J Biol Chem, Vol. 274, Issue 30, 21209-21216, July 23, 1999

Expression of the H Type 1 Blood Group Antigen during Enterocytic Differentiation of Caco-2 Cells^*

Junko Amano and Mieko Oshima

From the Division of Biochemistry and Nutrition, Research Institute, International Medical Center of Japan, Tokyo 162-8655, Japan

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

We made a comparative study of the structures of the oligosaccharides on the glycoproteins from Caco-2 human colonic adenocarcinoma cells, before and after differentiation. Enterocytic differentiated Caco-2 cells highly express H type 1 blood group antigen on the cell surface as well as activities of brush border membrane hydrolases, such as dipeptidyl peptidase IV and alkaline phosphatase. A strong correlation was observed between the amounts of H type 1 blood group antigen and the degrees of differentiation. Structural analysis with use of lectin affinity high performance liquid chromatography revealed that typical mucin-type sugar chains of the glycoproteins from undifferentiated cells have H type 2 group, linear polylactosamines, and core 1 structure. On the other hand, differentiated cells newly contain H type 1 and Le^b groups and core 2 structure. Mucins with H type 1 group make contact with brush border membrane enzymes on differentiated cells. Furthermore benzyl 2-acetamide-2-deoxy--D-galactopyranoside inhibited both expression of H type 1 group on the cell surface and enhancement of brush border membrane enzyme activities even in the presence of a differentiating inducer. These results suggest that the mucin-type sugar chains with H type 1 group have important functions regarding differentiation of Caco-2 cells.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Caco-2 cells derived from a human colonic adenocarcinoma differentiate into enterocytes-like cells spontaneously (1) or by induction with sodium butyrate (2). During enterocytic differentiation, in addition to morphological change with acquisition of a brush border, various biological changes have been noted, for example, expression of brush border-associated enzymes (1), mucin synthesis (3), and glycosylation. However, little is known of detailed oligosaccharide structures before and after differentiation of Caco-2 cells. Decrease in polylactosaminoglycans of lysosomal membrane glycoprotein h-Lamp-1 was observed in spontaneously differentiated Caco-2 cells (4), but, unexpectedly, the glycosyltransferases directly involved in polylactosaminoglycan biosynthesis remain essentially unchanged (5). Findings that increased activities of branching enzymes and decreased activity of mucin-type sugar chain core 1 enzyme were also obtained (5). Although these results suggest a change in glycosylation with differentiation of Caco-2 cells, no information on outer chains that contain blood group antigens and interact with other cells has been available. Only an increase in the 2,6-sialylation after differentiation has been reported (6). Differentiated Caco-2 cells are used for a model of adherence of bacteria to the intestinal epithelium (7, 8). This protein-carbohydrate interaction is critical for bacterial infection and involves microbial lectin-like adhesins and specific oligosaccharides present on the intestinal epithelium. To elucidate oligosaccharide structures on the surface glycoproteins of Caco-2 cells is essential for a better understanding of the mechanism of microbial infections. For this purpose a comparative study of the oligosaccharide structures of Caco-2 cells before and after the differentiation was done.

In comparison with N-linked oligosaccharides, systematic analysis of mucin-type sugar chains was not established mostly because of the variety and complexity. In this paper, a simple and easy structural analysis of the mucin-type sugar chains was developed, using lectin affinity HPLC.¹ The oligosaccharides were separated into three nonreducing termini, repeating units, and cores by digestion with endo--galactosidase. An LAS-AAL column that recognizes fucosylated oligosaccharides was used for investigation of fucosylation at the nonreducing termini, and an LAS-PVL column with an affinity to N-acetylglucosamine was used for analysis of the repeating units and the cores because these parts contain N-acetylglucosamine. Both an AAL-agarose column (9, 10) and a PVL-Affi Gel column (11, 12) have been already characterized and used for analysis of oligosaccharides, but application to HPLC progressively improved the separation and identification for this analysis.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Cell Culture-- Caco-2 cells obtained from the American Type Culture Collection were cultured (no data shown). For differentiation 2 mM sodium butyrate (Wako Chemicals, Osaka, Japan) was added to the medium after confluence. For some cases, 2 mM GalNAc--O-benzyl (Sigma) was added to the medium together with 2 mM sodium butyrate.

Antibodies and Lectins-- Mouse monoclonal antibodies against H type 1; Fuc1-2Gal1-3GlcNAc, Le^a; Gal1-3(Fuc1-4)GlcNAc, Le^b; Fuc1-2Gal1-3(Fuc1-4)GlcNAc, Y; Fuc1-2Gal1-4(Fuc1-3)GlcNAc were purchased from Signet Laboratories, Inc. (Dedham, MA). Mouse monoclonal antibody, KM380, which recognizes X; Gal1-4(Fuc1-3)GlcNAc was a gift of Dr. N. Hanai (Kyowa Hakko, Tokyo, Japan). The following mouse monoclonal antibodies were also used: against blood group A and B from Biomeda Corporation (Foster City, CA), specific for synthesized peptides of MUC-2 (clone Ccp58) and MUC-3 (clone M3.1) from Biogenesis Ltd., against alkaline phosphatase (clone ZAP 1) from Zymed Laboratories Inc. (San Francisco, CA). Biotin-Ricinus communis (RCA120, CAS 172304-66-4), Biotin-Ulex europaeus agglutinin I (UEA-I), and FITC-UEA-I were from Seikagaku Corporation (Tokyo, Japan). Biotin-SP-conjugated Affinipure goat anti-mouse IgG + IgM, FITC-conjugated Affinipure F(ab')₂ fragment donkey anti-mouse IgG and FITC-streptavidin were obtained from Jackson ImmunoResearch Laboratories, Inc. (West Grove, PA). R-Phycoerythrin (R-PE)-conjugated mouse IgG, R-PE-conjugated mouse anti-human CD26 (dipeptidyl peptidase IV) antibody, and mouse anti-human CD26 antibody were from PharMingen (San Diego, CA).

