasmin and -fetoprotein. While the binding with E-PHA of Lamp-1, -glucuronidase, and -GTP was increased at 12 h after forskolin treatment, the binding with L-PHA was decreased (Fig. 7, A-C), suggesting that the enhanced GnT-III activity synthesized the bisecting GlcNAc structure and inhibited the processing of the 1-6 branch (a product of GnT-V) in N-glycans. The binding with E-PHA of the secretory glycoproteins (Fig. 7, D and E) was increased at 12 h after forskolin treatment, supporting the finding that the bisecting GlcNAc structure was synthesized by forskolin-enhanced GnT-III. To determine whether or not these structural changes of oligosaccharides affect the expression levels of Lamp-1 and -GTP on the cell surface (Fig. 8, A and B), cell surface proteins were labeled with biotin, followed by immunoprecipitation with anti-Lamp-1 or anti--GTP antibodies. This showed that the expression levels of both Lamp-1 and -GTP, on the cell surface, were markedly decreased in M31 cells which had been treated with forskolin for 12 h (Fig. 8, A and B). The changes in the expression of Lamp-1 and -GTP on the cell surface were consistent with changes in GnT-III activity, suggesting that the addition of bisecting structures to Lamp-1 and -GTP inhibited their sorting onto the cell surface. In contrast to these results, forskolin-enhanced GnT-III has no effect on the secretion of secretory glycoproteins such as ceruloplasmin and -fetoprotein (Fig. 7, D and E), although bisecting GlcNAc structures were increased in these proteins.

Cell surface biotinylation and immunoprecipitation of Lamp-1 and -GTP.

To determine where the bisecting GlcNAc residues are mainly localized inside the M31 cells, fractionation of M31 cells treated with or without 10 µM forskolin for 12 h was performed (Fig. 9). The binding capacity as to E-PHA was markedly increased in glycoproteins of Golgi (B), lysosomes (C), and slightly in case of endoplasmic reticulum (D), especially the bands between 66 and 97 kDa by 12 h of forskolin treatment. In case of mitochondria, the binding capacity as to E-PHA showed that, some bands between 66 and 46 kDa disappeared by 12 h of forskolin treatment (A).

DISCUSSION

The oligosaccharides of glycoproteins are regulated by various glycosyltransferases, and GnT-III is a particularly unique enzyme because it reacts with the core region of N-glycans and inhibits their further processing (2, 40). Although GnT-III in the liver is markedly induced in various conditions such as transformation, carcinogenesis, and the clinical stage of liver disease (11, 16, 41, 42), the mechanism by which this enzyme is induced remains unclear. While we were searching for some factors that induce GnT-III, we found that forskolin strongly enhanced the activity and mRNA expression of GnT-III in a rat hepatoma cell line, M31. This effect was maximum at 12 h after 10 µM forskolin treatment, and the same effect was observed for human hepatoma cell line Huh6, and rat and mouse hepatocytes in primary culture, but no effect was noticed in the cases of the Huh7 and Hep3B human hepatoma cell lines. This effect of forskolin was not seen with any of the five other agents: interleukins 1 and 2, transforming growth factors  and , the phorbol ester 12-O-tetradecanoylphorbol-13-acetate, protein kinase C inhibitor, and protein kinase A inhibitor. The mechanism underlying the up-regulation appears to be via the cAMP pathway, but it remains unknown. It was reported that the cAMP pathway controls the transcription of many genes, as in the case of the human chorionic gonadotropin- subunit gene (43), and the transcription activation of many proteins is achieved within 1 h, as is the case for the insulin precursor (44), growth hormone (45), and tyrosine aminotransferase (46). In contrast, our results showed that forskolin enhanced GnT-III after 12 h of treatment, but not within 1 h. These facts suggest that forskolin may enhance the gene expression of GnT-III via a secondary pathway or probably requires ongoing protein biosynthesis (43). Consistent with our results, it was found in hepatoma HepG2 human hepatoma cells that forskolin induced the gene expression of insulin-like growth factor-binding proteins within 24 h via cAMP, suggesting that stimulation of insulin-like growth factor-binding protein gene expression by cAMP is transcriptional, via a protein recognizing the cAMP-responsive element consensus sequence (47). This indicated that GnT-III enhanced by forskolin is more complex, probably involving several levels of regulation.

