The expression of α(1,2) fucosyltransferases that catalyze the fucose transfer to galactose of the N-acetyl(iso)lactosamine chain is decreased in human metastatic pancreatic cancer cells.α(2,3) Sialyltransferases catalyze the transfer of sialic acid to the same substrate to form, with α(1,3/1,4) fucosyltransferases,sialyl-Lewis a and sialyl-Lewis x determinants on cell surface that are involved in pancreatic metastatic invasion. The aim of this study was to determine whether this decrease of α(1,2) fucosyltransferase expression can favor the α(2,3) sialyltransferase activity to form metastatic sialyl-Lewis antigens. Restoration of α(1,2)fucosyltransferase activity in the human pancreatic cancer cell line BxPC-3 was obtained by selecting stable transfectants expressing FUT1. Overexpression of FUT1 in BxPC-3 cells resulted in a substantial reduction of sialyl-Lewis antigen expression that correlated with an increase of expression of Lewis y and H-type antigens on cell surface. The modified oligosaccharide structures were preferentially restricted to three major glycoproteins,which could in part be related to mucin-type glycoproteins. The reduction of sialyl-Lewis antigen expression was associated with an inhibition of adhesive properties to E-selectin and a decrease of gastrointestinal metastatic power of BxPC-3 cells after xenograft transplantation into nude mice. This study provides evidence that the expression level of α(1,2) fucosyltransferase may regulate the expression of sialyl-Lewis a and sialyl-Lewis x antigens and consequently could play an important role in metastatic properties of human pancreatic cancer cells.
Adenocarcinoma of the pancreas is the fourth highest cause of cancer-related deaths in the United States and has the highest mortality rate in the Western world (1). It is extremely aggressive and very resistant to current treatments, including surgery(2). Its prognosis still remains a virtual death sentence for the patient because of the delayed diagnosis of the disease, at which time liver and peritoneal metastasis generally are present(3, 4). The molecular mechanism regulating pancreatic tumor cell invasion and metastasis still remains poorly understood. One crucial step is the attachment of tumor cells to activated endothelial cells (5), which involves the expression of adhesion molecules such as selectins on the plasma membrane of these latter cells. Sialylated and fucosylated oligosaccharide determinants such as sialyl-Lewis structures expressed preferentially on the surface of circulating adenocarcinoma cells have been shown to bind to endothelial E-selectin (5, 6, 7, 8). In particular, the expression of sialyl-Lewis a may be an important mediator of the metastasis formation consecutive to pancreatic carcinoma (9, 10, 11, 12). The level of surface sialyl-Lewis a expression is significantly associated with an increased number of metastatic colonies in the liver (13).
The key glycosyltransferases that regulate the synthesis of sialyl-Lewis antigens are either α(2,3)STs,3which add sialic acid onto the galactose of type I (Galβ1,3GlcNAc-R)or type II (Galβ1,4GlcNAc-R) disaccharide, or α(1,3) Fuc-Ts and α(1,4) fucosyltransferase, which transfer fucose onto the GlcNAc for sialyl-Lewis x and sialyl-Lewis a synthesis, respectively (Fig. 1). The genes encoding the α(2,3) ST family, ST3Gal I(14), ST3Gal II (15), ST3Gal III (16) and ST3Gal IV(14, 17), have been cloned to date. We have detected the presence of ST3Gal I, ST3Gal III,and ST3Gal IV in pancreatic cancer cells.4To date, five human α(1,3) Fuc-Ts—FUT3 (18),FUT4 (19, 20), FUT5 (21), FUT6 (22, 23), and FUT7 (24, 25)—have been cloned and characterized. FUT3, which corresponds to the Lewis type Fuc-T, is also an α(1,4) Fuc-T(18). In a previous study, we demonstrated that these five Fuc-Ts were expressed in normal and tumoral human pancreas(26).
A second family of Fuc-T, which transfers a fucose onto the galactose of type I or type II disaccharide in α(1,2) linkage to form H-type I or H-type II structures, has also been characterized (Fig. 1). Two human α(1,2) Fuc-T genes, FUT1 and FUT2,which encoded the H and Secretor enzymes, respectively, have been cloned (27, 28). Lewis b, Lewis y, and H antigens were expressed in normal pancreas, whereas Lewis x, sialyl-Lewis x, and sialyl-Lewis a were detected principally in pancreatic cancer tissues(29, 30, 31, 32).
