Adhesion of human colon carcinoma variant cell lines expressing different levels of the cell surface sialyl Lewis X (sLeX)antigen to frozen sections of mouse liver was examined. KM12-HX cells that bound the monoclonal antibody (mAb) FH6 (anti-sLeX)and thus expressed a high level of sLeX demonstrated a greater degree of adhesion to liver sections than their low-binding counterparts, KM12-LX cells. The adhesion of KM12-HX cells to liver sections was partially blocked by mAb FH6, but not by another anti-sLeX mAb, KM93. The adhesion was Ca2+dependent but was not inhibited by anti-E-selectin. Endo-β-galactosidase treatment significantly reduced adhesion and resulted in the loss of cell surface binding sites for mAb FH6. O-linked oligosaccharides from KM12-HX cells incubated in the presence of p-nitrophenyl-N-acetylgalactosaminide were fractionated by a combination of gel filtration, anion exchange chromatography, and normal phase high-performance liquid chromatography. The structure of a mAb FH6-reactive and endo-β-galactosidase-sensitive glycan was estimated by matrix-assisted laser desorption ionization time-of-flight mass spectrometry in a post source decay mode and by glycosidase digestions to be NeuAcα2–3Galβ1-4GlcNAcβ1-3Galβ1-4GlcNAcβ1-3Galβ1-4(±Fucα1-3)GlcNAcβ1-6(NeuAcα2-3Galβ1-3)GalNAc-pNP. Mild detergent lysates of mouse liver surface-labeled with sulfo-NHS biotin were incubated with glutaraldehyde-fixed monolayers of KM12-HX cells, and bound components were isolated after EDTA treatment. A Mr 49,000 component that bound only to KM12-HX cells and not to KM12-LX cells was identified.
Interactions of cell surface glycans and endogenous lectins from diverse families regulate trafficking of a variety of mobile cells. Metastatic tumor cells are thought to use similar mechanisms to disseminate to and colonize specific organs (1). Glycans containing terminal sLeX 3function as ligands for selectins (2, 3, 4) and are postulated to influence metastatic processes. Furthermore, the fact that tumor cell expression of a subtype of sLeXthat is recognized by mAb FH6 (5) inversely correlates with the survival of colorectal carcinoma patients strongly suggests that this carbohydrate epitope plays a biological role in malignant behavior (6, 7, 8). An important and unanswered question has been whether or not selectins are involved in this process(9). Although counterligands for L- and P-selectins have been identified on the surface of endothelial cells and granulocytes,respectively, the counterligands for E-selectin remain to be determined(10, 11). Furthermore, mucins containing sLeX seem to function as selectin ligands due to their multivalent configurations (12, 13). The issue is further complicated by the fact that there are structural variations of extended lactosamine-type selectin ligands that do not contain the sLeX tetrasaccharide (14, 15). Similar variations are apparently present in sLeX-specific mAbs. We have previously studied the ability of two anti-sLeX mAbs, FH6 and KM93,to bind to 16 human colon carcinoma cell lines by flow cytometry(16). The binding profiles of these two anti-sLeX mAbs did not correspond to each other. Three cell lines reactive with mAb FH6 but not with mAb KM93 did not adhere to E-selectin (16). Thus, a subtype of sLeX carbohydrate epitopes recognized by mAb FH6 does not always function as a ligand for E-selectin.
In the present study, we selected human colon carcinoma variant cells for expression of high or low levels of cell surface mAb FH6-reactive sLeX (17). These cells were used to elucidate the involvement of cell surface mAb FH6-reactive sLeX in the adhesion of human colon carcinoma cells to liver tissues by the Stamper-Woodruff assay (18). The adhesion was shown to be dependent on endo-β-galactosidase-sensitive surface glycans on KM12-HX cells. From KM12-HX cells, an endo-β-galactosidase-sensitive and mAb FH6-reactive O-glycan was isolated. This suggests the presence of a novel recognition molecule in the liver that binds to this glycan. A putative recognition molecule binding these glycans was identified in the liver.
MATERIALS AND METHODS
Cell Lines and Cell Culture.
