Article

• The EMBO Journal (1999) 18, 1516 - 1525
• doi:10.1093/emboj/18.6.1516

Dual roles of sialyl Lewis X oligosaccharides in tumor metastasis and rejection by natural killer cells

Chikara Ohyama¹, Shigeru Tsuboi¹ and Minoru Fukuda¹

1. Cancer Research Center, The Burnham Institute, 10901 North Torrey Pines Road, La Jolla, CA 92037, USA

Correspondence to:

Minoru Fukuda, E-mail: minoru@burnham-inst.org

Received 10 December 1998; Accepted 26 January 1999; Revised 26 January 1999

• Keywords:

  • NK cells,
  • selectin,
  • sialyl Lewis X,
  • tumor cell rejection,
  • tumor metastasis

Introduction

Cell surface carbohydrates are characteristic of different stages of differentiation, and distinct carbohydrates are expressed in tissue- and cell-specific manners during development and in adulthood (Feizi, 1985; Rademacher et al., 1988; Fukuda, 1996). Although the physiological roles for these carbohydrates are not understood, many of these carbohydrates are thought to play a role in cell–cell interaction. Among various cell type-specific carbohydrates, sialyl Lewis X, NeuNAc23Gal1 4(Fuc13)GlcNAcR, is expressed on neutrophils, monocytes and certain T lymphocytes (Fukuda et al., 1984; Mizoguchi et al., 1984) and plays a key role in the recruitment of leukocytes. E- and P-selectin expressed on activated endothelial cells capture these leukocytes, allowing them to roll, which leads to extravasation of leukocytes (Lowe, 1994; Rosen and Bertozzi, 1996). When lymphocytes recirculate from the intravascular compartment to the lymphatic compartment, circulating lymphocytes are captured to high endothelial venules. This process is initiated by binding of L-selectin on lymphocytes to sulfated sialyl Lewis X present in receptors that are restricted to high endothelial venules (Imai et al., 1993; Tsuboi et al., 1996). These results clearly indicate that sialyl Lewis X and its sulfated form play critical roles in the interaction between leukocytes and endothelial cells (Lowe, 1994; Rosen and Bertozzi, 1996).

In parallel with these studies, several laboratories have reported that expression levels of sialyl Lewis X are increased substantially in tumor cells such as carcinomas and leukemias (Fukushima et al., 1984; Fukuda et al., 1985; Saitoh et al., 1992; Shimodaira et al., 1997). In many carcinoma cells, the amount of sialyl Lewis A, NeuNAc23Gal13(Fuc14)GlcNAcR, is also increased (Magnani et al., 1982; Itzkowitz et al., 1988). In breast and colonic carcinoma patients, the presence of sialyl Lewis X and sialyl Lewis A was correlated with poor prognosis (Nakamori et al., 1993; Renkonen et al., 1997; Shimodaira et al., 1997). A recent study demonstrated that expression of sialyl Lewis X and sialyl Lewis A in mucin-type O-glycans is highly correlated with lymphatic and venous invasion (Shimodaira et al., 1997). Considering that sialyl Lewis A is also a ligand for E- and P-selectin (Berg et al., 1991; Takada et al., 1991), these results strongly suggest that blood-borne tumor cells may utilize selectin–carbohydrate interaction when tumor cells adhere to the endothelia distant from the site of primary tumor growth. In fact, we and others have shown previously that highly metastatic tumor cells adhere more strongly to E-selectin than do their poorly metastatic counterparts (Sawada et al., 1994; Martín-Satué et al, 1998). Tumor formation in lung was diverted to livers by making transgenic mice expressing E-selectin in livers (Biancone et al., 1996). These results strongly suggest that carbohydrates on tumor cell surfaces may be recognized by selectins on the endothelium. However, no studies have determined how different amounts of sialyl Lewis X play different roles in tumor metastasis.

Sialyl Lewis X is formed at termini of N- and O-glycan precursor carbohydrates that were pre-assembled. In particular, poly-N-acetyllactosamines are often modified to express sialyl Lewis X (Lee et al., 1990; Saitoh et al., 1992; Wilkins et al., 1996). Poly-N-acetyllactosamine is composed of N-acetyllactosamine repeats and the increased amount of poly-N-acetyllactosamine is parallel with the increase of side chain synthesized by N-acetylglucosaminyltransferase V (GnT-V) (Sasaki et al., 1987; Van den Eijnden et al., 1988). Because of this biosynthetic relationship, it has been observed repeatedly that transformed cells and metastatic tumor cells contain more poly-N-acetyllactosamine and side chains synthesized by GnT-V than the parent cells or low metastatic counterparts (Yamashita et al., 1984; Saitoh et al., 1992; Demetriou et al., 1995). These results indicate that it is important to understand the formation of terminal structures such as sialyl Lewis X in conjunction with the synthesis of underlying backbone carbohydrates (Fukuda, 1994).

To determine whether sialyl Lewis X and underlying carbohydrates play a role in tumor metastasis, we tested the ability of mouse melanoma B16 cells (Fidler, 1973) to form tumors after injection into a tail vein. We reasoned that adhesion of tumor cells to the endothelium in this system plays a primary role in establishing tumor nodules, since tumor cells are already in the bloodstream and no escape from the primary tumor site is involved. B16-FI cells were thus transfected with a cDNA encoding 1,3-fucosyltransferase III (Fuc-TIII) (Kukowska-Latallo et al., 1990), which is distributed widely in various tissues and directs the synthesis of both sialyl Lewis X and sialyl Lewis A. Transfected B16-FI cells expressing a moderate amount of sialyl Lewis X in long N-glycans (B16-FTIIIM) produced significantly increased lung tumor nodules in C57BL/6 mice. On the other hand, those transfected cells expressing the highest amount of sialyl Lewis X in short N-glycans (B16-FTIIIH) scarcely produced lung nodules as did those expressing no sialyl Lewis X in wild-type and SCID mice. B16-FTIIIH cells were found to undergo apoptosis in those mice. In contrast, B16-FTIIIH cells produced more lung nodules in beige mice or C57BL/6 mice depleted of natural killer (NK) cells than did B16-FTIIIM cells. These results demonstrate for the first time that acquisition of sialyl Lewis X leads to lung tumor formation or rejection by NK cells, depending on the amount of sialyl Lewis X and underlying glycans expressing sialyl Lewis X.