α(2,3) ST and α(1,2) Fuc-T use the same substrate (Fig. 1), and their expression levels in tumor cells could be an important factor in the formation of Lewis antigens. Some investigators have reported that the gene transfection of α(1,2) Fuc-T results in a decrease ofα(2,3) sialylation of (poly)N-acetyl(iso)lactosamine structures (33, 34, 35), suggesting a competition betweenα(1,2) Fuc-T and α(2,3) ST for their common acceptor substrate, N-acetyl(iso)lactosamine of glycoproteins and glycolipids. However, to date no study has been performed on human cancer cells in which sialyl-Lewis expression was controlled by a still unknown sophisticated system involving many glycosyltransferases. We recently observed a significant decrease in α(1,2) Fuc-T activity in tumoral pancreatic cell lines compared with normal tissue(26). These data suggested that decreased expression ofα(1,2) Fuc-T activity in correlation with α(2,3) ST, α(1,3)Fuc-T, and α(1,4) Fuc-T activities could favor the expression of sialyl-Lewis x and sialyl-Lewis a antigens on the cell surface and consecutively cell adhesion and metastasis.
To resolve this problem, we chose a reverse approach by overexpressing FUT1 in the human pancreatic tumor cell line BxPC-3, which expresses very low α(1,2) Fuc-T activity and harbors the sialyl-Lewis a and sialyl-Lewis x antigens on cell surface (26). We have shown that the restoration of FUT1 expression in human pancreatic tumor cells resulted in the decrease of sialyl-Lewis a and sialyl-Lewis x antigen expression, which correlated to an increase of Lewis y and H-type antigen expression. Consequently, a significant inhibition of E-selectin-mediated cell adhesion and a decrease of metastatic properties were observed in vivo.
MATERIALS AND METHODS
The human pancreatic carcinoma BxPC-3 cell line, obtained from the American Type Culture Collection (Rockville, MD), was grown as described previously (36).
Cells were harvested with a rubber policeman, washed in PBS, and pelleted by centrifugation. Pellets were washed twice with PBS and lysed for 1 h at 4°C in 100 μl of lysis buffer [10 mm Tris-HCl (pH 7.0), 150 mm NaCl, 1% Triton X-100] containing a mixture of protease inhibitors (Complete-EDTA free; Boehringer Mannheim, Germany). These homogenates were clarified by centrifugation at 14,000 × g for 30 min at 4°C. The concentration of protein was determined using the bicinchoninic acid assay (Pierce, Rockford, IL).(Glyco)proteins in reduced SDS buffer were loaded onto 7.5%polyacrylamide gels, separated electrophoretically, and transferred onto nitrocellulose membrane. (Glyco)proteins were stained with Ponceau red, detected by lectin, or immunodetected using mAbs to specific oligosaccharide structures as primary antibody. Detection and development were performed as described previously (37). To remove sialic acid residues, cell lysates (15 μg of proteins) were treated with 2.5 milliunits of Arthrobacter ureafaciens sialidase (Calbiochem, San Diego, CA) for 6 h at 37°C in 50 mm sodium acetate buffer (pH 5.5).
Details of the fucosyltransferase assays using N-acetyllactosamine, lacto-N-biose, and phenylβ-d-galactoside (Sigma, St. Louis, MO) as acceptor substrates have been described elsewhere (26). The different radiolabeled products were subjected to HPLC analysis on a Supelcosil LC-NH2 (4.6 × 250 mm; 5 μm) column (Supelco, Bellefonte, PA) equilibrated with acetonitrile/water (75:25, v/v) at a flow rate of 1 ml/min (LC 200 binary system; Perkin-Elmer, Norwalk, CT).
Anti-sialyl-Lewis x (clone KM-93) was from Seikagaku (Tokyo, Japan),anti-sialyl-Lewis x (clone 2H5) was from Becton Dickinson (Le Pont de Claix, France), and anti-sialyl-Lewis a (clone C241) was a generous gift from Dr. Ke Zhang. Anti-sialyl-Lewis a (clone 121SLE), anti-Lewis b (clone 2-25LE), anti-Lewis a (clone 7LE), anti-Lewis y (clone 12-4LE), and anti-H-type II (clone 19-0LE) were generously donated by Dr. Jacques Bara. Anti-Lewis x (clone SH1) was a gift from Dr. Else K. Philipsen. Anti-H-type (clone HMS2 1101A4) was obtained from Sanofi Diagnostic Pasteur (Marne-la-Coquette, France). Anti-CA19/9 (clone NCL-CA19/9) was from Novacastra (Newcastle, United Kingdom), and anti-MUC-1 (clone 4058) was from Euromedex (Souffelmeyersheim, France).
