Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Am J Physiol Gastrointest Liver Physiol 278: G915–G923, 2000. Accessibility of glycolipid and oligosaccharide epitopes on rabbit villus and follicle-associated epithelium NICHOLAS J. MANTIS, ANDREAS FREY, AND MARIAN R. NEUTRA Department of Pediatrics, Harvard Medical School, and Gastrointestinal Cell Biology Laboratory, Children’s Hospital, Boston, Massachusetts 02115 Peyer’s patch; glycocalyx; adhesion; pathogen; M cells ENTERIC PATHOGENS SELECTIVELY infect specific epithelial cell types in the gastrointestinal tract (1, 6, 25). For example, rotavirus replicates within villus enterocytes (39), whereas reovirus selectively binds to, and is endocytosed by, M cells (51). Adherence, the initial step in the infection process, is postulated to be mediated by lectinlike microbial adhesins that recognize defined glycoproteins and/or glycolipid epitopes on the apical surfaces of intestinal epithelial cells (6, 22, 25, 27). Lectinlike adhesins have been identified on a wide array of enteroviruses and pathogenic bacteria (1, 6, 19, 27). For example, rotavirus binds the glycosphingolipid asialo-GM1 and O-linked sialylglycoproteins in overlay assays (49, 50), and reovirus attachment to mouse fibroblasts in vitro depends on the presence of host cell The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked ‘‘advertisement’’ in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. http://www.ajpgi.org sialoglycoconjugates (14). However, because most of these glycolipids and galactosyl- or sialic acid-containing oligosaccharides are widely distributed on intestinal epithelial cells, receptor distribution alone cannot explain the cell type-specific tropisms of viral and bacterial pathogens that are observed in vivo. Thus additional factors must exist that influence pathogen attachment to specific cells or regions of the gastrointestinal mucosa (26, 27). The intestinal epithelium is comprised primarily of absorptive villus enterocytes (28). Enterocyte apical surfaces in vivo are highly differentiated structures consisting of rigid, closely packed microvilli whose membranes contain stalked glycoprotein enzymes (33, 45). In addition, the tips of enterocyte microvilli are coated with a 400- to 500-nm-thick meshwork called the filamentous brush border glycocalyx (FBBG) (23), which is composed of highly glycosylated transmembrane mucins (29, 30). The epithelium overlying organized mucosal lymphoid nodules, the so-called follicle-associated epithelium (FAE), is comprised of enterocyte-like cells interspersed with M cells. The M cell is a morphologically distinct epithelial cell type whose primary function is the transport of macromolecules, particles, and microorganisms from the lumen to underlying lymphoid tissue (11, 35). M cells generally lack an organized brush border and a well-defined FBBG, although their apical membranes have a thin (20–30 nm) glycoprotein coat (10). We have recently shown that the FBBG on the apical surfaces of rabbit villus and FAE enterocytes is a size-selective barrier that can prevent particles from gaining access to membrane glycolipids (10). This was demonstrated using the B subunit of cholera toxin (CTB), which specifically binds to ganglioside GM1, a glycolipid whose carbohydrate head extends just 2.5 nm from the plasma membrane (47). When applied in soluble form to rabbit mucosa, CTB (diameter 6.8 nm) had free access to GM1 on all epithelial cell types and bound to villus enterocytes, FAE, and M cells (10). When CTB was coupled to colloidal gold to form a particulate probe 28 nm in diameter, it was no longer able to penetrate the FBBG and was therefore unable to bind to enterocytes, but the particles did adhere to the apical surfaces of M cells. The thin glycoprotein coat on M cells can prevent larger particles from gaining access to ganglioside GM1, however, because CTB-coated, 1-µm particles were unable to bind. Our previous studies did not test whether the M cell glyco- 0193-1857/00 $5.00 Copyright r 2000 the American Physiological Society G915 Downloaded from http://ajpgi.physiology.org/ by 10.220.32.246 on June 18, 2017 Mantis, Nicholas J., Andreas Frey, and Marian R. Neutra. Accessibility of glycolipid and oligosaccharide epitopes on rabbit villus and follicle-associated epithelium. Am J Physiol Gastrointest Liver Physiol 278: G915–G923, 2000.— The initial step in many mucosal infections is pathogen attachment to glycoconjugates on the apical surfaces of intestinal epithelial cells. We examined the ability of virussized (120-nm) and bacterium-sized (1-µm) particles to adhere to specific glycolipids and protein-linked oligosaccharides on the apical surfaces of rabbit Peyer’s patch villus enterocytes, follicle-associated enterocytes, and M cells. Particles coated with the B subunit of cholera toxin, which binds the ubiquitous glycolipid GM1, were unable to adhere to enterocytes or M cells. This confirms that both the filamentous brush border glycocalyx on enterocytes and the thin glycoprotein coat on M cells can function as size-selective barriers. Oligosaccharides containing terminal b(1,4)-linked galactose were accessible to soluble lectin Ricinus communis type I on all epithelial cells but were not accessible to lectin immobilized on beads. Oligosaccharides containing a(2,3)linked sialic acid were recognized on all epithelial cells by soluble Maackia amurensis lectin II (Mal II). Mal II coated 120-nm (but not 1-µm) particles adhered to follicle-associated enterocytes and M cells but not to villus enterocytes. The differences in receptor availability observed may explain in part the selective attachment of viruses and bacteria to specific cell types in the intestinal mucosa. G916 GLYCOCONJUGATE ACCESSIBILITY ON INTESTINAL EPITHELIUM MATERIALS AND METHODS Chemicals and reagents. All lectins, lectin derivatives, and fluorophore-conjugated streptavidin reagents were purchased from Vector Laboratories (Burlingame, CA). Biotinylated CTB was purchased from List Laboratories (Campbell, CA). BSA, biotinylated BSA, and biocytin were from Sigma (St. Louis, MO). Neutravidin-coated, fluorescent, carboxylatemodified polystyrene particles (FluoSpheres) were purchased from