Measurements of Alkaline Phosphatase and Dipeptidyl Peptidase IV Activities-- The cells were harvested with 0.25% trypsin-1 mM EDTA, washed with Tris-buffered saline and subjected to ultrasonication in 0.25 M sucrose in 10 mM Tris-HCl, pH 7.2. The obtained cell homogenates were stored at 80 °C until assayed. Alkaline phosphatase activity was determined in cell homogenates using p-nitrophenylphosphate as a substrate. Dipeptidyl peptidase IV (DPP-IV) activity was determined in cell homogenates according to Nagatsu et al. (13) using glycyl-L-proline-4-nitroanilide as a substrate. Results are expressed as milliunits/mg of protein. One unit is defined as the activity that hydrolyzes 1 µmol of substrate/min at 37 °C. Proteins were measured using a Bio-Rad protein assay reagent and BSA as a standard.

Preparation of Glycoproteins-- After washing with PBS, cells were detached with a cell scraper, collected by centrifugation, and stored at 80 °C until use. Cell pellets were suspended in 1 ml of water and 2 ml of methanol and homogenized by ultrasonication on ice. After adding 3 ml of chloroform, lipids were extracted by shaking. The delipidated proteins were obtained by repeated extraction with chloroform/methanol/water (3:2:1 and 10:10:3) and stored at 80 °C until use.

Determination of Contents of Fucose-- The delipidated proteins (about 10 µg) were heated in 2 M trifluoroacetic acid at 100 °C for 3 h and dried on Speed Vac (model AES1010, Savant Instruments, Inc., Farmingdale, NY) with repeated addition of distilled water. The hydrolysate was incubated with 10 units/ml L-fucose dehydrogenase (Kikkoman Corp., Noda, Japan) (14) and 1 mM NADP, and the produced NADPH was determined by fluorescence intensity at 450 nm for emission and 360 nm for excitation with use of a micro plate reader (model MTP-100F, Corona, Hitachi, Japan). The amount of released fucose was calculated based on the produced NADPH. The minimal amount of fucose determined using this method is 0.1 nmol.

Enzyme Immunoassay of Glycoproteins-- The delipidated proteins prepared from Caco-2 cells were solubilized in 0.15 M ammonium bicarbonate, pH 7.8, and coated on a Maxi Sorp plate (Nunc, Japan Inter Med, Tokyo, Japan). For detection of MUC2 and 3 the proteins were coated after mild alkaline treatment (0.05 N NaOH at 37 °C for 4 h). After blocking with BSA, the coated proteins were either incubated with monoclonal antibodies and then biotin-SP-conjugated Affinipure goat anti-mouse IgG+IgM or incubated directly with biotin-conjugated lectins followed by avidin-biotin-peroxidase kit (Vectastain, ABC, Vector, Funakoshi, Tokyo, Japan). The peroxidase reaction was performed at room temperature using H₂O₂ and O-phenylenediamine as substrates. Optical density at 490 nm was measured using an EIA reader (model 3550, Bio-Rad, Richmond, CA).

Lectin Affinity High Performance Liquid Chromatography of Mucin-type Sugar Chains of Glycoproteins-- For determination of the nonreducing termini and the repeating units, mucin-type sugar chains were released from the delipidated proteins by alkaline-borohydride treatment (15) and digested with Escherichia freundii endo--galactosidase (Seikagaku Corporation, Tokyo, Japan) for 3 days at 37 °C by adding 30 milliunits of enzyme every day. The oligosaccharide fragments were reacted with 0.1 mmol aminobenzoyl ethyl ester (ABEE) in 50 µl of methanol containing 5 mg of sodium cyanoborohydride at 80 °C for 20 min. The ABEE-labeled oligosaccharides were obtained as aqueous solution by chloroform extraction. For determination of the core structures, mucin-type sugar chains were released from the delipidated proteins by hydrazinolysis at 60 °C for 5 h. After re-N-acetylation, oligosaccharides were labeled with ABEE and then digested with endo--galactosidase to obtain labeled core structures. Aliquots were subjected to HPLC on an immobilized Aleuria aurentia lectin column, LAS- AAL (Honen Corporation, Tokyo, Japan) for analysis of the nonreducing termini and an immobilized Psathyrella velutina lectin column, LAS-PVL (Honen Corporation, Tokyo, Japan) for the repeating units and the cores. HPLC was done using a Shimadzu LC-10A HPLC apparatus (Shimadzu, Kyoto, Japan). The LAS-AAL column (inner diameter, 4 × 50 mm) was run in 10 mM sodium acetate, pH 7.0 at 0.2 ml/min at 25 °C. Elution was performed by a linear gradient of concentration of fucose from 0 to 20 mM for 55 min after holding 0 mM for 5 min. The LAS-PVL column (inner diameter, 4 × 50 mm) was run in 10 mM sodium acetate, pH 7.0, at 0.5 ml/min at 25 °C. The elution was performed by a linear gradient of concentration of N-acetylglucosamine from 0 to 20 mM for 35 min after holding 0 mM for 5 min. Detection was made using a fluorescence detector, RF-10AXL (Shimadzu, Kyoto, Japan) at 360 nm for emission and 305 nm for excitation. Standard oligosaccharides for the LAS-AAL column were prepared from oligosaccharides of pig gastric mucin and keratan sulfate by endo--galactosidase treatment.