Isolation of B16-FTIII cells expressing different amounts of sialyl Lewis X

Since Fuc-TIII is distributed widely in various tissues and directs expression of both sialyl Lewis X and sialyl Lewis A, B16-F1 cells were transfected with Fuc-TIII. Stably transfected B16-FTIII cells were subjected to cell sorting with anti-sialyl Lewis X (CSLEX1) antibody. As shown in Figure 1A, B16-FTIII cells can be sorted into three groups: those highly positive for sialyl Lewis X (B16-FTIIIH), those moderately positive for sialyl Lewis X (B16-FTIIIM) and those negative for sialyl Lewis X (B16-FTIIIN). As shown in Figure 1B, B16-FTIIIH cells express as much sialyl Lewis X as promyelocytic HL-60 cells. B16-FTIIIM cells express a moderate amount of sialyl Lewis X, showing five times the intensity of anti-sialyl Lewis X antibody staining compared with B16-FTIIIN (Figure 1B). The expression levels of sialyl Lewis X in different B16-FTIII cells were stable even after five passages (see also Discussion). To ensure that the phenotype was maintained, fluorescence-activated cell sorting (FACS) analysis was always carried out before further experiments.

Figure 1.

Cell sorting and flow cytometric analysis of B16-FTIII cells. (A) B16-FTIII cells were stained with anti-sialyl Lewis X antibody followed by FITC-conjugated secondary antibody and sorted by FACStar. Cells indicated by open bars were pooled and designated B16-FTIIIN (negative), B16-FTIIIM (moderate) and B16-FTIIIH (highly positive), respectively. (B–F) Cultured B16-FTIIIN (N), B16-FTIIIM (M) and B16-FTIIIH (H) cells were subjected to flow cytometry after staining with anti-sialyl Lewis X (B), anti-sialyl Lewis A (C) or anti-Lewis X antibody (D), followed by FITC-conjugated secondary antibody, or FITC-conjugated SNA (E) or FITC-conjugated LEA (F). HL-60 cells were also stained in (B). The staining by FITC-conjugated MAA gave the same results as FITC-conjugated SNA.

View full figure (100 KB)

Metastatic properties of B16-FTIII cells expressing different amounts of sialyl Lewis X

To determine if the metastatic properties of these B16-FTIII cells differ, the three cell populations were injected into tail veins of C57BL/6 mice. We found that only a few lung nodules were formed 3 weeks after injection of wild-type B16-F1 cells or B16-FTIIIN cells (Figure 2A, top). In contrast, B16-FTIIIM cells produced a large number of lung nodules under the same conditions (Figure 2A, middle). These results, reproduced more than five times using different B16-FTIIIM clones, indicate that the metastatic potential of B16-F1 cells was remarkably increased after transfection of Fuc-TIII. To determine if sialyl Lewis X plays a critical role in this process, B16-FTIIIM cells were incubated with anti-sialyl Lewis X antibody and then injected into tail veins of mice. Tumor formation was suppressed completely by this pretreatment, while it was not suppressed by pre-incubation with an antibody specific to Lewis X (Figure 2B and C), which is a negligible ligand for selectins (Lowe, 1994; Rosen and Bertozzi, 1996).

Figure 2.

Representative lungs and number and weight of lung tumor nodules in BALB/c mice after intravenous injection of B16-FTIIIN, B16-FTIIIM or B16-FTIIIH cells. (A) The lungs were examined 3 weeks after injection. (B and C) B16-FTIIIN (N), B16-FTIIIM (M) and B16-FTIIIH (H) cells were injected intravenously, and tumors formed after 3 weeks were counted (B) and the weight of lungs was measured (C). B16-FTIIIM cells were incubated with anti-sialyl Lewis X antibody (anti-sLe^x) or anti-Lewis X antibody (anti-Le^x) at 4°C for 30 min and injected intravenously under the same conditions. The differences in both tumor nodule number and lung weight, denoted by asterisks, are statistically significant by the Mann–Whitney U-test (p <0.01 and p <0.05, respectively).

View full figure (63 KB)

Surprisingly, B16-FTIIIH cells, which were highly positive for sialyl Lewis X, produced as few tumor nodules as did B16-FTIIIN or the parent B16-F1 cells (Figure 2A, bottom). This observation was confirmed by comparing the number of lung nodules and weight of the lungs (Figure 2B and C). The same experiments also showed that melanin synthesis was not influenced by transfection of Fuc-TIII, since a good correlation exists between the number of lung nodules, determined by counting pigmented B16-FTIII cells, and the weight of lungs. These results indicate that B16-FTIIIM cells became highly metastatic after expressing a moderate amount of sialyl Lewis X, while B16-FTIIIH cells expressing the highest levels of sialyl Lewis X were poorly metastatic.

Expression of sialyl Lewis X does not increase cell growth

It is possible that B16-FTIIIM cells are more metastatic because they grow faster than B16-FTIIIN or B16-FTIIIH cells. To address this question, the cells were plated at low density and the growth rate was determined. The results clearly demonstrate that expression of sialyl Lewis X did not alter the rate of tumor cell growth when assayed by a cell proliferation assay (Figure 3). These results exclude the possibility that B16-FTIIIM cells grow faster than the other two cell lines, demonstrating that a difference other than growth rate contributed to the formation of lung tumor nodules.

Figure 3.

Comparison of tumor cell growth. Growth kinetics of B16-FTIIIN (), B16-FTIIIM () and B16-FTIIIH () cells are shown. Living cells were measured each day by the MTT assay.

View full figure (49 KB)

Binding of an E-selectin chimeric protein to different B16-FTIII cells

To determine whether these different B16-FTIII cells differ in E-selectin binding, cells were incubated with an E-selectin–IgM chimeric protein and the binding of E-selectin was assayed by flow cytometry.

As shown in Figure 4A and B, much more E-selectin chimeric protein was bound to B16-FTIIIM cells than to B16-FTIIIN cells. The highest amount of E-selectin binding was, however, observed in B16-FTIIIH cells (Figure 4C). When E-selectin–IgG chimeric protein (Sueyoshi et al., 1994) was used, the difference in E-selectin binding was less significant, but greater binding was observed to B16-FTIIIH cells than to B16-FTIIIM cells (data not shown). These results indicate that binding of E-selectin chimeric proteins was proportional to the amount of sialyl Lewis X as estimated by anti-sialyl Lewis X antibody staining.

Figure 4.

Flow cytometric analysis of E-selectin binding to different B16-FTIII cells. Monodispersed different B16-FTIII cells were incubated with E-selectin–IgM chimeric protein followed by FITC-conjugated goat anti-human IgM, and then subjected to flow cytometric analysis (closed figures). The open figures indicate control experiments in the presence of 10 mM EDTA. (A) B16-FTIIIN; (B) B16-FTIIIM; (C) B16-FTIIIH.

