1-Deoxymannojirimycin Inhibits Capillary Tube Formation in Vitro

Transcription

1 THE JOURNAL OF BIOLOGICAL CHEMISTRY by The American Society for Biochemistry and Molecular Biology, Inc. Vol. 267, No. 36, Issue of December 25, pp ,1992 Printed in U. S.A. 1-Deoxymannojirimycin Inhibits Capillary Tube Formation in Vitro ANALYSIS OF N-LINKED OLIGOSACCHARIDES IN BOVINE CAPILLARY ENDOTHELIAL CELLS* (Received for publication, April 22, 1992) Mai Nguyen$#ll, Judah Folkman$((, and Joyce BischoffS ** $4 From the $Surgical Research Laboratory, Children s Hospital, Brigham and Women s Hospital, the 11 Departments of Surgery and Anatomy and Cell Biology, and the **Departments of Surgery and Cellular and Molecular Physiology, Harvard Medical School, Boston, Massachusetts Capillary endothelial cells can be induced to form capillary-like structures in vitro by plating on fibronectin-coated dishes (Ingber, D. E., and Folkman, J. (1989) J. Cell Biol. 109, ), thereby mimicking angiogenesis. To assess the role of glycoproteins bearing asparagine-linked oligosaccharides in this process, we tested the effect of oligosaccharide processing inhibitors on the formation of capillary tubes. Deoxymannojirimycin, a compound that prevents synthesis of hybrid and complex-type oligosaccharides, inhibited the formation of capillary tubes. In contrast, swainsonine, an inhibitor that blocks synthesis of complex- but not hybrid-type oligosaccharides, did not inhibit tube formation. Lectin affinity chromatography of 2-[3H] mannose-labeled glycopeptides from endothelial cells induced to form tubes did not reveal a striking difference in the spectrum of oligosaccharides compared to uninduced cells. Since endothelial cells formed tubes normally in the presence of swainsonine, we analyzed glycopeptides from swainsonine-treated induced and uninduced cells. Cells induced to form tubes were enriched in monosialylated hybrid-type oligosaccharides sensitive to a-fucosidase, &galactosidase, suggestive of sialyl Lewis-X determinants. We used an enzyme-linked immunoassay to measure sialyl Lewis-X epitopes on capillary endothelial cells and found that both induced and uninduced cells expressed sialyl Lewis-X epitopes. Deoxymanno- jirimycin and, to a lesser extent, swainsonine reduced the level of sialyl Lewis-X epitopes in cells induced to form capillary tubes, but neither compound affected the level of epitopes in cell monolayers. We conclude that synthesis of at least hybrid-type oligosaccharides is required for capillary tube formation in vitro and that an increase in monosialylated, fucosylated glycans on asparagine-linked oligosaccharides occurs during this process. Growth of new capillaries, i.e. angiogenesis, is a complex and highly regulated process which occurs normally during specific physiologic events and during development. Abnor- * This work was supported by a grant-in-aid from the American Heart Association (to J. B.) and a grant from Takeda Chemical Industries Ltd., Osaka, Japan, to Harvard University. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked aduertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. n Supported by National Institutes of Health Grant 5 T32 CA $$To whom correspondence and reprint requests should be addressed mal capillary growth, however, underlies many disease states such as diabetic retinopathy, rheumatoid arthritis, and hemangiomas (1). In addition, neovascularization of tumors occurs by an aberrant stimulation of normally quiescent endothelial cells to migrate, proliferate, and form new capillary blood vessels. When grown under the appropriate culture conditions, endothelial cells isolated from both large and small vessels can form three-dimensional structures that resemble capillaries (2-5). For example, bovine capillary endothelial (BCE) cells isolated from the adrenal cortex can be induced to form capillary-like tubes similar to capillary blood vessels in uiuo and thus provide a means to study angiogenesis in uitro (6). Studies using BCE cells as well as other endothelial cell types have led to the discovery of growth factors, growth inhibitors, and extracellular matrix molecules involved in angiogenesis (for review, see Ref. 7). Less is known, however, about the cell surface molecules that mediate the recognition and adhesion events that must occur during the formation of new capillaries. Because carbohydrates are known to mediate many cellular interactions in mammalian tissues (8), we set out to examine the role of glycoproteins bearing asparagine-linked (N-linked) oligosaccharides in capillary tube formation using BCE cells and the method described by Ingber and Folkman (6) to induce capillary tube formation in uitro. Proteins containing N-linked oligosaccharides are glycosylated at their site of synthesis in the rough endoplasmic reticulum by the transfer of Glc3Man9GlcNAcP to acceptor asparagine residues. The oligosaccharide chains are processed by a variety of glycosidases and glycosyltransferases as the newly synthesized glycoproteins are transported to the Golgi apparatus and then to their final destinations in the cell (for review, see Ref. 9). The oligosaccharide structures found on mature glycoproteins can be sorted into three classes: high mannose-, hybrid-, and complex-type. High mannose-type oligosaccharides, besides comprising a class of N-linked oligosaccharides, are also intermediates in the biosynthesis of hybrid- and complex-type oligosaccharides. Therefore, inhibitors that block processing of high mannose-type oligosaccharides also block synthesis of hybrid- and complex-type oligosaccharides and can be used to assess the functions of these types of N-linked oligosaccharides. Two distinct inhibitors of Golgi a-mannosidase I, 1- deoxymannojirimycin (dmm) and kifunensine, block synthesis of hybrid- and complex-type oligosaccharides (10-12). In The abbreviations used are: BCE, bovine capillary endothelial; N-linked, asparagine-linked; dmm, 1-deoxymannojirimycin; SLAC, serial lectin affinity chromatography; ConA, concanavalin A, QAE, quaternary aminoethyl; HPLC, high pressure liquid chromatography; bfgf, basic fibroblast growth factor; DME, Dulbecco s modified Eagle s media; PHA, phytohemagglutinin; PBS, phosphate-buffered saline.