Human colon carcinoma variant cells with different mAb FH6 binding levels were selected from the parental cell line KM12 C(19) using a fluorescence-activated cell sorter(17). After four selection cycles, stable variant cells with high and low expression of sLeX were isolated and designated as KM12-HX and KM12-LX cells, respectively(17). CHO cells transfected with mouse E-selectin cDNA were provided by Drs. Katsunari Tezuka and Takuya Tamatani(Pharmaceutical Frontier Research Laboratories, Japan Tabacco Inc.,Yokohama, Japan; Ref. 20). All cell lines were grown in a 1:1 mixture of DMEM:F12 containing 10% FCS in a humidified atmosphere with 5% CO2 at 37°C.
mAb KM93 (mouse IgM, specific for sLeX) was from Kyowa Hakko Kogyo (Machida, Japan). mAb FH6 (mouse IgM, specific for sLeX) was provided by Dr. Sen-itiroh Hakomori(Northwestern Research Institute, Seattle, WA) and used after purification by gel filtration. Monoclonal anti-mouse E-selectin antibody (mouse IgM) was provided by Drs. Katsunari Tezuka and Takuya Tamatani (20).
Flow Cytometric Analysis.
Flow cytometric analysis was performed using Cyto ACE-150 (Jasco Co.,Tokyo, Japan) or Epics XL (Beckman Coulter, Inc., Fullerton, CA). The indirect immunofluorescence method was applied for the staining of colon carcinoma cells with mAb FH6. The cells (1 × 106) were incubated with mAb FH6 for 30 min at 4°C at a concentration of 10 μg/ml. Fluorescein-conjugated goat affinity-purified antibody to mouse immunoglobulins (Cappel, Inc., West Chester, PA) was used as the secondary antibody.
Cell Adhesion Assays.
The frozen section adhesion assay was adopted from the procedure of Stamper and Woodruff (18). Normal mouse (male, specific pathogen-free BALB/c) liver tissues were surgically resected and rapidly frozen in liquid N2. Cryostat sections(10 μm thick) were mounted on glass slides and stored at −20°C until use. Cultured colon carcinoma cells were detached from culture dishes by treatment with a mixture of 0.02% EDTA and 0.05% trypsin and suspended in DMEM:F12 containing 10% FCS at a concentration of 107 cells/ml. BCECF-AM (Wako Pure Chemical,Tokyo, Japan) was added to the cell suspension at a final concentration of 3 μm. The cell suspensions were incubated at 37°C for 30 min and then rinsed to remove excess BCECF-AM. Labeled KM12-HX or KM12-LX cells were suspended in assay medium (DMEM:F12) containing 0.1% BSA, and the pH was adjusted to 7.4 using HEPES at a concentration of 1 × 106cells/ml. The assay was initiated by layering 100 μl of BCECF-AM labeled colon carcinoma cell suspension (1 × 106 cells/ml) kept at 4°C in assay medium onto the liver sections. Slides were rotated at 60 rpm for 30 min at 4°C. Unbound cells were removed by decantation, and the slides were rinsed by dipping them repeatedly in cold PBS. Sections were then fixed in 3%glutaraldehyde diluted in Dulbecco’s modified PBS. Cells bound to liver sections were counted under a fluorescence microscope (×100, six fields/section). All assays were performed in triplicate. Data points represent mean ± SE. The effect of mAb FH6 or KM93 on the adhesion of colon carcinoma cells to liver tissue was tested by preincubation of the colon carcinoma cells with one of these antibodies at 50 μg/ml for 30 min at 4°C before the assays. The effect of EDTA treatment was tested by treating BCECF-AM labeled cells with EDTA (50 mm) for 30 min before the adhesion experiment. To test the effect of modification of cell surface carbohydrate chains, the cells were treated with endo-β-galactosidase from Escherichia freundii (Seikagaku Kogyo,Tokyo, Japan) at a final concentration of 83 milliunits/ml for 4 h at 37°C before BCECF-AM labeling of the cells. Monoclonal anti-mouse E-selectin antibody was preincubated with mouse liver sections or CHO cells transfected with mouse E-selectin cDNA and grown to confluence on chamber slides at 50 μg/ml for 30 min at 4°C before the adhesion assays to determine inhibitory activity.