cDNA Probes and Northern Blots
The cDNA probes for FUT1 and actin were obtained by reverse transcription and PCR, using specific primers as described previously (26). Purified cDNA probes were[32P]-labeled by random priming, using[α-32P]-dCTP (NEN, Les Ullis,France) and the random primed DNA labeling kit (Life Technologies,Inc.) to a specific radioactivity of ∼4 × 108 cpm/μg. Total RNA from cultured cells was isolated following a standard protocol (38). mRNA quantitation was performed as described by Sbarra et al.(39) The amount of specific mRNA was estimated by scanning the autoradiogram on a Macintosh power PC computer, using the public domain NIH Image program. The slope of each regression line was obtained from the densitometric data. It was then possible to appreciate the relative abundance of mRNA specific for FUT1(in arbitrary units) in a mixture of total RNA.
The cDNA coding for the sequence of FUT1 was obtained by PCR from pCDM7 plasmid containing human α(1,2) Fuc-T cDNA(FUT1; Ref. 27) and was cloned into EcoRI/XhoI sites of pCDNA3-neo vector(Invitrogen, Leek, The Netherlands) to obtain pCDNA3-FUT1 plasmid. The BxPC-3 cell line was transfected with the pCDNA3-FUT1, using DAC 30 reagent (Eurogentec, Seraing, Belgium). The latter plasmid confers neomycin resistance to transfected cells. Cells were dispensed into 6-well culture plates and grown to ∼50% confluence. Growth medium was removed, and the cells were washed twice with OptiMEM (Life Technologies, Cergy-Pontoise, France) and then incubated 6 h in 1 ml of OptiMEM with 5 μg of DAC 30 reagent and 1–5 μg of plasmid DNA. Then transfection medium was replaced for 48 h with RPMI medium (Life Technologies) and with fresh medium containing 1 mg/ml of neomycin (Life Technologies). After 5–6 weeks, individual colonies were isolated using cloning cylinders. Selected clones were referred to as BxPC-3 FUT1 cells. Control cells (BxPC-3-neo) were obtained by transfection of BxPC-3 cells with the empty vector pCDNA3-neo. The cDNA for human E-selectin cloned into pCDM8 vector and obtained from R&D Systems (Abingdon, United Kingdom) was cotransfected with pCDNA3-neo vector into CHO-K1 cells, and the selected transfectant was designated as CHO-ES cells. The CHO-K1 cell line transfected with the empty pCDNA3-neo was also selected and termed CHO-neo.
Detection of fucosylated oligosaccharide epitopes on the surface of BxPC-3 cells was carried out by indirect fluorescence under the following conditions. Cells were released from culture plates by treatment with nonenzymatic cell dissociation solution (Sigma) for 15 min at 37°C. All subsequent steps were carried out at 4°C. The cells were washed three times in PBS, fixed with 2% paraformaldehyde in PBS for 10 min, and washed extensively with 1% BSA in PBS. Oligosaccharide epitopes were exposed for 1 h to specific antibodies (see above), washed three times with PBS, and finally incubated for 30 min with fluorescein-labeled antimouse IgG as secondary antibody (Sigma). Alternatively, cells were incubated with FITC-conjugated lectin for 1 h on ice. The cells were then washed,resuspended in Isoflow buffer, and analyzed on an EPICS Profile II flow cytometer (Coulter, Hialeah, FL).
Cell Adhesion Assays
The cell adhesion assays were performed using a BrdUrd labeling and detection kit (Roche Diagnostic, Meylan, France) with the following modifications. Control BxPC-3-neo and BxPC-3 FUT1 cells were labeled 24 h with BrdUrd. Excess label was removed with PBS buffer, and cells were harvested by cell dissociation solution, washed twice with PBS, and resuspended in PBS buffer supplemented with 1 mmCaCl2. Labeled cells were added to CHO-ES and control CHO-neo cells grown in 96-well plates and saturated previously with 1% BSA in PBS. After a 30-min incubation at 4°C on a rocking platform, the plates were washed three times with PBS to remove nonadherent cells. After cell fixation and denaturation with the buffers of manufacturer, BrdUrd-labeled DNA was detected using anti-BrdUrd-POD antibody according to the manufacturer’s instructions. Specific binding was determined by the measure of absorbance at 370 nm using an MR 5000 microplate spectrophotometer (Dynatech, Billingohurst,United Kingdom).