View full figure (45 KB)

Characterization of cell surface oligosaccharides of different B16-FTIII cells

The results above indicate that B16-FTIIIM and B16-FTIIIH cells may differ in glycan structures and that such a difference may contribute to different metastatic potentials.

To address this question, B16-FTIIIN, B16-FTIIIM and B16-FTIIIH cells were stained with anti-sialyl Lewis A antibody or various lectins. As shown in Figure 1B and C, the expression profiles of sialyl Lewis X and sialyl Lewis A in different cell lines were similar, and B16-FTIIIH showed the strongest expression of sialyl Lewis A. These results exclude the possibility that sialyl Lewis A may be more abundant in B16-FTIIIM cells, contributing to higher metastatic potentials.

We then stained cells with anti-Lewis X antibody, Maackia amurensis agglutinin (MAA), Sambucus nigra agglutinin (SNA) and tomato lectin (LEA). However, different B16-FTIII cell lines did not show detectable differences in staining either by anti-Lewis X antibody (Figure 1D), SNA (Figure 1E) or MAA (data not shown). The results indicate that these different B16-FTIII cells did not vary in their expression of Gal14(Fuc13) GlcNAcR (Lewis X), NeuNAc23GalR (detected by MAA; Wang and Cummings, 1988) or NeuNAc2 6GalR (detected by SNA; Shibuya et al., 1987). In contrast, B16-FTIIIM cells were stained more strongly by fluorescein isothiocyanate (FITC)–LEA than were B16-FTIIIH or B16-FTIIIN cells (Figure 1F). LEA was shown to bind preferentially to poly-N-acetyllactosamines (Merkle and Cummings, 1987; Lee et al., 1990). It was also demonstrated that sialylation or fucosylation does not weaken this binding (Lee et al., 1990; Saitoh et al., 1992). These results thus strongly suggest that B16-FTIIIM and B16-FTIIIH cells differ in their amount of poly-N-acetyllactosamines.

To corroborate the findings above, the carbohydrate structures of glycoproteins present in different B16-FTIII cells were examined. The cells were metabolically labeled with [³H]galactose, from which glycopeptides were prepared. The glycopeptides were then fractionated by serial lectin affinity chromatography. Both B16-FTIIIN and B16-FTIIIM cells produced high levels of glycopeptides that bound to and eluted from LEA–agarose (Figure 5B and E). When glycopeptides which did not bind to LEA–agarose were applied to a Datura stramonium agglutinin (DSA)–agarose column, a large quantity of those glycopeptides bound to DSA–agarose (Figure 5F). In contrast, glycopeptides from B16-FTIIIH cells were barely bound to either LEA–agarose or DSA–agarose (Figure 5H and I).

Figure 5.

Serial lectin affinity chromatography of [³H]galactose-labeled glycopeptides. B16-FTIIIN (A, B and C), B16-FTIIIM (D, E and F) and B16-FTIIIH (G, H and I) cells were metabolically labeled with [³H]galactose and glycopeptides prepared were applied sequentially to ConA–Sepharose, LEA–agarose and DSA–agarose. The glycopeptides subjected to the next step are indicated by horizontal bars. Vertical arrows indicate the initiation of elutions by a mixture of chitotriose and chitobiose.

View full figure (100 KB)

The glycopeptides bound to LEA–agarose (IB) and DSA–agarose (IA2) were digested by endo--galactosidase before and after defucosylation as described previously (Lee et al., 1990; Saitoh et al., 1992). The results indicate that IB and IA2 fractions from B16-FTIIIM cells contained highly sialylated glycopeptides having two to three N-acetyllactosamine repeats. Moreover, IB contained substantial amounts of sialyl Lewis X as determined by the method described (Sueyoshi et al., 1994). IA1 contained the largest amount of radioactively labeled glycopeptides, which represent all of the O-glycans and the rest of the N-glycans. The amount of IB and IA2 glycopeptides thus constitutes a substantial portion of N-glycans, although they do not appear that way in Figure 5. These results, taken together, indicate that B16-FTIIIM cells contained more sialyl Lewis X and possibly sialyl Lewis A in poly-N-acetyllactosamines than did B16-FTIIIH cells. B16-FTIIIH cells, in contrast, express the highest amount of sialyl Lewis X in short N-glycans.

B16-FTIIIM but not B16-FTIIIH cells extravasate in lung tissues

The results above demonstrate that B16-FTIIIH and B16-FTIIIM cells do not differ in presentation of selectin ligands but differ in metastatic potential and glycan structures. These results suggest that both of these cell types may attach to endothelial cells in lungs but the fate of those cell types may differ.

To follow the fate of these cells, animals were sacrificed 3, 24 and 48 h after injection of different B16 cells, and lung tissue sections were examined using antibodies specific to a melanoma antigen (HMB-45) (Kapur et al., 1992). As shown in Figure 6A, B16-FTIIIN cells expressing no detectable amount of sialyl Lewis X barely arrested within lung blood vessels 3 h after the injection and did not show any extravasation into the interstitial stroma 24 or 48 h after injection (Figure 6B and C). In contrast, B16-FTIIIM cells were found to arrest in lung blood vessels 3 h after injection, apparently invading parenchyma cells and extravasating into the interstitial stroma (Figure 6D, insert). The tumor cells started dividing and growing 24 and 48 h after the injection, respectively (Figure 6E and F).

Figure 6.

Immunostaining of tumor cells in C57/BL6 mice by anti-melanoma antibody. Lungs were examined after injection of B16-FTIIIN (A, B and C), B16-FTIIIM (D, E and F) and B16-FTIIIH (G, H and I) cells by anti-melanoma antibody (HMB-45), followed by biotinylated anti-mouse IgG and avidin–biotin reagent. The sections were prepared at 3 (A, D and G), 24 (B, E and H) and 48 h (C, F and H) after injection. Lung endothelial cells are shown by the arrow in (A). Bar, 20 m.

View full figure (98 KB)

B16-FTIIIH cells displayed a more complex profile. The cells appeared to arrest within lung blood vessels and extravasate into the interstitial stroma 3 h after the injection (Figure 6G). These cells, however, were apoptotic, displaying condensed nuclei, which were even broken in some cells (see the cell indicated by an arrowhead and that in the left insert, Figure 6G). The cells then disappeared 24 h after the injection (Figure 6H and I). These results strongly suggest that B16-FTIIIH cells died during or after penetration of lung tissue, while B16-FTIIIM cells established tumor growth during the same process.

