Characteristics of a Glycoprotein in the Ocular Surface Glycocalyx

llene K. Gipson,*t Michelle Yankauckas,* Sandra J. Spurr-Michaud,* Ann S. Tisdale,* and William Rineharr*

A monoclonal antibody has been produced that binds to the apical squames (flattened cells) of the rat ocular surface epithelium and to the goblet cells of the conjunctiva. Immunoelectron microscopic localization of the antigen indicates that in apical cells it is present along the apical-microplical membrane in the region of the glycocalyx. In subapical squames, the antigen is in cytoplasmic vesicles. In some goblet cells, the antigen is in the Golgi network, and in others, it is located primarily in the membrane of the mucous granules. SDS-PAGE and immunoblot analysis demonstrate that the molecular weight of the antigen is greater than 205 kD, and the electrophoretic band stains with Alcian blue followed by silver stain. Periodate oxidation of immunoblots and cryostat sections removes antibody binding. Neuraminidase treatment of cryostat sections does not remove antibody binding, whereas N-glycanase does. Taken together, these data indicate that the antigen recognized by the monoclonal antibody is a carbohydrate epitope on a high-molecular-weight, highly glycosylated glycoprotein in the glycocalyx of the ocular surface epithelium and goblet cell mucin granule membrane. The antigen appears to be stored within cytoplasmic vesicles and reaches the glycocalyx when cells differentiate to the apical-most position where the glycocalyx interfaces with the mucin layer of the tear film. Invest Ophthalmol Vis Sci 33:218-227,1992

The ocular surface epithelium that covers the con- tissue stained with tannic acid demonstrates that the junctiva and cornea is nonkeratinizing, stratified, and glycocalyx is a fine, filamentous layer. Each filament squamous and is made up of three to seven cell layers. inserts into the cell membrane and has angular bends The flattenedoute r squames of the epithelium are cov- and branches distally.2 These filaments are particu- ered at their apical membrane by the tear film, which larly prominent at the tips of the microplicae. is generally considered to be subdivided—from air in- Very little is known about the biochemical nature terface to epithelial apical membrane—into oil, of the glycocalyx, and even less is known of its interac- aqueous, and mucus layers. The mucus layer is se- tion with or role in the spread of mucus over the api- creted onto the ocular surface epithelium by goblet cal cells. That the glycocalyx contains many highly cells in the conjunctival region. In guinea pigs, the charged polyanions is demonstrated by the intense mucus layer varies in thickness from 1.0 nm over cor- binding of ruthenium red to fixed tissue.23 Other stud- nea to 2-7 jim over conjunctiva.1 As has been ele- ies demonstrate binding of Alcian blue, dialyzed iron, gantly demonstrated in the electron micrographs of cationized ferritin, periodic acid-Schiff reagent, and the rapid-freeze, freeze-substitution-prepared ocular several lectins to the ocular surface.4"8 These studies surface epithelium of the guinea pig, the mucus layer indicate that the ocular surface is rich in carbohydrate is intimately associated with the glycocalyx of the api- moieties, but they do not give specific molecular in- cal cell.' The glycocalyx is a carbohydrate-rich, extrin- formation nor do they differentiate totally between sic cell surface coat that forms a layer along the apical glycocalyx and mucus layers. membrane to which the mucus layer binds, presum- We have produced a monoclonal antibody that ably loosely. Electron microscopy of ocular surface binds to apical cells of the ocular surface epithelium of the rat and that appears to recognize a component of the glycocalyx. We have begun to characterize the gly- From the *Eye Research Institute, and the f Department of Oph- coprotein recognized by this antibody. thalmology, Harvard Medical School, Boston, Massachusetts. Supported by grant R37-EY-03306 from the National Eye Insti- tute, National Institutes of Health, Bethesda, Maryland. Materials and Methods Submitted for publication: April 29, 1991; accepted July 22, 1991. All investigations involving animals reported in Reprint requests: llene K. Gipson, Eye Research Institute, 20 this study conform to the ARVO Resolution on the Staniford Street, Boston, MA 02114. Use of Animals in Research. Adult male Sprague-

218

Downloaded from iovs.arvojournals.org on 09/27/2021 No. 1 OCULAR SURFACE GLYCOCALYX GLYCOPROTEIN / Gipson er ol 219