Metastasis Formation Assays in Nude Mice
Female NMRI nu/nu mice (8 weeks old) were obtained from Janvier(le Genest-St-Isle, France) and were kept in pathogen-free conditions. All surgical procedures and animal care were carried out according to accreditation number 04333 given by the French Ministère de l’Agriculture. Details of orthotopic transplantation in nude mice, which is the appropriate method for the in vivometastasis assays, have been described elsewhere (40, 41). Briefly, the human pancreatic cancer cell line BxPC-3 was cultured to 90% confluence, washed twice with cold PBS buffer, and harvested with the cell-dissociation solution. The cells were washed three times with PBS buffer and kept on ice until injection. The mice were anesthetized with i.p. injections of a mixture of xylazine (Bayer, Sens, France) and ketamine (Rhône-Mérieux, Lyon, France), the peritoneal cavity was opened, and the tumor cell suspension(106 cells in 10 μl of PBS) was injected into the pancreas, taking care to ensure that all tumor cells remained within the pancreas. Four weeks after tumor implantation, the mice were sacrificed and examined for the pancreatic primary tumor and the metastatic foci observed mainly in the peritoneum, mesenteric lymphatic duct, small intestine, mesentery, stomach, and liver. The colonization of these different tissues by pancreatic tumoral cells have been described by several investigators (13, 40, 41, 42, 43).
Development of Stable BxPC-3 Cell Clones Overexpressing α(1,2)Fuc-T, FUT1
BxPC-3 is a moderately differentiated human pancreatic adenocarcinoma cell line and was chosen for this study because it expresses a very low level of endogenous α(1,2) Fuc-T activity and high levels of endogenous α(1,3) and α(1,4) Fuc-T activities as well as sialyl-Lewis a and sialyl-Lewis x antigens on its cell surface. The cDNA encoding the human H-type α(1,2) Fuc-T (FUT1) was transfected into the BxPC-3 cells, and several stable cell clones were selected in the presence of neomycin. One of these clones, named BxPC-3 FUT1-A, was selected for further studies for the higher expression of FUT1 mRNA and α(1,2) Fuc-T activities. Control cells(BxPC-3-neo) were concomitantly transfected with the empty pCDNA3. Northern blot analyses were used to assess the relative mRNA abundance in the BxPC-3-neo and BxPC-3 FUT1-A clones. As shown in Fig. 2, quantitation of the Northern blot using FUT1-specific probe indicated that the abundance of mRNA of FUT1 was 4-fold higher in BxPC-3 FUT1-A than in BxPC-3-neo cells. The amount of dotted RNA was normalized with the actin probe. Northern blot analysis with actin probe indicated no mRNA degradation (data not shown).
Analysis of α(1,2) Fuc-T Activities in BxPC-3 FUT1-A and BxPC-3-neo Cells
To further confirm the specific overexpression of FUT1, the activity of α(1,2) Fuc-T was measured using three different acceptors: phenyl β-d-galactoside, N-acetyllactosamine, and lacto-N-biose. As shown in Fig. 3,A, the amount of [14C]fucose transferred to phenyl β-d-galactoside was significantly increased in BxPC-3 FUT1-A cells compared with BxPC-3-neo cells. N-Acetyllactosamine may also be the acceptor for fucose transfer catalyzed by an α(1,2) Fuc-T, which allows the formation of H-type II structures, or by an α(1,3) Fuc-T, which leads to the formation of Lewis x structures (Fig. 1). As shown in Fig. 3,B, we observed a significant increase of H-type II antigens(Fig. 3,B, open columns) in BxPC-3 FUT1-A compared with BxPC-3-neo cell values. A comparable increase of theα(1,2) Fuc-T activity was also observed in BxPC-3 FUT1-A cells when we used lacto-N-biose as acceptor that can be fucosylated byα(1,2) Fuc-T to generate H-type I structures (Fig. 3,C, open columns) or by α(1,4) Fuc-T to generate Lewis a structures (Fig. 3,C, hatched columns). The formation of Lewis x and Lewis a structures (Fig. 3, B and C, hatched columns) was not significantly modified between these two cell clones.
Flow Cytometric Analysis of BxPC-3 FUT1-A and BxPC-3-neo Cells with Anticarbohydrate Antibodies and Lectin
α(1,2) Fuc-T can transfer a fucose residue onto the terminal galactose of type I and type II disaccharide (Galβ1,3/1,4GlcNAc-R)and can block the accessibility of other sugar residues to these galactoside structures. In particular, sialic acid residue, as fucose residue, is known to terminate elongation of the carbohydrate chain(Fig. 1). Thus, flow cytometric analyses with anticarbohydrate antibodies and lectin were performed to compare fucosylated and sialylated oligosaccharides expressed by BxPC-3 FUT1-A and BxPC-3-neo cells. As shown in Fig. 4, B and C, a strong expression of H-type antigens was observed on the cell surface of BxPC-3 FUT1-A cells, which correlated with the increased activity of α(1,2) Fuc-T. These antigens were undetectable or poorly detectable in parental BxPC-3 cells and in control BxPC-3-neo cells. Similar profiles were obtained with two different anti-H antibodies. The same result was also obtained with UEA I lectin, which recognizes the terminal fucose of Fucα1,2Gal-R motif (Fig. 4,D). Lewis y epitopes were more or less detected in two subpopulations of BxPC-3-neo cells. However, an homogeneous increased binding of anti-Lewis y was observed in the BxPC-3 FUT1-A clone (Fig. 4,E). On the other hand, a weak modification of Lewis b expression between these two cell clones was observed (Fig. 4 F).