Molecular Probes (Eugene, OR). Liquid paraformaldehyde (16% solution) was obtained from Electron Microscopy Sciences (Fort Washington, PA). Preparation of ligand-coated particles. The fluorescent red or green virus-sized particles used in this study had an estimated diameter of 0.093 µm 6 7.5%. The manufacturer coated them with Neutravidin, resulting in ,10 nmol biotin binding sites per milligram of polystyrene. To derivitize particles with ligands, 1 3 1012 particles were suspended in 500 µl of PBS (50 mM sodium phosphate, pH 7.4, 50 mM NaCl) containing 0.02% Tween-20, 0.5% BSA, 2 mM NaN3, and 50 µg/ml of biotinylated ligand or biocytin. The particle mixtures were gently rocked overnight at 4°C and then transferred to a 0.5-ml Dispodialyzer dialysis tube (Spectrum, Houston, TX) with a molecular weight cut-off of 300,000. Excess unbound ligand was removed by dialysis against 4 l of PBS containing 2 mM NaN3 changed daily for 4 days. In control experiments, this dialysis protocol was sufficient to remove .95% of soluble ligand. After dialysis, particles were transferred to siliconized microcentrifuge tubes and stored at 4°C in the dark. The bacterium-sized (1.0 µm 6 2.5%) particles were coated with biotinylated ligand as described previously (10). Briefly, particles and biotinylated ligand (400 µg/ml) were combined in 1 ml of PBS and incubated with gentle rocking overnight at 4°C. The particles were then collected by centrifugation (3,500 g) and washed with PBS several times to remove excess, unbound ligand. The particles were resuspended in 400 µl PBS containing 2 mM NaN3 and then stored in the dark at 4°C. Before use in adhesion assays, particles were vortexed for 2 min at room temperature and then bath sonicated (Laboratory Supply, Hicksville, NY) in an ice slurry for 10 min. Ligand-coated, fluorescent green particles and biocytin-quenched (or BSA-coated), fluorescent red particles were diluted at a 1:1 ratio as determined by fluorescence microscopy into sterile PBS containing BSA (0.1% wt/vol) and stored on ice until ready to use. BSA-coated and biocytin-coated particles were used interchangeably as controls. The final diameter of the virus-sized, CTB-coated particles was estimated to be 120 nm. This value was calculated knowing that the diameter of the carboxylatemodified latex beads was 93 nm and assuming a uniform coat of Neutravidin (7 nm) and CTB-biotin (6.5 nm), as described previously (10). Although the hydrodynamic diameters of RCA-I [relative molecular weight (Mr) 120,000] and Mal II (Mr 140,000) are undetermined, we expect that they are slightly greater than that of avidin (Mr 60,000). Cell culture. The Caco-2BBe cell line was a generous gift from Dr. Mark Mooseker (Yale University). Caco-2 cells were grown in high-glucose DMEM (GIBCO BRL, Rockville, MD) supplemented with 10% FCS (HyClone Labs, Logan, UT), penicillin (100 IU/ml), streptomycin (100 µg/ml), and glutamine (2 µg/ml) at 37°C with an atmosphere containing 5% CO2, as described previously (41). For particle adhesion assays, cells were seeded at 2 3 105 cells/ml onto acetonewashed, 12-mm-diameter glass coverslips (Bellco Glass, Vineland, NJ) in 12-well plates (Costar, Cambridge, MA), as done previously (10). Seven days after confluence, cells were washed three times with PBS and then overlaid with soluble, biotinylated lectins (100 µg/ml) or particle mixtures and incubated with gentle rocking for 1 h at room temperature. Cells on coverslips were washed three times with PBS, fixed by immersion in formaldehyde (4% vol/vol in PBS), then inverted and mounted using Mowiol 50.5 g/ml glycerol, 0.1 g/ml Mowiol 4–88 (Calbiochem, San Diego, CA), 10 mg/ml diazabicylo[2.2.2]octane (Sigma) in 0.1 M Tris · HCl (pH 8.5)6 Downloaded from http://ajpgi.physiology.org/ by 10.220.32.246 on June 18, 2017 protein coat can exclude particles in the size range of most viruses from contact with membrane glycolipids in vivo. This is important because it has been shown that HIV-1 (120 nm in diameter) attaches to intestinal cells in culture via galactosylceramide (9, 12). HIV can adhere to M cells in rabbit intestine (2) and may possibly exploit M cell transport activity to enter the rectal mucosa in humans (34). Protein-linked oligosaccharides would be expected to provide more accessible binding sites for microorganisms because they extend up to 500 nm from enterocyte plasma membranes and 20–30 nm from M cell membranes (10). They would thus be the first structures encountered by intestinal pathogens (25). Integral membrane mucins such as those of the enterocyte FBBG consist of up to 80% heterogeneous O-linked carbohydrates that form dense, negatively charged gels thought to function in cytoprotection (48). Oligosaccharide side chains on both mucin and nonmucin membrane glycoproteins are notoriously heterogeneous, and the exact composition, structure, and topology of only a small minority have been defined (25). There is evidence that glycosylation patterns not only of M cells but of the entire FAE differ from those of villus epithelium (5, 13, 17, 18, 46), but the exact topology or accessibility of specific oligosaccharide structures on epithelial cells of small intestinal villi and FAE has not been explored. Such information could be important for understanding microbial tropism in the intestine and for designing strategies for targeting of particulate vaccines and vaccine vectors to the FAE and M cells. For this study we selected three ligands that recognize glycolipid or protein-linked oligosaccharide epitopes that are known to be exploited by enteric pathogens. We examined the ability of the ligands immobilized on virus-sized (120-nm) and bacterium-sized (1-µm) particles to adhere to apical surfaces of rabbit Peyer’s patch villus enterocytes, FAE, and M cells. Particles were coated with CTB to probe accessibility of the glycolipid GM1, the lectin Ricinus communis agglutinin type I (RCA-I) specific for terminal b(1, 4)-linked galactose, and Maackia amurensis lectin II (Mal II), which recognizes a hierarchy of oligosaccharides containing a(2,3)-linked sialic acid. Data presented indicate that the relatively thin glycoprotein coat on the apical surfaces of M cells is sufficient to prevent