Fucosylation of Disaccharide Precursors of Sialyl LewisX Inhibit Selectin-mediated Cell Adhesion*

We showed previously that HL-60 and F9 mouse embryonal carcinoma cells will take up and deblock peracetylated Galβ1–4GlcNAcβ-O-naphthalenemethanol (Galβ1–4GlcNAc-NM) and use the disaccharide as a primer of oligosaccharide chains (Sarkar, A. K., Fritz, T. A., Taylor, W. H., and Esko, J. D. (1995) Proc. Natl. Acad. Sci. U. S. A. 92, 3323–3327). We now report that another disaccharide, acetylated GlcNAcβ1–3Gal-naphthalenemethanol (GlcNAcβ1–3Gal-NM), has even greater potency and that both compounds will inhibit sialyl LewisX(sLex)-dependent cell adhesion. When fed to U937 cells, acetylated forms of Galβ1–4GlcNAc-NM and GlcNAcβ1–3Gal-NM primed oligosaccharides in a dose-dependent manner. Analysis of compounds assembled on Galβ1–4GlcNAc-NM showed only one product, namely Galβ1–4(Fucα1–3)GlcNAc-NM. In contrast, GlcNAcβ1–3Gal-NM generated Galβ1–4GlcNAcβ1–3Gal-NM, Galβ1–4(Fucα1–3)GlcNAcβ1–3Gal-NM, NeuAcα2–3Galβ1–4GlcNAcβ1–3Gal-NM, and NeuAcα2–3Galβ1–4(Fucα1–3)GlcNAcβ1–3Gal-NM. Both compounds decreased the incorporation of [3H]fucose into cellular glycoconjugates, without affecting the incorporation of [3H]mannosamine, a precursor of sialic acid residues. Moreover, the overall extent of sialylation was not affected based on the reactivity of cells to fluorescein isothiocyanate-conjugatedMaackia amurensis lectin. Priming inhibited expression of sLex on cell surface glycoconjugates, which reduced E-selectin-dependent cell adhesion to tumor necrosis factor-α-activated human umbilical vein endothelial cells. GlcNAcβ1–3Gal-NM and Galβ1–4GlcNAc-NM represent starting points for making enzyme-specific, site-directed inhibitors of glycosyltransferases that could act in living cells.

Selectins are part of a family of cell adhesion molecules involved in immune cell trafficking (1)(2)(3)(4)(5)(6)(7). Cytokine activated endothelial cells express on their lumenal surface E-and Pselectins, which bind to carbohydrate ligands related to the Lewis blood group antigens expressed on monocytes, neutro-phils, and certain subsets of T-cells. When engaged, the receptor-ligand complexes facilitate leukocyte rolling on the endothelium, which later gives way to strong adhesion and extravasation of cells into the underlying tissue. L-selectin on the surface of the leukocytes plays a similar role, possibly by facilitating aggregation of neutrophils (8 -10). Although these adhesion events help defend against infection and facilitate tissue repair, they sometimes go awry and cause inadvertent tissue destruction (e.g. during ischemia-reperfusion or traumainduced tissue damage). Therefore, much interest exists in developing agents to inhibit the adherence of leukocytes to endothelium as a way of modulating deleterious inflammatory reactions.
Glycoside-based primers represent another class of potential inhibitors. These compounds resemble biosynthetic intermediates and act as substrates for oligosaccharide assembly, thereby diverting the synthesis of chains from endogenous proteins and lipids. Examples include N-acetylgalactosaminides, which prime oligosaccharide chains similar to those found O-glycosidically linked to serine and threonine residues of glycoproteins (35,36). Incubation of HL-60 promyelocytic leukemia cells with GalNAc␣-O-benzyl 2 inhibited the synthesis of sLe x on O-linked glycoproteins, which in turn blocked selectin-mediated adhesion to activated endothelial cells (37). Peracetylated fluorinated analogs of N-acetylglucosamine had similar effects on selectin-mediated adhesion of tumor cells to endothelial cells (38,39). Recently, we showed that a peracetylated disaccharide ((Ac) 4 Gal␤1-4(Ac) 2 GlcNAc␤-O-naphthalenemethanol) inhibited sLe x expression on HL-60 cells and worked at a low concentration (50 M) (40). In this study, we have extended this finding to another disaccharide, peracetylated GlcNAc␤1-3Gal␤-O-naphthalenemethanol, and to human U937 cells. We show that the acetylated disaccharides inhibit the E-selectin-dependent adhesion of U937 cells to cytokine-activated endothelial cells by blocking the formation of sLe x .