When the expression of sialyl-Lewis x and sialyl-Lewis a antigens was investigated (Fig. 5), we observed a substantial decrease in expression of these epitopes on BxPC-3 FUT1-A compared with BxPC-3-neo cell surface. The same results were obtained with two different anti-sialyl-Lewis x antibodies (Fig. 5, B and C), two different anti-sialyl-Lewis a antibodies (Fig. 5, D and E), and with CA19/9 antibody, which recognizes an oncofetal gastrointestinal marker associated with sialyl-Lewis a antigen (Fig. 5,F). These data indicate that the cell surface expression of NeuAcα2,3Galβ1,3/1,4[Fucα1,3/1,4]GlcNAc-R structure was inversely related to that of the Fucα1,2Galβ1,3/1,4[Fucα1,3/1,4]GlcNAc-R structure. This most likely results from a competition between the α(1,2) Fuc-T and theα(2,3) ST for the same acceptor substrate. Interestingly, the expression of Lewis x and Lewis a was not modified, indicating an unchanged level of α(1,3)- and α(1,4)-linked fucose (Fig. 5, G and H).
Glycoproteins Carrying the Modified Oligosaccharide Structures in BxPC-3 FUT1-A and BxPC-3-neo Cells
We attempted to further characterize the glycoproteins carrying modified oligosaccharide structures. Western-blot analysis of glycoproteins present in cellular lysate of BxPC-3-neo cells revealed at least three major glycoproteins immunostained with antibodies directed against sialyl-Lewis a antigens (Fig. 6). These bands, detected by the antibodies 121SLE and CA19/9, exhibited an apparent molecular size of 195 and 175 kDa (Fig. 6, B and C). Another band, migrating at 155 kDa, was detected with CA19/9 antibody (Fig. 6,C). On the other hand, we observed the lack of reactivity of sialyl-Lewis a antibodies in the BxPC-3 FUT1-A cells, whereas the Ponceau red staining of (glyco)proteins electrotransferred onto nitrocellulose membrane from BxPC-3 FUT1-A and BxPC-3-neo cell lysates did not show differences (Fig. 6,A). Treatment of BxPC-3-neo cell lysate with sialidase abolished the sialyl-Lewis a antibody reactivity, thus confirming the specificity of antibodies against sialic acid residues. Interestingly, these glycoproteins were not detected by Lewis y antibodies and UEA I lectin in BxPC-3-neo cell lysate (Fig. 6, D and E). On the other hand, these glycoproteins were recognized by Lewis y mAb and UEA I lectin in BxPC-3 FUT1-A cellular lysate (Fig. 6, D and E). These results corroborate the decreased binding of sialyl-Lewis a antibodies and the increased binding of Lewis y antibodies and UEA I lectin on BxPC-3 FUT1-A cells observed by cytofluorometry (see Fig. 5). These data suggested that the α(1,2)Fuc-T and α(2,3) ST compete for sugar transfer on the same glycoproteins of BxPC-3 cells.
At this stage it was difficult to identify these glycoproteins. However, in pancreatic cancer cells, the high molecular mass compounds present on cell surface often are related to mucin-type glycoprotein,in particular MUC-1 (44). To verify whether one of these bands corresponds to MUC-1, Western blots were performed with lysates of BxPC-3-neo and BxPC-3 FUT1-A cells. Immunodetection with the mAb disected against MUC-1 revealed the presence of two main bands with apparent molecular sizes of 195 and 155 kDa in lysate of BxPC-3-neo cells (Fig. 7), probably corresponding to the glycoproteins detected with sialyl-Lewis a antibodies (Fig. 6, B and C). On the other hand, no reaction was observed with BxPC-3 FUT1-A cells (Fig. 7). Because the reactivity of mAb MUC-1 could be increased by sialic acid residues, the lysates were treated with sialidase. As also shown in Fig. 7, this treatment abolished the reactivity of the 195-kDa protein and promoted a substantial decrease in the reactivity of the 155-kDa protein in the lysate of BxPC-3-neo cells. These data strongly suggest that glycoproteins carrying modified oligosaccharide structures could be, in part in BxPC-3 cells, related to MUC-1.