2 Deoxymannojirimycin Inhibits contrast, swainsonine inhibits Golgi cy-mannosidase 11, a point downstream from the key intermediate for synthesis of hybrid-type oligosaccharides, and therefore allows synthesis of hybrid- but not complex-type oligosaccharides (13). We examined the effect of these three inhibitors on capillary tube formation in vitro and found that dmm and kifunensine, but not swainsonine, blocked capillary tube formation. To determine if the array of N-linked oligosaccharides synthesized by BCE cells changed upon tube induction, we analyzed 2-[3H]mannose-labeled glycopeptides from induced and uninduced BCE cells by serial lectin affinity chromatography (SLAC) (14) and high pressure liquid chromatography (HPLC). In addition, we analyzed the oligosaccharides synthesized in swainsonine-treated BCE cells either induced to form tubes or grown in monolayer. Using a combination of concanavalin A (ConA)-Sepharose, quaternary aminoethyl (QAE)-Sephadex, HPLC, and exoglycosidase digestions, we found that a subset of hybrid-type oligosaccharides synthesized by BCE tubes were monosialylated, fucosylated structures containing the four monosaccharides that comprise sialyl Lewis-X and sialyl Lewis-A glycans. BCE cells grown in monolayer, however, contained -10-fold less of this oligosaccharide species. Sialyl Lewis-X (Siaa2, 3GalP1, 4(Fuccy1,3) GlcNAc) structures as well as Lewis-X (Galpl, 4(Fuccul,3)GlcNAc) structures have been identified on N- linked oligosaccharides isolated from human neutrophilic granulocytes (15) and human colon carcinoma cells (16). Sialyl Lewis-X (17-19) and sialyl Lewis-A (20-22) have also been shown to be ligands for endothelial-leukocyte adhesion molecule-1, now designated E-selectin, and are thoughto be involved in binding of leukocytes to activated endothelium during inflammation (for review, see Ref. 23). Antibodies directed against sialyl Lewis-X and sialyl Lewis-A glycans inhibit capillary tube formation, and BCE cells express a novel form of E-selectin.' Thus, it is interesting to speculate that asparagine-linked oligosaccharides bearing sialyl Lewis- X or sialyl Lewis-A structures may be directly involved in capillary tube formation. EXPERIMENTAL PROCEDURES Materials-Materials were obtained from the following sources: [2-3H]mannose (specific activity = 23 Ci/mmol), Du Pont-New England Nuclear; Ecolume, ICN Biomedicals; 1-deoxymannojirimycin and endoglycosidase H, Genzyme; swainsonine, pronase, and bovine testis P-galactosidase, Boehringer Mannheim; Vibrio cholerae neuraminidase, Calbiochem; jack bean meal P-D-N-acetylhexosaminidase, V- Labs; bovine epididymal a-l-fucosidase, a-methylmannoside, a- methylglucoside, Amberlite MB-3, anti-mouse IgM (p-chain-specific)-alkaline phosphatase conjugate, and p-nitrophenyl phosphate, Sigma; ConA-Sepharose 4B, QAE-Sephadex A-25, Sephadex G-25 medium, Pharmacia LKB Biotechnology Inc.; AX-5 MicroPak column, Varian; HPLC-grade acetonitrile, Pierce; wheat germagglutinin-agarose, pea lectin-agarose, leukoagglutinating phytohemagglutinin-agarose (L-PHA-agarose), erythroagglutinating phytohemagglutinin-agarose (E-PHA-agarose), and N-acetylglucosamine, E-Y Laboratories; human CR-transferrin (iron-saturated), Collaborative Research; human plasma fibronectin, Chemicon International; human high density lipoprotein, Organon Teknika; Dulbecco's modified Eagle's medium (DME) and calf serum, Hazelton; and 25% glutaraldehyde, Electron Microscopy Sciences; anti-sialyl Lewis-A ascites, UCLA Tissue Typing Laboratory, Los Angeles, CA; and anti-sialyl Lewis-X (CSLEXl), American Tissue Culture Collection, Rockville, MD. All other chemicals were of reagent grade. The following reagents were kindly provided to us: recombinant basic fibroblast growth factor (bfgf), Takeda Chemical Industries, Osaka, Japan; anti-sialyl Lewis-X ascites, Dr. Chris Corless, Brigham and Women's Hospital; kifunensine, Dr. Alan Elbein, University of Arkansas, Little Rock, AR; a [3H]galactose-labeled E-PHA oligosac- M. H. Nguyen, N. A. Strubel, and J. Bischoff, submitted for publication. Capillary Formation in Vitro charide standard, Dr. Richard Cummings, University of Georgia, Athens, GA; a [3H]mannose-labeled pea lectin oligosaccharide standard, Dr. Rosalind Kornfeld, Washington University, St. Louis, MO; and jack bean a-mannosidase, Dr. Chris Gabel, Pfizer Pharmaceuticals, Groton, CT. 2-[3H]Mannose-labeled high mannose-type oligosaccharide standards were prepared as described (24). Oligosaccharide standards for all other lectin columns were prepared from a BW5147 mouse lymphoma cell line as described (14). Cell Culture-Bovine capillary endothelial (BCE) cells were isolated from adrenal cortex, cloned, and passaged as previously described (25) except that 5 ng/ml basic fibroblast growth factor (bfgf) was used instead of tumor-conditioned medium. BCE cells from passage numbers 10 through 13 were used for experiments. Tube formation was induced as described by Ingber and Folkman (6) with the following modifications. 1) Bacteriologic plates were coated with fibronectin at 37 "C and then washed with phosphate-buffered saline (PBS). 2) For tube induction, BCE cells were serum-starved for 16 h in DME, 1% calf serum, 2 mm glutamine, 100 units/ml penicillin, 100 units/ml streptomycin. 3) Tube induction medium contained 5 ng/ml bfgf instead of 2 ng/ml bfgf. To determine cell number, a single-cell suspension was prepared by trypsinization and subjected to Coulter counter analysis. The oligosaccharide processing inhibitors were added to the cell cultures as indicated. Control BCE cells were trypsinized, but then plated on gelatin-coated tissue culture dishes (rather than fibronectin-coated dishes) in DME, 10% calf serum, 2 mm glutamine, 100 units/ml penicillin, 100 units/ml streptomycin, and 5 ng/ml bfgf (growth media). Photography-Phase-contrast images of cells were obtained using a Diaphot Nikon inverted microscope and 400ASA Kodak TMY film. Cells were fixed by the addition of 1/10 volume of 25% glutaraldehyde to the culture medium for 5 min at room temperature followed by washing three times with PBS. Metabolic Labeling-BCE cell cultures (-10' cells) were metabolically labeled with 100 pci/ml 2-[3H]mannose in DME with reduced D-glucose (100 mg/liter). The other components of growth or tube induction media were the same. At the end of the labeling period, the radioactive medium was removed, and the cells were washed with PBS. Total cell-associated glycopeptides were isolated as described below. Preparation of 2-fH]Mannose-labeled Glycopeptides from BCE Cells-Cellswere solubilized in 1 mlof 50 mm Tris, ph 7.5, 1% sodium dodecyl