Preparation of O-Glycans.
KM12-HX and KM12-LX cells were cultured in the presence of 2 mm pNP-GalNAc and 5 μCi/ml[3H]glucosamine for 48 h.
The culture supernatant was filtered through CentriPrep 10 (Amicon,Beverly, MA). The filtrate was concentrated to 1 ml by a rotary evaporator and fractionated on a column of Bio-Gel P-2 (extrafine,1.6 × 100 cm) with 0.2 m ammonium acetate at a flow rate of 4 ml/h. An amide-80 column (SenshuPak Amide-80, φ0.5 × 15 cm; Senshu-kagaku, Japan) was then loaded, and the sample was eluted at a flow rate of 1 ml/min with acetonitrile for 5 min and then eluted with a gradient of acetonitrile and 0.2 m acetate-triethylamine buffer (pH 7.4) diluted at an equal ratio with acetonitrile (0:100 to 65:35 for the first 25 min and 65:35 to 90:10 for the next 30 min). The effluent was monitored by absorbance at 303 nm with a Gulliver UV-975. One-ml fractions were collected, and aliquots were counted for radioactivity. The purified oligosaccharides were dissolved in 100 μl of H2O, applied to a Mono Q HR5/5 column (0.5 × 5 cm; 1 ml; Pharmacia,Uppsala, Sweden), and eluted with a gradient of 0–0.1 mNaCl in 2 mm Tris-HCl (pH 7.4). The effluent was monitored by absorbance at 303 nm and for radioactivity. Sepharose 4B resins conjugated with mAb FH6 or mAb KM93 were prepared with cyanogen bromide-activated Sepharose 4B (3 mg purified antibody/ml resins). Twenty μl of oligosaccharides were applied to each column, followed by elution with PBS. Fractions (50 μl each) were collected, and the radioactivity was measured.
Structural Characterization of O-Glycans.
To determine the structure of an endo-β-galactosidase-sensitive and mAb FH6-binding oligosaccharide, MALDI-TOF MS was carried out on a Voyager Elite Biospectrometry Workstation equipped with a reflector (PE Biosystems, Foster City, CA). Ionization was accomplished by a 337 nm beam from a nitrogen laser. Ten mg of 2,5-dihydroxybenzoic acid were dissolved in 1 ml of an ethanol/water solution at a 1:1 ratio in the presence of 0.1% trifluoracetic acid. Samples were prepared by mixing oligosaccharide solutions with a matrix solution at a ratio of 1:1. Under the PSD mode, fragmentation was induced by increasing the laser intensity. By adjusting the voltage applied to the reflector (PSD Mirror Ratio setting), different fragments were focused and detected. By using the timed ion selector feature, the ion of interest can be analyzed selectively without interference from other compounds.
Endo- and exoglycosidase digestion of O-glycans was performed as follows. pNP-oligosaccharide was dissolved in 50 μl of 0.2 m sodium acetate (pH 4.5) containing 10 milliunits of endo-β-galactosidase (Seikagaku Corp., Tokyo, Japan)and 1% BSA, and the solution was incubated at 37°C for 18 h. The resulting oligosaccharides were analyzed on a column of SenshuPak Amide-80 under the same conditions described above or on Bio-Gel P-4(0.5 × 100 cm) kept at 50°C and eluted with water at 0.2 ml/min. Sialidase treatment was performed in 50 μl of 0.2 m sodium acetate (pH 4.5) containing pNP-oligosaccharide, 100 milliunits of sialidase (Arthrobacter urecasis; Nacalai Tesque, Kyoto, Japan), and 1% BSA at 37°C for 18 h. After digestion, the resulting oligosaccharides were analyzed on a column of SenshuPak Amide-80 or Bio-Gel P-4. A mixture of 10 milliunits of β-N-acetylhexosaminidase from jack bean(Seikagaku Kogyo) and 10 milliunits of β-galactosidase from Diplococus pneumonia (Seikagaku Kogyo) was applied to pNP-oligosaccharides in 0.2 m sodium acetate buffer (pH 4.5). Incubations were performed at 37°C for 18 h. After digestion of the oligosaccharide, the resulting oligosaccharides were analyzed on a column of SenshuPak Amide-80.