Adhesion of BxPC-3 FUT1-A and BxPC-3-neo Cells to E-Selectin-expressing CHO Cells
Previous results indicated that the binding of human pancreatic BxPC-3 cells to E-selectin is depen-dent on sialic acid residues of sialyl-Lewis structures (45). To determine the effects of overexpression of α(1,2) Fuc-T on adhesion to E-selectin, we recorded the binding of BxPC-3 FUT1-A and BxPC-3-neo cells to E-selectin-expressing CHO cells (CHO-ES). As shown in Fig. 8, we observed a significant increase of adhesion of BxPC-3-neo cells(Fig. 8, open columns) to CHO-ES cells compared with control CHO-neo cells. On the other hand, a significant inhibition of adhesion of BxPC-3 FUT1-A cells (Fig. 8, hatched columns) to CHO-ES cells was observed when compared with BxPC-3-neo cells (Fig. 8, open columns), demonstrating that adhesion of these pancreatic cancer cells to E-selectin depends on the terminal sialic acid residue linked to (poly)N-acetyl(iso)lactosamine structures.
Effects of FUT1 Overexpression on BxPC-3 Cells In Vivo
To determine whether the decrease of sialyl-Lewis structures can affect the colonization of adjacent tissues by pancreatic tumoral cells, the metastatic abilities of BxPC-3 FUT1-A and BxPC-3-neo cells were examined in nude mice. Pancreatic injection with these cells was performed, and the formation of secondary tumors was evaluated after 4 weeks. As expected, all injected mice presented tumors in the pancreas. As shown in Fig. 9, the number of mice with metastatic foci in the lymph mesentery, small intestine, and stomach was reduced in the experimental group injected with BxPC-3 FUT1-A cells compared with the control group of mice injected with BxPC-3-neo cells. None of the mice in the experimental group developed tumor nodules in the liver and mesentery. Associated with the decrease in the number of foci detected, we also noticed a reduced size of these secondary tumors, in particular the peritoneal tumors.
The expression of cell surface carbohydrate antigens in normal pancreas and their alteration in malignant neoplasms have been reported in several studies. Lewis b, Lewis y, and H-type antigens were expressed in normal pancreas, whereas Lewis x, sialyl-Lewis x, and sialyl-Lewis a were detected principally in pancreatic cancer tissues(29, 30, 31, 32). α(2,3) ST and Fuc-T were the key glycosyltransferases that regulated the Lewis antigen expression. In a previous study, we showed a significant decrease in α(1,2) Fuc-T activity in pancreatic cell lines compared with normal tissue. Twoα(1,2) Fuc-Ts, FUT1 and FUT2, which are expressed in pancreatic tissue, can be responsible for this decrease. Related to this decrease is a reduced expression of Lewis b, Lewis y, and H-type antigens on the cell surface. On the other hand, tumoral pancreatic cells presented an enhanced expression of sialyl-Lewis a and sialyl-Lewis x antigens,which are essential factors in adhesion and metastasis. This decrease of α(1,2) Fuc-T activity in cell lines, in correlation with increasedα(2,3) ST and α(1,3/1,4) Fuc-T activities, could favor the expression of sialyl-Lewis x and sialyl-Lewis a determinants on the cell surface (26). To determine whether the low level ofα(1,2) Fuc-T expression can promote α(2,3) sialylation of N-acetyl(iso)lactosamine to form the precursor of the sialyl-Lewis antigens, we overexpressed FUT1 cDNA in the human metastatic pancreatic cell line, BxPC-3.
The data presented here show that the restoration of α(1,2) Fuc-T activity in BxPC-3 cells results in increased α(1,2) fucosylation associated with decreased α(2,3) sialylation on terminal galactosyl residues of cell surface carbohydrate structures. Indeed,our results showed, for the first time in human cancer cells, a marked reduction of sialyl-Lewis a and sialyl-Lewis x antigens that correlated to an increase of Lewis y and H-type antigens, suggesting a competition between FUT1 and α(2,3) ST for the same acceptor substrate. This phenomenon has been observed in animal models by some investigators. A study comparing the types of glycans synthesized by CHO cells expressing FUT1 with those synthesized by parental CHO cells lacking FUT1 showed that FUT1 preferentially fucosylates polylactosamine structures expressed by these cells, resulting in decreased α(2,3) sialylation of these structures (33). Gorelik et al. (34) and Goupille et al. (35) transfected mouse BL6 melanoma cells and rat REGb colon carcinoma cells, respectively, with FUT1 cDNA and observed a decrease of N-acetyllactosamine sialylation.