Immobilization on a Micro Plate and Identification of Oligosaccharide Fractions Obtained by LAS-AAL High Performance Liquid Chromatography-- The ABEE-labeled oligosaccharide fractions eluted from the LAS-AAL column were passed through a small AG3 (OH form) column, and the column was washed with distilled water. The effluent was dried on Speed Vac, and the residue was incubated in 0.1 N NaOH at 37 °C for 4 h, and the mixture was neutralized by adding 0.1 N HCl. To covalently bind to amino groups on a micro plate, the obtained amino benzoate-conjugated oligosaccharide solution (0.01-1 nmol/10 µl) and 30 µl of 2.5 mg/ml ethylenediamine and N-hydroxysuccinimide were added to a well of CovaLink (Nunc, Japan Inter Med, Tokyo, Japan) and reacted on a heating block at 60 °C for 5 h and then at room temperature overnight. The plate was washed with 2 M NaCl and blocked with 1% BSA in PBS. The oligosaccharides immobilized on the plate were subjected to enzyme immunoassay using the same procedures described under "Enzyme Immunoassay of Glycoproteins."

After labeling with ABEE, a mixture with 0.5 nmol of each of five kinds of human milk oligosaccharides (Oxford GlycoSystems Ltd., Abingdon, UK) was applied to the LAS-AAL column and separated into five peaks (Fig. 1A). Fifth parts/well of these fractions were immobilized on a micro plate as mentioned above and detected using lectin and antibodies. Because peaks 1, 2, 3, 4, and 5 were reacted with RCA120, anti-H type 1, anti-Le^a, anti-X, and anti-Le^b antibodies, respectively, the structures were confirmed to be lacto-N-neotetraose, lacto-N-fucopentaose-I, lacto-N-fucopentaose-II, lacto-N-fucopentaose-III, and lacto-N-difucohexaose-I (Fig. 1B). The binding character of LAS-AAL showed the same tendency as that seen using the AAL-agarose column (9), but LAS-AAL has a high resolution.

View larger version (19K): [in this window] [in a new window]   Fig. 1.   Separation and identification of milk oligosaccharides. A, chromatogram of a mixture of ABEE-labeled standard milk oligosaccharides, lacto-N-neotetraose, lacto-N-fucopentaose-I, lacto-N-fucopentaose-II, lacto-N-fucopentaose-III, and lacto-N-difucohexaose-I on LAS-AAL column by HPLC. B, results of detection by enzyme-linked immunosorbent assay of immobilized oligosaccharides obtained from A. Each peak was immobilized on a micro plate and subjected to enzyme-linked immunosorbent assay, as described in the text. Peaks detected by the antibody or not detected are shown as + or .

Flow Cytometry-- Cells were detached by trypsin-EDTA, washed with Tris-buffered saline, and incubated with control mouse IgG, anti-H type 1, or anti-alkaline phosphatase antibody in 0.1% BSA in Tris-buffered saline, washed, and then with FITC-conjugated Affinipure F(ab')₂ fragment donkey anti-mouse IgG for 30 min on ice. Other cells were incubated with R-PE-conjugated mouse IgG, R-PE-conjugated anti-CD26 antibody, FITC-conjugated streptavidin, or FITC-conjugated UEA-I. The labeled cells analyzed using a FACscan flow cytometer (Becton Dickinson, San Jose, CA).

Separation of Cells Expressing H Type 1 Antigen-- Cells were harvested with trypsin-EDTA treatment and incubated with anti-H type 1 at 4 °C for 15 min, washed, and then incubated with goat anti-mouse IgG micro beads at 4 °C for 15 min. The mixture was applied to a magnetic cell sorter (Miltenyi Biotec, Daiichi Purechemicals Co., Ltd., Tokyo, Japan). Cell separation was done according to attached instructions. The obtained negative and positive cells were stored at 80 °C until use.