virussized particles from gaining access to glycolipids like GM1. In addition, we found that protein-linked oligosaccharides vary in their accessibility to particulate ligands and that differences in accessibility exist between villus and FAE. We conclude that successful attachment of pathogens to glycolipids and glycoproteins on apical surfaces of intestinal epithelial cells is dependent on both receptor availability and receptor accessibility. GLYCOCONJUGATE ACCESSIBILITY ON INTESTINAL EPITHELIUM have a variable degree of nonspecific binding to intestinal mucosa, 1:1 mixtures of ligand-coated and control particles were applied in all experiments. For each experimental condition we analyzed 40 tissue sections, determining the numbers of ligand-coated particles (fluorescent green) and control particles (fluorescent red) bound to villus-associated epithelium and FAE. In specimens in which specific binding occurred, there were on average .5,000 total particles bound to 40 sections. For each tissue section we divided the number of ligand-coated particles by the number of control particles to obtain a relative index of binding. Binding indices were then averaged. All experiments were done at least twice using two independently prepared batches of ligand and ligand-coated particles and two different rabbits. RESULTS CTB-coated, 120-nm particles did not adhere to live Peyer’s patch mucosa but adhered selectively to M cells on aldehyde-fixed tissues. To probe glycolipids on the apical surfaces of intestinal epithelial cells, we applied soluble, biotinylated CTB to rabbit Peyer’s patch mucosa. As observed previously (10), we found that CTB bound to GM1 in apical plasma membranes of all intestinal epithelial cells on live and formaldehydefixed Peyer’s patch mucosa (Fig. 1A). Biotinylated BSA applied at identical concentrations to Peyer’s patch mucosa did not label epithelial cells (Fig. 1B). To test whether virus-sized particles have access to GM1, we applied CTB-coated 120-nm particles (fluorescent green) mixed 1:1 with biocytin-coated control particles (fluorescent red) to live Peyer’s patch explants. We observed that very few particles bound to either villus or FAE and that equal numbers of CTB-coated and control particles were associated with the mucosal surface (Table 1). Thus the FBBG on villus and FAE enterocytes as well as the thin glycoprotein coat on M cells were sufficient to prevent these virus-sized particles from gaining access to glycolipid receptors. Tissue fixation with formaldehyde cross-links glycoproteins and may unmask certain glycoconjugate epitopes not normally available in vivo (20). We therefore predicted that aldehyde fixation might sufficiently alter the thin glycoprotein coat on M cells to render membrane glycolipids accessible to particles but that this might not be the case for the thick FBBG on enterocytes. To test this hypothesis, CTB-coated 120-nm particles and control particles were applied to aldehydefixed Peyer’s patches, and binding was visualized by fluorescence microscopy. On fixed tissue we observed that CTB-coated particles adhered in greater numbers than control particles and that CTB-coated particles adhered preferentially to M cells and not to enterocytes on FAE or adjacent villi (Fig. 1C; Table 1). RCA-I binding sites on intestinal epithelial cells were inaccessible to 120-nm particles. Galactose-containing glycoproteins are proposed to be important attachment sites for enteric bacteria. To confirm the distribution of carbohydrates containing terminal b(1,4)-linked galactose, we applied soluble, biotinylated RCA-I to live or fixed Peyer’s patch mucosa. RCA-I labeled the surfaces of all epithelial cells strongly and uniformly on live Downloaded from http://ajpgi.physiology.org/ by 10.220.32.246 on June 18, 2017 onto Fisher Superfrost microscope slides (Fisher Scientific, Pittsburgh, PA) and viewed by fluorescence or phase-contrast microscopy. Application of ligand-coated particles to live or fixed rabbit Peyer’s patch mucosa. New Zealand White female rabbits weighing ,1.5 kg each were purchased from Charles River Laboratories (Wilmington, MA) and maintained in the animal resource facility at the Children’s Hospital. All animal procedures were conducted in strict compliance with Guidelines for Animal Experimentation established by Harvard Medical School, the Children’s Hospital, and the National Institutes of Health. Animals were fasted overnight with water ad libitum before surgery. To obtain Peyer’s patches, rabbits were sedated with an intramuscular injection of 0.5 ml acetopromazine maleate (10 mg/ml) followed 45 min later by an intramuscular injection of ketamine (80 mg/kg) and xylazine (20 mg/kg). Animals were laparotomized and jejunoileal Peyer’s patches were excised and treated as described below. Rabbits were euthanized by a single intravenous injection of pentobarbital sodium (100 mg/kg) into the marginal ear vein. Freshly excised Peyer’s patches were washed in PBS (25°C), and the luminal surfaces were blotted lightly with Whatman paper (3 MM) to remove excess mucus. Peyer’s patches were then pinned to surgical cork board, and the mucosa was removed from the muscularis externa using a single-edged razor blade. For experiments using live tissue, the mucosae were cut into 2 3 2 mm pieces and each was placed in a single well of a 12-well tissue culture plate (Costar), which had been previously treated with RPMI 1640 (GIBCO BRL) and BSA (0.5% wt/vol). Taking special care not to damage the epithelium, the mucosa was washed with RPMI 1640 and overlaid with control and experimental particles (4 3 1010 total particles in 0.5 ml) or soluble, biotinylated lectins (100 µg/ml) and incubated at room temperature for 40 min. The tissue was carefully rinsed three times with PBS and either fixed for cryosectioning by immersion in formaldehyde (4% vol/vol) in PBS or for plastic embedding by immersion in formaldehyde (4% vol/vol) in 0.1 M cacodylate buffer (pH 7.2) (17). Lectin-treated tissue destined for cryosectioning was washed with PBS, incubated with glycine (0.1 M in PBS) to quench residual reactive aldehydes, blocked in PBS containing BSA (0.5% wt/vol) for 1 h, and labeled with fluorophore-conjugated