Labeling Studies-The disaccharides were dissolved in Me 2 SO and added to growth medium to achieve the concentrations indicated in the figures and tables. The concentration of Me 2 SO was adjusted to 0.5% in all samples, and then cells were added to a starting density of ϳ1 ϫ 10 5 cells/ml. For labeling studies, the cells were incubated with the disaccharides and 10 Ci/ml [6-3 H] Hmethyl]thymidine. The cells were sedimented by centrifugation, and the pellets were treated with 10% trichloroacetic acid to precipitate glycolipids and glycoproteins. The pellets were dissolved in 0.1 M NaOH and counted by liquid scintillation spectrometry using UltimaGold (Packard). In priming studies, the conditioned medium was adjusted to 0.5 M NaCl with a stock solution of 1.5 M NaCl. Supernatants were then applied to 0.2-cc Sep-Pak Vac RC C18 cartridges (Waters), which had been prewashed with 100% methanol, water, and 0.5 M NaCl. After applying the samples to the columns, they were washed sequentially with 0.5 M NaCl (2.5 ml) and water (25 ml). Radioactive oligosaccharides were eluted with 40% methanol in water (2.5 ml). This fraction was concentrated by vacuum centrifugation (Savant SpeedVac concentrator), dissolved in water, and counted by liquid scintillation. Total cell protein was estimated from the cell pellet using the Bio-Rad protein assay kit and bovine serum albumin (BSA) as standard.
Separation of Neutral and Charged Species by QAE-Sephadex Anion Exchange Chromatography-Primed oligosaccharides that eluted in the 40% methanol fraction were analyzed by QAE-Sephadex (Sigma) anionexchange chromatography to separate charged and neutral species (45). Samples were dissolved in 2 mM Tris base and applied to a column of QAE-Sephadex (0.5 ml) pre-equilibrated with the same buffer. The column was first washed with 2 mM Tris base (10 ml) and then sequentially with solutions of 2 mM Tris base containing 5 mM NaCl, 10 mM NaCl, 20 mM NaCl, 50 mM NaCl, 100 mM NaCl, or 200 mM NaCl (2.5 ml each). Radioactive oligosaccharides were recovered in the fraction containing 2 mM Tris base (neutral) and the 20 mM NaCl and 2 mM Tris base wash (Ϫ1 charged). These samples were concentrated, dissolved in 2.5 ml of 0.5 M NaCl solution, and applied to a Sep-Pak C18 cartridge, which was pre-equilibrated with 0.5 M NaCl. The column was washed with water (5 ml), and primed material was eluted with 40% methanol in water (1 ml). This material was then used for subsequent analytical work.
Reverse Phase HPLC-Samples were dissolved in water (200 l) and injected into a reverse phase C18 column (TosoHaas RP18, 4.6 mm ϫ 25 cm) connected to a Rainin HPXL solvent delivery system. The column was first washed with water (flow rate 0.5 ml/min) and then with increasing concentrations of acetonitrile in water, as shown by the dashed line in Fig. 4. Radioactivity in the eluant was monitored by Radiomatic Flo-one Beta detector connected in line to the column.
Cell Surface Staining-To measure sLe x on the cell surface, U937 cells were grown in the presence and absence of primers for 42-48 h, washed with PBS, and fixed with 1% p-formaldehyde in PBS (5 min, room temperature). The fixed cells were rinsed once with PBS and suspended (1 ϫ 10 5 ) in 100 l of PBS containing 1% (w/v) BSA. CSLEX1 mAb (Becton-Dickinson) was added (5 g/ml). A control incubation contained nonspecific mouse IgM (5 g/ml) instead of CSLEX1. Human IgG (500 g/ml) was added to prevent reaction of CSLEX1 with Fc receptors on the cells. Anti-mouse IgM coupled to horseradish peroxidase was added (1:100). After 30 min at room temperature, the cells were washed with BSA/PBS buffer three times by centrifugation and resuspended in 250 l of reaction buffer containing 25 mM citric acid, 50 mM Na 2 HPO 4 , 3.7 mM o-phenylenediamine, and 30 l of 30% H 2 O 2 . The reaction was stopped after 5 min at room temperature by adding 250 l of 3 M sulfuric acid, and the absorbance at 490 nm was recorded.