Dawley rats, New Zealand white rabbits, and Hartley immunofluorescence (IF) microscopy. Hybridomas guinea pigs were used. Animals were sacrificed with with apical cell binding by IF were cloned by limiting an overdose of sodium pentobarbital. dilution (0.5 cell/well) two consecutive times. Anti- bodies from tissue culture medium were concentrated Monoclonal Antibody Production by ammonium sulfate precipitation. Preparation of immunogen and immunization: Apical cells of the corneas of adult Sprague-Dawley Immunofluorescence Localization rats (175-225 g) were obtained by gentle brushing of corneas that had been excised, pinned on paraffin Six-micrometer cryostat sections of rat cornea, eye- posts, and incubated overnight in low-Ca2+ minimum lid, skin, esophagus, lacrimal gland, oral mucosa, essential medium (MEM) (Gibco, Grand Island, liver, pancreas, ileum, lung, and colon were placed on NY).9 The cells were centrifuged at 1,000 X g for 15 gelatin-coated slides and dried overnight at 37 °C. Sec- min. The cell pellet was resuspended in MEM with tions were similarly prepared from guinea pig and 10% dimethyl sulfoxide (Sigma, St. Louis, MO), fro- rabbit corneas and human corneas obtained from Na- zen, and stored in liquid nitrogen until enough cells tional Disease Research Interchange. Sections were were obtained for the immunizations. Prior to immu- rehydrated in PBS, pH 7.2, and blocked in PBS with nization, the cells were thawed in a 37°C water bath 1% bovine serum albumin (BSA) for 10 min. Primary and washed two or three times in MEM and once in antibody (hybridoma tissue culture media or mono- phosphate-buffered saline (PBS). The cells were resus- clonal antibody) was applied for 1 hr at room tempera- pended in equal amounts of PBS and RIBI adjuvant ture in a moist chamber. Sections were rinsed with (RIBI; Immunochem Research, Hamilton, MT). One PBS followed by 10 min in PBS with 1% BSA. Fluores- times 107 apical cells prepared in this way were in- cein isothiocyanate (FITC)-goat anti-mouse IgG (Cal- jected intraperitoneally into 6-week-old female biochem, La Jolla, CA) was similarly applied for 1 h at BALB/cByJ mice (Jackson Laboratories, Bar Harbor, room temperature. After a PBS wash, coverslips were ME). A booster injection of 7.5 X 105 cells in PBS/ mounted with a medium consisting of PBS, glycerol, RIBI adjuvant was given one month later. and para-phenylenediamine." Negative control tis- Cell fusion and hybridoma cloning: Four days after sue sections (primary antibody omitted) were rou- the boost, cell fusion was carried out according to a tinely included in each antibody-binding study. The modification of the procedure of Kohler and Mil- sections were viewed and photographed on a Zeiss stein.10 Briefly, spleen cells from an immunized photomicroscope III (AZI, Avon, MA) equipped for mouse were mixed with P3/NSl/l-Ag4-l(NS-l) epi-illumination. (ATCC, Rockville, MD) myeloma cells in a ratio of Neuraminidase treatment: Cryostat sections of rat 5:1 in serum-free medium. Cells were centrifuged at cornea were treated with 1.25 U/ml neuraminidase 200 X g for 10 min at room temperature. The super- isolated from Clostridium perfringens (Sigma, St. natant was gently removed and the tube transferred to Louis, MO) in PBS, pH 5.5, for 20, 40, or 60 min at a 37°C water bath where 1 ml of 50% polyethylene 37°C.4 As a control, adjacent serial sections were si- glycol (PEG) (Boehringer Mannheim Biochem, In- multaneously incubated in PBS, pH 5.5. Following dianapolis, IN) in 75 raMHEPE S (Gibco) was added. the neuraminidase or control buffer incubation, the After 1 min, the PEG was diluted out by adding 1, 2, sections were rinsed in PBS, pH 7.2, and the normal and 4 ml of serum-free Iscove's Modified Dulbecco's immunofluorescence labeling procedure was fol- Medium (IMDM; Gibco) after 1,2, and 4 min, respec- lowed. The neuraminidase used for this treatment tively. The dilution was completed by adding 8 ml of was checked for contaminating protease activity using IMDM with 10% fetal calf serum (FCS). The cells the Bio-Rad Protease Detection Kit. Dispase II were centrifuged at 200 X g for 10 min and resus- (Boehringer-Mannheim, Indianapolis, IN), a neutral pended in 50 ml of IMDM plus FCS and HAT bacterial protease, was used as the standard for the (Sigma). One hundred microliters per well of this sus- assay. The neuraminidase was found to be free of pro- pension was plated in flat-bottom, 96-well plates con- tease activity at 62.5 U/ml (50 times the concentra- taining a feeder layer of BALB/cByJ mouse peritoneal tion used in the treatment). macrophages. After 7 days, 100 /A of HAT was added Periodate treatment: The effect of strong, moder- to each well. After 2 weeks, the cultures were fed with ate, and mild oxidation on antibody binding to cryo- HT-containing medium and screened by ELISA for stat sections was examined using sodium periodate 12 IgG production using the Bio-Rad (Richmond, CA) (NaIO4), according to the method of Basbaum et al. Clone Selector Mouse Monoclonal Antibody Screen- Cryostat sections of rat cornea were incubated at ing Kit. Positive cultures were screened for hybrid- room temperature in the dark, overnight, in 100 mM omas of interest on cryostat sections of rat corneas by NaIO4 (strong oxidation; Sigma), at 4°C for 1 hr in 50