streptavidin (50 µg/ml). For experiments using prefixed tissue, freshly excised mucosae were immersed in four volumes of formaldehyde (4% vol/vol) in PBS for 12 h. The tissue samples were then cut into 2 3 2 mm blocks, washed with PBS, incubated in 0.1 M glycine in PBS to quench residual reactive aldehydes, and blocked in PBS containing BSA (0.5% wt/vol) for 1 h. Blocks of tissue were then overlaid with lectins or particles for 40 min, rinsed, fixed again, and processed as described above for live tissue. Tissue sectioning, microscopy, and quantitation of particle binding. For cryosectioning, tissue samples were incubated for 1 h in sucrose (15% wt/vol in PBS), followed by 10 min in Tissue-Tek optimum cutting temperature embedding medium (Sakura Finetek, Torrance, CA), frozen in liquid nitrogen-cooled isopentane, and stored at 220°C. Frozen tissue sections were cut at 8–10 µm using a Leica cryostat model CM3050 (Nussloch, Germany). Sections were captured on Fisher Superfrost microscope slides and mounted with coverslips using Mowiol. Slides were viewed under oil immersion using a Zeiss Axiophot microscope (Carl Zeiss, Thornwood, NY) equipped for epifluorescence and photographed using a 35 mm camera. Because protein-coated polystyrene particles G917 G918 GLYCOCONJUGATE ACCESSIBILITY ON INTESTINAL EPITHELIUM Fig. 1. Adhesion of soluble B subunit of cholera toxin (CTB) and CTB-coated, 120-nm particles to Peyer’s patch epithelium. A: aldehyde-fixed rabbit Peyer’s patch explants were exposed to soluble, biotinylated CTB (100 µg/ml) for 40 min, labeled with streptavidinrhodamine, and viewed by fluorescence microscopy. Soluble CTB labeled apical surfaces of cells in follicle-associated epithelium (FAE) (bottom) and villus epithelium (top). B: on same tissue, soluble, biotinylated BSA (100 µg/ml) failed to label apical surfaces of epithelial cells. C: aldehyde-fixed Peyer’s patch explants were exposed to an equal number of CTB-coated, 120-nm particles (green) and biocytin-quenched particles (red), cryosectioned, and viewed by fluorescence microscopy. Representative micrograph shows CTBcoated particles bound in greater numbers than control particles to apical surface of M cells, which were identified by dark areas representing M cell pockets. Scale bars 5 100 µm. (Fig. 2, A and B) and fixed tissue (data not shown). To test whether the galactose epitopes recognized by RCA-I are accessible to virus-sized particles in vivo, we applied a 1:1 mixture of RCA-I-coated 120-nm particles (fluorescent green) and biocytin-quenched control par- Table 1. Binding of ligand-coated particles to rabbit Peyer’s patch villus and follicle-associated epithelium Live Tissue Fixed Tissue Particles, Diameter Villus FAE Villus FAE CTB, 120 nm RCA-1, 120 nm Mal II, 120 nm Mal II, 1 µm 1.1 (0.08) 0.9 (0.06) 1.5 (0.2) 1.1 (0.8) 1.0 (0.08) 1.0 (0.1) 4.5 (0.3) 1.5 (0.9) 1.1 (0.05) n.d. 37 (4.7) 25 (1.8) 2.9 (0.2) n.d. 54 (7.1) 28 (5.0) Live or fixed tissue explants were exposed to a one-to-one mixture of ligand-coated and control particles for 40 min, washed, cryosectioned, and then viewed by fluorescence microscopy as described in MATERIALS AND METHODS. Number of ligand-coated and control particles bound to either villus epithelium or follicle-associated epithelium (FAE) was determined for 40 sections. For each section, number of ligand-coated particles was divided by number of control particles, and these values were averaged to obtain a relative index of binding. These indices are shown with standard errors in parentheses. CTB, B subunit of cholera toxin; RCA-I, Ricinus communis type 1; Mal II, Maackia amurensis lectin II; n.d., not determined. Downloaded from http://ajpgi.physiology.org/ by 10.220.32.246 on June 18, 2017 ticles (fluorescent red) to live rabbit Peyer’s patch mucosa. The RCA-I particles did not show specific binding to any epithelial cells (Table 1). Both RCA-I and control particles adhered sparsely (Fig. 2C) and RCA-I-coated particles did not bind to the intestinal epithelium any better than control particles. On the FAE, nonspecific particle binding occurred primarily on the apical surfaces of M cells. This was not surprising since rabbit Peyer’s patch M cells have been shown previously to actively phagocytose uncoated polystyrene particles (10, 37). Aldehyde fixation did not enhance the availability of terminal b(1, 4) galactose sites to RCA-I-coated particles (data not shown). To exclude the possibility that the particle-coating procedure destroyed the receptor-binding activity of RCA-I, the same green-red particle mixture was applied to confluent monolayers of Caco-2 cells, an enterocyte-like cell line derived from a human colonic adenocarcinoma (41). We previously established that within 7 days after confluence, the cloned Caco-2 cells used in this study express RCA-I recognition epitopes on their apical surfaces (15) but lack a thick FBBG (10). We observed that RCA-I-coated, 120-nm particles adhered to live or fixed Caco-2 cells 39 times better than control particles (Fig. 2, E and F). RCA-I particle binding could be inhibited by coadministration of 250 µg/ml soluble RCA-I (data not shown). These data indicate that the particles were sufficiently coated with active lectin to impart RCA-I-dependent binding to cells. Moreover, this result underscores the fact that the accessibility of protein-linked oligosaccharide epitopes on epithelial cells grown in vitro may not be the same as on enterocytes in vivo. Mal II-coated 120-nm particles preferentially adhered to the FAE. We next examined the accessibility of carbohydrate structures containing a(2,3)-linked sialic acid to virus-sized particles. On live and fixed tissue, soluble, biotinylated Mal II adhered to the luminal surfaces of villus enterocytes, FAE, and M cells (Fig. 3). On the FAE, Mal II labeled the apical surfaces of enterocytes somewhat more intensely than the apical GLYCOCONJUGATE ACCESSIBILITY ON INTESTINAL EPITHELIUM G919 surfaces of M cells, possibly reflecting an abundance of lectin-binding sites in the thick FBBG (data not shown). To examine to what degree the Mal II epitopes are accessible to 120-nm particles, we applied a 1:1 mixture of Mal II-coated particles (fluorescent