Immunofluorescence-- Cells were grown in glass bottom dishes Matsunami, Osaka, Japan) and after confluence were cultured in medium containing 2 mM sodium butyrate in the presence or absence of 2 mM GalNAc--O-benzyl. 6 or 8 days past confluence, the cells were rinsed with PBS and fixed with cold ethanol for 15 min at 4 °C. After washing with PBS, the fixed cells were incubated with anti-H type 1 antibody or mouse IgG in 1% BSA in PBS at 4 °C overnight, washed, and then incubated with fluorescein-conjugated Affinipure F(ab')₂ fragment donkey anti-mouse IgG. Other fixed cells were incubated with R-PE-conjugated mouse anti-human CD 26 antibody, R-PE-conjugated mouse IgG, or FITC-UEA-I for 15 min at 4 °C. For double staining, cells were stained by a series of anti-H type 1/FITC-conjugated secondary antibodies and then R-PE-conjugated anti-CD26 antibody. The stained preparations were analyzed using confocal microscope, Fluoroview (Olympus, Tokyo, Japan).

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Differentiation and Fucosylated Blood Group Antigens-- Delipidated glycoproteins were prepared from the undifferentiated cells (2 days before confluence) and the differentiated cells (6 days past confluence), and the fucose contents were measured. The values were 83 and 130 nmol fucose/mg protein for the undifferentiated cells and the differentiated cells, respectively. Because the amount of obtained proteins from the differentiated cells exceeded that of the undifferentiated cells, the actual amount of fucose on the differentiated cells (116 nmol/10⁶ cells) was twice that for undifferentiated cells (54 nmol/10⁶ cells). To clarify which fucose-containing blood group antigens are expressed on the Caco-2 cells before and after differentiation, the obtained proteins were studied by an enzyme immunoassay using lectins, RCA120 and UEA-I, and monoclonal antibodies against blood type A and B, X, Y, Le^a, Le^b, and H type 1. The reactivities of the glycoproteins from the differentiated cells to the antibodies against H type 1 and Le^b were obviously higher than for the undifferentiated cells (Fig. 2, A and B), yet the difference was small between the two kinds of cells regarding reactivities to UEA-I and anti-Y antibody (Fig. 2, C and D). The glycoproteins from both cells showed a high reactivity to RCA120 and almost no activity to antibodies against blood group A and B, X and Le^a (data not shown).

View larger version (18K): [in this window] [in a new window]   Fig. 2.   Enzyme immunoassay of the glycoproteins prepared from undifferentiated (open circles) and differentiated (closed circles) Caco-2 cells. Caco-2 cells 2 days before confluence and 6 days past confluence served as undifferentiated and differentiated cells, respectively. Detection was made using anti-H type 1 (A), anti-Leb (B), UEA-I (C), and anti-Y (D). Results represent the means ± S.D. of three determinations.

Expression of H Type 1 on the Glycoproteins from the Caco-2 Cells in Differentiation-- Because the amount of H type 1 antigen on glycoproteins from the differentiated cells exceeded that from the undifferentiated cells, change in the amount of this antigen expressed on the Caco-2 cells in differentiation was examined. The early differentiation marker, dipeptidyl peptidase IV activity, was gradually elevated during growth in the presence of butyrate, and elevation of the late differentiation marker, alkaline phosphatase activity, was behind (Fig. 3A). Antibody against MUC3 reacted after but not before differentiation, although expression of MUC2 decreased after differentiation (Fig. 3B). Glycoproteins were prepared from the cells on the day of confluence and also after 2, 4, 6, and 8 days past confluence and subjected to enzyme immunoassay by using anti-H type 1 antibody. The reactivity of the glycoproteins to this antibody increased according to the degree of differentiation, as shown in Fig. 3C. The reactivity to UEA-I, which recognizes H type 2 structure, however, was almost the same through the differentiation (data not shown). These findings mean that the expression of H type 1 but not H type 2 antigen correlates with differentiation of the Caco-2 cells.

View larger version (18K): [in this window] [in a new window]   Fig. 3.   Differentiation marker enzyme activities (A) and expression of MUC2 and MUC3 (B) and H type 1 antigen (C) during growth in culture of Caco-2 cells. A, dipeptidyl peptidase (closed circles) and alkaline phosphatase (closed triangles). B, enzyme immunoassay of the glycoproteins prepared from undifferentiated (open symbols) and differentiated (closed symbols) Caco-2 cells with use of anti-MUC2 (circles) and anti-MUC3 (diamonds). C, enzyme immunoassay of the glycoproteins prepared from Caco-2 cells immediately after confluence () or 2 days (), 4 days (), 6 days (), or 8 days () past confluence by using with anti-H type 1. Results represent the means ± S.D. of three determinations.

To determine whether the expression of H type 1 antigen occurs on the surface of the cells, cells at different stages of differentiation were examined using flow cytometry. Because the confluent cells already possess a significant DPP-IV activity (Fig. 3A), cells immediately following confluence as well as post-confluent cells show expression of CD26, which is the same molecule as DPP-IV as shown in Fig. 4 (A and E). The confluent cells showed no reactivity to anti-alkaline phosphatase, but at the late stage of differentiation, the cells acquired reactivity to anti-alkaline phosphatase (Fig. 4, B and F), and this result agrees with the expression of enzyme activity (Fig. 3A). The H type 2 structure was always expressed on the cell surface during the culture (Fig. 4, C and G). Although H type 1-positive cells were not evident prior to confluence (data not shown), the cells on and after confluence did express the H type 1 structure corresponding to a degree of differentiation (Fig. 4, D and H).