To determine whether B16-FTIIIH cells underwent apoptosis in lung tissues, TUNEL assays were performed on tissue sections. A very small number of TUNEL-positive cells were observed 3 or 24 h after injection of B16-FTIIIN or B16-FTIIIM cells (Figure 7A–D). In contrast, a large number of TUNEL-positive cells were detected 3 h after injection of B16-FTIIIH cells (Figure 7E), but they were no longer detected after 24 h (Figure 7F). These TUNEL-positive cells were also positive for the melanoma antigen as described above. Moreover, some of those apoptotic cells were engulfed by macrophages 3 h after injection (data not shown).

Figure 7.

Apoptosis of B16-FTIII cells after injection into mice. B16-FTIIIN (A and B), B16-FTIIIM (C and D) and B16-FTIIIH (E and F) cells were injected into tail veins of C57BL/6 mice, and lung sections were prepared 3 (A, C and E) and 24 h (B, D and F) after injection. Apoptotic cells were detected by Frag EL DNA Fragmentation Detection Kits. Bar, 20 m.

View full figure (56 KB)

B16-FTIIIH cells were eliminated due to attack by NK cells

The results above demonstrate that overexpression of sialyl Lewis X on B16-F1 melanoma cells leads to apoptosis in lung tissues where they arrest, suggesting that those cells may be attacked by immune cells. To address this question, the three different B16-FTIII transfectants were injected into SCID and beige mice in the same manner as described for injection into C57BL/6 mice.

The results shown in Figure 8A illustrate that B16-FTIIIH and B16-FTIIIN cells barely formed lung tumor nodules when injected into SCID mice (deficient in both B and T cells), while B16-FTIIIM cells produced a large number of lung tumor nodules. In contrast, B16-FTIIIH cells produced a large number of lung tumor nodules in beige mice which are deficient in NK cells and macrophages (Figure 8A and C). Finally, B16-FTIIIH cells produced a large number of lung tumor nodules in C57BL/6 mice, which had been depleted of NK cells by anti-asialo–GM1 (Figure 8A and D). In beige mice and NK cell-depleted mice, the number of tumor nodules produced by different B16-FTIII cells was proportional to the amount of sialyl Lewis X expressed on the cell surface. These results strongly suggest that B16-FTIIIH cells underwent apoptosis subsequent to attack by NK cells in C57BL/6 mice and SCID mice.

Figure 8.

Lung tumor nodules formed in various immunodeficient mice. (A) B16-FTIIIN (N), B16-FTIIIM (M) and B16-FTIIIH (H) cells were injected into tail veins of C57BL/6 SCID mice (SCID), C57BL/6 bg/bg mice (beige) and C57BL/6 mice that had been depleted of NK cells by anti-asialo-GM1 antibody (NK^-). (B–D) The number of lung tumor nodules formed in SCID (B), beige (C) and NK cell-depleted C57BL/6 mice (D).

View full figure (82 KB)

This study demonstrates that B16-FI melanoma cells become highly metastatic by acquisition of sialyl Lewis X through expression of fucosyltransferase III (Figures 1 and 2). This increase of metastatic capability was due to expression of the E-selectin ligand, but not due to increased growth rate (Figures 3 and 4). Such expression of E-selectin ligand resulted in the arrest in lung blood vessels (Figure 6). Moreover, tumor formation was inhibited by pre-incubation of those cells with anti-sialyl Lewis X antibody but not by incubation with a control antibody specific to Lewis X. These results demonstrate that acquisition of sialyl Lewis X resulted in the increased metastatic capability of B16-F1 melanoma cells.

This study also demonstrates that highly metastatic B16-FTIIIM cells express more sialyl Lewis X in poly-N-acetyllactosamines than do poorly metastatic B16-FTIIIH cells (Figures 1 and 5). B16-FTIIIH cells, however, contain more sialyl Lewis X in short N-glycans than do B16-FTIIIM cells. Similarly, it was shown previously that highly metastatic colonic carcinoma cells express more sialyl Lewis X in poly-N-acetyllactosamines than do their poorly metastatic counterparts, when both were derived from the same patient (Saitoh et al., 1992). Moreover, the total amount of sialyl Lewis X does not differ between these carcinoma cells, and sialyl Lewis X in short N-glycans is more abundant in poorly metastatic colon carcinoma cells (Saitoh et al., 1992). These findings are in agreement with those obtained in the present study.

Our preliminary studies indicate that almost identical results were obtained using human MeWo melanoma cells. Cells expressing high levels of sialyl Lewis X, similar to B16-FTIIIH cells, were not metastatic, while those corresponding to B16-FTIIIM cells produced a large number of lung tumor nodules. Under the same conditions, the parent MeWo cells barely formed tumors. Structural characterization of N-glycans attached to MeWo transfectants also showed a similar difference among those three different cells as for the different B16-FTIII cells.

The present study provides direct evidence that sialyl Lewis X plays a critical role in tumor metastasis. Preferential arrest of sialyl Lewis X-expressing tumor cells on lung epithelium is probaby due to their binding to E- and possibly P-selectin, although this mechanism has not been shown in the present study. In support of this hypothesis, it has been demonstrated that sialyl Lewis X-expressing B16-F10 cells lodged in liver that expresses an E-selectin transgene, while the same cells formed tumors in the lungs of wild-type mice (Biancone et al., 1996). In breast carcinoma patients, carcinoma cells expressing sialyl Lewis X were found to arrest to endothelial cells expressing E-selectin (Renkonen et al., 1997). These results, together with the results obtained in the present study, strongly suggest that the interaction between sialyl Lewis X and selectin plays a key role in tumor metastasis.

One of the striking discoveries presented in our study is that overexpression of sialyl Lewis X on tumor cells, as seen in B16-FTIIIH cells, can be unfavorable to tumor metastasis in wild-type and SCID mice (Figures 6 and 7). In contrast, B16-FTIIIH cells formed tumor nodules more efficiently in beige mice or NK cell-depleted mice than did B-FTIIIM cells (Figure 8). As discussed above, B16-FTIIIH cells express significant quantities of sialyl Lewis X in short N-glycans. These results, taken together, indicate that overexpression of sialyl Lewis X in short N-glycans may be recognized by NK cells. Our histochemical analysis demonstrated that the mice used in the present study express small but detectable amounts of sialyl Lewis X as determined by mouse monoclonal 2H5 antibody (Sawada et al., 1993) (data not shown). We thus conclude that NK cells do not recognize low levels of sialyl Lewis X as foreign and suggest that only sialyl Lewis X overexpressed in short N-glycans is recognized by NK cells. This conclusion is corroborated by the experiments employing B16-FTIIIH cells after >15 passages of culture. Those cells tend to lose sialyl Lewis X expression and became stable once they expressed a moderate amount of sialyl Lewis X, which corresponds to B16-FTIIIM cells. Such a change in sialyl Lewis X expression was also accompanied by a conversion from poorly to highly metastatic potentials in wild-type mice. These results, taken together, strongly suggest the proposal that the difference in metastatic potentials in these populations is caused solely by the difference in sialyl Lewis X content.