green) and biocytin-quenched control particles (fluorescent red) to live rabbit Peyer’s patch explants. On villus epithelium, Mal II-coated particles bound in very low numbers and the ratio of Mal II to control particles was ,1:1 (Table 1). However, on the FAE Mal II particles adhered 4.5 times better than control particles (Fig. 4A and Table 1). Fluorescence microscopy of plastic-embedded thin sections confirmed that Mal II particles bound to both FAE and M cells (data not shown). These data suggest that although glycoconjugates recognized by Mal II epitopes are present on the apical surfaces of all intestinal epithelial cells in vivo, they are only accessible to virus-sized particles on the FAE. To test whether cross-linking of the FBBG could expose additional a(2,3)-linked sialic acid-containing sites to virus-sized particles, Mal II-coated particles were applied to aldehyde-fixed Peyer’s patch mucosa. On fixed tissue, specific binding of Mal II particles to FAE and M cells was enhanced (Fig. 4B and Table 1). In addition, Mal II particles now adhered to villus enterocytes. From these data we conclude that the a(2,3)-linked sialic acid-containing oligosaccharides in the FBBG of villus enterocytes, although inaccessible to virus-sized particles on living tissue, are readily exposed by fixation. This is in contrast to the glycolipids that lie deep to the glycocalyx in the plasma membrane. Accessibility of oligosaccharides to bacterium-sized (1-mm) particles. In vitro binding assays have revealed bacterial adhesins that recognize a diversity of oligosaccharides and glycolipids (1, 19, 22, 43). Adhesins are often located on the tips of fimbriae and pili that extend hundreds of nanometers from the bacterial surface, suggesting that they may be able to penetrate the surface coats of host cells. This could be particularly important in the intestine, where the thick FBBG of Downloaded from http://ajpgi.physiology.org/ by 10.220.32.246 on June 18, 2017 Fig. 2. Binding of soluble Ricinus communis type I (RCA-I) and 120-nm RCA-I particles to Peyer’s patch epithelium and Caco-2 cell monolayers. Live Peyer’s patch explants were exposed to RCA-I biotin (A and B) or a 1:1 mixture of RCA-I particles (green) and biocytin particles (red) (C and D) for 40 min. A: RCA-I biotin (100 µg/ml) visualized by streptavidin-rhodamine-labeled apical surfaces of all epithelial cells present in FAE. B: RCA-I biotin (100 µg/ml) visualized by streptavidinfluorescein-labeled apical surfaces of all villus epithelial cells. C: RCA-I particles and biocytin particles (red) adhered in equal numbers to M cells present in FAE (arrow) and occasionally to goblet cells on villi (top). D: phasecontrast image of C. Arrow indicates apical surface of an M cell. E–F: same particle mixture used in experiment shown in C was applied to a monolayer of Caco-2 cells. E: monolayer viewed by confocal laser scanning fluorescence microscopy. RCA-I particles (green) bound .20 times better than biocytin particles (red). F: phase-contrast image of Caco-2 monolayer used in experiment described in E. Scale bar 5 100 µm. G920 GLYCOCONJUGATE ACCESSIBILITY ON INTESTINAL EPITHELIUM cated that the a(2,3)-linked sialic acid-containing determinants present on the FAE that were accessible to virus-sized particles were not accessible to particles as large as bacteria. When the same particle mixture was applied to fixed tissue, Mal II particles adhered to both FAE and villus epithelium .25 times better than control particles (Fig. 4D, Table 1). This confirmed that cross-linking of cell surface glycoprotein coats resulted in exposure of the normally sequestered sialic acidcontaining epitopes. DISCUSSION enterocytes faces an environment rich in microorganisms. On this basis one might predict that many of the epithelial glycolipid and oligosaccharide structures that are potential microbial binding sites might be inaccessible to lectins or toxins immobilized on the surfaces of inert bacterium-sized particles that lack appendages. Indeed, we previously showed that ganglioside GM1 on live rabbit Peyer’s patch mucosa is not accessible to CTB-coated, 1-µm beads (10). To test the accessibility of oligosaccharides containing terminal b(1,4)-linked galactose or a(2,3)-linked sialic acid, we coated 1-µm particles with RCA-I or Mal II, mixed them with BSA-coated control particles at a ratio of 1:1, and applied the mixture to live or fixed rabbit Peyer’s patch explants. We found that 1-µm RCA-I-coated particles failed to adhere to villus enterocytes, FAE, or M cells (data not shown), consistent with our results obtained using 120-nm particles (Fig. 2 and Table 1). Similarly, 1-µm Mal II-coated particles were unable to bind specifically to any cell surface, including villus epithelium or FAE (Fig. 4, Table 1). This indi- Downloaded from http://ajpgi.physiology.org/ by 10.220.32.246 on June 18, 2017 Fig. 3. Adhesion of soluble Maackia amurensis lectin II (Mal II) to both villus and FAE cells. Live rabbit Peyer’s patch explants were incubated with soluble, biotinylated Mal II (100 µg/ml) for 40 min and then labeled with streptavidin-fluorescein. Mal II labeled apical surfaces of epithelial cells in villus epithelium (A) and FAE (B). Scale bar 5 50 µm. For the majority of enteric pathogens, the primary event in mucosal infection is attachment to glycoconjugates on apical surfaces of intestinal epithelial cells. Indeed, bacteria and viruses express lectin-like adhesins on their surfaces (1, 19, 22, 27), and intestinal absorptive cells are coated with a heterogeneous array of glycolipids and protein-linked oligosaccharides (30, 45). The fact that glycoconjugates may extend from a few nanometers to hundreds of nanometers from the plasma membrane lead to the proposal that potential glycoconjugate binding sites may not be equally accessible to pathogens present in the intestinal lumen (25). In this study we have shown that certain glycolipid and protein-linked oligosaccharide epitopes that are present on the surfaces of all villus and FAE cells are not uniformly available for binding of ligands immobilized on virus- and bacterium-sized particles. The regional and cellular differences in receptor availability that we observed