sulfate, 0.1% p-mercaptoethanol and precipitated with 10% trichloroacetic acid in the presence of 500 pg/ml carrier bovine serum albumin. Precipitated proteins were incubated with 10 mg/ml pronase in 100 mm Tris, ph 8, 2 mm CaC12 at 56 "c for 24 h, followed by heat inactivation of the pronase. Unincorporated [3H] mannose was removed by Sephadex G-25 chromatography. Glycopeptide fractions were pooled, dried by evaporation under reduced pressure, and subjected to ConA-Sepharose chromatography as described (24). Here, and in all chromatography steps to follow, an aliquot of each column fraction was counted in Ecolume scintillation fluid in a Beckman LS-3801 scintillation counter. Serial Lectin Affinity Chromatography (SLAC)-Chromatography on wheat germ agglutinin-agarose, pea lectin-agarose, E-PHA-agarose, and L-PHA-agarose was carried out exactly as described (14). Each lectin-agarose column was tested for its ability to bind or retard the appropriate test glycopeptide. To prepare samples for subsequent lectin-agarose chromatography steps, fractions were pooled, dried by evaporation, desalted on Sephadex G-25, and dried again. Structural Analysis of Oligosaccharides-Glycopeptides eluted from ConA-Sepharose with 100 mm a-methylmannoside were desalted on Sephadex G-25, dried by evaporation, resuspended in 50 mm sodium acetate, ph 5.5, and digested with endoglycosidase H (1 milliunit/pl) for 16 hat 37 "C. For oligosaccharides from dmm-treatedor untreated BCE cells, the endoglycosidase H-released oligosaccharides were desalted on Amberlite MB-3, dried by evaporation, and fractionated according to size on a 4 mm X 30-cm MicroPak AX-5 HPLC column (26). High mannose oligosaccharides ranging in size from Man,GlcNAc to Man9GlcNAc were used as standards. Endoglycosidase H-released oligosaccharides from swainsonine-treated cells were desalted by Sephadex G-25, dried by evaporation, and fractionated according to charge by QAE-Sephadex anion exchange chromatography (27). Neutral (unbound) oligosaccharides recovered from QAE- Sephadex were subjected to HPLC either before or after digestion with jack bean a-mannosidase. Jack bean a-mannosidase digestion resulted in the removal of high mannose oligosaccharides from hybrid-type oligosaccharides since high mannose structures were de-

3 1-Deoxymannojirimycin Inhibits Capillary Formation in graded to freemannoseandmanj31,4glcnac.negativelycharged oligosaccharides were treated with neuraminidase, desalted, andrechromatographed on QAE-Sephadex. Neuraminidase-sensitive, neutral oligosaccharideswere then fractionated according to sizeby HPLC. Individual peaks from HPLCwere digested with exoglycosidases, desalted on Amberlite MB-3, dried by evaporation, and fractionated by HPLC. Exoglycosidase Digestions-Neuraminidase digestions were carried out in 50 mm sodium acetate, ph 4.6, 150 mm NaCl, 10 mm CaClz with 54 milliunits/ml Vibrio cholerae neuraminidase for 48 h at 37 "C. Jack bean a-mannosidasedigestionswerecarriedoutin 50 mm sodium citrate, ph 4.5, with 5 units/ml for 16 h at 37 "C. 8-Galactosidase digestionswere carried out in 0.1M sodium citrate phosphate, ph 4.3, with 30 milliunits/ml bovine testis 8-galactosidase for 16 h at 37 "C. a-fucosidase digestions were carried out in 0.1 M sodium citrate phosphate buffer, ph 6.0, with 0.3unit/ml of bovine epididymis a-fucosidase for 48 h at 37 "C. 8-N-Acetylhexosaminidasedigestions were carried out in50 mm sodium citrate, ph 5.0, with 10 units/ ml jack bean meal 8-N-acetylhexosaminidase for 24 h at 37 "C.Mock digestions were carried out in parallel for each enzyme digestion. Enzyme-linked Immunoassay to MeasureSialyl Lewis-X and Sialyl Lewis-A Binding Sites-semm-starved BCE cells were trypsinized and plated onto gelatin-coated dishes in growth media (monolayers) or induced to form tubes by plating on fibronectin-coated dishes in tube induction media (tubes). Swainsonine (0.1 mm) or dmm(1.0 mm) was addedto cells at the time of plating. Forty-eighthours later, the BCE monolayers and BCE tubes were washedand gently scraped 5 mm EDTA. Thecellswere from theirdishesinicecoldpbs, sedimentedonto a poly-(l-lysine)-coated96-wellmicrotiter plate (88,000 cells/well) and fixed with 0.25% glutaraldehyde/pbs (28). et al. (29) Antibody bindingsites were measured as described by Zhou by incubating firstwith saturating amounts of anti-sialyllewis-x or anti-sialyl Lewis-A ascites (1/50dilution),then with anti-mouseigmalkaline phosphatase (1/1000 dilution), and then with 1 mg/ml p nitrophenyl phosphate in0.1 M NaHC03, ph9.6. Vitro FIG Deoxymannojirimycin inhibits capillary tubeformation in a reversible manner. BCE cells induced to form tubes were treated with 0 dmm (PanelA ), with 1 mm dmm from 0 to 42 h of induction (Panel B ),with 1 mm dmm from 0 to 20 h of induction (Panel C), orfrom 20 to 42 h of induction (Panel D). Ineach or without 1mM dmm condition, fresh induction medium either with was added to cultures at 20 h. Cells were fixed with glutaraldehyde 42 h after plating and photographed. RESULTS 1 -DeoxymunmjirimycinInhibits Capillary Tube Formation in a Reversible Manner-BCE cells induced to form capillary tubes in vitro undergo dramatic changes incell architecture. T h e cells extend processes, form cell-cell contacts, and establish a branched network. Cellular cordsare formed after approximately 24 h, which are anchored to the culture dish through contacts withcell aggregates. Depending on the microscope plane of focus, a capillary lumen can be observed as a translucent slit within the cell (6). We tested the effect of exposure to 1 mm dmm on this capillary tubeformation process by adding dmm to induced cultures at the time of plating and examined cell morphology 42 h later. Cells induced toform tubes in the absence of dmm areshown in Fig. 1, Panel A. Compared to untreated cells which formed capillary tube-like structures, dmm-treatedcells exhibited an adherent, flattened morphology and did not form tubes(fig. 1, Panel B ). In Panel C,BCE cells were exposed to 1mM dmm for the first 20 h of tube induction, and then the inhibitorcontaining medium was removed and replaced with tube induction medium. At 42 h, the cells had formed capillary structures which resembled the untreated control cultures (Fig. 1, Panel C). This result indicated that dmm was not toxic to BCE cells and that its effect was reversible. Addition of 1 mm dmm to cultures 20 h after plating also inhibited tube formation (Fig. 1, Panel D ). By 42 h, thecells exhibited a flattened morphology with very few, if any, capillary tube structures indicating that d" was inhibitory even when added 20 hoursafter plating. These