Downloaded from iovs.arvojournals.org on 09/27/2021 220 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / January 1992 Vol. 03

raM NaIO4 (moderate oxidation), or at 4°C for 10 30-sec pulses of the polytron (Brinkmann Instru- min in 10 mM NaIO4 (mild oxidation). The NaIO4 ments, Westbury, NY). The solution was diluted to a solutions were made up in 50 mM sodium acetate, pH protein concentration of 1.6 /ig/Vl (via Bio-Rad pro- 4.5. As controls, adjacent sections were similarly in- tein assay) with 2X reaction mix (60 mM Tris, 0.25% cubated in the NaIO4 solution plus 0.1 M glucose or glycerol, 0.5% SDS, 45 mM dithiothreitol, 4 mM 0.015 (volume/volume) ethylene glycol. Following urea). SDS-PAGE was performed using the buffer sys- 13 the incubations in NaIO4, the sections were incubated tem described by Miles Laboratories, which is an in 10 mM sodium borohydride (Sigma) in PBS, pH adaptation of Jovin's discontinuous (multiphasic) 14 7.2, for 30 min at room temperature. The sections buffer system using the mini-protean II dual slab gel were rinsed well in five changes of PBS, and the usual apparatus from Bio-Rad. Gels 0.75 mm thick, 6% (6% immunofluorescence labeling procedure was fol- T, 2.75% C), were run at constant voltage of 200 V for lowed. 1.25 hr using reagents from Bio-Rad. Prestained mo- N-glycanase treatment: Cryostat sections of rat cor- lecular weight markers included myosin (205 kD) and nea were incubated in 60 U/ml N-glycanase peptide b-galactosidase (118 kD). After the SDS-PAGE, gels —N-glycosidase F (Genzyme, Boston, MA) in 0.55 M were stained by a modification of an Alcian blue- sodium phosphate, pH 8.6, overnight, at 44°C. As a silver staining method designed to stain highly glycosylated glycoproteins that will not stain by Coomassie control, adjacent sections were incubated in 60 U/ml 15 N-glycanase in 30 mM citrate buffer, which inhibits Blue or silver. Gels were fixed and stained with Al- cian blue15 followed by silver staining using the proto- glycanase activity. Following the overnight incuba- 16 tion, the sections were rinsed well in PBS, pH 7.2, and col of Wray et al. the usual immunofluorescence localization protocol Proteins in gels were transferred to nitrocellulose was followed. paper as described by Towbin et al.17 Blotted antigens then were detected using the Vectastain elite mouse Immunoelectron Microscopy (IEM) IgG kit (Vector Labs, Burlingame, CA) and the manu- Following fixation in 4% paraformaldehyde and facturer's protocol with several exceptions. Tween-20 0.2% glutaraldehyde in 0.1 M PO4 buffer, pH 7.4, for (Sigma) was eliminated from the blocking buffer be- 1 h at 4°C, corneas were rinsed in 0.1 M PO4 buffer cause it was determined that it interfered with anti- and cut into quarters. They were dehydrated in a body binding. Ten percent horse serum (Gibco) in graded ethanol series, then were embedded in me- Tris-buffered saline, pH 7.5, was used as a blocking dium-grade LR White resin (London Resin Co.; Er- agent during initial blocking of unbound nitrocellu- nest F. Fullam, Latham, NY) in gelatin capsules fol- lose binding sites, as well as during primary and sec- lowed by heat-curing at 50°C. Sections 1 nm thick ondary antibody incubations. The NaCl concentra- were stained with toluidine blue for orientation. tion in the ABC reagents was increased to 0.5 M to Thick sections also were mounted on gelatin-coated decrease nonspecific staining. slides for immunofluorescence localization (see pro- Periodate oxidation: To determine if periodate oxi- tocol above) to verify antibody binding to the LR dation removed the antibody binding to immuno- White fixed and embedded tissue. Thin sections were blots, blotted nitrocellulose strips were soaked in 50 mounted on 3-mm, 200-mesh nickel grids (Ernest F. mM sodium acetate, pH 4.5, for 5 min and trans- Fullam). Rat apical monoclonal antibodies were local- ferred to 0.1, 1,5, 10, or 20 mM NaIO4 (Sigma), pH ized using post-embedding immuno-gold labeling fol- 4.5, for 1 hr in the dark at 23°C, according to the lowing the protocol described by the manufacturers of technique of Woodward et al.18 Antigens then were the gold-conjugated secondary antibody (Janssen, detected using the Vectastain kit as already described. Ted Pella, Redding, CA). Janssen Auroprobe One af- finity-purified, goat anti-mouse IgG antibody linked Results to 1-nm colloidal-gold particles (Ted Pella) was used as the secondary antibody. Gold signal was visualized Two fusions (one did not yield hybridomas of inter- by silver amplification with IntenSE M according to est) yielded 12 cell lines. Supernatant from these lines the Janssen protocol (Ted Pella). yielded several types of antibodies. Eight of the 12 supernatants bound apical cells. We selected one hy- Electrophoresis and Immunoblotting bridoma specific for apical cells for further cloning. Limbal to limbal corneal epithelium was scraped Several clones from the hybridoma were selected for from rat eyes and solubilized in 500 /xl of 7.5 mM characterization. Each clone from the same hybrid- Tris, pH 8.9, 12.5% glycerol, 0.05% SDS, 5 mM urea oma had similar characteristics, and we believe they solution run through 18 G, 20 G, 21 G, 25 G, and 27 all recognize the same epitope. The data reported here G needles, and then homogenized with two or three are from monoclonal antibody 13G5.339. In the re-