View larger version (20K): [in this window] [in a new window]   Fig. 4.   Flow cytometric analysis of undifferentiated and differentiated Caco-2 cells. Cells immediately following confluence served as undifferentiated cells (A-D), and the cells 6 days past confluence served as differentiated cells (E-H). A and E, R-PE-anti-CD26 (black) and R-PE-mouse IgG (white); B and F, anti-alkaline phosphatase/FITC-anti-mouse IgG (black) and mouse IgG/FITC-anti-mouse IgG (white); C and G, FITC-UEA-I (black) and FITC-streptavidin (white); D and H, anti-H type 1/FITC-anti-mouse IgG (black) and mouse IgG/FITC-anti-mouse IgG (white). Detailed procedures were described under "Experimental Procedures."

Differentiation Marker Enzyme Activities of the Caco-2 Cells Expressing H Type 1 Group on the Surface-- The flow cytometric study showed that not all the cells express H type 1, even at late differentiation. Cells selected using a magnetic cell sorter with antibody against H type 1 were obtained from Caco-2 cells 2, 6, and 8 days past confluence, and differentiation marker enzyme activities were compared between H type 1-negative and positive cells. 6 and 8 days past confluence H type 1-positive cells showed higher activity of dipeptidyl peptidase IV than H type 1-negative cells (Fig. 5A). As well as dipeptidyl peptidase IV, a higher activity of alkaline phosphatase in H type 1-positive cells compared with negative cells was obtained 8 days post confluence (Fig. 5B). These observations clearly indicate that cells highly expressing H type 1 are in a state of higher differentiation than the cells expressing a lesser amount of H type 1. 

View larger version (15K): [in this window] [in a new window]   Fig. 5.   Differentiation marker enzyme activities of Caco-2 cells selected by anti-H type 1 (shaded bars) and not selected by anti-H type 1 (white bars). Caco-2 cells 2, 6, and 8 days past confluence were subjected to magnetic cell sorting using anti-H type 1 antibody, and DPP-IV (A) and alkaline phosphatase (B) activities were determined for negative and positive fractions. Results represent the means ± S.D. of three determinations.

Structure Analysis of Mucin-type Sugar Chains of the Glycoproteins from Undifferentiated and Differentiated Caco-2 Cells-- Because expression of H type 1 group on the surface associates with differentiation of Caco-2 cells, the structures of the mucin-type sugar chains of the glycoproteins were compared between undifferentiated and differentiated Caco-2 cells. Structures of mucin-type sugar chains are so complicated that systematic methods for analysis have not been available. We divided the oligosaccharides into three portions obtained by digestion with endo--gatactosidase; the nonreducing termini that contain blood group antigens, the repeating units, and the cores were studied separately. Because each portion has characteristic sugars like fucose and N-acetylglucosamine, application on HPLC on LAS-AAL and LAS-PVL columns is feasible. At first, the oligosaccharide fragments digested by endo--galactosidase from the mucin-type sugar chains and labeled with ABEE were analyzed on lectin affinity HPLC using an immobilized AAL column (Fig. 6A). Three peaks (a, b, and c) were collected, immobilized on a micro plate, and detected using UEA-I and antibodies against H type 1, Le^a, Le^b, X, and Y, respectively, as described under "Experimental Procedures." Peaks a, b, and c were identified to contain H type 1, H type 2, and Le^b antigens, respectively. The differentiated Caco-2 cells contained H type 1 group and Le^b group, but the undifferentiated cells showed a trace amount (Fig. 6A). The fractions passing through the AAL column were analyzed using HPLC on an LAS-PVL column. The amount of GlcNAc1-3Gal fragment from undifferentiated cells was 1. 8 times more than that from differentiated cells. Neither sample contained GlcNAc1-6(GlcNAc1-3)Gal fragment, and this means that they have only linear repeating units but not branches (Fig. 6B). The core fragments obtained from the mucin-type sugar chains by hydrazinolysis, labeled with ABEE and digested by endo--galactosidase were also analyzed using HPLC on an LAS-PVL column. The sample from differentiated Caco-2 cells showed the core 2 structure in addition to core 1 structure (Fig. 6C); thus, it was clear that Caco-2 cells produce the core 2 structure after differentiation. From these results the following structures were proposed as typical structures of the mucin-type sugar chains of the glycoproteins from undifferentiated and differentiated Caco-2 cells: undifferentiated, Fuc1-2Gal1-4GlcNAc1-3Gal1-(4GlcNAc1-3Gal1-)_n4GlcNAc1-3Gal1-3GalNAc, and differentiated, Fuc1-2Gal1-3GlcNAc1-3Gal1-(4GlcNAc1-3Gal1-)_n4GlcNAc1-6(Gal1-3)GalNAc.