In Situ Biotinylation of the Hepatic Microvasculature.
The liver of mice (male, specific pathogen-free BALB/c) anesthetized with Avertin was exposed. The portal vein was cannulated with a 24-gauge Surflo i.v. catheter (Terumo, Tokyo, Japan) and perfused with 40 ml of PBS supplemented with 1 unit/ml heparin, 1%MEM, and 15 mm glucose at a flow rate of 3 ml/min to remove circulating cells, leukocytes, and plasma protein. Ten ml of saline were subsequently applied to remove the MEM. Sulfo-NHS-biotin (5 mg/10 ml; Pierce, Rockford, IL) was dissolved in saline and injected into the portal vein to biotinylate all surface proteins in the hepatic microvasculature. The liver was excised and minced in 50 ml of Gey’s balanced salt solution supplemented with 0.05% collagenase (Wako) and 0.005% DNase I (Boehringer Mannheim Biochemica, Mannheim, Germany) for 30 min at 37°C. The minced tissue suspensions were rinsed and filtered through a 75-μm mesh nylon screen. The cell suspensions were rinsed twice in 0.25 m sucrose, 10 mm Tris-HCl(pH 7.2), 0.05 mm CaCl2, and 10μ m phenylmethylsulfonyl fluoride and then solubilized in the same solution containing 0.5% NP40 at 4°C for 2 h. The supernatants were collected by centrifugation at 13,000 × g for 5 min and used as lysates of mouse livers.
Identification of a Recognition Molecule for mAb FH6-binding Sites.
Biotin-labeled lysates of mouse livers were overlaid onto monolayers of KM12-HX or KM12-LX cells (4 × 106cells/well) previously fixed with 0.25% glutaraldehyde for 30 min at room temperature. Liver lysates corresponding to 3 mg of protein were used in a single well of 12-well multiwell plates. Incubation was performed at 4°C overnight. Unbound components were removed, and the plates were rinsed gently four times with desalting buffer. Bound components were eluted with 200 μl of 100 mm EDTA solution (pH 7.2). The bound and eluted fractions were analyzed by SDS-PAGE (10% running gels) under reducing conditions and blotted onto polyvinylidene difluoride membranes (Millipore, Bedford, MA) using a Milli Blot-SDE system (Millipore). The membranes were soaked in PBS containing 2% BSA at 4°C overnight to block nonspecific binding. The membrane was incubated with alkaline phosphatase-conjugated streptavidin (diluted at 1:1000) for 45 min. After washing three times with PBS containing 0.1% Tween 20, the membranes were reacted with alkaline phosphatase-conjugated streptavidin. An alkaline phosphatase substrate kit II (Vector Laboratories) was used to visualize bound components.
Expression of sLeX on KM12-HX and KM12-LX Cells.
KM12-HX and KM12-LX cells selected previously for high and low levels of cell surface binding of mAb FH6, respectively, were tested for the binding of two different anti-sLeX mAbs with a flow cytometer. KM12-HX cells expressed high levels of the mAb KM93 epitope and the mAb FH6 epitope (presumably an extended sLeX) as shown in Fig. 1 A, although the cells were heterogeneous in terms of the levels of expression of these epitopes.
Adhesion of KM12-HX and KM12-LX Cells to Mouse Liver Frozen Sections in the Stamper-Woodruff Assay.