It has been shown that NK cells express the C-type lectin-like receptor Ly-49A (Phillips et al., 1996; Matsumoto et al., 1998). C-type lectin is the collective name for lectins dependent on Ca²⁺ for their activity which share a conserved domain structure (Weis et al., 1992). Selectins are also members of this gene family (Lowe, 1994; Rosen and Bertozzi, 1996). The C-type lectin domain of Ly-49A was shown to bind to MHC class I antigen in a carbohydrate-independent manner (Matsumoto et al., 1998). The same domain also bound well to fucoidin, an inhibitor of L-selectin, and such binding was shown to inhibit the binding of Ly-49A to H-2d, triggering an attack by NK cells (Matsumoto et al., 1998). Crystallographic studies on the tertiary structure of mannose-binding protein and E-selectin demonstrated that these two C-type lectins share conserved tertiary structures (Weis et al., 1992; Graves et al., 1994). It is thus likely that the NK cell receptor Ly-49A also shares a tertiary structure with other C-type lectins and binds to sialyl Lewis X structures. This hypothesis is supported by a report that an NK cell lectin-like receptor recognizes sialyl Lewis X (Bezouska et al., 1994), although its avidity is not as strong as originally reported (Bezouska et al., 1996). It is tempting to speculate that binding of Ly-49A to sialyl Lewis X interferes with its recognition of MHC class I antigen and that such interference triggers an attack on tumor cells by NK cells.

Previously, it was demonstrated that T cells were involved in the rejection of B16-F10 melanoma cells when interleukin-12 (IL-12) was administered (Brunda et al., 1993). In contrast, IL-10-stimulated tumor rejection was found to be due to an NK cell-mediated attack (Zheng et al., 1996). Moreover, it was demonstrated that NK cells together with CD8-positive cells attack tumor cells (Iwanuma et al., 1997). These findings are similar to the results presented in this study and indicating a critical role for NK cells in tumor rejection. Moreover, the results obtained in this study provide an insight into the mechanism of NK cells recognition of tumor cells. The present study, as a whole, demonstrates that the amount of sialyl Lewis X and the structure of backbone glycans containing sialyl Lewis X determine whether tumor cells form tumor metastasis or are rejected by NK cells, pointing toward an exciting possibility that tumor metastasis may be inhibited by carbohydrate-based antagonists or up-regulation of an 1,3-fucosyltransferase.

Transfection of B16-FI cells with pcDNA3-Fuc-TIII

The cDNA encoding Fuc-TIII was excised by HindIII and XbaI digestion from pcDNAI-Fuc-TIII (Kukowska-Latallo et al., 1990; Sueyoshi et al., 1994) and was cloned into the same sites of pcDNA3(neo), resulting in pcDNA3(neo)-Fuc-TIII. B16-F1 cells, completely negative for sialyl Lewis X expression as assessed by immunostaining, were transfected with pcDNA3(neo)-Fuc-TIII by electroporation as described previously (Fukuda et al., 1995). B16-F1 cells were purchased from the American Type Culture Collection. Transfected cells were cultured in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal calf serum (FCS) and G418 (1 mg/ml, Gibco-BRL) for 2 weeks and several single colonies were selected for screening. The isolated clones were dissociated into monodispersed cells using an enzyme-free cell dissociation solution (Hank's based, purchased from Cell and Molecular Technologies, Lavellette, NJ). Monodispersed cells were incubated with 10 g/ml of mouse anti-sialyl Lewis X antibody (CSLEX1, Becton Dickinson), followed by FITC-conjugated goat affinity-purified (Fab')₂ fragment specific to mouse IgM (Cappel). The cells were then separated by FACS using FACStar (Becton Dickinson) into cells negative for sialyl Lewis X expression (B16-FTIIIN), cells expressing moderate amounts of sialyl Lewis X (B16-FTIIIM) and cells strongly positive for sialyl Lewis X (B16-FTIIIH).

Tumor formation in mice

The above cell populations were cultured for 1 week to obtain sufficient numbers of cells. The cloned cells were dissociated into monodispersed cells as described above. A total of 110⁵ of cells exhibiting >90% viability were suspended in 100 l of serum-free DMEM and injected into a tail vein of C57BL/6, SCID (C.B.-17 scid) and beige (C.B.-17 scid-beige) mice (6–8 weeks, female). After 2 weeks (for beige and NK cell-depleted mice, see below) or 3 weeks (for C57BL/6 and SCID), mice were sacrificed, lung, liver, spleen and lymph nodes were fixed with Bouin's solution, and tumor nodules were counted under a dissecting microscope. Since the majority of tumors were formed in lung, the weight of the lung was also measured. In this protocol, wild-type B16–F1 cells or mock-transfected B16-F1 cells barely formed tumors. C57BL/6 mice were purchased from B&K Universal, Fremont, CA, and SCID and beige mice were from Taconic, Germantown, NY.

Monoclonal antibody treatment of B16-FTIIIM cells and NK cytotoxic activity depletion

B16-FTIIIM cells were pre-incubated with CSLEX 1 (10 g/ml) for 30 min at 4°C before tail vein injection into C57BL/6 mice. For control, 10 g/ml of anti-Lewis X monoclonal antibody (CD15 monoclonal antibody, Immunotech) was used. For experiments with mice depleted of NK cells, C57BL/6 mice were injected intraperitoneally with 25 l of anti-asialo GM1 1 day before tumor injection and every 5 days during the experiment, as described previously (Brunda et al., 1993). Anti-asialo GM1 antibody was purchased from Wako Chemicals.

Cell proliferation assay

B16-FTIIIN, B16-FTIIIM and B16-FTIIIH cells were seeded in 96-well plates at 10⁵ cells/ml in DMEM containing 10% FCS and 1 mg/ml of G418, and cultured for various times. The number of living cells was measured each day using Cell Titer 96^® Aqueous One Solution Cell Proliferation Assay (Promega). Triplicate cultures were used for each sample (Qian et al., 1994).

Binding of E-selectin to different B16-FTIII cells

A cDNA encoding a chimeric protein of mouse E-selectin fused with hinge and constant regions of human IgM was kindly provided by Dr John Lowe (Maly et al., 1996). This vector, pcDNAI-E-selectin–IgM, was transfected into COS-1 cells using Lipofectamine™. Three days after transfection, the culture medium was used directly without dilution as a source of E-selectin–IgM. B16-FTIIIN, B16-FTIIIM and B16-FTIIIH cells were dissociated using the cell dissociation solution (see above) and incubated with E-selectin–IgM at 4°C for 30 min followed by FITC-conjugated goat anti-human IgM (Pierce). After washing the cells in DMEM, labeled cells were analyzed by flow cytometry as described using a FACScan flow cytometer (Maly et al., 1996).

FACS analysis of carbohydrate antigens expressed on different B16-FTIII cells

In order to determine the expression levels of carbohydrate antigens on different B16-FTIII cells, the same cell samples used for tumor assays were dissociated and incubated with FITC-conjugated SNA, FITC-conjugated MAA or FITC-conjugated LEA. Analyses were carried out using FACSort flow cytometry using the CellQuest program (Becton-Dickinson) as described (Tsuboi and Fukuda, 1997). Similarly, the cells were incubated with CSLEX-1, anti-Lewis X antibody or mouse anti-sialyl Lewis A antibody (CSLEA1, UCLA Tissue Typing Laboratory) followed by FITC-conjugated secondary antibodies, and analyzed by flow cytometer. FITC-conjugated lectins were purchased from EY Laboratories.

Lectin affinity chromatography and characterization of glycopeptides

B16-FTIIIN, B16-FTIIIM and B16-FTIIIH cells were metabolically labeled with [³H]galactose (20 Ci/ml, DuPont-NEN) in glucose-free DMEM containing 10% dialyzed FCS supplemented with 2% regular DMEM containing 10% FCS for 24 h as described (Saitoh et al., 1992). The glycopeptides prepared by pronase digestion of [³H]galactose-labeled cells were desalted in a small column of Sephadex G-25 equilibrated with water and then fractionated sequentially by serial lectin affinity chromatography employing ConA–Sepharose, LEA–agarose and DSA–agarose (Saitoh et al., 1992). Briefly, the glycopeptides were first applied to a column of ConA–Sepharose, and unbound glycopeptides containing tri-antennary and tetra-antennary N- and O-glycans (Fraction I) were separated from the others. Fraction I was then applied to a column of LEA–agarose and separated into unbound (IA) and bound fractions directly (IB). The unbound fraction (IA) was fractionated further by a column of DSA–agarose into unbound (IA1) and bound (IA2) fractions. LEA–agarose binds to glycopeptides containing three or more N-acetyllactosamine units (Merkle and Cummings, 1987; Lee et al., 1990; Saitoh et al., 1992). DSA–agarose binds to glycopeptides containing two or more N-acetyllactosamine units (Cummings and Kornfeld, 1984; Lee et al., 1990; Saitoh et al., 1992). Highly fucosylated N-glycans, however, are not bound to DSA–agarose even when they contain poly-N-acetyllactosamines (Yamashita et al., 1987; Lee et al., 1990; Saitoh et al., 1992). IB and IA2 glycopeptides were subjected to endo--galactosidase digestion (Fukuda, 1981) before and after defucosylation as described previously (Saitoh et al., 1992).

Immunocytochemical detection of tumor cells and apoptotic cells

The lungs of mice injected with tumor cells were removed 3, 24 and 48 h after injection and fixed. Tissue was embedded in paraffin and 5 m thick sections were cut and mounted on glass slides. After deparaffination and rehydration, the sections were incubated with mouse anti-melanoma antibody (HMB-45, Dako) followed by biotinylated anti-mouse IgG and avidin–biotin reagent (Vectastain ABC Kit, Vector) as described (Shimodaira et al., 1997). HMB-45 antibody was shown to bind to a melanoma-specific antigen (Kapur et al., 1992). TUNEL assays were performed on paraffin sections of lungs using Frag EL DNA Fragmentation Detection Kits (Oncogene Research Products) as described previously (Yu et al., 1997).

The authors wish to thank Drs John Lowe, Reiji Kannagi and Michiko Fukuda for the gifts of pcDNAI-FucTIII and pcDNAI-E-selectin IgM, 2H5 antibody and endo--galactosidase, respectively, Dr Jiunn-Chern Yeh for useful discussion, Dr Edgar Ong for critical reading of the manuscript, and Ms Susan Greaney for organizing the manuscript. This work was supported by grants R37 CA33000 (MERIT award) and RO1 CA48737 awarded by the National Cancer Institute.

Berg EL, Robinson MK, Mansson O, Butcher EC and Magnani JL (1991) A carbohydrate domain common to both sialyl Le (A) and sialyl Le (X) is recognized by the endothelial cell leukocyte adhesion molecule ELAM-1. J Biol Chem, 266, 14869–14872. | PubMed | ISI | ChemPort |

Bezouska K et al. (1994) Oligosaccharide ligands for NKR-P1 protein activate NK cells and cytotoxicity. Nature, 372, 150–157. | Article | PubMed | ISI | ChemPort |

Bezouska K et al. (1996) Oligosaccharide ligands for NKR-P1 protein activate NK cells and cytotoxicity. Nature, 380, 559. | Article | ISI | ChemPort |

Biancone L, Araki M, Araki K, Vassalli P and Stamenkovic I (1996) Redirection of tumor metastasis by expression of E-selectin in vivo. J Exp Med, 183, 581–587. | Article | PubMed | ISI | ChemPort |

Brunda MJ, Luistro L, Warrier RR, Wright RB, Hubbard BR, Murphy M, Wolf SF and Gately MK (1993) Antitumor and antimetastatic activity of interleukin 12 against murine tumors. J Exp Med, 178, 1223–1230. | Article | PubMed | ISI | ChemPort |

Cummings RD and Kornfeld S (1984) The distribution of repeating [Gal 1,4GlcNAc 1,3] sequences in asparagine-linked oligo- saccharides of the mouse lymphoma cell lines BW5147 and PHAR 2.1. J Biol Chem, 259, 6253–6260. | PubMed | ISI | ChemPort |

Demetriou M, Nabi IR, Coppolino M, Dedhar S and Dennis JW (1995) Reduced contact-inhibition and substratum adhesion in epithelial cells expressing GlcNAc-transferase V. J Cell Biol, 130, 383–392. | Article | PubMed | ISI | ChemPort |

Feizi T (1985) Demonstration by monoclonal antibodies that carbohydrate structures of glycoproteins and glycolipids are onco-developmental antigens. Nature, 314, 53–57. | Article | PubMed | ISI | ChemPort |

Fidler IJ (1973) Selection of successive tumor lines for metastasis. Nature, 242, 148–149. | ISI | ChemPort |

Fukuda MN (1981) Purification and characterization of endo--galactosidase from Escherichia freundii induced by hog gastric mucin. J Biol Chem, 256, 3900–3905. | PubMed | ISI | ChemPort |

Fukuda M (1994) Cell type-specific expression of cell surface carbohydrates. In Fukuda,M. and Hindsgaul,O. (eds), Molecular Glycobiology. Oxford University Press, Oxford, UK, pp. 1–52.

Fukuda M (1996) Possible roles of tumor-associated carbohydrate antigens. Cancer Res, 56, 2237–2244. | PubMed | ISI | ChemPort |

Fukuda M, Spooncer E, Oates JE, Dell A and Klock JC (1984) Structure of sialylated fucosyllactosaminoglycan isolated from human granulocytes. J Biol Chem, 259, 10925–10935. | PubMed | ChemPort |

Fukuda M, Bothner B, Ramasamooj P, Dell A, Tiller PR, Varki A and Klock JC (1985) Structures of sialylated fucosyl polylactos- aminoglycans isolated from chronic myelogenous leukemia cells. J Biol Chem, 260, 12957–12967. | PubMed | ChemPort |

Fukuda MN, Sato T, Nakayama J, Klier G, Mikami M, Aoki D and Nozawa S (1995) Trophinin and tastin, a novel cell adhesion molecule complex with potential involvement in embryo implantation. Genes Dev, 9, 1199–1210. | PubMed | ISI | ChemPort |

Fukushima K et al. (1984) Characterization of sialosylated Lewis X as a new tumor-associated antigen. Cancer Res, 44, 5279–5285. | PubMed | ISI | ChemPort |

Graves BJ et al. (1994) Insight into E-selectin/ligand interaction from the crystal structure and mutagenesis of the lec/EGF domains. Nature, 367, 532–538. | Article | PubMed | ISI | ChemPort |

Imai Y, Lasky LA and Rosen SD (1993) Sulphation requirement for GlyCAM-1, an endothelial ligand for L-selectin. Nature, 361, 555–557. | Article | PubMed | ISI | ChemPort |

Itzkowitz SH, Yuan M, Fukushi Y, Lee H, Shi ZR, Zurawski V,Jr, Hakomori S and Kim YS (1988) Immunohistochemical comparison of Le^a, monosialosyl Le^a (CA 19-9) and disialosyl Le^a antigens in human colorectal and pancreatic tissues. Cancer Res, 48, 3834–3842. | PubMed | ISI | ChemPort |

Iwanuma Y, Chen F-A, Egilmez NK, Takita H and Bankert RB (1997) Antitumor immune response of human peripheral blood lymphoctyes coengrafted with tumor into severe combined immunodeficient mice. Cancer Res, 57, 2937–2942. | PubMed | ISI | ChemPort |

Kapur RP, Bigler SA, Skelly M and Gown AM (1992) Anti-melanoma monoclonal antibody HMB45 identifies an oncofetal glycoconjugate associated with immature melanosomes. J Histochem Cytochem, 40, 207–212. | PubMed | ISI | ChemPort |

Kukowska-Latallo JF, Larsen RD, Nair RP and Lowe JB (1990) A cloned human cDNA determines expression of a mouse stage-specific embryonic antigen and the Lewis blood group (1,3/1,4)fucosyl- transferase. Genes Dev, 4, 1288–1303. | PubMed | ChemPort |

Lee N, Wang W-C and Fukuda M (1990) Granulocytic differentiation of HL-60 cells is associated with increase of poly-N-acetyllactosamine in Asn-linked oligosaccharides attached to human lysosomal membrane glycoproteins. J Biol Chem, 265, 20476–20487. | PubMed | ISI | ChemPort |

Lowe BJ (1994) Carbohydrate recognition in cell–cell interaction. In Fukuda,M. and Hindsgaul,O. (eds), Molecular Glycobiology. Oxford University Press, Oxford, UK, pp. 163–205.

Magnani JL, Nilsson B, Brockhaus M, Zopf D, Steplewski Z, Koprowski H and Ginsburg V (1982) A monoclonal antibody-defined antigen associated with gastrointestinal cancer is a ganglioside containing sialylated lacto-N-fucopentaose II. J Biol Chem, 257, 14365–14369. | PubMed | ChemPort |

Maly P et al. (1996) The (1,3)fucosyltransferase Fuc-TVII controls leukocyte trafficking through an essential role in L-, E- and P-selectin ligand biosynthesis. Cell, 86, 643–653. | Article | PubMed | ISI | ChemPort |

Martín-Satué M, Marrugat R, Cancelas JA and Blanco J (1998) Enhanced expression of (1,3)-fucosyltransferase genes correlates with E-selectin-mediated adhesion and metastatic potential of human lung adenocarcinoma cells. Cancer Res, 58, 1544–1550. | PubMed | ISI |

Matsumoto N, Ribaudo RK, Abastado J-P, Margulies DH and Yokoyama WM (1998) The lectin-like NK cell receptor Ly-49A recognizes a carbohydrate-independent epitope on its MHC class I ligand. Immunity, 8, 245–254. | Article | PubMed | ISI | ChemPort |

Merkle RK and Cummings RD (1987) Relationship of the terminal sequences to the length of poly-N-acetyllactosamine chains in asparagine-linked oligosaccharides from the mouse lymphoma cell line BW5147. Immobilized tomato lectin interacts with high affinity with glycopeptides containing long poly-N-acetyllactosamine chains. J Biol Chem, 262, 8179–8189. | PubMed | ISI | ChemPort |

Mizoguchi A, Takasaki S, Maeda S and Kobata A (1984) Changes in asparagine-linked sugar chains of human promyelocytic leukemic cells (HL-60) during monocytoid differentiation and myeloid differentiation. Decrease of high-molecular-weight oligosaccharides in acidic fraction. J Biol Chem, 259, 11949–11957. | PubMed | ChemPort |

Nakamori S et al. (1993) Increased expression of sialyl Lewis X antigen correlates with poor survival in patients with colorectal carcinoma: clinicopathological and immunohistochemical study. Cancer Res, 53, 3632–3637. | PubMed | ISI | ChemPort |

Phillips JH, Chang C, Mattson J, Gumperz JE, Parham P and Lanier LL (1996) CD94 and a novel associated protein (94AP) form a NK cell receptor involved in the recognition of HLA-A, HLA-B and HLA-C allotypes. Immunity, 5, 163–172. | Article | PubMed | ISI | ChemPort |

Qian F, Vaux DL and Weissman IL (1994) Expression of the integrin 41 on melanoma cells can inhibit the invasive stage of metastasis formation. Cell, 77, 335–347. | Article | PubMed | ISI | ChemPort |

Rademacher TW, Parekh RB and Dwek RA (1988) Glycobiology. Annu Rev Biochem, 57, 785–838. | Article | PubMed | ISI | ChemPort |

Renkonen J, Paavonen T and Renkonen R (1997) Endothelial and epithelial expression of sialyl Le^x and sialyl Le^a in lesions of breast carcinoma. Int J Cancer, 74, 296–300. | Article | PubMed | ISI | ChemPort |

Rosen SD and Bertozzi CR (1996) Two selectins converge on sulphate. Leukocyte adhesion. Curr Biol, 261, 261–264.

Saitoh O, Wang W-C, Lotan R and Fukuda M (1992) Differential glycosylation and cell surface expression of lysosomal membrane glycoproteins in sublines of a human colon cancer exhibiting distinct metastatic potentials. J Biol Chem, 267, 5700–5711. | PubMed | ISI | ChemPort |

Sasaki H, Bothner B, Dell A and Fukuda M (1987) Carbohydrate structure of erythropoietin expressed in Chinese hamster ovary cells by a human erythropoietin cDNA. J Biol Chem, 262, 12059–12076. | PubMed | ISI | ChemPort |

Sawada M, Takada A, Ohwaki I, Takahashi N, Tateno H, Sakamoto J and Kannagi R (1993) Specific expression of a complex sialyl Lewis X antigen on high endothelial venules of human lymph nodes: possible candidate for L-selectin ligand. Biochem Biophys Res Commun, 193, 337–347. | Article | PubMed | ISI | ChemPort |

Sawada R, Tsuboi S and Fukuda M (1994) Differential E-selectin-dependent adhesion efficiency in sublines of a human colon cancer exhibiting distinct metastatic potentials. J Biol Chem, 269, 1425–1431. | PubMed | ISI | ChemPort |

Shibuya N, Goldstein IJ, Broekaert WF, Nsimba-Lubaki M, Peeters B and Peumans WJ (1987) The elderberry (Sambucus nigra L.) bark lectin recognizes the Neu5Ac (2–6)Gal/GalNAc sequence. J Biol Chem, 262, 1596–1601. | PubMed | ISI | ChemPort |

Shimodaira K, Nakayama J, Nakamura N, Hasebe O, Katsuyama T and Fukuda M (1997) Carcinoma-associated expression of core 2 -1,6-N-acetylglucosaminyltransferase gene in human colorectal cancer: role of O-glycans in tumor progression. Cancer Res, 57, 5201–5206. | PubMed | ISI | ChemPort |

Sueyoshi S, Tsuboi S, Sawada-Hirai R, Dang UN, Lowe JB and Fukuda M (1994) Expression of distinct fucosylated oligosaccharides and carbohydrate-mediated adhesion efficiency directed by two different -1,3-fucosyltransferases. Comparison of E- and L-selectin-mediated adhesion. J Biol Chem, 269, 32342–32350. | PubMed | ISI | ChemPort |

Takada A, Ohmori K, Takahashi N, Tsuyuoka K, Yago A, Zenita K, Hasegawa A and Kannagi R (1991) Adhesion of human cancer cells to vascular endothelium mediated by a carbohydrate antigen, sialyl Lewis A. Biochem Biophys Res Commun, 179, 713–719. | Article | PubMed | ISI | ChemPort |

Tsuboi S and Fukuda M (1997) Branched O-linked oligosaccharides aberrantly expressed in the Wiskott–Aldrich syndrome reduce primary T-cell immune responses. EMBO J, 16, 6364–6373. | Article | PubMed | ChemPort |

Van den Eijnden DH, Koenderman AH and Schiphorst WE (1988) Biosynthesis of blood group i-active polylactosaminoglycans. Partial purification and properties of an UDP-GlcNAc:N-acetyllactosaminide beta 1,3-N-acetylglucosaminyltransferase from Novikoff tumor cell ascites fluid. J Biol Chem, 263, 12461–12471. | PubMed | ChemPort |

Wang W-C and Cummings RD (1988) The immobilized leukoagglutinin from the seeds of Maackia amurensis binds with high affinity to complex-type Asn-linked oligosaccharides containing terminal sialic acid-linked 2,3 to penultimate galactose residues. J Biol Chem, 263, 4576–4585. | PubMed | ISI | ChemPort |

Weis WI, Drickamer K and Hendrickson WA (1992) Structure of a C-type mannose-binding protein complexed with an oligosaccharide. Nature, 360, 127–134. | Article | PubMed | ISI | ChemPort |

Wilkins PP, McEver RP and Cummings RD (1996) Structures of the O-glycans on P-selectin glycoprotein ligand-1 from HL-60 cells. J Biol Chem, 271, 18732–18742. | Article | PubMed | ISI | ChemPort |

Yamashita K, Ohkura T, Tachibana Y, Takasaki S and Kobata A (1984) Comparative study of the oligosaccharides released from baby hamster kidney cells and their polyoma transformant by hydrazinolysis. J Biol Chem, 259, 10834–10840. | PubMed | ChemPort |

Yamashita K, Totani K, Ohkura T, Takasaki S, Goldstein IJ and Kobata A (1987) Carbohydrate binding properties of complex-type oligosaccharides on immobilized Datura stramonium lectin. J Biol Chem, 262, 1602–1607. | PubMed | ISI | ChemPort |

Yu Q, Toole BP and Stamenkovic I (1997) Induction of apoptosis of metastatic mammary carcinoma cells in vivo by disruption of tumor cell surface CD44 function. J Exp Med, 186, 1985–1996. | Article | PubMed | ISI | ChemPort |

Zheng LM et al. (1996) Interleukin-10 inhibits tumor metastasis through an NK cell-dependent mechanism. J Exp Med, 184, 579–584. | Article | PubMed | ISI | ChemPort |

Main navigation

• Search EMBO Journal

  Search

• Advanced search

• The EMBO Journal

  • Home
  • Aims and scope
  • Current issue
  • Advance online publication
  • Web focuses
  • Archive:
  • Browse by issue
  • Browse by subject
  • Browse by category
  • Free online sample issue
  • Press releases

• Authors and Referees

  • Editorial process
  • Guide for authors
  • Submit an article
  • Guide for referees
  • Editors and Editorial Board
  • Contact editorial office

• Customer Service

  • Subscribe
  • Order sample copy
  • Purchase articles
  • Reprints and permissions
  • Contact NPG
  • Advertising