may explain in part the selective attachment of pathogenic viruses and bacteria to specific cell types in the intestinal mucosa. Membrane glycolipids such as GM1, whose carbohydrate head extends only 2.5 nm from the membrane surface (47), should be relatively inaccessible to particles and most readily masked by membrane-associated glycoproteins. Since the thick FBBG on the microvilli of villus and FAE enterocytes had previously been shown to exclude CTB-coated particles as small as 28 nm in diameter, it was not surprising that 120-nm particles were also prevented from contacting enterocyte GM1. In contrast, the apical glycoprotein coat on rabbit Peyer’s patch M cells is a relatively thin, irregular layer that extends only 20–30 nm from the membrane and allows binding of CTB-coated 28-nm particles to GM1 (10). This observation had raised hopes that virus-sized or even larger particles coated with CTB could be used to target vaccines to M cells and enhance the efficiency of mucosal immunization (16, 24). Unexpectedly, we found that the glycoprotein coat on live M cells was sufficient to prevent 120-nm particles from gaining access to GM1. This implies that CTB may not be useful for M cell targeting of particulate vaccine carriers such as copolymer microspheres but may be most effective as a component of macromolecular antigen complexes, as has been previously observed (31). Our results also have implications for understanding rectal transmission of HIV-1. It was observed that cell-free HIV-1 adheres to, and is endocytosed by, M GLYCOCONJUGATE ACCESSIBILITY ON INTESTINAL EPITHELIUM G921 cells in mucosal explants from experimental animals (2) and humans (P. D. Smith, personal communication). Although it has been proposed that HIV-1 initially adheres to the glycosphingolipid galactosylceramide (9, 12), our data would suggest that it is unlikely that this molecule would be accessible to virus particles. Rather, we propose that the initial attachment of HIV-1 to the apical surfaces of M cells may involve an interaction of gp120 with more accessible membrane surface components (4, 32) or the interaction of host-derived adhesion molecules in the viral envelope with relevant ligands on M cells (21). Mucinlike glycoproteins, including those that comprise the FBBG on the apical surfaces of intestinal epithelial cells, contain abundant heterogeneous oligosaccharide side chains that tend to form thick, negatively charged gels (48). We hypothesized that the accessibility of certain carbohydrate determinants in the enterocyte glycocalyx to lectins immobilized on virus- or bacterium-sized particles may vary, depending on their relative position within the FBBG. We found that soluble RCA-I, which preferentially recognizes oligosaccharides containing terminal b(1,4)linked galactose, adhered to rabbit villus epithelium and FAE, but RCA-I immobilized on 120-nm or 1-µm particles did not. The particles did bind specifically to our Caco-2 cells that have abundant terminal galactose moieties and that lack a thick FBBG (10). Although the exact topological locations of the epitopes recognized by RCA-I on rabbit enterocytes and M cells are unknown, our results suggest that they are sequestered within the FBBG or perhaps located on smaller brush-border membrane glycoproteins. Since oligosaccharides are commonly terminated by sialic acid, we expected that glycoconjugates containing a(2,3)-linked sialic acid recognized by Mal II on both FAE and villus epithelium would be readily available to lectin immobilized on particles. Surprisingly, on live tissue 120-nm particles had access to Mal II epitopes only on the FAE and not on villus epithelium. Access to the Mal II epitopes on the FAE was size restricted, since 1-µm particles were unable to bind. Both 120-nm and 1-µm Mal II-coated particles adhered to aldehydefixed tissue, indicating that the failure of the particles to bind to villus epithelium was due to limited accessibility and not to a lack of Mal II binding sites. These data add to the list of phenotypic differences that are known to exist between the FAE and villus epithelium. For example, all cells of the FAE lack expression of polymeric immunoglobulin receptors (38) and show reduced brush border hydrolase activity compared with villus enterocytes (44). Lectin and antibody-binding studies in humans, mice, and rabbits revealed that FAE may differ from villus epithelium in glycosylation Downloaded from http://ajpgi.physiology.org/ by 10.220.32.246 on June 18, 2017 Fig. 4. Adherence of Mal II-coated 120-nm and 1-µm particles to FAE and villus (v) epithelium. Live or aldehydefixed Peyer’s patch explants were exposed to a 1:1 mixture of Mal II-coated particles (green) and biocytin-quenched particles (red) for 40 min. A: on live tissue, Mal II-coated 120-nm particles bound preferentially to FAE, including enterocytes and M cells. B: on fixed tissue, Mal II-coated 120-nm particles bound to both FAE and villus epithelium. C: on live tissue, Mal II-coated 1-µm particles bound to FAE, but this binding was nonspecific since control particles were present in equal numbers. D: on fixed tissue, Mal II-coated particles bound to both FAE and villus epithelium. Scale bar 5 100 µm. G922 GLYCOCONJUGATE ACCESSIBILITY ON INTESTINAL EPITHELIUM We thank Drs. Paul Giannasca, Karen Giannasca, and Larry Silbart for their helpful comments. N. J. Mantis gratefully acknowledges the receipt of a postdoctoral fellowship from the Harvard AIDS Institute, Cambridge, MA [National Institutes of Health (NIH) Grant T32-A107387-07] and a NIH Individual National Research Service Award (NIH F32-AI10009-02). A. Frey was supported by a personal grant from the Bundesministerium fur Bildung, Wissenschaft, Forschung und Technologie, and research grant FR958/2-1 from the Deutsche Forschungsgemeinschaft. M. R. Neutra is supported by NIH research grants (HD-17557 and AI-34757) and by a NIH Center grant to the Harvard Digestive Diseases Center (DK-34854). Present address for A. Frey: Center for Molecular Biology of Inflammation, University of Muenster, Germany. Address for reprint requests and other correspondence: M. R. Neutra, GI Cell Biology Laboratory, Enders 1220, Children’s Hospital, 300 Longwood Ave., Boston, MA 02115 (E-mail: [email protected]). Received 24 June 1999; accepted in final form 14 January 2000. REFERENCES 1. Abraham AN, Sharon N, and Ofek I. Adhesion of bacteria to mucosal surfaces. In: Mucosal Immunology (2nd ed.), edited by Ogra P, Mestecky J, Lamm ME, Strober W, Bienenstock J, and McGhee JR. San Diego: Academic, 1999, p. 31–42. 