Downloaded from iovs.arvojournals.org on 09/27/2021 No. 1 OCULAR SURFACE GLYCOCALYX GLYCOPROTEIN / Gipson er al 221

suits described below, we refer to the antibody as rat (Figs. 2A, B). In the cytoplasmic vesicles, the binding ocular surface glycocalyx antibody, or ROSG anti- appeared to be along the internal face of the vesicle body. membrane (Fig. 2C). These vesicles that bind the anti- By immunofluorescence microscopy, we found body occur only in flattened squames of the ocular that binding of antibody was to all the flattened cell surface epithelium. There was variability in the layers or squames along the entire ocular surface epi- amount of gold label present on apical cells (compare thelium (Fig. 1A). Binding ended abruptly where the Figs. 2A and D). In a loosened cell that appears to be epithelium becomes keratinized at the lid margin desquamating (Fig. 2D), there is no binding to its api- (Fig. 1C). Goblet cells within the conjunctival epithe- cal surface. The cell beneath the loosened cell shows lium also bound the antibody (Fig. IB). Lacrimal binding along its apical membrane. gland did not bind the antibody, but an occasional In rats, goblet cells occur in groups or clusters of duct cell did (data not shown). Binding was not found cells that in some respects have the appearance of in the other species tested (rabbit, guinea pig, human) acini. Immunoelectron microscopy of the goblet cell nor in other rat tissues tested, including skin, esopha- cluster showed that only some of the goblet cells of the gus, oral mucosa, liver, pancreas, ileum, lung, and cluster bind the antibody. In Figure 3, three adjacent colon (data not shown). cells of a cluster of goblet cells show different binding Immunoelectron microscopic localization studies patterns. A cell in the center of the cluster shows no showed that in the outer squames adjacent to the tear binding; an adjacent cell shows binding in the Golgi film, binding was in the outer apical membrane (Fig. region outside the mucin granules, and adjacent to 2A). At higher magnification, binding was particu- this cell near the outer edge of the cluster, binding is larly prominent at the tips of the microplicae (Fig. present on the mucin granules. Study of the localiza- 2B). The antigen was present within small vesicles in tion along mucin granules showed a prevalence of the cytoplasm of the squames below the apical cell binding to the membrane region of the granule (Fig.

Fig. 1. Immunofluorescence micrographs demonstrating localization of the ROSG antibody in the corneal epithelium (A) and conjunctival epithelium (B). Binding in the cornea is present on the several layers of apical flattened squames. In the conjunctiva, in addition to apical squame binding, intense binding to goblet cells is present. (C) The abrupt end to binding is seen where the nonkeratin- ized ocular surface epithelium joins keratinized epidermis at the lid margin. The hair fol- licle of the eyelash is seen at the upper left between arrows (A-C, X300).

Downloaded from iovs.arvojournals.org on 09/27/2021 222 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / January 1992 Vol. 03

•••.

Fig. 2. Immunoelectron microscopic localization of the ROSG antibody in corneal epithelium. (A) At low magnification, the silver-en- hanced 1 -nm immunogold can be seen along the apical tear-facing membrane of the apical cell. In subapical cells, the label is present within the cytoplasm of the cells. Arrows indicate cell boundaries of subapical cells (X7200). (B) Higher magnification electron micrograph demonstrating prevalence of antibody binding on microplicae of apical cells. The membrane of the abutting subapical cell is indicated by the arrows (X21,300). (C) Higher magnification electron micrograph showing labeled cytoplasmic vesicles of subapical cells; the label is particularly prevalent on the vesicle membrane (arrows) (X21,300). (D) Variation in amount of binding to apical cells was noted. In this electron micrograph of a loosely adherent cell, there is no apical membrane binding. The cell beneath has binding in its apical membrane (x21,300).