results indicate that synthesis of one ormore glycoproteinsbearing properlyprocessedn-linked oligosaccharide chainsis requiredfor the formation of capillary tubes in vitro. Since Akiyama et al. (30)reported that fibroblasts grown in the presence of 1 mm dmm for 40 h exhibited a 70% reduction in fibroblast attachmentto fibronectin-coated FIG. 2. Effect of kifunensine and swainsonineon capillary tube formation. In Panel A, BCE cells were plated on gelatin-coated dishes in growth media with 4.6 p~ kifunensine. In Panels B-D, BCE cellswereinduced to form tubes onfibronectin-coateddishes,as described under "Experimental Procedures," inthe absence of inhibitor (Panel B ), in the presence of 4.6 p~ kifunensine (Panel C),or in the presence of 0.1 mm swainsonine (PanelD). Cells were fixed with glutaraldehyde 48 h after plating. dishes, we measured the number of BCE cells attached to fibronectin-coated dishes in thepresence or absence of 1 mm dmm. dmm-treated BCEs exhibited a 40% increase in attachment to fibronectin-coated dishes ( pg of fibronectin/cm2) compared to untreated cells that had been subjected to the same tube inductionprotocol (data not shown). These results indicate that dmm does not affect the ability of BCE cells to attach fibronectin to under theseexperimental conditions. Kifunensine and swainsonine, two different inhibitors of oligosaccharide processing, were also tested for their effect on capillary tube formation. As withdmmand swainsonine, incubation of BCE cells under growth conditions in thepresence of kifunensine did not result in a visible effect on BCE cell morphology (Fig. 2, Panel A). Addition of kifunensine to BCE cells induced to formtubes, however, blocked tube formation (Fig. 2, Panel C)compared to untreated cells (Fig. 2, Panel B ). Kifunensine is a potent inhibitor of Golgi a-

4 Deoxymannojirimycin Inhibits Capillary Formation in Vitro mannosidase I (12) and therefore would be expected to exert the same inhibitory effect as dmm. In contrast, swainsonine (0.1 mm) had no effect on BCE cell capillary tube formation (Fig. 2, Panel D). Swainsonine blocks synthesis of complexbut not hybrid-type oligosaccharides by inhibiting Golgi a- mannosidase I1 (13). Since swainsonine did not inhibit tube formation, synthesis of glycoproteins bearing hybrid- and high mannose-type oligosaccharides must be sufficient for tube formation. Effect of the Inhibitors on Oligosaccharide Processing-BCE cells plated in growth medium on gelatin-coated dishes (monolayers) or induced to form tubes (tubes) were incubated in the presence or absence of inhibitor beginning at 0 h and then labeled with 2-[3H]mann~~e from h after plating. We chose this labeling period since the results presented above (Fig. 1) suggested that proper processing of N-linked oligosaccharides is most critical 20 h after plating the cells to form capillary tubes. Glycopeptides were isolated and analyzed by ConA-Sepharose chromatography. The N-linked glycopeptides were separated into three fractions: the unbound fraction, ConA I, contains tri- and tetra-antennary complextype oligosaccharides; the 10 mm a-methylglucoside eluate, ConA 11, contains biantennary complex-type glycopeptides; and the 100 mm a-methylmannoside eluate, ConA 111, contains hybrid- and high mannose-type glycopeptides. The results, summarized in Table I, demonstrated that all three inhibitors reduced synthesis of bi-, tri-, and tetra-antennary complex-type oligosaccharides. The concentrations of inhibitors used are similar to concentrations used by others to examine oligosaccharide synthesis and function in mammalian cells (10, 24, 31). Analysis of N-Linked Oligosaccharides by SLAC-To determine if induction of capillary-like structures alters the array of N-linked oligosaccharides synthesized by BCE cells, we examined these structures by metabolically labeling induced and uninduced cells with 2-[3H]mannose and then preparing total cellular glycopeptides for analysis. We used a rapid and quantitative method developed by Cummings and Kornfeld (14) in which radiolabeled glycopeptides are subjected to sequential lectin affinity chromatography steps. The structural information obtained from this analysis reveals the overall spectrum of N-linked oligosaccharides synthesized by the cell, especially pertaining to the branching pattern of the trimannosyl core. To extend the analysis, the high mannosetype glycopeptides eluted from ConA-Sepharose with 100 mm a-methylmannoside were digested with endoglycosidase H TABLE I Effect of processing inhibitors on asparagine-linked oligosaccharide biosynthesis in BCE cells Cells were labeled with 2-[3H]mannose from 20 to 48 h after plating. Cell-associated glycopeptides were isolated and subircted to ConA- Sepharose chromatography. The counts/min incorporated into glycopeptides from each culture condition are shown. The radioactivity (counts/min) in each eluate pool, defined in the text, is expressed as a percent of total counts/min recovered. ConA I ConA I1 ConA 111 % total cpm Monolayers No addition , mm dmm , pm kifunensine , mm swainsonine ,912 Tube No addition , mm dmm nn.* ' 43, pm kifunensine , mm swainsonine ,005 and fractionated according to size by AX-5 HPLC (26). The structures that can be identified by these techniques are shown in Fig. 3. BCE cells, either induced to form tubes or controls in which cells were plated in growth medium on gelatin-coated dishes (monolayers), were labeled with 2-[3H] mannose for the following periods of time after plating: 6-20 h, h, and h. These time periods were chosen to examine potential changes in the array of oligosaccharide structures during the capillary differentiation process. The results of the SLAC analysis are summarized in Table 11, and the structures identified are shown in Fig. 3. The relative amounts of each oligosaccharide were calculated by correcting for the number of [3H]mannose residues in each structure. The array of structures synthesized by induced or control BCE cells in the three time frames was similar. Some small differences, however, were observed. For example, BCE cells induced to form tubes were enriched %fold in the high mannose oligosaccharide ManaGlcNAc during the and h labeling periods compared to BCE cells in monolayer. In addition, high mannose-type glycopeptides were more abundant in tubes (38%) uersus monolayers (28%). Interestingly, in both monolayer and tube BCE cells, a high percentage (35-40%) of the complex-type glycopeptides were retarded by E-PHA-agarose suggesting the presence of a bisecting GlcNAc residue (Fig. 3, Cl ). Furthermore, a relatively small percent of the glycopeptides in induced or