View larger version (23K): [in this window] [in a new window]   Fig. 6.   Lectin affinity HPLC on LAS-AAL and LAS-PVL columns of the mucin-type sugar chains of the glycoproteins from undifferentiated (top panels) and differentiated (bottom panels) Caco-2 cells. A, analysis of nonreducing termini obtained by endo--galactosidase digestion and ABEE-label by HPLC on an LAS-AAL column. Peaks a, b, and c were immobilized and subjected to enzyme-linked immunosorbent assay as described under "Experimental Procedures." Details are in the text. B, analysis of repeating units in the fraction passed through the LAS-AAL column of endo--galactosidase fragments by HPLC on an LAS-PVL column. C, analysis of cores obtained by ABEE-label at reducing termini and endo--galactosidase digestion by HPLC on an LAS-PVL column. The eluting conditions were described under "Experimental Procedures." Arrows indicate the positions of standard oligosaccharides: 1, GlcNAc1-3Gal-ABEE; 2, GlcNAc1-6(GlcNAc1-3)Gal-ABEE: 3, GlcNAc1-6(Gal1-3)GalNAc-ABEE; 4, GlcNAc1-3Gal1-3GalNAc-ABEE.

Localization of H Type 1 Structure on Differentiated Caco-2 Cells-- To examine localization of the H type 1 group on differentiated Caco-2 cells, the well differentiated Caco-2 cells were analyzed using a confocal microscope after binding to FITC-conjugated antibody or lectin. As shown in Fig. 7A, H type 1 structure was observed in the form of a sheet covering the surface of the cells. The localization of Le^b antigen structure on the cells was the same as H type 1 structure, although fluorescence intensity was low under a confocal microscope (data not shown). On the contrary, H type 2 structures occur around the cells (Fig. 7B). These results mean that H type 1 and Le^b structure is considered to be expressed on membrane-bound or secreted mucins, whereas the H type 2 structures are located in the glycoproteins on the cell membrane.

View larger version (52K): [in this window] [in a new window]   Fig. 7.   Localization of H type 1 (A) and H type 2 (B) structures in differentiated Caco-2 cells. Caco-2 cells were cultured on a glass bottom dish with 2 mM sodium butyrate after confluence and analyzed 6 days past confluence by immunofluorescence labeling with anti-H type 1 antibody followed by FITC-conjugated second antibody or FITC-conjugated UEA-I. No fluorescence was observed by using with FITC-conjugated mouse IgG or FITC-conjugated streptavidin as controls. Bar, 20 µm.

Interaction of DPP-IV with Mucin Expressing H Type 1 Structure-- Because it was shown that the H type 1 structure occurs in mucins on the cell surface of differentiated Caco-2 cells, localization of DPP-IV and H type 1 structure was studied using double-labeled immunofluorescence. Differentiated Caco-2 cells on 6 days past confluence were fixed and incubated with anti-H type 1 antibody followed by FITC-conjugated anti-mouse IgG and then PE-conjugated anti-CD26 antibody. The H type 1 group shown in green covers the cells (Fig. 8A, bottom panel). DPP-IV shown in red exists on the cell surface (Fig. 8A, middle panel), although differentiated Caco-2 cells become piled up, sometimes in a dome formation. When both kinds of fluorescence occurred at the same area, double-labeled regions are shown in yellow (Fig. 8A, top panel). DPP-IV expresses the area where the H type 1 group exists. The vertical view shown in Fig. 8B indicates that DPP-IV is adjacent to the H type 1 group. Because the H type 1 group itself is small, it is not known if the H type 1 group directly interacts with DPP-IV molecule.

View larger version (72K): [in this window] [in a new window]   Fig. 8.   Coexistence of DPP-IV and H type 1 group in differentiated Caco-2 cells. A, Caco-2 cells double labeled with R-PE-conjugated anti-CD26 and anti-H type 1 antibody followed by FITC-conjugated second antibody. Detailed procedures are described under "Experimental Procedures." The results are shown by a combination of double labeling (top panel), a single observation for R-PE-labeling (middle panel), and a combination of FITC-labeling and light microscopy (bottom panel). Bar, 20 µm. B, vertical analysis of Caco-2 cells double labeled with R-PE-conjugated anti-CD26 and anti-H type 1 antibody followed by FITC-conjugated second antibody. The results were shown by a combination of double labeling (top panel), a single observation for R-PE-labeling (middle panel), and a single observation of FITC-labeling (bottom panel). Bar, 10 µm.