Distribution patterns of KM12-HX and KM12-LX cells to mouse liver frozen sections visualized under a fluorescence microscope (×100) are shown in Fig. 1,B. KM12-HX cells had a higher degree of adhesion than KM-LX cells on all portions of mouse liver sections. Quantitative adhesion of BCECF-AM-labeled KM12-HX and KM12-LX cells to mouse liver sections are shown in Fig. 2,A. The number of KM12-HX cells that adhered to frozen sections of mouse livers was greater than that of KM12-LX cells(P < 0.05). When KM12-HX cells were treated with EDTA (50 mm), the adhesion of KM12-HX cells was reduced to approximately 65% of the adhesion obtained with untreated cells (P < 0.05; Fig. 2,A). When KM12-HX cells were treated with mAb FH6, their adhesion to mouse liver sections was reduced to ∼58%(P < 0.05). Another anti-sLeX mAb, KM93, did not show such effects(Fig. 2,B). Subsequently, the effect of modification of cell surface carbohydrate chains on adhesion was examined. KM12-HX cells treated with endo-β-galactosidase were analyzed for sLeX expression by a flow cytometer using mAb FH6 or mAb KM93. Treatment of KM12-HX cells with endo-β-galactosidase significantly reduced the binding of mAb FH6 without affecting the binding of mAb KM93 (Fig. 3,A; P < 0.05). Adhesion of KM12-HX cells treated with endo-β -galactosidase to liver sections under the same conditions was reduced to ∼73% of the adhesion obtained with untreated cells (Fig. 3,B; P < 0.05). The effect of a mAb specific for mouse E-selectin was tested. Adhesion of KM12-HX cells to mouse liver sections was not influenced by preincubation of the liver sections with 50 μg/ml of this mAb (Fig. 4). Adhesion of KM12-HX cells to CHO cells stably transfected with mouse E-selectin cDNA was also investigated. KM12-HX cells adhered to CHO cells stably transfected with mouse E-selectin cDNA; however,monoclonal anti-mouse E-selectin antibodies inhibited this adhesion.
Preparation of O-linked Oligosaccharides from KM12-HX and KM12-LX Cells.
As we have shown previously with benzyl-N-acetylgalactosaminide (21), KM12-HX cells cultured in the presence of pNP-GalNAc were devoid of mAb FH6-binding sites on the cell surfaces. Thus, spent media of KM12-HX cells cultured in the presence of pNP-GalNAc and[3H]glucosamine were used as a source of O-glycans containing mAb FH6-binding sites.
Bio-Gel P-2 gel permeation chromatography with 0.2 mammonium acetate resulted in fractions H0, H1, and H2 from KM12-HX cells and fractions L1 and L2 from KM12-LX cells according to the order of elution. Fractions H1 and H2 from KM12-HX cells and fractions L1 and L2 from KM12-LX cells were further purified by Amide-80 absorption chromatography. Each fraction from Bio-Gel P-2 (H1, H2, L1, and L2)eluted as one major peak in each case. Fractions H0, H1, H2, L1, and L2 were further fractionated by Mono Q anion-exchange column chromatography. Fraction H0 from KM12-HX cells separated into three peaks on Mono Q chromatography and were designated H0-a, H0-b, and H0-c according to the order of negative charge. Fraction H1-2 was separated into two peaks, neutral H1-a and acidic H1-b. Similarly, fraction H2-2 was separated into H2-a and H2-b. Fraction L1-2 was separated into L1-a and L1-b. Fraction L2-2 was separated into L2-a and L2-b.
Acidic fractions H1-b, H2-b, L1-b, and L2-b were rechromatographed on Amide-80. Fraction H1-b was further separated into two peaks (H1-αand H1-β). Fractions L1-b, H2-b, and L2-b were also separated into L1-α and L1-β, H2-α and H2-β, and L2-α and L2-β,respectively. The elution profiles of these fractions on Bio-Gel P-2,Mono Q, and Amide-80 chromatography suggested that the fractions from KM12-HX cells designated as H1-α, H2-β, H2-α, and H2-β were the same oligosaccharides from KM12-LX cells, i.e., L1-α,L1-β, L2-α, and L2-β. We assumed that mAb FH6 bound acidic oligosaccharides and processed acidic fractions. The purification scheme and the relative contents of the acidic fractions estimated by the radioactivity are shown in Fig. 5.
mAb FH6-Sepharose Affinity Chromatography of Fractions from KM12-HX and KM12-LX Cells.
O-Glycans from KM12-HX cells (fractions H0-c, H1-α,H1-β, H2-α, and H2-β) were subjected to affinity chromatography with mAb FH6-Sepharose. After treatment with sialidase, these fractions were used as negative controls. As shown in Fig. 6, fractions H0-c, H1-β, and H2-β were slightly retarded by a mAb FH6-Sepharose column. Such elution patterns were not observed with any other O-glycans from KM12-HX cells or KM12-LX cells.