The expression of sialic acids on the cell surface was estimated by lectin binding and flow cytometry. Cells were washed three times with cold PBS containing 2 mg/ml BSA and 0.05% sodium azide, and then treated for 15 min at room temperature with fluorescein isothiocyanateconjugated Maackia amurensis lectin (10 g/ml, E-Y Laboratories). After rinsing the cells at 4°C, they were fixed with 1% (w/v) p-formaldehyde and analyzed by flow cytometry (FACScan, Becton Dickinson).
Cell Adhesion Assays-HUVEC (1 ϫ 10 5 cells/well) at the second to fourth passage were grown for 48 h in a 24-well tissue culture plate (Corning). The monolayers were washed three times in Medium 199 (1 ml) without growth supplement and heparin. Medium 199 (0.5 ml) containing 20% fetal bovine serum with and without 20 ng/ml TNF-␣ (R&D Systems) was added next (51). After 5 h, the monolayers were washed three times with minimal essential medium (MEM).
U937 cells (2 ϫ 10 5 /well) in RPMI 1640 medium were incubated with and without acetylated disaccharide primers and [ 3 H-methyl]thymidine (1 Ci/ml) for 2 days in a six-well plate. The cells were washed three times with MEM containing 2% heat-inactivated fetal bovine serum (adjusted to pH 7.4 with Tricine) and resuspended at 1 ϫ 10 5 cells/ml. The cells were then added to TNF-␣-activated HUVEC and incubated for 35 min at 4°C. The plates were washed gently three times with 1 ml of cold MEM, and the cells were solubilized with 0.1 M NaOH (250 l).

RESULTS
Previous studies showed that F9 cells take up and rapidly deacetylate peracetylated Gal␤1-4GlcNAc-NM ( Fig. 1), resulting in the stimulation of oligosaccharide synthesis on the exogenous disaccharide (40). Priming of oligosaccharides in this way inhibited the expression of sLe x on the surface of HL-60 cells, presumably due to diversion of chain assembly on endogenous glycoconjugates. To determine whether other disaccharides would behave similarly, we prepared acetylated Glc-NAc␤1-3Gal-NM and compared it to Gal␤1-4GlcNAc-NM. Both compounds were quite hydrophobic and limited in solubility in aqueous growth medium to Յ0.2 mM. U937, a human histiocytic cell line that expresses sLe x , grew well in the presence of the disaccharides, up to 50 M (Fig. 2). Above this concentration, some diminution in growth rate was observed, possibly due to the detergent properties of the compounds at high concentration.
Acetylated Disaccharides Act as Primers-To test whether the disaccharides primed oligosaccharide chains, U937 cells were incubated with acetylated Gal␤1-4GlcNAc-NM or Glc-NAc␤1-3Gal-NM and [6-3 H]Fuc, [6-3 H]Gal, or [6-3 H]GlcN. Radioactive oligosaccharides assembled on the primer and secreted into the growth medium were collected on Sep-Pak C18 cartridges, which bound the compounds due to the hydrophobic aglycone (naphthalenemethanol). In addition, some endogenous material produced by the cells bound to the resin (Table I). However, these counts were not recoverable after analytical C18 reverse phase HPLC and probably represented unincorporated sugars or cellular glycoconjugates. Over 95% of the radioactive products generated on the primer were secreted into the growth medium. Therefore, in all subsequent experiments, the conditioned medium was the source of material for further analysis.
As shown in Table I, both acetylated Gal␤1-4GlcNAc-NM and acetylated GlcNAc␤1-3Gal-NM stimulated the incorporation of [6-3 H]fucose into products that bound to Sep-Pak C18 cartridges compared with the control which did not contain added disaccharides. The incorporation of [ 3 H]Fuc into oligosaccharide products rose with increasing concentration of added primer (Fig. 3). Priming on GlcNAc␤1-3Gal-NM showed a trend toward saturability above 50 M, whereas priming on Gal␤1-4GlcNAc-NM did not. Interestingly, the incorporation of [ 3 H]Fuc into oligosaccharides assembled on GlcNAc␤1-3Gal-NM was greater at all concentrations than those assembled on Gal␤1-4GlcNAc-NM. This difference was also observed in F9 teratocarcinoma cells (data not shown).