The levels of cAMP in hepatocytes are also increased in regenerating liver (48, 49), suggesting that cAMP is one of the factors that induce GnT-III in vivo. Recently, we reported that overexpression of GnT-III suppresses lung metastasis of melanoma cells (50) and expression of the hepatitis B virus (51), suggesting that the bisecting GlcNAc structure, which is a product of GnT-III, has some biological meaning. However, in those studies, bisecting GlcNAc structures were synthesized with ectopically expressed GnT-III by transfection of the gene into cancer cells. The present study demonstrated the role of instrinsic GnT-III in M31 cells. GnT-III activity up-regulated by forskolin resulted in an increase of the total cellular bisecting GlcNAc structures, as determined by studying the structures of different glycoproteins, such as Lamp-1 and -glucuronidase (lysosomal enzyme), -GTP (membranous protein), ceruloplasmin, and -fetoprotein (secretory proteins), but decreases the levels of expression of the glycoproteins expressed on the cell surface such as Lamp-1 and -GTP, and accumulates them in Golgi and lysosome. In contrast, the addition of the bisecting GlcNAc structure does not affect the levels of secretory glycoproteins.

This controversial phenomenon was not observed in most hepatoma cells (data not shown). The role of N-glycans in the secretory pathway has been reported in many articles (6); for instance when N-glycan addition to proteins is blocked, most nonglycosylated forms of the proteins accumulate in the endoplasmic reticulum and aggregate, and thus do not exit (52). The introduction of N-glycosylation sites into the rat growth hormone leads to apical sorting and delivery of this secretory protein in Madin-Darby canine kidney cells (5). On the other hand, non-glycosylated secretory proteins are usually secreted both apically and basolaterally (53-55), presumably being included by default into post-Golgi transport vesicles. The sorting efficiency could depend on the number of available glycan chains. When the N-glycans of some glycoproteins are modified, they cannot be correctly sorted (56-58); furthermore, underglycosylated glycoproteins and defective N-glycan assembly were noticed in patients with carbohydrate-deficient glycoprotein syndromes (59). It was reported that N-linked glycosylation is required for both cell surface expression and immunogenicity of the rabies virus glycoprotein (60). Lamp-1, a representative glycoprotein possessing many N-glycans, is associated with tumor metastasis (61). The authors argued about the significance of their oligosaccharides, especially the 1-6 branching. There have been many reports that suggested the importance of Lamp-1 as a useful indicator for some diseases, such as systemic lupus erythematosus and scleroderma patients (62, 63). Our studies demonstrated a possible functioning of the bisecting GlcNAc structure as an inhibitory factor for the sorting of some glycoproteins on the cell surface, including Lamp-1 as an excellent model formed in large amounts, possessing N-glycans of different structures, and expressed on the cell surface membrane (64-66); and also -GTP, as a plasma membrane glycoprotein (67). Cell fractionation analyses showed that many glycoproteins bearing bisecting GlcNAc residues were mainly localized in Golgi and lysosomal membranes (Fig. 9). This accumulation may be due to the presence of special lectin that binds to bisecting GlcNAc residues, or it may be due to some conformational structural changes of these glycoproteins after addition of the bisecting GlcNAc residues leads to decrease their normal sorting to the cell membrane. Most secretory glycoproteins are secreted through coated pits, without any special conformational changes for passage through the membrane (68, 69).

Membranous glycoproteins contain many adhesion molecules and receptors for growth factors. If structural changes of oligosaccharides in these proteins result in some changes in their distribution on the cell surface, the bisecting GlcNAc structure may be an important tool for changing a cellular phenotype. The means of decreased expression levels of branched N-linked oligosaccharides on the cell surface of M31 hepatoma cells on modification of the expression of the GnT-III gene may play roles in a variety of phenomena such as cell-cell adhesion and the capacity for metastasis. Further investigation is needed to figure out which kinds of glycoproteins are influenced by the addition of GlcNAc structures, post-transcriptional modifications, and what kind of the lectin, if any, leads to their accumulation in Golgi and Lysosome. Taken together, the present study is the first report on glycoprotein sorting and a specific structure of oligosaccharide.