Structure of Fraction H0-c.
The structure of fraction H0-c was further investigated because this fraction was endo-β-galactosidase sensitive and had affinity with mAb FH6. Fraction H0-c treated with sialidase was analyzed by MALDI-TOF MS in the positive ion reflector mode. The MALDI-TOF MS spectrum of this fraction is shown in Fig. 7,A. The predominant pseudomolecular ions of fraction H0-c were detected at m/z 1758.08 and 1608.8. Spectra of metastable decompositions and the PSD of this fraction were recorded at various reflector potentials. These spectra were assembled to form a reconstructed PSD mass spectrum by computer software, the Grams,resulting in the detection of fragment ions at m/z 1758.52,1255.95, and 714.467 (Fig. 7 B). These spectra indicated that fraction H0-c contained one methylpentose (likely to be fucose) residue and three repetitive units of hexose-N-acetylhexosamine.
When fraction H0-c was treated with endo-β-galactosidase, the major portion was converted into peaks corresponding to oligosaccharides that had higher molecular weights than Galβ1-4(Fucα1-3)GlcNAcβ1-6(Galβ1-3)GalNAc-pNP (desialized H2-b;its structural determination will be published separately) on Amide-80(Fig. 8 A). This suggested that fraction H0-c contained internalβ1-4 galactosyl linkages and linear repeats of Gal-GlcNAc. A fucose should be linked to the innermost N-acetylglucosamine residue because the major product was estimated to correspond to GlcNAcβ1-3Galβ1-4(Fucα1-3)GlcNAcβ1-6(NeuAcα2-3Galβ1-3)GalNAc-pNP but not GlcNAcβ1-6(NeuAcα2-3Galβ1-3)GalNAc-pNP. The latter should be the digestion product of NeuAcα2-3Galβ1-4GlcNAcβ1-3Galβ1-4(Fucα1-3)GlcNAcβ1-3Galβ1-4GlcNAc(NeuAcα2-3Galβ1-3)GalNAc or the digestion product of another derivative with a fucose linked to the outermost N-acetylglucosamine residue. These products were estimated by using the previously published specificity of this enzyme (22).
As shown in Fig. 8 B, the Amide-80 chromatography was also performed to analyze H0-c after treatment with sialidase,β-galactosidase from Diplococus pneumonia, andβ-N-acetylhexsosaminidase. After this treatment, fraction H0-c was converted into two peaks, corresponding to Galβ1-4(Fucα1-3)GlcNAcβ1-6(Galβ1-3)GalNAc-pNP and Galβ1-3GalNAc-pNP. Thus, the structure of fraction H0-c was proposed to be NeuAcα2-3Galβ1-4GlcNAcβ1-3Galβ1-4GlcNAcβ1-3Galβ1-4(±Fucα1-3)GlcNAcβ1-6(NeuAcα2-3Galβ1-3)GalNAc-pNP.
Mouse Liver Proteins That Recognize mAb FH6-binding Oligosaccharides.
Lysates of livers were obtained after surface biotinylation. Proteins from the lysates that bound KM12-HX were isolated by overnight incubation. The bound components were eluted with EDTA and electrophoretically separated on 10% gels under reducing conditions(Fig. 9). The bound fraction contained multiple components and had a wide range of molecular weights. Among these components, a biotinylated protein from mouse liver that had a Mr of 49,000 was found to bind KM12-HX cells to a greater extent than KM12-LX cells. Thus, this component may represent a unique molecule that recognizes cell surface structures expressed on KM12-HX cells but not on KM12-LX cells. A candidate of such structures specific for KM12-HX cells is the oligosaccharide H0-c.
Carbohydrate structures having NeuAcα2-3Galβ1-4 (Fucα1-3)GlcNAc as the terminal residue are called sLeX antigens. We demonstrated previously that the expression of sLeX antigens recognized by mAb FH6, specific for a putative subtype of sLeX, was greater at the region of hepatic metastasis of human colon carcinoma than at the primary site (6, 7). The content of sLeX carbohydrate antigen as recognized by mAb FH6 was shown to correlate with the progression of human colon carcinomas to the advanced stages (6, 8).