In order to elucidate interactions of DPP-IV and mucins with the H type 1 group, the following experiments were performed. First, the effect of the presence of anti-H type 1 antibody on DPP-IV activity in cell homogenates was studied. Addition of anti-H type 1 antibody to homogenates prepared from the cells 6 days past confluence enhanced DPP-IV activity up to 1.3 times (Table I). This effect is specific for the case of anti-H type 1, because another antibody against Le^a, which does not react to Caco-2 cells, and BSA showed no effect (Table I). On the contrary, DPP-IV activity remained unchanged in the presence of anti-H type 1 in homogenates from the preconfluent cells, which had little expression of H type 1 structure (data not shown). In experiment 2, after addition of anti-H type 1 antibody or BSA to the cell homogenates, the mixtures were passed through a filter, and the activity of DPP-IV in the filtrates that contain less than 300 kDa was determined. Because the molecular masses of DPP-IV and MUC3 were 110 kDa and over 500 kDa, respectively, only free DPP-IV should be detected in the filtrate. With the addition of BSA only 25% of DPP-IV activity was recovered in the filtrate, but the addition of anti-H type 1 antibody enhanced the yield of the activity up to 40% (Table I). In experiment 3, after incubation of anti-H type 1 antibody or BSA protein A-Sepharose 4B was added to the mixture, and DPP-IV activity adsorbed on the beads was determined. When incubated with anti-H type 1 antibody, 30% of activity was obtained in the adsorbed fraction by protein A-Sepharose 4B, but BSA did not trap any activity (Table I). These results suggest the following. All the H type 1 groups are not covered by antibodies because of huge amounts of mucin-type sugar chains of the mucins. Some antibodies interfere with interactions between DPP-IV and the mucins and make DPP-IV free, resulting in the appearance of DPP-IV in the filtrate. Other antibodies bind to the mucins at the portion far from DPP-IV and precipitate DPP-IV together with the mucins. The activity of DPP-IV coprecipitated with mucins by anti-Le^b antibody showed half of the activity by using anti-H type 1 antibody. This result suggests that the mucins that interact with DPP-IV express Le^b antigen in addition to H type 1 antigen.

Differentiation Marker Enzyme Activities of Cells Cultured with GalNAc--O-benzyl-- Because DPP-IV interacts with mucins with the H type 1 groups on the surface of the differentiated Caco-2 cells, the DPP-IV activity was studied when the cells were cultured with GalNAc--O-benzyl, which inhibits biosynthesis of mucin-type sugar chains. Enzyme immunoassay and immunofluorescence studies showed that although a half of H type 2 still remained to the cells with GalNAc--O-benzyl, H type 1 structure practically disappeared in the proteins from GalNAc--O-benzyl-treated cells (data not shown). The DPP-IV and alkaline phosphatase activities in cell homogenates from cells in the presence of GalNAc--O-benzyl were lower than those from the cells in the absence of the reagent (Fig. 9).

View larger version (15K): [in this window] [in a new window]   Fig. 9.   Differentiation marker enzyme activities of Caco-2 cells treated with (shaded bars) or without (white bars) GalNAc--O-benzyl. Caco-2 cells 2, 6, and 8 days past confluence cultured in 2 mM sodium butyrate with or without 2 mM GalNAc--O-benzyl were sonicated and DPP-IV (A) and alkaline phosphatase (B) activities in the homogenates were determined. Results represent the means ± S.D. of three determinations.

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

We have described herein expression of the H type 1 blood group antigen in mucins as an additional character of enterocytic differentiation of Caco-2 cells. The evidence was obtained using various methods, including enzyme immunoassay of the glycoproteins, analysis of the mucin-type sugar chains, flow cytometry, and immunofluorescence. Increase in the amount of the H type 1 group in the glycoproteins from Caco-2 cells during culture is proportional to the days past confluence (Fig. 3C). This finding also correlates to the hydrolase activities associated with the brush border membrane (Fig. 3A). On the contrary, the expression of the isomeric structure, H type 2, does not change during differentiation (Fig. 2C). In this study we used sodium butyrate as an inducer of enterocytic differentiation and obtained the same result on the Caco-2 cells spontaneously differentiated (data not shown). Caco-2 cells that express H type 1 structures on the surface have higher hydrolase activities than Caco-2 cells with poor expression of H type 1 structure (Fig. 5). Expression of the H type 1 group, as well as early differentiation marker, DPP-IV, begins at a very early state of differentiation. This suggests that the H type 1 group plays an important role in differentiation of Caco-2 cells. The structural study of oligosaccharides elucidated that the H type 1 group occurs on the mucins as O-linked sugar chains (Fig. 6), and GalNAc--O-benzyl inhibited the expression of the H type 1 group on the glycoproteins. Differentiated Caco-2 cells produce MUC3 (Fig. 3B), and mRNA emerges shortly after confluence (3). Early expression of the H type 1 group and MUC3 is considered to positively contribute to ongoing processes of differentiation of Caco-2 cells. In fact, Caco-2 cells treated with GalNAc--O-benzyl do not express mucin-type sugar chains including H type 1 antigen, and these cells retain low levels of hydrolase activity (Fig. 9). In the case of another human colon cancer cell line, HT-29, which differentiates to enterocyte and mucus-secreting cells, GalNAc--O-benzyl inhibits 2,3-sialylation and blocks the intracellular transport of DPP-IV and mucins to the apical side. As a result, the cell swells because of a decrease in mucin secretion, and DPP-IV is not distributed on the brush border membrane, although cell lyzate does have this enzyme activity (16). No change of appearance between Caco-2 cells with and without GalNAc--O-benzyl treatment was observed under a microscope. Obviously GalNAc--O-benzyl works on Caco-2 cells in a manner different from that of HT-29 cells. Factors to be considered are kinds of mucins produced by these cells and their glycosylation processes. MUC3 is the most abundant mucin in Caco-2 cells, although mucin biosynthesis in this cell is generally low (17). On the contrary HT-29 cells produce MUC1 and MUC5AC (16, 18). The detailed structure, biosynthesis, and glycosylation of MUC3 are unknown but seem to be different from those of MUC1. 2,6-sialylation instead of 2,3-sialylation, which should be a target for transport to the apical side in HT-29 cells, increases in Caco-2 cells after differentiation (6). We also detected a higher level of 2,6-sialylation but little of 2,3-sialylation in endo--galactosidase-digested fragments from differentiated Caco-2 cells when using RCA120 column HPLC (data not shown).