Structure of Oligosaccharides Formed on Primers-To characterize the primed products, the [ 3 H]Fuc-labeled oligosaccharides eluted from Sep-Pak C18 cartridges were separated into  charged and uncharged species by passage through a QAE-Sephadex column (45). Essentially all of the products generated on Gal␤1-4GlcNAc-NM were neutral (Table II), whereas about one-half of the material generated on GlcNAc␤1-3Gal-NM was neutral and the remainder eluted at a salt concentration indicative of a Ϫ1 charge (45). Analysis by reversed-phase HPLC of the neutral fucosylated products formed on Gal␤1-4GlcNAc-NM revealed a single peak of material eluting exactly at the same position as standard Le X glycoside (Gal␤1-4([ 14 C]Fuc␣1-3)GlcNAc␤-O-NM; Fig. 4B). The linkage stereochemistry of the fucose residue was established by two criteria. First, treatment of the radioactive material with ␣-fucosidase from Streptomyces sp. (which cleaves ␣1,3/␣1,4-linked fucose) (47) liberated all of the radioactivity, as measured by step elution from a C18 cartridge. In contrast, treating the material with F. oxyporum ␣-fucosidase (specific for ␣1,2/␣1,4-linked fucose) (46) had no effect, indicating that the linkage was ␣1-3. Second, feeding cells a modified disaccharide, in which the 3-OH of the GlcNAc residue was blocked by methylation (Gal␤1-4(3-OMe)GlcNAc-NM), did not result in [ 3 H]Fuc addition to oligosaccharides (Table I). These findings demonstrated that the primary oligosaccharide assembled on Gal␤1-4GlcNAc-NM was Le x , i.e. Gal␤1-4(Fuc␣1-3)-GlcNAc-NM. The lack of any [ 3 H]Gal or [ 3 H]GlcN incorporation into oligosaccharides above the control confirmed the absence of any chain extension products (Table I). Furthermore, treatment of the sample with mild base did not alter the elution of the product by reverse phase HPLC (data not shown), indicating that all of the acetyl groups in the original primer had been removed.
The fucosylated products generated on GlcNAc␤1-3Gal-NM were more complex inasmuch as both neutral and charged species were detected (Table II). The neutral species eluted from a C18 reverse-phase HPLC column as a single peak ϳ2 min earlier than the Le x glycoside standard, suggesting that it was more polar possibly due to the presence of another sugar residue (Fig. 4C). Labeling cells with [ 3 H]Gal yielded a peak with similar retention time (large peak in Fig. 5A (Fig. 5C), but after ␣-fucosidase treatment the galactose residue was released quantitatively (data not shown). Together, these results suggest the structure Gal␤1-4(Fuc␣1-3)GlcNAc␤1-3Gal-NM for the major neutral product generated on Glc-NAc␤1-3Gal-NM. The more retarded peak of [ 3 H]Gal-labeled material shown in Fig. 5A co-eluted with the standard [ 3 H]Gal␤1-4GlcNAc␤1-3Gal-NM, and it was sensitive to ␤galactosidase (Fig. 5C), suggesting that it was the trisaccharide, Gal␤1-4GlcNAc␤1-3Gal-NM.
The [ 3 H]Fuc-labeled species with a Ϫ1 charge eluted from the C18 reverse phase column as a single peak at an earlier time than the neutral products described above (Fig. 4D). After digestion with NDV sialidase, the charged material comigrated with the neutral [ 3 H]Fuc-labeled species (Fig. 4, compare E and C), and the neutralized material was sensitive to ␣1,3/␣1,4fucosidase (data not shown). These findings suggested the structure NeuAc␣2-3Gal␤1-4(Fuc␣1-3)GlcNAc␤1-3Gal␤-NM. To confirm its identity, we isolated the charged material labeled biosynthetically with [ 3 H]Gal. This material eluted from the C18 column at the same position as the charged species labeled with [ 3 H]Fuc (compare Figs. 5D and 4D). However, the [ 3 H]Gal-labeled material eluted in a somewhat broader peak than the [ 3 H]Fuc-labeled material, suggesting the presence of more than one charged species in the [ 3 H]Gallabeled products. Treating the [ 3 H]Gal-labeled samples with NDV sialidase followed by reverse phase HPLC yielded two peaks, which comigrated with the two neutral [ 3 H]Gal-labeled oligosaccharides (Fig. 5E). Fucosidase and galactosidase treatment had the same effects on these oligosaccharides as they had on the neutral species that were generated on this primer. Together, these findings suggested that the charged oligosaccharides were NeuAc␣2-3Gal␤1-4GlcNAc␤1-3Gal-NM and NeuAc␣2-3Gal␤1-4(Fuc␣1-3)GlcNAc␤1-3Gal-NM. These data and the recovery of material in the individual peaks are summarized in Fig. 6.