The expression of sialyl-Lewis antigens in human cancer tissues and cell lines has been studied exhaustively. However, few studies showed a direct reciprocal relation between Lewis b, Lewis y, and H antigens, and sialyl-Lewis a and sialyl-Lewis x antigens in the same tissues or tumoral cell lines. It has been shown that the Lewis b and H antigens were present in normal human gastric cells (46),whereas sialyl-Lewis antigens were detected in the primary tumor in patients with advanced gastric cancer (47). Bryne et al. (48) and Kurahara et al.(49) showed, respectively, that the loss of expression of H antigen and the high expression of sialyl-Lewis a is associated with the metastatic potential of oral squamous cell carcinomas. Conversely,an increase of Lewis y expression associated with an increase of sialyl-Lewis a were often observed in human colorectal cancer tissues (50, 51). On the other hand, a recent study showed that Lewis b antigen expression was inversely related to Lewis a and sialyl-Lewis a expression in human colon cancer tissues, suggesting that FUT2 could compete with α(2,3) ST for the type I acceptor substrates (52). Our results suggest that in pancreatic cells, FUT1 competes preferentially with α(2,3) ST for the type II acceptor substrate because we observed a substantial increase of Lewis y antigen associated with a weak increase of Lewis b antigen expression on the cell surface (Fig. 4). This corroborates with the FUT1 specificity, which efficiently uses type II chain oligosaccharides as acceptors (27). Interestingly, the expression of Lewis a and Lewis x was not modified, indicating an unchanged level ofα(1,3)- and α(1,4)-linked fucose but also that the glycoconjugates carrying these structures are not acceptor substrates for α(1,2)Fuc-T or α(2,3) ST.
The Golgi subcompartmentation of these glycosyltransferases,which is difficult to examine, and their levels of expression seem to be an important factor in the aberrant expression of cell surface carbohydrates during tumor formation (53). Interestingly, our results showed that the competition between α(1,2)Fuc-T and α(2,3) ST is highly selective in terminal fucosylation and sialylation during glycoprotein biosynthesis. Indeed, α(1,2)fucosylation and α(2,3) sialylation in BxPC-3 cells are preferentially restricted to three major glycoproteins (Fig. 6). This selective α(1,2) fucosylation was also observed on only two glycoproteins, lysosome-associated membrane protein-1 and -2, in CHO cells expressing FUT1 (33) and on a variant of the CD44 adhesion molecule in rat colon carcinoma cells expressing FUT1 (35). In our study, we tentatively identified these glycoproteins as mucin-type glycoprotein MUC-1. Immunodetection with the mAb for MUC-1 that we have used seems to be affected by terminal sialic acid and fucose residues. This observation was supported by Sawada et al. (54), and Ho et al. (55), who have shown that the accessibility of different antibodies against MUC-1 could be altered by terminal sialic acid. It has been shown that sialyl-Lewis a and sialyl-Lewis x antigens are present on MUC-1 expressed by human pancreatic cancer cells, SW1990 (55), and could be implicated in tumor cell binding to the endothelial cell adhesion molecule E-selectin and in cellular extravasion during metastasis(56). Purified SW1990 mucin as well as pancreatic cancer sera rich in sialylated Lewis antigens can also inhibit the binding SW1990 cells to E-selectin (10). The modifications of sialyl-Lewis antigens, specifically on MUC-1, could be very useful in the development of antiadhesion therapy for cancer metastatic cells. In addition, these modifications of the glycosylation of MUC-1, which now is an attractive target for immunotherapy, could enhance the generation of both anti-MUC-1 antibodies and CTL response (57). Our results suggested a complete inhibition of adhesion of BxPC-3 FUT1-A cells on a CHO-ES cell monolayer compared with control cells (Fig. 8). These data confirm studies that have shown (a) that the expression of sialyl-Lewis a on cell surface proteins and/or lipids could be essential for the adhesion of human pancreatic cancer cells to activated endothelium mediated by E-selectin (9, 12), and(b) that the level of surface sialyl-Lewis a expression of pancreatic cancer cells correlates with the number of metastatic colonies in the liver (13).