Our data also prove that in well differentiated Caco-2 cells DPP-IV interacts the mucins with the H type 1 group. Immunofluorescence studies revealed that DPP-IV distribution is the same as H type 1 (Fig. 8). Furthermore, vertical observation revealed that the both molecules exist in close proximity. Studies on the effect of antibodies against sugar chains on DPP-IV activity elucidated that anti-H type 1 antibody interferes with interactions of the mucins with DPP-IV and coprecipitates DPP-IV with the mucins. These effects are specific for anti-H type 1 antibody but not for anti-Le^a antibody (Table I). A similar effect was observed by using anti-Le^b antibody, but it was weak because the reactivity of anti-Le^b antibody to the mucins was weaker than that of anti-H type 1 as shown in Fig. 2. In vivo glycocalyx including the mucins is thought to play an important role in the digestion and absorption of food by disturbing diffusion of digestives in the intestine. Also glycocalyx probably tightly retains hydrolases on the cell surface for effective digestion and adsorption. It may be that DPP-IV interacts with mucin through the Xaa-Pro sequence rather than carbohydrate moiety, such as cell-matrix interaction in liver between DPP-IV and collagen (19).

The sugar chains on glycocalyx or mucins first contact pathogens in the intestines. Recently we found that H type 1 group antigen occurs in the small intestine but not large intestine of the human with blood type O(H).² We also observed that differentiated Caco-2 cells can bind at a significantly high rate to some kinds of Vibrio cholerae³ that infect the human with blood type O(H) (20, 21). Other groups reported that blood type O(H) associates with diarrhea due to heat-labile-enterotoxic Escherichia coli (22) and that this bacteria interacts with differentiated Caco-2 cells (23). Consequently differentiated Caco-2 cells are a pertinent model for studies of interactions with these bacteria, because the cells exhibit the H type 1 group as well as human small intestine, which should be a target of the bacteria.

In differentiated Caco-2 cells, unlike Y structure, Le^b structure also increases in the amount of expression compared with undifferentiated cells. The occurrence of oligosaccharide structures containing H type 1 should rely on the change in relative activities of the glycosyltransferases, Gal1-3GlcNAc:1,2-fucosyltransferase and GlcNAc:1,3-galactosyltransferase. From analysis of the endo--galactosidase fragments by PVL-HPLC (Fig. 6B), mucin-type sugar chains from differentiated Caco-2 cells contained smaller amounts of the repeating units. This may indicate that elongation of sugar chains is reduced in mucin-type sugars as well as in N-linked sugars (4). Enhancement of fucosylation and sialylation, which compete with elongation, can explain termination of elongation even though the activities of elongation enzymes are not altered before and after differentiation (5). In addition, changes in produced mucins and differences in core structures of sugar chains (core 1 versus core 2) may affect modification of the external structures of the sugar chains. In vivo each mucin from the different cells seems to have typical sugar chain structures reflecting their functions. Further studies are required to elucidate the relationship between detailed glycosylation patterns of various mucins and the functions. Systematic analysis of mucin-type sugar chains using lectin affinity HPLC, which was developed in this study, will pave the way for such investigations.

	ACKNOWLEDGEMENTS

We express our gratitude to M. Nakasuji for skillful technical assistance, Dr. N. Hanai (Kyowa Hakko) for providing an antibody against X, M. Kamei and S. Yasuno (Honen Corporation) for providing an LAS-PVL column, and T. Furukawa and T. Ishii (Olympus) for assistance with microscope analysis.

	FOOTNOTES

^* This work was supported by a Sasakawa scientific research grant from the Japan Science Society.The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

To whom correspondence should be addressed. Laboratory of Cancer Biology and Molecular Immunology, Graduate School of Pharmaceutical Sciences, University of Tokyo, 7-3-1, Hongo, Bunkyo-ku, Tokyo 113-0033, Japan. Tel.: 81-3-3812-2111; Fax: 81-3-5841-4879.

² J. Amano, H. Morita, and M. Oshima, manuscript in preparation.

³ S. Takizawa, J. Amano, and M. Oshima, unpublished data.

	ABBREVIATIONS

The abbreviations used are: HPLC, high performance liquid chromatography; AAL, A. aurentia lectin; PVL, P. velutina lectin; GalNAc--O-benzyl, benzyl 2-acetamide-2-deoxy--D-galactopyranoside; RCA120, R. communis agglutinin 120; UEA-I, U. europaeus lectin-I; FITC, fluorescein isothiocyanate; R-PE, R-phycoerythrin; ABEE, aminobenzoyl ethyl ester; DPP-IV, dipeptidyl peptidase-IV; BSA, bovine serum albumin; PBS, phosphate-buffered saline.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES