Cell Surface Glycoproteins of Corneal Epithelium

Noorjahan Panjwani, Sameer Ahmad, and Michael B. Raizman

Purpose. To identify those plasma membrane glycoproteins of corneal epithelial cells that are synthesized in a higher amount or are downregulated during cell migration. Methods. Primary cell cultures of rabbit corneal epithelium were used. Sialic acid and terminal galactose/N-acetylgalactosamine residues of plasma membrane glycoproteins of migrating and nonmigrating corneal epithelial cells were labeled using two well-characterized cell surface carbohydrate labeling techniques. The labeled glycoproteins were extracted in phosphate- buffered saline containing nonionic detergents and various protease inhibitors, and then they were analyzed by sodium dodecyl sulfate polyacrylamide gel electrophoresis and fluorography. Results. At least 12 to 13 radiolabeled components (molecular weight, 240 kd to 21 kd) were detected in the fluorographs of both sialic acid and galactose-labeled cells. Regardless of the labeling technique used, one sialoglycoprotein (CP1, 240 kd) was found in a higher amount in the extracts of nonmigrating cells than in migrating cells, and another glycoprotein (GP12, 28 kd) was present in a higher amount in migrating than in nonmigrating cells. Furthermore, one sialoglycoprotein (GP13, 21 kd) was detected only in migrating cells, and two glycoproteins (GP10, 42 kd; GPU, 32 kd) with terminal galactose/N-acetylgalactosamine residues were present in a higher amount in migrating than in nonmigrating cells. Conclusions. This study has demonstrated that during corneal epithelial cell migration, the level of one membrane glycoprotein is markedly reduced, and the levels of four membrane glycoproteins are elevated. Further characterization of these glycoproteins should contribute to a better understanding of the mechanisms that modulate corneal epithelial sheet migration and wound healing. Invest Ophthalmol Vis Sci. 1995;36:355-363.

After accidental injury to the cornea and various sur- Studies aimed at identifying individual membrane gical procedures, initial healing of the wound takes glycoproteins of corneal epithelium have shown that place by the migration of adjacent cells to the injured integrins, which are widely expressed transmembrane area.12 The failure of epithelium to migrate to the glycoproteins and are known for their role as recep- wound surface or the failure of migrated epithelium tors that mediate attachment of cells to extracellular to remain adherent to the substratum may result in matrix and cell-cell adhesion, are localized in the severe clinical problems, leading to the development cornea along the basal cell membrane of corneal epi- of persistent epithelial defects and corneal ulcer- thelium at the site of cell-substrate interactions and ation.3'4 Molecular events that mediate migration of cell-cell contact.14"17 Other known plasma membrane epithelium to the wound and adhesion of epithelium glycoproteins of corneal epithelium include bullous to the underlying substrata have not been fully eluci- pemphigoid antigens (BPAG 1+2), found along the dated, but a number of studies have suggested that basal membrane of corneal epithelium,'819 and uvo- plasma membrane glycoconjugates of corneal epithe- morvulin,20 also known as E-cadherin, which is a mem- lium play a central role in these events.5"13 ber of a family of cell adhesion molecules that regulate cell-cell contact.

From the New England Eye Center,and the Department of Ophthalmology, Tufts It is well established that plasma membrane glyco- University School of Medicine, Boston, Massachusetts. conjugates of different cell types are diverse and cell- Supported by National Institutes of Health grants EY07088 and EY09349. Submitted for publication April 15, 1991; revised July 22, 1994; accepted October type specific. Integrins, BPAGs, and cadherins most 3, 1994. probably constitute only a portion of cell surface glyco- Proprietary interest category: N. Reprint requests: Noorjahan Panjwani, Department of Ophthalmology, Tufts proteins of corneal epithelium, and it is most likely University School of Medicine, 136 Harrison Avenue, Boston, MA 02111. that other glycoproteins unrelated to any known com-

Downloaded from iovs.arvojournals.org on 09/24/2021 356 Investigative Ophthalmology & Visual Science, February 1995, Vol. 36, No. 2

3 ponents are also present and play an important role groups with NaB H4. Various reaction times and re- in the regulation of cell function in health and disease. agent concentrations were obtained from parameters To understand better the cell surface architecture determined by Panjwani and Baum.23 Cells were la- of the corneal epithelium with respect to glycoprotein beled directly on the monolayer in 100-mm dishes by constituents, two well-characterized cell surface carbo- sequential treatment with 45 /zmols of NaBH4 for 15 hydrate labeling techniques were used in this study to minutes at room temperature, NaIO4 at a final concen- identify the plasma membrane glycoproteins of pri- tration of 1 mM at 4°C for 10 minutes, and 2 or 3 mCi 3 mary cell cultures of rabbit corneal epithelium. To of NaB H4 (5 to 15 Ci/mmol) for 5 minutes at room identify the specific molecules that may be involved temperature. The cell membranes were solubilized by in cell-cell and cell-matrix interactions, electropho- incubating the labeled cells with 0.5% CHAPS or 0.5% retic patterns of plasma membrane glycoproteins of Triton X-100 in phosphate-buffered saline containing confluent-nonmigrating cell cultures were compared various protease inhibitors for 1 hour at 4°C with gen- with those of migrating cell cultures, and it was shown tle shaking. The solubilized membrane extracts were that during cell migration, levels of four 3-[(3-Cho then centrifuged at 12,000gfor 15 minutes in a Sorvall lamidopropyl) dimethylammonio]-l-propanesulfo- centrifuge to remove any insoluble material, the su- nate (CHAPS)-extractable glycoproteins are elevated, pernatants were transferred to Centricon-10 tubes and the level of one glycoprotein is markedly reduced. (Amicon, Beverly, MA) for further centrifugation at 6000g to remove detergent and salts, and the labeled glycoproteins were analyzed by sodium dodecyl sulfate MATERIALS AND METHODS (SDS) polyacrylamide gel electrophoresis. Control cells underwent the same labeling procedure except Preparation of Rabbit Corneal Epithelial Cell that the periodate step was omitted. Because similar Cultures results were obtained regardless of whether the mem- Rabbit corneal epithelial cells were grown in tissue branes were solubilized by CHAPS or Triton X-100, culture using eyes from Pel-Freeze Biologicals (Rog- with the exception of a few initial experiments, the ers, AR),)" as described by Jumblatt and Neufeld.21 membranes of the labeled cells were solubilized with Epithelial fragments from corneas were cultured in buffers containing CHAPS in all studies, mainly be- 100-mm dishes using supplemental hormonal epithe- cause it can be removed more efficiently than Triton lial medium. Within 3 days of starting the culture, X-100 by dialysis or centrifugation in Centricon tubes. approximately 30% to 50% of the dish was populated with cells migrating away from the explants. These Neuraminidase Treatment of Cell Surface cells were designated as migrating (M) corneal epithe- Sialoglycoproteins of Corneal Epithelial Cells lial cells. Within 10 to 12 days, 90% to 95% of the To ensure that the label was introduced specifically to dish was populated with tightly packed polygonal cells, cell surface sialic acid residues, membrane sialoglyco- designated as nonmigrating (N) corneal epithelial 3 proteins were radiolabeled with NaIO4/NaB H4 treat- cells. Migrating cells were also obtained by treating ment as described above and were then treated with monolayer, nonmigrating cultures with Dispase II (2 neuraminidase (1 U, Type VI from Clostridium per- mg/ml) (Boehringer Mannheim, Indianapolis, IN) fringens; Sigma, St. Louis, MO) or with enzyme buffer for 20 minutes at 37°C. Cells from one dish were re- alone for 1 hour at room temperature. After enzyme plated onto three dishes and were allowed to spread treatment, cell membranes were solubilized, as de- for 16 to 18 hours, at which time approximately 50% scribed above, using the CHAPS buffer and were elec- to 70% of the dish was populated with migrating cells. trophoresed. Radiolabeling of Plasma Membrane Affinity Chromatography of Membrane Sialoglycoproteins of Corneal Epithelial Cells Glycoproteins of Corneal Epithelial Cells on Sialic acid residues of epithelial membrane glycopro- Agarose-Bound Wheat Germ Agglutinin 3 teins were radiolabeled by sodium periodate (NaIO4)- To ensure further that the NaIO4/NaB H4 treatment 3 tritiated sodium borohydride (NaB H4) treatment us- specifically introduced the label to sialic acid residues, ing the Gahmberg and Andersson method,22 with an the radiolabeled sialoglycoproteins were chromato- additional prereduction step to minimize nonspecific graphed on a wheat germ agglutinin (WGA)-agarose labeling.23 This procedure is known to label externally column expected to bind specifically to glycoproteins exposed cell surface sialic acid residues specifically containing sialic acid and N-acetylglucosamine resi- and involves cleavage of carbons 8 and 9 of the sialic dues.24 For this, CHAPS-solubilized material from 3 acid molecule by NaIO4, leaving an aldehyde group at NaIO4-NaB H4-treated corneal epithelial cells was ap- carbon 7 followed by reduction of CHO to HC3HO3H plied to the WGA-agarose column (0.7 X 2.0 cm) and

Downloaded from iovs.arvojournals.org on 09/24/2021 Corneal Epithelial Cell Surface Glycoproteiiis 357

Protein Determination The total protein concentration for each preparation was determined by the bicinchoninic acid protein assay (Pierce, Rbckford, IL) using bovine serum albumin as a standard.

Polyacrylamide Gel Electrophoresis in Sodium Dodecyl Sulfate

Electrophoresis was carried out according to the pro: cedure of Laemmli*6 in reducing conditions using 10% separating gels (70-mm X 80-mm X 1.5-mm) and 4% stacking gels. The gels were stained with Coomas- sie blue to visualize protein components and then were processed for fluorography. The following proteins were added as molecular weight standards: myo- sin (200 kd), phosphorylase b (97 kd), bovine serum albumin (69 kd), ovalbumin (46 kd), carbonic anhy- drase (30 kd), soybean trypsin inhibitor (22 kd), and lysozyme (14 kd).

RESULTS Cell Cultures Under phase-contrast microscope, confluent/nonmigrating cultures appeared to contain polygonal cells with marked contact inhibition, whereas sparse/mi- FIGURE 1. {lop) Migrating corneal epithelial cell cultures. Bar grating cell cultures contained mostly nonpolarized = 100 /x. (bottom) Nonmigrating corneal epithelial cell cul- cells (Fig. 1). tures. Bar = 450 fi. Radiolabeling of Membrane Glycoproteins of Migrating and Nonmigrating Corneal was allowed to incubate for 20 minutes. The columns Epithelium were washed with 15 ml of equilibration buffer (50 Cell viability, by trypan blue exclusion test, was at least mM Tris-HCl, pH 7.8, 0.15 M NaCl, 0.2% CHAPS), 99% at the end of labeling. Treatment with 0.5% and subsequently the bound glycoproteins were eluted with 0.3 M N-acetylglucosamine made in the equilibration buffer. Fractions of 1 ml were collected TABLE l. Incorporation of Tritium From and analyzed for radioactivity using 10-//1 aliquots. 3 NaB H4 Into Periodate- or Galactose Fractions comprising peaks were pooled, dialyzed, and Oxidase-Treated Migrating and electrophoresed as described below. The entire affin- Nonmigrating Rabbit Corneal ity chromatography procedure was carried out at 4°C. Epithelial Cells Radiolabeling of Terminal Galactose Residues Cells Treatment DPM X Kf/fig Protein of Cell Surface Glycoproteins of Corneal 3 Nonmigrating NaiO4/NaB H4* 7.0 ± 1.3 (N = 6) Epithelial Cells Migrating NaIO4/NaB*H4* 14.0 ± 1.8 (N = 4) Nonmigrating Galactose oxidase/ 4.9 ± 0.8 (N = 3) To label terminal galactose-N-acetylgalactosamine 3 NaB N4t residues of the plasma membrane glycoproteins of mi- Migrating Galactose oxidase/ 10:6 ± 3.9 (N = 5) grating and nonmigrating rabbit corneal epithelial NaB^f 3 cells in culture, the procedure described above to la- Nonmigrating NaB H4* 1.0 ± 0.3 (N = 3) 3 bel the cell surface sialoglycoproteins was used. How- Migrating NaB H4f 3.9 (N = 2) ever, instead of treatment with NaIO4, the cells were Values are given as means ± SEM. N = number of preparations. incubated with galactose oxidase (5 U, Dactylium den- 3 * 2 mCi of NaB H4 was used for reduction. 25 3 droides; Sigma) at 37°C for 1 hour. f 3 mCi of NaB H4 was used for reduction.

Downloaded from iovs.arvojournals.org on 09/24/2021 358 Investigative Ophthalmology & Visual Science, February 1995, Vol. 36, No. 2 CHAPS solubilized approximately 45% of total pro- C N M C N M tein. The radioactivity incorporated in the NaIO4/ 3 3 NaB H4- and galactose oxidase/NaB H4-treated cells 200- was always greater than in the control cells treated with 3 NaB H4 alone (Table 1). In spite of prior reduction of the cells with cold NaBH4, some nonspecific incorporation of radioactivity was always observed in the control cells, which, however, were found to contain no radiolabeled components upon electrophoresis (see below).