The sLeX antigens expressed on the surface of various human carcinomas are known to be involved in the adhesion of these cells to endothelial cells, and the adhesion is thought to be mediated by E-selectin (23). Thus, it might be hypothesized that sLeX antigens recognized by mAb FH6 on the surface of colon carcinoma cells play an important role in the process of metastasis formation such as adhesion to endothelial cells in the liver, although such an argument is quite controversial(9). Furthermore, sLeX antigen is not a single entity but is considered to be a group of structurally related carbohydrate chains. In our previous report, mAb FH6 and another anti-sLeX mAb, KM93, were used in flow cytometry to determine their ability to bind to 16 human colon carcinoma cells. The binding profile of mAb KM93 did not correspond to that of mAb FH6. Three of the 16 cell lines were reactive with mAb FH6 but not with mAb KM93. These three cell lines did not adhere to CHO cells that were stably transfected with human E-selectin cDNA. In contrast, almost all human colon carcinoma cell lines that bound to mAb KM93 also adhered to cells that expressed E-selectin. These results suggest that a subtype of sLeX carbohydrate epitopes recognized by mAb FH6 does not function as a ligand for E-selectin. From the present study, we propose that the structure recognized by mAb FH6 functions as a ligand for a recognition molecule present in the liver.
We have been investigating the possible structure and biological function of sLeX and related carbohydrates using human colon carcinoma variant cells that were selected for high or low cell surface expression of sLeX antigens. These variant cells, i.e., KM12-HX cells and KM12-LX cells, were obtained by fluorescence-activated cell sorting using mAb FH6(17). When these cells were tested for adhesion to sections of livers (mice or humans), more KM12-HX cells adhered than KM12-LX cells. The present study is concerned with the mechanism of adhesion to mouse livers. A similar observation with human liver will be published separately. Our findings indicate that endo-β-galactosidase-sensitive carbohydrate chains on the surface of KM12-HX cells are responsible for the adhesion of these cells to mouse liver sections. Because the number of adherent KM12-HX cells was much greater than that of KM12-LX cells, the carbohydrate chains are also likely to be recognized by mAb FH6 (17).
The mAb FH6 was originally prepared against sialyl dimeric LeX glycolipids (5). This mAb was subsequently reported to have affinity for other sLeX-related antigens that have extended poly-N-acetyllactosamine backbones (24). In a study using glycolipid acceptors (25), it was shown that the specificity of mAb FH6 was different from that of other anti-sLeX mAbs, such as mAb KM93 that recognizes the NeuAcα2-3Galβ1-4 (Fucα1-3) GlcNAc tetrasaccharide. When KM12-HX cells and KM12-LX cells were compared for their binding of mAb FH6 and mAb KM93, KM12-HX cells had a greater number of binding sites for both of these mAbs compared with KM12-LX cells. Thus, it is reasonable to speculate that mAb KM93 recognized and bound to terminal portions of the same carbohydrate chains recognized by mAb FH6(25). However, this does not always seem to be the case,as shown in the present study. One of the most striking findings was that the epitope for mAb KM93 was insensitive to endo-β-galactosidase treatment (Fig. 3). In other words, the mAb KM93 epitope on KM12-HX cells is not at the nonreducing termini of endo-β-galactosidase-sensitive backbones, despite the fact that this backbone structure is apparently a portion of the mAb FH6 epitope.