Inhibition of sLe x Expression-To test whether the priming activity of the disaccharides inhibited glycosylation of endogenous glycoconjugates, cells were labeled in the presence and absence of the primers with  Because the extent of mannosamine labeling was considerably lower than with the other radioactive sugars, we also examined the reactivity of the cells toward M. amurensis lectin, which binds terminal ␣2-3-linked sialic acid residues (52). Cell sorting with fluorescein isothiocyanate-conjugated lectin showed that disaccharide-treated and untreated cells had identical profiles (Fig. 7). As expected, the reactivity was diminished by previous treatment with NDV sialidase. Thus, the primers selectively reduced fucose incorporation without altering the addition of sialic acid. These findings suggested that the primers might reduce expression of fucose containing determinants on endogenous glycoconjugates, such as sLe x . To test this hypothesis, we measured sLe x expression on U937 cells using an ELISA (see "Ex-perimental Procedures"). As shown in Fig. 8, the primers inhibited sLe x expression in a dose-dependent manner. Acetylated Gal␤1-4GlcNAc-NM at its highest concentration (50 M) inhibited expression of sLe x by ϳ30%, whereas the more potent primer, acetylated GlcNAc␤1-3Gal-NM, inhibited sLe x expression by ϳ4-fold at 25 M. In general, the relative inhibitory activity of the disaccharides paralleled their priming activity (cf. Fig. 3).
Inhibition of Adhesion of U937 Cells to Activated Endothelial Cells-sLe x on O-linked glycoproteins is thought to be responsible for adhesion of U937 cells to E-selectin expressed on TNF-␣-activated endothelial cells (37). Thus, the inhibition of sLe x expression by the disaccharides should reduce their adhesion to activated endothelial cells. To test this possibility, we grew U937 cells in the presence of acetylated Gal␤1-4Glc-NAc-NM and GlcNAc␤1-3Gal-NM at various concentrations and challenged them to adhere to TNF-␣-activated HUVEC. Adhesion was dependent on previous activation of the endothelial cells with TNF-␣, and attachment under these conditions was inhibited by 80% with anti-ELAM-1 (E-selectin) mAb (Fig.   FIG. 4. Reverse phase HPLC of [6-3 H]Fuc-labeled oligosaccharides. U937 cells (1 ϫ 10 5 /ml) were incubated with 50 M disaccharides in the presence of 10 Ci/ml [6-3 H]Fuc. Radioactive oligosaccharides assembled on the primers were collected from the growth medium by chromatography on Sep-Pak C18 cartridges (see "Experimental Procedures"). Neutral and charged species were separated by chromatography on a column of QAE-Sephadex. Samples (2000 -4000 cpm) were dissolved in 0.2 ml of water and injected into the C18 reverse phase column, and products were eluted from the column using a gradient of acetonitrile in water (dashed line). Portions of each sample (1000 -2000 cpm) were treated with glycosidases and subjected to reverse phase HPLC as described above. A, radioactive standards; B, fucosylated oligosaccharide assembled on Gal␤1-4GlcNAc-NM; C, neutral fucosylated oligosaccharide built on GlcNAc␤1-3Gal-NM; D, charged fucosylated oligosaccharide assembled on GlcNAc␤1-3Gal-NM; E, charged fucosylated oligosaccharide assembled on GlcNAc␤1-3Gal-NM after treatment with NDV sialidase. The products assembled on the primer and secreted into the growth medium were collected by Sep-Pak C18 chromatography (see "Experimental Procedures"). Neutral and charged species were separated by chromatography on QAE-Sephadex. Some samples were treated with glycosidases and then subjected to reverse phase HPLC as described in the legend of Fig. 4. A, neutral oligosaccharides; B, after treatment with streptococcus sp. fucosidase; C, after treatment with D. pneumonia ␤-galactosidase; D, charged oligosaccharides; E, charged oligosaccharides after treatment with NDV sialidase. The arrow indicates the elution position of authentic Gal␤1-4GlcNAc␤1-3Gal-NM. 9). Pretreatment of U937 cells with NDV sialidase showed that adhesion depended on sialic residues. Cells treated with acetylated disaccharides adhered to a lesser extent than untreated cells and the extent of inhibition depended on concentration. Acetylated Gal␤1-4GlcNAc-NM inhibited cell adhesion by ϳ50% at 50 M, whereas acetylated GlcNAc␤1-3Gal-NM inhibited adhesion more dramatically. In general, the more potent primer (acetylated GlcNAc␤1-3Gal-NM) inhibited adhesion to a greater extent at all concentrations tested.