Downloaded from iovs.arvojournals.org on 09/27/2021 No. 1 OCULAR SURFACE GLYCOCALYX GLYCOPROTEIN / Gipson er ol 223

B Fig. 3. (A) Immunoelectron microscopic localization of the ROSG antibody on a cluster of goblet cells. The cell in the center of the cluster (1) is not labeled, even though it has mucin granules. In the adjacent cell (2), the antigen is localized to the Golgi apparatus (arrows) between mucin granules. The cell at the periphery of the cluster (3) has intensely labeled mucin granules (X9000). The inset shows these mucin granules at higher magnification. The label appears to be more prevalent along the peripheral membrane region of the granule (X36.000). (B) Electron micrograph of epithelium processed as a secondary antibody control in which incubation with primary antibody was omitted (X9000).

3A, inset). Some binding is in the "lumen" region In developing the post-embedding technique for above the goblet cell cluster. Because secondary anti- immunoelectron microscopic localization of the antibody controls (Fig. 3B) appear free of gold, the lu- gen, we found that although we preserved antigenicity minal binding may be from secretory products in the after fixation and embedding in LR White resin (as lumen. judged by immunofluorescence localization on the

Downloaded from iovs.arvojournals.org on 09/27/2021 224 INVESTIGATIVE OPHTHALMOLOGY G VISUAL SCIENCE / January 1992 Vol. 33

1-^m sections), we could not label with secondary antibodies conjugated to 5-nm gold particles. Only secondary antibodies conjugated to 1-nm immuno-gold followed by silver enhancement allowed localization at the electron microscope level.

Electrophoretic Mobility and Immunoblots By immunoblot analysis, the ROSG antibody reacted with a prominent band that has a molecular weight greater than 205,000 (Fig. 4). Coomassie or silver staining of the gel in that region demonstrated no band of similar molecular weight. Periodic acid- Schiff reagent did stain a band in the same region O.I 1.0 5.0 10*0 (data not shown) as did Alcian blue followed by silver staining (Fig. 4). These data plus the observation that mM Periodate the reactive band was more diffuse at lower acrylam- ide concentrations (6% compared to 10-12%) suggest Fig. 5. Blots of 6% SDS-PAGE treated with increasing concentrations of sodium periodate to remove carbohydrate. The arrow indi- that the antigen is a highly glycosylated glycoprotein. cates remaining reactive bands in the control and low periodate Because highly glycosylated glycoproteins run anoma- concentration lanes. At higher concentrations, binding is com- lously on SDS-PAGE, an accurate molecular weight pletely lost. determination was not possible.

Periodate Incubation tion time also showed a dose-, and treatment time-, dependent removal of antibody binding (Fig. 6). To determine whether the epitope of the antigen Under mild periodate oxidation conditions (10 mM, recognized by the ocular-surface monoclonal anti- 1 hr incubation), partial binding remained. Under body was to the carbohydrate portion of the glycopro- moderate (50 mM, 1 hr) or strong (100 mM, over- tein, periodate oxidation of immunoblots and frozen night) oxidation conditions, all binding was lost. sections were done. The effect of increasing concentrations of periodate on binding of ROSG monoclo- Enzyme Treatments nal antibody to immunoblots is shown in Fig. 5. Treatment of immunoblot with 5 and 10 mM perio- Having evidence that the epitope recognized by date for 1 hr completely removed antibody binding. ROSG monoclonal antibody is carbohydrate, two At 1 mM periodate, binding was partially removed, glycosidases were used on cryostat sections to deter- and at 0.1 mM, binding was similar to that of the mine whether antibody binding was lost after incuba- control incubation. Treatment of cryostat sections tion in their presence. Neuraminidase treatment at with increasing periodate concentrations and incuba- 1.25 U/ml for increasing periods of incubation up to 1 hr did not affect antibody binding (data not shown). Incubation of sections with N-glycanase to remove asparagine-linked glycoproteins abolished antibody binding as compared to controls (Fig. 7). 205 — Discussion Our immunohistochemical and immunoelectron 118. microscopy findings suggest that the antigen recognized by the ROSG antibody is a component of the 77 — glycocalyx of the ocular surface epithelium. That it is a glycocalyx component rather than a goblet cell mucin product is supported by these lines of evidence: (1) the antigen is localized in the cytoplasm in subapical Fig. 4. Left lane, 6% SDS-PAGE of cornea! epithelium stained squames; (2) fixation does not remove the antigen with alcian blue followed by silver. Right lane, immunoblot show- from the surface of the eye. Nichols et al1 have shown ing antibody binding to a band of similar molecular weight. Molecu- lar masses (in kilodaltons) determined from standard proteins that the mucus layer is not preserved by conventional (myosin, 205; /3-galactosidase, 118; bovine serum albumin, 77) are fixation for electron microscopy; and (3) incubation noted. of cryostat sections with N-glycanase destroys anti-

Downloaded from iovs.arvojournals.org on 09/27/2021 No. 1 OCULAR SURFACE GLYCOCALYX GLYCOPROTEIN / Gipson er ol 225

Fig. 6. Immunofluorescence micrographs of cryostat sections treated with 100 mM (A; region between arrows is epithelium) and 10 mM (B) sodium periodate. Control incubations had ethylene glycol (C) or glucose (D) added to incubation solutions. (A-D, X75O).

body binding. These data suggest that the epitope rec- ule, loose association would be required to allow mu- ognized by the ROSG antibody is on an N-linked su- cin discharge from the granule and its membrane. gar chain. In mucins, sugar chains are attached to the Loose association of the mucus coat to the apical central core protein primarily through O-glycosidic membrane is presumed necessary to allow movement linkage.19 of the mucus to act as a debris removal system for the An argument could be made that the binding of the ocular surface much as it acts in the trachea.20 Per- antibody to goblet cells refutes the claim that the anti- haps antigen recognized by our antibody is a glycoca- gen recognizes an N-linked glycocalyx component. lyx component that facilitates such loose association. Maybe the sugar epitope is common to cell surface If mucins and glycocalyx glycoproteins bound the an- glycoprotein and a goblet cell mucin. The pattern of tibody, binding to several protein bands could be an- immunoelectron microscopic localization of the anti- ticipated. By Western blot analysis, only one reactive gen in the goblet cell and the Western blot data sug- band was present. gest that the goblet cell binding is to a nonmucin com- In the electron microscopic localizations of the an- ponent of the mucin granule. Binding of the antibody tigen in the apical-most cell glycocalyx, we noted a in clusters of goblet cells varies among cells of the difference among cells regarding the amount of anti- cluster, with Golgi binding and no mucin packet bind- gen detected. The cells that were less adherent to un- ing in some cells and mucin packet membrane bind- derlying cells and that had the appearance of cells ing in others. This suggests that the antigen is added to about to desquamate bound less antibody. Perhaps as the mucin granule late in the maturation of the goblet cells age, glycocalyx components are lost from the cell cell, after mucin production and packaging into gran- surface by movement of the mucus layer along the ules. The predilection to bind to the granule mem- microplicae during blinking or by active shedding brane may be similar to that on the apical membrane from the cell surface. As the cell ages and loses its of the apical cells of the stratified epithelium. The mu- mucin-interacting glycoproteins of the glycocalyx, cin granule membrane and the apical membrane of perhaps the mucus layer sticks more tightly to the cell, the epithelium are assumed to require a loose associa- inducing or facilitating desquamation. Wells and tion of mucus to them. In the case of the mucin gran- Hazlett have reported an increase in mucus on "dark"

Downloaded from iovs.arvojournals.org on 09/27/2021 226 INVESTIGATIVE OPHTHALMOLOGY & VI5UAL SCIENCE / January 1992 Vol. 33

mation. Formation of the tight junction is known to be responsible for the targeting of secretory products or membrane components to the appropriate pole of the cell.23 Numerous examples of such targeting have been provided for products of simple epithelium.23 To our knowledge, no such movement has been demonstrated for stratified squamous epithelia. Apically targeted membrane components of simple epithelia, including MDCK cells, thyroid epithelia, and colonic epithelia, are stored within a novel vacuolar apical compartment termed VAC.24 Initiation of cell-cell contact induces rapid formation of tight junc- tions formation in cultured MDCK cells. Upon cell- cell contact, the VAC is exocytosed toward the region of cell-cell contact, where it contributes significantly to the formation of the apical surface.24 Culture conditions that prevent cell-cell contact, ie, culture in low Ca+2, prevent VAC exocytosis; with increased Ca"1"2 concentrations, cell-cell contact is initiated and VAC exocytosis ensues.25 Perhaps the subapical squames of the ocular surface epithelium store apical membrane or glycocalyx components in a VAC, and perhaps movement of the VAC to the apical membrane occurs as the tight junc- Fig. 7. Immunofluorescence micrographs of section treated with tion forms between apical squames. Vinculin, a tight N-glycanase (A) or N-glycanase plus citrate (B) as control. Anti- junction component, has been localized to the region body binding is lost in (A), where epithelium is barely visible be- of contact between apical cells of the corneal epithe- tween arrows (A, B, X370). lium of the rat.26 Verification of tight junction induc- tion of movement of the glycocalyx glycoprotein rec- cells of the ocular surface.4 By scanning electron mi- ognized by the ROSG antibody to the ocular surface croscopy, "light, medium, and dark" cells have been awaits double-labeling experiments that will allow noted,421 and these authors postulate that dark cells correlation of junction formation with movement of are the "oldest" cells on the ocular surface. If there is vacuoles to the apical surface. loss of glycocalyx components from cells into the tear Movement of vesicles to the surface of apical con- film as the cells age, their presence in tear fluid pre- junctival epithelial cells has been proposed.52627 sumably would be detectable. In preliminary studies, Greiner et al5 suggested that vesicle delivery provided we have detected antibody binding to rat tear samples a second source of mucin to the ocular surface. The obtained with a capillary pipette and dried onto glass stains these investigators used to follow the vesicles slides. In another tissue, surface glycoconjugates from bind highly glycosylated molecules of either N- or O- ciliated cells have been demonstrated to be compo- linkage. Whether the products carried within the vesi- 22 nents of the tracheal "mucus" secretions. cles are cell surface glycocalyx components or mucins The immunoelectron microscopy data on localiza- remains to be determined. Possibly the antigen de- tion in the stratified ocular surface epithelium show tected by our ROSG antibody is the highly glycosy- that most of the antigen in the apical-most cell is in lated molecule detected by these investigators. the glycocalyx with little cytoplasmic binding. Be- The antigen detected by immunoblot analysis in cause the antigen appears within small, membrane- our study has a molecular weight greater than 205 kD, bound vesicles within subapical squames, a signal to just entering a 6% gel. In an SDS-PAGE analysis of move the vesicles to the membrane must be generated individual ocular mucus samples from normal and as the cell takes up its apical-most position adjacent diseased human conjunctivas, high-molecular-weight to the tear film. That signal may be tight junction for- glycoproteins were found in 2-16% gradient gels.28 mation. In simple columnar epithelium, the tight Although the two gel systems in these studies are dif- junction is known to be the structure along the cell ferent, comparing the data is tempting. The most prev- membrane that segregates the apical membrane com- alent band within the human gels, GP2, appears at 23 ponents from basolateral components. The estab- approximately the same region of the gel as the reac- lishment of apical-basal polarity through cell-sub- tive band in our immunoblots. strate and cell-cell contact leads to tight junction for- The Western blot data, the Alcian blue-silver stain

Downloaded from iovs.arvojournals.org on 09/27/2021 No. 1 OCULAR SURFACE GLYCOCALYX GLYCOPROTEIN / Gipson er ol 227