Our previous observations with KM12-HX cells indicated that O-linked carbohydrate chains carry the majority of epitope carbohydrate chains for mAb FH6. Furthermore, these cell surface epitopes could be depleted by incubating the cells with aryl-α-GalNAc(1, 21). Thus, we used this method to obtain O-glycans produced by these cells. We obtained an endo-β-galactosidase-sensitive and mAb FH6-reactive O-glycan, H0-c. The structure was estimated to be NeuAcα2-3Galβ1-4GlcNAcβ1-3Galβ1-4GlcNAcβ1-3Galβ1-4(±Fucα1-3)GlcNAcβ1-6(NeuAcα2-3Galβ1-3)GalNAc-pNP. Surprisingly, it did not seem to contain a sLeXat its terminus. Its estimated structure was an extended poly-N-acetyllactosamine on the core two branches with a fucose at the innermost GlcNAc that attached to the GalNAc linked to the pNP residue. Although its affinity with mAb FH6 seemed to be low,no other oligosaccharide fractions had both characteristics of sensitivity to endo-β-galactosidase and affinity with mAb FH6. We demonstrated that KM12-HX cells express an increased amount of fucosyltransferase VI (data to be published separately). This enzyme has recently been found to transfer fucose at position 3 of GlcNAc residues such that the preferential site was the innermost GlcNAc(26). Therefore, it is not surprising to find a structure like H0-c having an extended polylactosamine and fucosylation at the innermost GlcNAc. It should be noted that carbohydrate chains that extend on pNP-GalNAc may not represent the entire glycosylation pattern on the cell surfaces.
Whether the mechanism of adhesion to the liver sections has any link to the process of metastasis formation in livers is unknown. In our previous work in which KM12 variant cells were established, KM12-HX cells showed increased metastatic potential to the liver in nude mice compared with KM12-LX cells (21). However, after passages,these two variant cell lines showed similar liver colonization ability,but KM12-HX cells expressed significantly higher levels of cell surface sLeX and E-selectin-dependent adhesion than KM12-LX cells. Our present study has been carried out using these cells at relatively high passages. The initial location of hepatic metastasis formation of colon carcinomas is known to be the area surrounding the portal vein, and the location of experimental metastasis from KM12-HX cells in nude mice is also in the same area. In contrast, the distribution of KM12-HX cells adherent to liver sections was uniform,suggesting that this adhesion might have little correlation to mechanisms of metastasis formation in livers. However, it should be emphasized that the level of expression of the mAb FH6 epitope on colon carcinoma cells has reverse correlation with the survival of colon carcinoma patients. The adhesive interactions may be involved in the process of establishment of metastatic foci from micrometastases formed by newly arrived cells. The significance of extended O-glycans in colon carcinoma progression has also been suggested, based on the expression of a core 2-N-acetylglucosaminyl-transferase (27).
To identify the molecule on liver cells that binds the colon carcinoma cells, we examined exposed proteins on the surface of the liver vasculature. When the localization of biotin residues was examined histochemically, the residues were found to lie primarily along the portal vein and sinusoids. The location corresponded roughly to the sites of carcinoma cell adhesion, although the precise subcellular localization was not clear. The lysates of biotinylated liver tissue were incubated with monolayers of KM12-HX cells and KM12-LX cells. Bound components were eluted with 100 mm EDTA and separated electrophoretically on polyacrylamide gels. Multiple biotinylated components that bound to KM12-HX and KM12-LX cells were revealed. One protein in particular was found to bind KM12-HX cells to a greater extent than KM12-LX cells. Thus, a putative carbohydrate recognition molecule involved in metastasis formation has been identified. Whether or not this molecule is involved in the natural immunity suggested for the putative function of sLeX by Ohyama et al. (28) is unknown.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Supported by Grants-in-Aid 07407063, 09254101,11557180, and 11672162 from the Ministry of Education, Science, Sports and Culture of Japan and by the Research Association for Biotechnology and the Program for Promotion of Basic Research Activities for Innovative Biosciences.
The abbreviations used are: sLeX,sialyl Lewis X; BCECF-AM, 3′-O-acetyl -2′,7′-bis(carboxy ethyl)-4 or 5-carboxyfluorescein diacetoxymethyl ester; DMEM:F12,DMEM:Ham’s F-12; mAb, monoclonal antibody; MALDI-TOF MS,matrix-assisted laser desorption ionization time-of-flight mass spectrometry; pNP, p-nitrophenyl; pNP-GalNAc, p-nitrophenyl-N-acetyl-α-d-galactosaminide;PSD, post source decay; CHO Chinese hamster ovary.
We thank Dr. Sen-itiroh Hakomori for providing mAb FH6. We also thank Drs. Katsunari Tezuka and Takuya Tamatani for providing CHO cells transfected with mouse E-selectin cDNA and anti-mouse E-selectin mAb. We thank Chizu Hiraiwa for assistance in preparing the manuscript.