2. Amerongen HM, Weltzin R, Farnet CM, Michetti P, Haseltine WA, and Neutra MR. Transepithelial transport of HIV-1 by intestinal M cells: a mechanism for transmission of AIDS. J Acquir Immune Defic Syndr Hum Retrovirol 4: 760–765, 1991. 3. Amerongen HM, Wilson GAR, Fields BN, and Neutra MR. Proteolytic processing of reovirus is required for adherence to intestinal M cells. J Virol 68: 8428–8432, 1994. 4. Batinic D and Robey FA. The V3 region of the envelope glycoprotein of human immunodeficiency virus type 1 binds sulfated polysaccharides and CD4-derived synthetic peptides. J Biol Chem 267: 6664–6671, 1992. 5. Clark MA, Jepson MA, Simmons NL, and Hirst BH. Differential expression of lectin-binding sites defines mouse intestinal M cells. J Histochem Cytochem 41: 1679–1687, 1993. 6. Compans RW and Herrler G. Virus infection of epithelial cells. In: Mucosal Immunology (2nd ed.), edited by Ogra P, Mestecky J, Lamm ME, Strober W, Bienenstock J, and McGhee JR. San Diego: Academic, 1999, p. 671–683. 7. Corfield AP, Wagner SA, Clamp JR, Kriaris M, and Hoskins LC. Mucin degradation in the human colon: production of sialidase, sialate O-acetyltransferase, N-acetylneuraminate lyase, arylesterase, and glycosulfatase activities by strains of fecal bacteria. Infect Immun 60: 3971–3978, 1992. 8. Falk P, Boren T, Haslam D, and Capron M. Bacterial adhesion and colonization assays. Methods Cell Biol 45: 165– 192, 1994. 9. Fantini J, Cook DG, Nathanson N, Spitalnik SL, and Gonzalez-Scarano F. Infection of colonic epithelial cell lines by type 1 human immunodeficiency virus is associated with the cell surface expression of galactosylceramide, a potential alternative gp120 receptor. Proc Natl Acad Sci USA 90: 2700–2704, 1993. 10. Frey A, Giannasca KT, Weltsin R, Giannasca PJ, Reggio H, Lencer WI, and Neutra MR. Role of the glycocalyx in regulating access of microparticles to apical plasma membranes of intestinal epithelial cells: implications for microbial attachment and vaccine targeting. J Exp Med 184: 1045–1059, 1996. 11. Fujimura Y and Owen RL. M cells are portals of infections: clinical and pathophysiological aspects. Infect Agents Dis 5: 144–156, 1996. 12. Furuta Y, Eriksson K, Svennerholm B, Fredman P, Horal P, Jeansson S, Vahlne A, Holmgren J, and Czerkinsky C. Infection of vaginal and colonic epithelial cells by the human immunodeficiency virus type 1 is neutralized by antibodies raised against conserved epitopes in the envelope glycoprotein. Proc Natl Acad Sci USA 91: 12559–12563, 1994. 13. Gebert A and Hach G. Differential binding of lectins to M cells and enterocytes in the rabbit cecum. Gastroenterology 105: 1350–1361, 1993. 14. Gentsch JR and Pacitti AF. Effect of neuraminidase treatment of cells and effect of soluble glycoproteins on type 3 reovirus attachment to murine L cells. J Virol 56: 356–364, 1985. 15. Giannasca KT, Giannasca PJ, and Neutra MR. Adherence of Salmonella typhimurium to Caco-2 cells: identification of a glycoconjugate receptor. Infect Immun 64: 135–145, 1996. 16. Giannasca P, Boden J, and Monath T. Targeted delivery of antigen to hamster nasal lymphoid tissue with M-cell-directed lectins. Infect Immun 65: 4288–4298, 1997. 17. Giannasca PJ, Giannasca KT, Falk P, Gordon JI, and Neutra MR. Regional differences in glycoconjugates of intestinal M cells in mice: potential targets for mucosal vaccines. Am J Physiol Gastrointest Liver Physiol 267: G1108–G1121, 1994. 18. Giannasca PJ, Giannasca KT, Leichtner AM, and Neutra MR. Human intestinal M cells display the sialyl Lewis A antigen. Infect Immun 67: 946–953, 1999. Downloaded from http://ajpgi.physiology.org/ by 10.220.32.246 on June 18, 2017 patterns as well (18, 46). Together, these data suggest that differences in both receptor availability and receptor accessibility could enable pathogens to discriminate FAE from villus epithelium. Viruses may overcome the glycocalyx barriers on enterocytes and M cells by at least two mechanisms. Karlsson et al. (26) proposed that an initial low-affinity, high-avidity interaction between viral surface protein(s) and readily accessible oligosaccharide epitopes in the glycocalyx might allow a virus to adhere to an epithelial cell and then progressively gain closer proximity to receptors in the host cell plasma membrane. Such a strategy might be used by rotavirus, which somehow penetrates the FBBG and infects villus epithelial cells. Alternatively, viruses could physically penetrate glycoprotein coats that are relatively thin, such as the 20- to 30-nm apical surface coat on M cells. This is suggested by the observation that extension of the reovirus outer capsid protein sigma 1, which forms a 45-nm-long fiber, is a prerequisite for selective adhesion to mouse Peyer’s patch M cells (3, 36). Despite the fact that both terminal galactose- and sialic acid-containing glycoconjugates are recognized by enteric pathogens in vitro, we found that these epitopes were inaccessible to lectin-coated, bacteriumsized particles in vivo. These results underscore the importance of the active strategies that bacteria have developed to overcome the problem of limited receptor accessibility (1, 22, 25). For example, Shigella flexneri secrete enzymes like a-galactosidase that effectively degrade mucinlike glycoproteins (7, 42), and enterotoxigenic Escherichia coli express fimbriae that could ‘‘reach’’ into the glycocalyx to attach to glycoconjugates containing b1-linked galactosyl residues (40). Although it has been shown previously that tissue fixation can alter accessibility of carbohydrate antigens to soluble lectins (20), our data demonstrate for the first time that aldehyde fixation may enhance accessibility of certain oligosaccharides to lectin-coated, bacterium-sized particles by more than 20-fold. In light of these findings, bacterial adherence studies on isolated glycoconjugates, tissue sections, or fixed tissue should be interpreted with caution (8). In conclusion, more detailed information about the accessibility of glycolipid and oligosaccharide epitopes present on specific intestinal epithelial cells could lead to a better understanding of mucosal infection and more effective antigen delivery strategies. GLYCOCONJUGATE ACCESSIBILITY ON INTESTINAL EPITHELIUM 35. Neutra MR, Pringault E, and Kraehenbuhl JP. Antigen sampling across epithelial barriers and induction of mucosal immune responses. Annu Rev Immunol 14: 275–300, 1996. 