To confirm that the inhibition of adhesion was due to priming per se as opposed to some other effect, cells were incubated with the methylated disaccharide, Gal␤1-4(3-OMe)GlcNAc-NM, in which the site for fucosylation (3-OH of GlcNAc) was blocked by a methyl group. The methylated derivative did not prime a fucosylated product (Table I). Furthermore, it did not inhibit expression of cell surface sLe x (Fig. 10A) or affect the adhesion of U937 cells to activated HUVEC (Fig. 10B). Methylated derivatives of GlcNAc␤1-3Gal-NM (i.e. (3-OMe)Glc-NAc␤1-3Gal-naphthalenemethanol and (4-OMe)GlcNAc␤1-3-Gal-naphthalenemethanol) were also tested. Both disaccharides failed to prime any oligosaccharides and also did not inhibit sLe x expression or cell adhesion (data not shown).

DISCUSSION
We reported previously that HL-60 and F9 cells readily take up acetylated Gal␤1-4GlcNAc-NM, remove the acetyl groups, and use the deblocked disaccharide for oligosaccharide synthesis. In this report we have extended this finding to U937 cells and to a second disaccharide, GlcNAc␤1-3Gal-NM, which resembles another intermediate in the assembly of lactosaminoglycans (Fig. 1). Priming by these disaccharides results in the formation and secretion of oligosaccharides, sufficient in mass to divert the assembly of carbohydrate chains from endogenous glycoproteins and glycolipids. Priming resulted in diminution of sLe x expression and a parallel decrease in adhesion dependent on sLe x . Thus, the disaccharides act as metabolic inhibitors and have the interesting property of affecting a key step involved in an inflammatory reaction.
Several important features of these compounds deserve additional attention. First, in order for the compounds to be effective, the cells had to remove the O-linked acetyl groups, which were required for passage of the compounds across the plasma membrane (40). All of the oligosaccharide products generated by the cells lacked the O-acetyl groups, suggesting that deacetylation occurred very efficiently. In other studies, we have found that deacetylation occurs rapidly in the cytoplasm and in microsomal membranes, 4 presumably mediated by one or more carboxylesterases (53)(54)(55)(56)(57). Because the glycosyltransferases responsible for oligosaccharide biosynthesis reside in the Golgi, only the deacetylated compounds arising in that compartment may be accessible to the enzymes. If correct, then the effective concentration of active, deacetylated primers may be much less than the disaccharide concentration added to the cells. However, a recent study of ␤-xyloside priming in isolated Golgi vesicles indicates that the primary neutral disaccharide product (Gal␤1-4Xyl␤-O-methylumbelliferol) could exit the Golgi and appear in the incubation medium (58). Furthermore, incubation of Golgi vesicles with non-acetylated Gal␤1-4Glc␤-O-methylumbelliferol gave rise to sialylated products trapped inside the vesicles (59). These studies suggest that the Golgi may be much more permeable to polar compounds than the plasma membrane. The permeability limit of the Golgi with respect to oligosaccharide size and structure deserves further study since it may be possible to design larger, more specific primers.
As shown in Figs. 4 and 5, Gal␤1-4GlcNAc-NM and Glc-NAc␤1-3Gal-NM primed the formation of specific oligosaccharides in U937 cells. Gal␤1-4GlcNAc-NM underwent selective fucosylation to form the Le x analog, Gal␤1-4(Fuc␣1-3)GlcNAc-NM, as determined by enzymatic defucosylation. That Gal␤1-4(Fuc␣1-3)GlcNAc-NM was the sole product formed indicates that the inhibition of sLe x expression and diminished cell adhesion by Gal␤1-4GlcNAc-NM was due to inhibition of fucosylation of endogenous glycoconjugates rather than inhibition at another biosynthetic step (e.g. sialylation). Analysis of endogenous glycoconjugates confirmed this idea (Table III and Fig.  7). A more complex array of oligosaccharides arose on Glc-NAc␤1-3Gal-NM, including fucosylated and sialylated oligosaccharides. Since sialylation of endogenous glycoconjugates also was not altered by GlcNAc␤1-3Gal-NM, both compounds appear to inhibit sLe x expression by altering fucosylation.