uninduced cells bound to pea lectin agarose (Fig. 3, C5, C6, and C10). This finding may be consistent with abundance of structures retarded by E-PHA-agarose since the presence of a bisecting GlcNAc on core-fucosylated oligosaccharides can reduce binding to pea lectin (32). Alternatively, since prior addition of a bisecting GlcNAc by N-acetylglucosaminyltransferase I11 precludes addition of a core fucose residue by the a-1,6-fucosyl- Complex-type glycopeptides GN - c5 -GNL M M-ON-ON-bdn High Hybrid-type Mannose-type r-7 I MTM,a -0-GN < - C6 FIG. 3. Structures of N-linked glycopeptides identified by SLAC. Abbreviations for the monosaccharides are as follows: g, galactose; GN, N-acetylglucosamine; M, mannose; Fuc, fucose; Asn, asparegine. The numbers indicate the linkage to hydroxyl groups on the monosaccharide to the right. Arrows indicate positions where additional substitutions may occur. High mannose-type glycopeptides are composed of structures with 5 to 9 mannose residues; variable sugar residues are indicated by boxes. 1

5 1 -Deoxymannojirimycin Inhibits Capillary Formation in Vitro TABLE I1 Summary of serial lection affinity chromatography of N-linked glycopeptides 2-[3H]Mannose-labeled glycopeptides from BCE cell monolayers or BCE cells induced to form tubes were subjected to SLAC and HPLC. HPLC separated high mannose-type oligosaccharides by size into species containing 5 to 9 mannose residues. Structures of each glycopeptide are shown in Fig. 3. The counts/min incorporated into total cellular N-linked glycopeptides are shown for each labeling period. Monolayers Tubes Glycopeptide 6-20 h h h 6-20 h h h % % c C2,3, C4, c C c c Hybrid ManeGN Man6GN Man7GN MansGN Man9GN Total cdm 3.1 X lo6 4.1 X lo6 3.7 X X x lo6 1.0 x lo6 / Con-A Sepharose. QAESephadex \ (1- &-e) nennminidase -. QAE -. HPU: -. exoglycosidnse digestions -. HPU: (neutral) Jack Bean emnnaridase -. HPU: -. exoglycosidnse digestions -. HPU: FIG. 4. Analysis of N-linked oligosaccharides in swainsonine-treated BCE cells. Total cell-associated N-linked glycopeptides were isolated from BCE cell monolayers and tubes that had been labeled with 2-[3H]mannose. The analysis was carried out as outlined and described under Experimental Procedures. transferase (33), one might expect fewer pea lectin-bound of the oligosaccharides were recovered in the unbound (neuglycopeptides. tral) and the 20 mm NaCl eluate (1- charge) fractions. Oli- Analysis of N-Linked Oligosaccharides in Swainsonine- gosaccharides recovered in the 20 mm NaCl eluate were ditreated BCE Cells-Since swainsonine did not inhibit capil- gested with neuraminidase and rechromatographed on QAElary tube formation, the hybrid- and high mannose-type oligosaccharides synthesized in its presence must be sufficient Sephadex (Fig. 5, Panels C and D). Although the same percent of counts/min (23%) from total cell-associated glycopeptides for capillary differentiation. Therefore, we analyzed the struc- were present initially in the 20 mm NaCl eluate in both BCE tures synthesized in swainsonine-treated BCE cell mono- monolayers and tubes, only 6% of these counts/min from layers and BCE cells induced to form tubes. The cells were monolayer cells were sensitive to neuraminidase versus 55% labeled for 28 h total beginning at 20 h after plating and in cells induced to form tubes (Fig. 5, Panel D uersus C). continuing to 48 h. The rationale for this labeling period was based on our finding that the presence of dmm during 0-20 h of tube induction did not affect capillary formation (Fig. 1) as long as the inhibitor was removed from the cells. Thus, to label oligosaccharides synthesized during the most critical Thus, the monosialylated, neuraminidase-sensitive oligosaccharides in BCE tubes represented 12.6% of total counts/min recovered but only 1.4% of the counts/min in BCE monolayers. The neuraminidase-sensitive oligosaccharides recovered in period of capillary tube induction, we began the labeling the neutral fractions upon rechromatography on QAE-Sephperiod 20 h after plating. The oligosaccharides were analyzed by ConA-Sepharose, QAE-Sephadex, and HPLC coupled with adex were then separated according to size by HPLC (Fig. 5, Panels E and F). Individual HPLC peaks were subjected to exoglycosid.ase digestions as outlined in Fig. 4. Total counts/ serial exoglycosidase digestions in the following order: a- min incorporated into N-linked oligosaccharides were 4.8 x fucosidase, P-galactosidase, and P-N-acetylhexosaminidase. lo6 cpm for BCE tubes and 7.5 x lo6 cpm for BCE monolayers. These results are summarized in Table 111. The major peak In Fig. 5, the QAE-Sephadex fractionations of ConA-Sepharose purified oligosaccharides from tube (Panel A) and monolayer (Panel B ) BCE cells are shown. In both cases, the bulk from BCE tubes (Panel E), which migrated slightly larger than the Man6GlcNAc standard, was resistant to a-fucosidase digestion but completely sensitive and then

6 Deoxymannojirimycin Inhibits Capillary Formation in Vitro P-N-acetylhexosaminidase indicating that this hybrid-type oligosaccharide contained sialic acid, galactose, and GlcNAc on the Manal,3Man branch. This oligosaccharide species represented 10.4% of the counts/min incorporated into cellular N-linked oligosaccharides in swainsonine-treated BCE tubes. Twenty-five percent of a minor oligosaccharide, which co-migrated with Man7GlcNAc standard (Fig. 5, Panel E, first open arrow), was sensitive to a-fucosidase. The a-fucosidasesensitive material was pooled and digested with P-galactosidase and then P-hexosaminidase. This species, which represented 0.2% of counts/min recovered, contained sialic acid, fucose, galactose, and GlcNAc on the Manal,3Man branch. The third peak (Fig. 5, Panel E, second open arrow) contained A NaCl(mM/: 20 7Q 140 2W 250 1oW Fraction No. I I Monolayers FIG. 5. Analysis of N-linked oligosaccharides from swainsonine-treated BCE cells induced to form tubes or grown in monolayers. BCE cells induced to form tubes were labeled with 2- [3H]mannose from h after plating. BCE cell monolayers were also labeled for 28 h under normal growth conditions. [3H]Mannoselabeled N-linked oligosaccharides from BCE tubes (Panel A ) and BCE monolayers (Panel B) were subjected to QAE-Sephadex chromatography. The bars in Panels A and B indicate the fractions pooled for incubation with neuraminidase. Panels C (tubes) and D (monolayers) show the QAE-Sephadex fractionation of 20 mm NaCl eluates from Panels A and B after neuraminidase digestion. The bars in Panels C and D indicate the fractions pooled for HPLC. The HPLC fractionation of the neutral N-linked oligosaccharides is shown in