In analogy to these in vitro experiments, we observed a decrease of the number and mass of metastatic foci in gastrointestinal tissues when BxPC-3 FUT1-A cells were xenografted in nude mice. Moreover, no difference was observed in the development of pancreatic tumors between mice xenotransplanted with BxPC-3 FUT1-A cells and BxPC-3-neo cells, suggesting that the decrease of sialyl-Lewis antigens could target only the invasiveness throughout activated endothelium mediated by E-selectin. These data corroborate results reported by Gorelik et al. (34), who reported that transfection of mouse BL6 melanoma cells with FUT1 cDNA resulted in a reduction of sialylation of N-acetyllactosamine structures and a decrease in the metastatic ability of the transfected cells. The impairment of sialyl-Lewis antigen expression could become a good challenge in the development of antiadhesion treatments of pancreatic cancer metastasis(58, 59).
It has been shown that the α(1,2) fucosylation of CD44v in rat colon carcinoma contributes to the tumorigenicity (35, 52, 60)and that the increase of Lewis b, Lewis y, and H antigens may be associated with metastatic and invasive properties of lung tumor cells(61) and bladder carcinoma cells (62). On the other hand, these same antigens were expressed in normal pancreas. Thus, Lewis b, Lewis y, and H-antigen expression could vary with tissues and with the differentiation and pathophysiological states of these tissues (29, 52, 63, 64). The amounts of specific messages for FUT1 and FUT2 are certainly a determinant factor in the regulation of sialyl-Lewis antigen expression. A recent study has identified several forms of the FUT1 transcript, generated by two transcription start codons and alternative splicing of 5′-untranslated exons in several tumor cell lines (64). These authors suggested that dual promotors regulated the stage- and tissue-specific expression of the FUT1 transcript and thus the expression of Lewis-related antigens in many human tissues. Hakomori (65) proposed a scheme for the involvement of glycosylation in tumor cell metastasis where the mechanism at each step is greatly influenced by a different glycosylation associated with expression of key adhesion molecules,such as CD44, integrin, and E-cadherin. The aberrant glycosylation may alter tumor cell adhesion or motility. In particular,the α(1,2) fucosylation of CD44 could be implicated in extracellular matrix-dependent cell adhesion and motility of tumoral cells, whereas sialyl-Lewis antigens could be implicated in the E-selectin-mediated adhesion of tumoral cells to activated endothelium (65). It has also been shown that sialyl-Lewis a and integrin mediate the process from adhesion to implantation of human pancreatic cancer SW1990 cells to endothelial cells and that CD44 and integrin play important roles in the initial attachment of these cells to mesothelium cells(66).
To modify the metastatic phenotype of pancreatic carcinoma cells, it is important to understand the mechanisms that regulate the biosynthesis of sialyl-Lewis a and sialyl-Lewis x. The expression of these structures in pancreas could be controlled by a sophisticated system involving α(1,2) Fuc-T and α(2,3) ST associated with α(1,3) Fuc-T and/or α(1,4) Fuc-T expression. For the first time, we have shown that the decrease of α(1,2) Fuc-T expression in human cancer cells could favor the α(2,3) ST activity to form metastatic sialyl-Lewis antigens. An investigation of the expression level of α(2,3) ST andα(1,3) and (1,4) Fuc-T gene products in normal and neoplastic human pancreatic tissues (and model cell lines) would be required to gain more information concerning the regulation of the expression of sialyl-Lewis determinants on the cell surface, which is implicated in cell adhesion and metastasis, and to develop antiadhesion cancer therapy.
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 9506 from the Association pour la Recherche sur le Cancer (Villejuif, France) and 930/25LC/64 from the Ligue Nationale Contre le Cancer (Comité des Régions PACA et Corse, France; to E. M.). M. A., and L. P. are recipients of doctoral fellowships from Ministère de l’Enseignement Supérieur, de la Recherche et de la Technologie.
The abbreviations used are: ST,sialyltransferase; GlcNAc, N-acetylglucosamine; Fuc-T,fucosyltransferase; mAb, monoclonal antibody; N-acetyllactosamine, type II precursor,Galβ1,4GlcNAc; lacto-N-biose, N-acetyl(iso)lactosamine, type I precursor,Galβ1,3GlcNAc; HPLC, high-pressure liquid chromatography; CHO,Chinese hamster ovary; BrdUrd, 5-bromo-2′-deoxyuridine.
We thank Dr. John B. Lowe (University of Michigan, Ann Arbor, MI) for the gift of pCDM7-FUT1 vector and Drs. Jacques Bara(Hospital St Antoine, Paris, France), Else K. Philipsen (Sankt Elisabeth Hospital, Kobenhaum, Denmark), and Ke Zhang (University of Göteborg, Sweden) for the gifts of different antibodies.