of the SDS-PAGE gels, and the periodate oxidation cific immunofluorescent findings with use of a para-phenylen- data in this report suggest that the antigen recognized ediamine mounting buffer. J Invest Dermatol 78:449, 1982. 12. Basbaum CB, Chow A, Macher BA, Finkbeiner D, Veissiere D, by the ROSG antibody is a highly glycosylated glyco- and Forsberg LS: Tracheal carbohydrate antigens identified by protein greater than 205 kD in molecular weight. In monoclonal antibodies. Arch Biochem Biophys 249:363, 1986. addition, the epitope recognized by the antibody is to 13. Canalco SAGE Kit. Miles Laboratories, Inc., Elkhart, IN, a sugar portion of the molecule. The immunohisto- 1980, pp. 1-16. chemical data suggest that the sugar epitope is not 14. Jovin TM, Dante ML, and Chrambach A: Multiphasic Buffer sialic acid and that it is an N-linked sugar. These later Systems Output. National Technical Information Service, Springfield, VA, Publication No. 196085-196091, 203016, studies are, however, not definitive. Attempts to con- 1970. firm the immunohistochemical data by Western blot 15. Jay GD, Culp DJ, and Jahnke R: Silver staining of extensively procedures were unsuccessful. Because such protocols glycosylated proteins on sodium dodecyl sulfate-polyacryl- are more successful with purified glycoproteins, de- amide gels: Enhancement by carbohydrate-binding dyes. Anal finitive classification of the glycoprotein into cell sur- Biochem 185:324, 1990. 16. Wray W, Boulikas T, Wray VP, and Hancock R: Silver staining face or mucin categories awaits purification of the an- of proteins in polyacrylamide gels. Anal Biochem 118:197, tigen. 1981. In summary, we have developed a monoclonal an- 17. Towbin H, Staehelin T, and Gordon J: Electrophoretic transfer tibody that recognizes a sugar epitope on a high-mole- of proteins from polyacrylamide gels to nitrocellulose sheets: cular-weight, highly glycosylated glycoprotein found Procedure and some applications. Proc Natl Acad Sci USA 76:4350, 1979. in the ocular surface glycocalyx of the rat. Further 18. Woodward MP, Young WW Jr, and Bloodgood RA: Detection characterization of the glycoprotein may yield infor- of monoclonal antibodies specific for carbohydrate epitopes mation relevant to the biological properties of the ocu- using periodate oxidation. J Immunol Methods 78:143, 1985. lar surface, such as mucin spread and resistance to 19. Carlstedt I, Sheehan JK, Corfield AP, and Gallagher JT: Mu- pathogen adherence. cous glycoproteins: A gel of a problem. Essays Biochem 20:40, 1985. Key words: ocular surface, glycocalyx, tear film spread, 20. Ringler NJ, Selvakumar R, Woodward HD, Bhavanandan VP, monoclonal antibody, goblet cells and Davidson EA: Protein components of human tracheo- bronchial mucin: Partial characterization of a closely asso- References ciated 65-kilodalton protein. Biochemistry 27:8056, 1988. 21. Pfister RR: The normal surface of corneal epithelium: A scan- 1. Nichols BA, Chiappino ML, and Dawson CR: Demonstration ning electron microscopic study. Invest Ophthalmol Vis Sci of the mucous layer of the tear film by electron microscopy. 12:654, 1973. Invest Ophthalmol Vis Sci 26:464, 1985. 22. Varsano S, Basbaum CB, Forsberg LS, Borson DB, Caughey 2. Nichols B, Dawson CR, and Togni B: Surface features of the BG, and Nadel JA: Dog tracheal epithelial cells in culture syn- conjunctiva and cornea. Invest Ophthalmol Vis Sci 24:570, thesize sulfated macromolecular glycoconjugates and release 1983. them from the cell surface upon extracellular proteinases. Exp 3. Dilly PN: On the nature and the role of the subsurface vesicles Lung Res 13:157, 1987. in the outer epithelial cells of the conjunctiva. Br J Ophthalmol 23. Rodriguez-Boulan E and Nelson WJ: Morphogenesis of the 69:477, 1985. polarized epithelial cell phenotype. Science 245:718, 1989. 4. Wells PA and Hazlett LD: Complex carbohydrates at the ocu- 24. Vega-Salas DE, Salas PJI, and Rodriguez-Boulan E: Modula- lar surface of the mouse: An ultrastructural and cytochemical tion of the expression of an apical plasma membrane protein of analysis. Exp Eye Res 39:19, 1984. Madin-Darby canine kidney epithelial cells: Cell-cell interac- 5. Greiner JV, Weidman TA, Korb DR, and Allansmith MR: tions control the appearance of a novel intracellular storage Histochemical analysis of secretory vesicles in nongoblet con- compartment. J Cell Biol 104:1249, 1987. junctival epithelial cells. Acta Ophthalmol 63:89, 1985. 25. Vega-Salas DE, Salas PJI, and Rodriguez-Boulan.E: Exocytosis 6. Gipson IK, Riddle CV, Kiorpes TC, and Spurr SJ: Lectin bind- of vacuolar apical compartment (VAC): A cell-cell contact ing to cell surfaces: Comparisons between normal and migrat- controlled mechanism for the establishment of the apical ing corneal epithelium. Dev Biol 96:337, 1983. plasma membrane domain in epithelial cells. J Cell Biol 7. Panjwani N, Moulton P, Alroy J, and Baum J: Localization of 107:1717, 1988. lectin binding sites in human, cat and rabbit corneas. Invest 26. Dilly PN and Mackie IA: Surface changes in the anaesthetic Ophthalmol Vis Sci 27:1280, 1986. conjunctiva in man, with special reference to the production of 8. Hazlett LD and Mathieu P: Glycoconjugates on corneal epithe- mucus from a non-goblet-cell source. Br J Ophthalmol 65:833, lial surface. Effect of neuraminidase treatment. J Histochem 1981. Cytochem 37:1215, 1989. 27. Greiner JV, Kenyon KR, Henriquez AS, Korb DR, Weidman 9. Trinkaus-Randall V and Gipson IK: A technique for obtaining TA, and Allansmith MR: Mucus secretory vesicles in conjunc- basal corneal epithelial cells. Invest Ophthalmol Vis Sci tival epithelial cells of wearers of contact lenses. Arch Ophthal- 26:233, 1985. mol 98:1843, 1980. 10. Kohler G and Milstein C: Continuous cultures of fused cells 28. Wells PA, Ashur ML, and Foster CS: SDS-gradient polyacryl- secreting antibody of predefined specificity. Nature 256:495, amide gel electrophoresis of individual ocular mucus samples 1975. from patients with normal and diseased conjunctiva. Curr Eye 11. Huff JC, Weston WL, and Wanda KD: Enhancement of spe- Res 5:823, 1986.

Downloaded from iovs.arvojournals.org on 09/27/2021