Panels E and F. The elution position of Man5-9GlcNAc oligosaccharide standards is indicated. Open arrows in Panels E and F identify minor peaks (see text). little a-fucosidase-sensitive material but was sensitive to P- galactosidase. The HPLC-separated oligosaccharides from monolayer BCE cells (Fig. 5, Panel F) were also analyzed by exoglycosidase digestion. The major peak (0.9% of the counts/min), which co-migrated with the Man7GlcNAc standard (Fig. 5, Panel F), was resistant to a-fucosidase digestion but was then digested completely (Table 111). The next detectable peak (Panel F, first open arrow) was partially sensitive to a-fucosidase and P-galactosidase (Table 111). This a-fucosidaselp-galactosidase-sensitive oligosaccharide represented less than 0.02% of the counts/min recovered from cellassociated glycopeptides. Thus, monolayer BCE cells contained about 10-fold less monosialylated, fucosylated hybridtype oligosaccharides compared to BCE cells induced to form tubes. Furthermore, hybrid oligosaccharides from BCE monolayers contained 11-fold less monosialylated oligosaccharides than BCE tubes (Fig. 5, Panels C and D). The neutral oligosaccharides recovered in the unbound fractions from QAE-Sephadex (Fig. 5, Panels A and B) were also analyzed. Jack bean a-mannosidase digestion was used to degrade terminal a-linked mannose residues so that the hybrid antenna linked to the al,3-linked mannose residue could be examined in isolation. In BCE monolayers, 65% of the counts/min were removed by the a-mannosidase while 74% were digested in BCE tubes (data not shown) consistent with the abundance of high mannose structures in tubes. Individual peaks separated by HPLC were subjected to a- fucosidase and P-galactosidase. Of the capillary tube oligosaccharides that co-fractionated with the Man5.7GlcNAc oligosaccharide standards, 0.5% of the counts/min recovered from total cellular glycopeptides were sensitive to a-fucosidase. In contrast, only 0.03% of the counts/min recovered from BCE monolayers were similar a-fucosidase-sensitive structures. Thus, nonsialylated hybrid-type oligosaccharides from BCE tubes were enriched in fucose-containing structures compared to BCE monolayers (data not shown). Binding of Anti-sialyl Lewis-X Antibody to Capillary Endo- thelial Cells-Since the monosialylated, fucosylated N-linked glycans synthesized in BCE cells induced to form tubes resembled sialyl Lewis-X (NeuAca2,3Gal/31, 4(Fucal,3)GlcNAc) and/or sialyl Lewis-A (NeuAca2,3Gal~l,3(Fucal,4)GlcNAc) determinants, we used anti-sialyl Lewis-X and sialyl Lewis- A antibodies to measure these carbohydrate structures in BCE capillary tubes and monolayers by an enzyme-linked immunoassay (28, 29). At saturating antibody concentrations, both epitopes were detected, but sialyl Lewis-X epitopes were 2-3- TABLE 111 Exoglycosidase digestions of HPLC-purified oligosaccharides Individual peaks from HPLC analysis were incubated with a-fucosidase, refractionated by HPLC, and prepared for subsequent exoglycosidase digestions. Oligosaccharides sensitive to the enzymes were detected as peaks with shorter retention times compared to mock-digested controls. The results are expressed as the percent of counts/min that shifted in size by 1 hexose residue. The counts/min from each HPLC peak used for this analysis are shown at the right. a-fucosidase &Galactosidase &Hexosaminidase % total cpm Tubes Peak 1 (Man6GlcNAc)" Peak 2 (Man7GlcNAc) Peak 3 (ManV4GlcNAc) Control Peak 1 (MamGlcNAc) Peak 2 (Man8GlcNAc) Peak 3 (MangGlcNAc) ND Peak co-migrated most closely with the oligosaccharide standard indicated in parentheses. ND = not determined , ,760 ND * 6,350 ND ND ND 31,000 3,760 5,700

7 1 -Deoxymannojirimycin Inhibits Capillary Formation in Vitro fold higher than sialyl Lewis-A epitopes (data not shown). The results obtained with anti-sialyl Lewis-X (CSLEX1) are presented in Fig. 6: sialyl Lewis-X epitopes were 1.6-fold higher in BCE cells induced to form tubes (C, open bar) compared to BCE cell monolayers (C, solid bar). In parallel, BCE cells induced to form tubes or grown in monolayers in the presence of dmm or swainsonine (Sw) for 48 h were analyzed to determine if any of the antibody binding activity could be attributed to sialyl Lewis-X structures present on N- linked oligosaccharides. dmm would inhibit synthesis of these epitopes on newly synthesized glycoproteins bearing N-linked oligosaccharides but not on 0-linked oligosaccharides or glycolipids. Swainsonine would inhibit synthesis if the epitopes are present on branches of the a1,g-linked mannose of the trimannosyl core but not if the epitopes are present on branches of the al,3-linked mannose residue. As seen in Fig. 6, dmm reduced the anti-sialyl Lewis-X binding in BCE cells induced to form tubes but had no effect on these epitopes in BCE monolayers. The level of antibody binding was restored to near the control level in swainsonine-treated BCE tubes. Swainsonine had no effect, however, on the level of sialyl Lewis-X epitopes in BCE cell monolayers. These results indicate that a statistically significant fraction of the sialyl Lewis-X binding sites on BCE tubes are present on N-linked oligosaccharides. The decrease in binding due to dmm may not reflect the true fraction of sialyl Lewis-X epitopes on N- linked oligosaccharides since some long-lived glycoproteins may still contain their native oligosaccharide structures and be detected in this assay. That is, it is unlikely that during the 48-h incubation 100% of the BCE cell glycoproteins bearing N-linked oligosaccharides would have been replaced by new proteins synthesized in the presence of dmm. These findings are consistent with the structural analysis of N- linked oligosaccharides from swainsonine-treated BCE tubes and monolayers presented above and suggest that some of the monosialylated, fucosylated N-linked glycans present in swainsonine-treated tubes may be sialyl Lewis-X structures. The sialyl Lewis-X epitopes expressed by BCE monolayers are not sensitive to dmm or swainsonine suggesting that the 0.25 g 0.20 v L a ; 0.15 a h dmm Sw C dmm Sw FIG. 6. Binding of anti-sialyl Lewis-X to capillary endothelial cells. BCE cells induced to form tubes (open bars) and grown in monolayers (solid bars) were incubated with no inhibitor (C), 1 mm dmm (dmm), or 0.1 mm swainsonine (Sw) for 48 h. The bargraphs represent the averages (n = 5) with error bars indicating the standard error of the mean. The absorbance obtained in the absence of primary antibody has been subtracted. The Student's t test statistical analysis was performed between each of the conditions. For capillary tubes (open bars), the p values are as follows: C versus dmm, p = 0.01; dmm versus Sw, p = 0.03; C versus Sw, p = For cells grown in monolayers (solid bars), the p values are as follows: C uersus dmm, p = 0.24; dmm versus Sw, p = 0.86; C versus Sw, p = A p value of less than 0.05 indicates that the differences in experimental values are statistically significant. bulk of these epitopes are attached to glycolipids or 0-linked oligosaccharides. DISCUSSION In these studies we have used three different inhibitors of N-linked oligosaccharide processing to examine the role of glycoproteins in capillary tube formation. 