36. Nibert M, Furlong D, and Fields B. Mechanisms of viral pathogenesis: distinct forms of reoviruses and their roles during replication in cells and host. J Clin Invest 88: 727–734, 1991. 37. Pappo J and Ermak TH. Uptake and translocation of fluorescent latex particles by rabbit Peyer’s patch follicle epithelium: a quantitative model for M cell uptake. Clin Exp Immunol 76: 144–148, 1989. 38. Pappo J and Owen RL. Absence of secretory component expression by epithelial cells overlying rabbit gut-associated lymphoid tissue. Gastroenterology 95: 1173–1174, 1988. 39. Parashar UD, Breese JS, Gentsch JR, and Glass RI. Rotavirus. Emerg Infect Dis 4: 561–570, 1998. 40. Payne D, O’Reilly M, and Williamson D. The K88 fimbrial adhesin of enterotoxigenic Escherichia coli binds b1-linked galactosyl residues in glycolipids. Infect Immun 61: 3673–3677, 1993. 41. Peterson MD and Mooseker MS. Characterization of the enterocyte-like brush border cytoskeleton of the Caco-2BBe clone of the human intestinal cell line Caco-2. J Cell Sci 102: 581–600, 1992. 42. Prizont R. Degradation of intestinal glycoproteins by pathogenic Shigella flexneri. Infect Immun 36: 615–620, 1982. 43. Rafiee P, Leffler H, Byrd JC, Cassels FJ, Boedeker EC, and Kim YS. A sialoglycoprotein complex linked to the microvillus cytoskeleton acts as a receptor for pilus (AF/R1) mediated adhesion of enteropathogenic Escherichia coli (RDEC-1) in rabbit small intestine. J Cell Biol 115: 1021–1029, 1991. 44. Savidge T, Smith M, Mayel-Afshar S, Collins A, and Freeman T. Selective regulation of epithelial gene expression in rabbit Peyer’s patch tissue. Pflügers Arch 428: 391–399, 1994. 45. Semenza G. Anchoring and biosynthesis of stalked brush border membrane proteins: glycosidases and peptidases of enterocytes and renal tubuli. Annu Rev Cell Biol 2: 255–313, 1986. 46. Sharma R, van Damme EJM, Peumans WJ, Sarsfield P, and Schumacher U. Lectin binding reveals divergent carbohydrate expression in human and mouse Peyer’s patches. Histochem Cell Biol 105: 459–465, 1996. 47. Spangler BD. Structure and function of cholera toxin and the related Escherichia coli heat labile enterotoxin. Microbiol Rev 56: 622–647, 1992. 48. Van Klinken BJ-W, Dekker J, Buller H, and Einerhand E. Mucin gene structure and expression: protection vs. adhesion. Am J Physiol Gastrointest Liver Physiol 269: G613–G627, 1995. 49. Willoughby RE. Rotaviruses preferentially bind O-linked sialylglycoconjugates and sialomucins. Glycobiology 3: 437–445, 1993. 50. Willoughby RE, Yolken RH, and Schnaar RL. Rotaviruses specifically bind to the neutral glycosphingolipid asialo-GM1. J Virol 64: 4830–4835, 1990. 51. Wolf J, Rubin D, Finberg R, Kauffman R, Sharpe A, Trier J, and Fields B. Intestinal M cells: a pathway for entry of reovirus into the host. Science 212: 471–472, 1981. Downloaded from http://ajpgi.physiology.org/ by 10.220.32.246 on June 18, 2017 19. Goldhar J. Bacterial lectinlike adhesins: determination and specificity. In: Bacterial Pathogenesis, edited by Clark VL and Bavoil PM. San Diego: Academic, 1994, p. 211–230. 20. Gonnella PA and Neutra MR. Glycoconjugate distribution and mobility on apical membranes of absorptive cells of suckling rat ileum in vivo. Anat Rec 213: 520–528, 1985. 21. Hioe CE, Bastiani L, Hildreth JE, and Zolla-Pazner S. Role of cellular adhesion molecules in HIV type 1 infection and their impact on virus neutralization. AIDS Res Hum Retroviruses 14, Suppl 3: S247–254, 1998. 22. Hultgren SJ, Abraham S, Caparon M, Falk P, St. Geme JW, and Normark S. Pilus and non-pilus bacterial adhesins: assembly and function in cell recognition. Cell 73: 887–901, 1993. 23. Ito S. Form and function of the glycocalyx on free cell surfaces. Philos Trans R Soc Lond B Biol Sci 268: 55–66, 1974. 24. Jepson MA, Clark MA, Foster N, Mason CM, Bennett MK, Simmons NL, and Hirst BH. Targeting to intestinal M cells. J Anat 189: 507–516, 1996. 25. Karlsson KA. Meaning and therapeutic potential of microbial recognition of host glycoconjugates. Mol Microbiol 29: 1–11, 1998. 26. Karlsson KA, Angstrom J, and Teneberg S. Characteristics of the recognition of host cell carbohydrates by viruses and bacteria. In: Molecular Pathogenesis of Gastrointestinal Infections, edited by Wadstrom T. New York: Plenum, 1991, p. 9–18. 27. Lentz TL. The recognition event between virus and host cell receptor: a target for antiviral agents. J Gen Virol 71: 751–766, 1990. 28. Madara JL. Epithelia: biological principles of organization. In: Textbook of Gastroenterology (2nd ed.), edited by Yamada T. Philadelphia: Lippincott, 1995, p. 141–159. 29. Maury J, Bernadac A, Rigal A, and Maroux S. Expression and glycosylation of the filamentous brush border glycocalyx (FBBG) during rabbit enterocyte differentiation along the cryptvillus axis. J Cell Sci 108: 2705–2713, 1995. 30. Maury J, Nicoletti C, Guzzo-Chambraud L, and Maroux S. The filamentous brush border glycocalyx, a mucin-like marker of enterocyte hyper-polarization. Eur J Biochem 228: 323–331, 1995. 31. McGhee JR, Mestecky J, Dertzbaugh MT, Eldridge JH, Hirasawa M, and Kiyono H. The mucosal immune system: from fundamental concepts to vaccine development. Vaccine 10: 75–88, 1992. 32. Mondor I, Ugolini S, and Sattentau QJ. Human immunodeficiency virus type 1 attachment to HeLa CD4 cells is CD4 independent and gp120 dependent and requires cell surface heparans. J Virol 72: 3623–3634, 1998. 33. Mooseker MS. Organization, chemistry, and assembly of the cytoskeletal apparatus of the intestinal brush border. Annu Rev Cell Biol 1: 209–241, 1985. 34. Neutra MR, Frey A, and Kraehenbuhl JP. Epithelial M cells: gateways for mucosal infection and immunization. Cell 86: 345–348, 1996. G923