The specificity of oligosaccharide synthesis on the primers provides insights into the substrate preference and capacity of the biosynthetic enzymes involved in sLe x formation in vivo. Myeloid cells express nearly equal amounts of Fuc TIV and Fuc TVII mRNA, and both enzymes play a role in forming fucosylated selectin ligands (60,61), although Fuc TVII may predominate (62). Interestingly, Gal␤1-4GlcNAc␤-O-R (where R ϭ a hydrophobic aglycone) serves as an acceptor in vitro for Fuc TIV, but not Fuc TVII because the latter requires prior sialylation for activity (e.g. NeuAc␣2-3Gal␤1-4GlcNAc␤-O-R is a substrate; Refs. 60, 62, and 63). These observations suggest that Gal␤1-4(Fuc␣1-3)GlcNAc-NM was produced on Gal␤1-4GlcNAc-NM by the action of Fuc TIV rather than Fuc TVII. Thus, Fuc TIV would appear to play a significant role in forming E-selectin ligands in U937 cells.
The absence of any sialylated products from Gal␤1-4Glc-NAc-NM suggests that the K m for the ␣2-3 sialyltransferase may be higher that the apparent K m for Fuc TIV. Since extension products also were not detected, the K m for the ␤1-3 GlcNAc transferase responsible for the formation of polylactosaminoglycan chains also may be high (64). These data do not take into account competition by endogenous fucosylated and were incubated for 48 h at 37°C with acetylated disaccharides at the indicated concentration. The cells were challenged to adhere to TNF-␣-activated HUVEC and the total number of adherent cells was measured (see "Experimental Procedures"). Control incubations were done using HUVEC without prior activation with TNF-␣ or with activated cells after treatment of U937 cells with sialidase, which destroys sLe x . Each point is the average of duplicate determinations that varied by Յ15%. sialylated glycoprotein and glycolipid intermediates which may differ in concentration and affinity. Analysis of the endogenous glycoconjugates that accumulate in the presence of the primers may provide more insight into this issue. Moreover, such studies would reveal whether the disaccharides inhibit specific glycoproteins and whether the pathways of N-linked and Olinked oligosaccharide formation are affected similarly.
Inasmuch as these disaccharides resemble biosynthetic intermediates involved in the formation of sLe x , they compete with endogenous substrates. Thus, they act like N-acetylgalactosaminides, which mimic ␣-GalNAc containing glycoprotein intermediates (35). However, the disaccharide primers reported here are nearly 50-fold more potent than the monosaccharide primer. This difference is not related to the aglycone, because GalNAc␣-O-naphthalenemethanol has the same dose profile as the benzyl derivative. 4 Interestingly, inhibition of sLe x expression occurred at lower doses of GlcNAc␤1-3Gal-NM compared with Gal␤1-4GlcNAc-NM, which correlated well with the fact that the former primed fucosylated oligosaccharide chains at lower concentration than the latter (Fig. 3). A priori, one could not have predicted these differences in priming efficacy by merely comparing kinetic constants. Thus, other disaccharide-based primers may prove even more potent than the ones reported here (e.g. GlcNAc␤1-6Gal␤-O-NM, Glc-NAc␤1-6GalNAc␣-O-NM, or Gal␤1-3GalNAc␣-O-NM). Use of higher order oligosaccharide primers also may prove useful, and varying the aglycone may be advantageous (70 -72).
Finally, the disaccharides described here represent starting points for designing specific inhibitors of glycosyltransferases and drugs for treating diseases or disorders that depend on protein-carbohydrate interactions (73). The selective addition of fucose to Gal␤1-4GlcNAc-NM suggests that modifying the disaccharide by deoxygenation, fluorination, or by the addition of reactive groups may convert a primer into an active-site directed inhibitor (32,33,74). It is unclear how these compounds will behave in animals and whether they have useful therapeutic value. Preliminary studies indicate that intraperitoneal injection of Gal␤1-4GlcNAc-NM into mice had no immediate deleterious effects, but more studies are needed to determine its bioavailability, clearance, and ability to block formation of sLe x in vivo.