1-Deoxymannojirimycin and kifunensine, two inhibitors of Golgi a-mannosidase I which cause an accumulation of MangGlcNAcz and MansGlcNAcz oligosaccharides and prevent synthesis of hybrid and complex-type oligosaccharides, blocked capillary tube formation in uitro. The inhibitory effect of dmm was studied in more detail and found to be reversible (Fig. 1). The finding that kifunensine, a second inhibitor of Golgi a-mannosidase I, also blocked tube formation (Fig. 2) supports the notion that dmm exerts its inhibitory effect on tube formation by affecting N-linked oligosaccharide processing. Swainsonine, an inhibitor which disrupts oligosaccharide processing at a later step in the biosynthetic pathway, had no observable effect on capillary tube formation (Fig. 2). Swainsonine differs from dmm and kifunensine in that it permits synthesis of Man5-7GlcNAc2 high mannose- and hybrid-type oligosaccharides. Thus, synthesis of glycoproteins bearing small high mannose- or hybrid-type oligosaccharides must be sufficient for tube formation. Since dmm affects the biosynthesis of all newly synthesized glycoproteins bearing N-linked oligosaccharides, which may include receptors, secreted factors, extracellular matrix molecules, cell adhesion molecules, and intracellular glycoproteins, one might propose several mechanisms by which dmm disrupts capillary formation. It is possible that one or more of the glycoproteins required for tube formation are inactive with Man9-8GlcNAc2 oligosaccharide chains, but that the partial processing that can occur in the presence of swainsonine is sufficient to produce functional glycoproteins. A second explanation for our observations would be that BCE cell N- linked oligosaccharides play a direct role in tube formation; that is, certain oligosaccharide structures may mediate cellcell or cell-matrix interactions required for this process. For this hypothesis to be correct, the partially processed high mannose or hybrid-type oligosaccharides that accumulate in cells incubated with swainsonine would have to contain the requisite structural information to mediate these cellular adhesion events. Because of these results which suggest an important role for glycoproteins in tube formation and the fact that very little is known about the N-linked oligosaccharide structures present in capillary endothelial cells, we analyzed the N- linked oligosaccharides synthesized by BCE cells both in control monolayers and induced to form tubes. 2-[3H]Mannose-labeled glycopeptides were fractionated by SLAC and HPLC to identify the major classes of oligosaccharide struc- tures, assign partial structures, and to determine relative abundance. SLAC analysis provides partial structural information in that the overall branching patterns can be determined, but less information is gained about terminal sugar residues. The general pattern of N-linked glycosylation in BCE cells did not change upon tube induction, given the limits of detection inherent in this analysis (Table 11). Two interesting findings, however, were obtained. First, 35-40% of N-linked oligosaccharides synthesized by BCE monolayers and tubes appear to contain a bisecting GlcNAc based upon the E-PHA-agarose chromatography (Table 11). The bisecting GlcNAc is a unique structure (Fig. 3, C1 ) found only in certain cell types. It is thought to play an important role in the control of oligosaccharide processing since addition of a bi-

8 Deoxymannojirimycin Inhibits Capillary Formation in Vitro secting GlcNAc prevents modification of the oligosaccharide by many other glycosyltransferases (32-34). The second observation is that BCE tubes are enriched in high mannosetype oligosaccharides, especially Man5GlcNAc2, compared to BCE monolayers (Table 11). The significance of this observation is not clear at this time. TO gain more structural information about the terminal sugar residues present on N-linked oligosaccharides in BCE cells, we analyzed the structures present in swainsoninetreated BCE cell monolayers and tubes. The rationale for this was 2-fold. First, since swainsonine had no effect on tube formation, the oligosaccharides synthesized in the presence of this inhibitor represent the simplest structures that can sustain capillary tube formation. Second, the structural analysis was simplified greatly since only high mannose- and hybrid-type oligosaccharides can be synthesized in the presence of swainsonine % of the counts/min recovered as [3H] mannose-labeled glycopeptides from swainsonine-treated BCE tubes or monolayers bound tightly to ConA-Sepharose indicating they consisted of high mannose- and hybrid-type structures (Table I). Approximately 80% of these counts/min were recovered in the neutral and 1- charge eluates by QAE- Sephadex chromatography (Fig. 5, Panels A and B). These two fractions from BCE tubes and monolayers were subjected to further analysis (Fig. 5). The highly anionic species most likely consist of sialylated, phosphorylated, and/or sulfated oligosaccharides (27, 35). Analysis of the oligosaccharides in the 20 mm NaCl eluate revealed a striking difference between the structures synthesized by BCE tubes and monolayers. Fifty-five percent of the 20 mm NaCl eluate (1- charged species) in BCE tubes was sensitive to neuraminidase while only 6% of the equivalent sample from BCE monolayers was sensitive (Fig. 5, Panels C and D). Thus, BCE cells induced to form tubes in the presence of swainsonine contain a significant fraction of monosialylated hybrid-type structures while BCE monolayers contain much less. These neuraminidaseresistant oligosaccharides could contain sulfated residues, phosphodiesters, or sialic acid residues resistant to hydrolysis by V. cholerae neuraminidase. Sulfated N-linked oligosaccharides have been identified in two types of endothelial cells, human umbilical vein endothelial cells (36) and a bovine pulmonary artery endothelial cell line (35). The neuraminidase-sensitive oligosaccharides from BCE cell tubes were further analyzed by HPLC and serial exoglycosidase digestions. In Fig. 5, Panel E, the major peak purified by HPLC eluted at a position slightly larger than the MansGlcNAc standard. This species was found to contain sialic acid, galactose, and GlcNAc residues (Table 111), three common monosaccharides found on hybrid-type oligosaccharides. In contrast, 25% of the minor HPLC peak that eluted at a position slightly larger than the ManTGlcNAc standard, was found to be sensitive to a-fucosidase, then completely sensitive to P-galactosidase and p-n-acetylhexosaminidase (Table 111). This minor species represented 0.2% of the counts/min recovered from BCE cells induced to form tubes. In BCE cell monolayers, the HPLC fractionation of neuraminidase-sensitive structures (Fig. 4, Panel F) and subsequent analysis by exoglycosidase digestions revealed that a much smaller percent (0.02%) of the counts/min represented similar sialylated, fucosylated oligosaccharides. Thus, we have identified a significant difference between the N-linked oligosaccharides synthesized in BCE cell monolayers and BCE cells induced to form tubes. One might speculate that an increase in sialyltransferase expression might be part of the differentiation program that BCE cells undergo during capillary formation. In addition, an increase in specific a-fucosyltransfer- ase activities, which modify sialylated and nonsialylated N- linked oligosaccharide precursors, may also occur (37-39). The finding that swainsonine-treated BCE tubes are enriched in hybrid N-linked oligosaccharides containing sialic acid, fucose, galactose, and GlcNAc suggested that these structures may represent sialyl Lewis-X and/or sialyl Lewis-A determinants. To examine this further, we measured sialyl Lewis-X (Fig. 6) and sialyl Lewis-A (data not shown) epitopes on BCE cells induced to form tubes or grown in monolayer. Since these epitopes are found on glycolipids as well as glycoproteins, we used the inhibitors dmm and swainsonine to determine the fraction of total epitopes that could be present on glycoproteins bearing N-linked oligosaccharides. Consistent with our oligosaccharide structural analysis (Fig. 5 and Table 111), we found that BCE tubes were enriched in sialyl Lewis-X epitopes whose synthesis was reduced in the presence of dmm but not swainsonine (Fig. 6). These findings may be relevant to our concomitant studies' in which we have found that antibodies directed against sialyl Lewis-X and sialyl Lewis-A inhibit capillary tube formation and that BCE cells express a novel form of E-selectin, a lectin known to bind sialyl Lewis-X (17-19) and sialyl Lewis-A glycans (20-22). It is intriguing to speculate that the mechanism by which anti-sialyl Lewis-X and anti-sialyl Lewis-A antibodies inhibit tube formation is by binding to these epitopes on BCE cell surface N-linked oligosaccharides. If this were the case, then one would predict our finding that dmm and kifunensine are effective inhibitors of capillary tube formation since each would block synthesis of sialyl Lewis-X/Acontaining N-linked oligosaccharides. Since swainsonine does not completely block synthesis of these epitopes, one would predict that swainsonine might not inhibit capillary formation. Studies to elucidate the role of these monosialylated, fucosylated N-linked glycans in capillary morphogenesis are in progress. REFERENCES 1. Klagsbrun, M., and Folkman, J. (1990) in Handbook of Experimental Pharmacology (Sporn, M. B., and Roberts, A. B., eds) Vol. 95/11, Springer-Verlag, New York Inc., New York 2. Folkman, J., and Haudenschild, C. (1980) Nature 288, Maciag, T., Kadish, J., Wilkins, L., Stemerman, M. B., and Weinstein, R. (1982) J. Cell Biol. 94, Montesano, R., and Orci, L. (1985) Cell 42, Kubota, Y., Kleinman, H. K., Martin, G. R., and Lawley, T. J. (1988) J. Cell Biol. 107, Ingber, D. E., and Folkman, J. (1989) J. Cell Biol. 109, Folkman, J., and Klagsbrun, M. (1987) Science 236, Paulson, J. C. (1989) Trends Biochem. Sci. 14, Kornfeld, K., and Kornfeld, S. (1985) Annu. Reu. Biochem. 64, Fuhrmann, U., Bause, E., Legler, G., and Ploegh, H. (1984) Nature 307, Bischoff, J., and Kornfeld, R. (1984) Biochem. Biophys. Res. Commun. 126, R9A-RRl ". "_ 12. Elbein, A. D., Tropea, J. E., Mitchell, M., and Kaushal, G. P. (1990) J. Biol. Chem. 266, Tulsiani, D. R. P., Harris, T. M., and Touster, 0. (1982) J. Biol. Chem. 267, Cummings, R. C., and Kornfeld, S. (1982) J. Biol. Chem. 267, Fukuda, M., Spooner, E., Oaks, J. E., Dell, A., and Klock, J. C. (1984) J. Biol. Chem. 269, Saitoh, O., Wang, W.-C., Lotan, R., and Fukuda, M. (1992) J. Biol. Chem. 267, Lowe, J. B., Stoolman, L. M., Nair, R. P., Larsen, R. D., Berherd, T. L., and Marks, R. M. (1990) Cell 63, Phillips, M. L., Nudelman, E., Gaeta, F. C. A., Percez, M., Singhal, A. K., Hakamori, S. I., and Paulson, J. C. (1990) Science 260, Walz, G., Aruffo, A., Kolanus, W., Bevilacqua, M., and Seed, B. (1990) Science 260, Berg, E. L., Robinson, M. K., Mansson, O., Butcher, E. C., and Magnani, J. L. (1991) J. Biol. Chem. 266, Takada, A,, Ohmori, K., Tadahashi, N., Tsuyuoka, K., Yago, A., Zentia, K., Hasegawa, A., and Kannagi, R. (1991) Biochem. Biophys. Res. Commun. 179, Tyrrell, D., James, P., Rao, N., Foxall, C., Abbas, L., Dasgupta, F., Nashed, M., Hasegawa, A., Kiso, M., Asa, D., Kidd, J., andbrandley, B. K. (1991) Prac. Natl. Acad. Sci. U. S. A. 88, McEver, R. P. (1991) Thromb. Haemostasls 66,8& Bischoff, J., Liscum, L., and Kornfeld, R. (1986) J. Biol. Chem. 261, Folkman, J., Haudenschild, C. C., and Zetter, B. R. (1979) Proc. Natl. Acad Sci. U. S. A. 76,

9 1 -Deoxymannojirimycin Inhibits Capillary Formation in Vitro Mellis, S. J., and Baenzuger, J. U. (1981) Ad. Biochem. 114, Longmore, G. D., and Schachter, H. (1982) Carbohydr. Res. 100, Varki, A,, and Kornfeld,.(1983) J. Bml. Chem. 268,280% Campbell, C., and Stanley, P. (1984) J. Biol. Chern. 261, Heussen, C. H., Stocker, J. W., and Gisler, R. H. (1981) Methods Enzymol. 35. Roux, L., Holojda, S., Sundblad, G., Freeze, H. H., and Varki, A. (1988) J. 73, Biol. Chem. 263, Zhou, Q., Moore, K. L., Smith, D. F., Varki, A., McEver, R. P., and 36. Heifetz, A., Watson, C., Johnson, R.R., and Roberts, M.K. (1982) J. Biol. Cummings, R. D. (1991) J. Cell Biol. 116, Chem. 267, Akiyama, S. K., Yamada, S. S., and Yamada, K. M. (1989) J. Biol. Chem. 37. Stanley, P., and Atkinson, P. H. (1988) J. Biol. Chem. 263, , Mollicone, R., Gibaud, A., Francois, A,, Ratcliffe, M., and Oriol, R. (1990) 31. Fuhrmann, U., Bause, E., and Ploegh, H. (1985) Biochim. Biophys. Acta Eur. J. Biochem. 191, , Macher, B. A., Holmes, E. H., Swiedler, Stults, S. J., C. L. M., Srnka, and 32. Narasimhan, S. (1982) J. Biol. Chem. 267, A. C. (1991) Glycobiology 1,