Polyacryiamide Gel Electrophoresis of Plasma Membrane Glycoproteins of Migrating and Nonmigrating Corneal Epithelium We analyzed electrophoretic patterns of CHAPS solu- 3 Coomassie Blue Fluorography bilized material of NaIO4/NaB H,rtreated migrating (M) and nonmigrating (N) cells to identify the major membrane sialoglycoproteins of rabbit corneal epithelium and to determine if the alteration in membrane sialoglycoproteins occurs during cell migration. Fluo- N M rography of the SDS-PAGE gels of material derived from M cells revealed at least 13 components (GP1- -GP12 GP13, molecular weight 21 kd to 240 kd, Fig. 2). Com- parison of the fluorography patterns of M and N cells revealed significant differences: component GPl (240 kd) was present in a higher amount in N cells than -GPl 3 in M cells; component GPl2 (28 kd) was present in a higher amount in M cells compared to N cells; and component GP13 (21 kd) was detected only in the fluorograph of M cells (Fig. 2). Differences between M and N cells, with respect to GP12 and GP13 were consistent in all four preparations of M and N cells FIGURE 2. Polyacryiamide gel electrophoresis in sodium do- analyzed in this study. GPl, which appeared as an decyl sulfate of radiolabeled membrane sialoglycoproteins intense band in fluorographs of N cells in all four of migrating and nonmigrating rabbit corneal epithelial preparations analyzed in this study, was either unde- cells in culture. Migrating (M) and nonmigrating (N) cells 8 tected (three preparations) or was detected only as a were treated with NaIO4/NaB H4,and control cells (C) were 3 weak band (one preparation) in M cells. Several other treated with NaB H4 only as described in the Methods sec- differences between the fluorography patterns of M tion. The radiolabeled cell surface glycoproteins were solubi- and N cells (for example, the presence of GP4 in lized in a buffer containing 0.5% CHAPS and were electro- higher amounts in N cells than in M cells; Fig. 2) phoresed on 10% gels using 54 /xg of protein in each sample. are not discussed here because they were not found Protein components were visualized by Coomassie blue consistently in fluorographs of all preparations ana- staining (top left), and the same gel was subsequendy pro- lyzed. Similar fluorography patterns of the labeled cessed for fluorography to visualize radioactive components (top right). The molecular weight scale is based on the elec- cells were obtained regardless of whether the cell trophoretic mobility of reduced standard proteins. The gly- membranes were solubilized in buffers containing Tri- coproteins GPl through GPl 3 are designated in order of ton X-100, or CHAPS, and whether the migrating cells increasing mobility on the gel. Note that: (1) no radiola- used were from explant cultures or were prepared by beled components were detected in fluorography patterns subculture of primary confluent cells with a split ratio of nonmigrating (C) or migrating (not shown) control cells 3 of 1:3 (passage 1). Fluorography of the control (C) treated with NaB H4 alone, (2) GPl is present only in N 3 cells that underwent only NaB H4 treatment did not cells, (3) GP12 is present in a higher amount in M cells reveal any bands, whereas Coomassie blue staining than in N cells, and (4) GPl3 is present only in M cells. To 3 3 visualize GP13 more clearly, some gels were overexposed pattern of NaB H4- as well as NaIO4/NaB H4-treated cells was identical (Fig. 2). These observations suggest (bottom). Note that GPl3 is detected only in M cells even that the components detected in the NaIO /NaB3H - when gels are overexposed. BPB = bromophenol blue 4 4 tracking dye; CHAPS = 3-[(3-Cholamidopropyl) dimethyl- treated cells were specifically labeled, and the C cell ammonio]-l-propanesulfonate.

Downloaded from iovs.arvojournals.org on 09/24/2021 Corneal Epithelial Cell Surface Glycoproteins 359

the enzyme buffer alone before extraction with the UJ CHAPS buffer and electrophoresis. Treatment with neuraminidase removed 50% of the radioactivity. All HN components, with the exception of a portion of GP- 12, were susceptible to the enzyme treatment (Fig. 3), 3 IS B suggesting that NaIO4/NaB H4 treatment specifically Origin — labeled the cell surface sialic acid residues of epithelial cells. The neuraminidase-resistant fraction of GP12 CO 200- also migrated faster on electrophoresis gels than the 6 fraction that was susceptible to the enzyme. The Coo- massie blue staining pattern of samples treated with 97- the enzyme and the control samples incubated with 69- the enzyme buffer alone were indistinguishable (not shown). Affinity Chromatography of Sialoglycoproteins O 46- of Corneal Epithelium on WGA-Agarose Lectin LU 3 To confirm that NaIO4/NaB H4 treatment specifically labeled the cell surface sialic acid residues, the CHAPS-solubilized material obtained from NaIO4/ oc 3 NaB H4-treated cells was chromatographed on an 30- agarose-bound lectin column, which is expected to bind specifically to glycoconjugates containing sialic acid and N-acetylglucosamine residues.24 Electropho- uLU resis and fluorography of the bound and unbound O material eluted from the WGA-agarose column revealed that, with the exception of a portion of GP-12, 22_ all components bound to the WGA-agarose column (Fig. 4), providing additional evidence that NalO4/ 3 BPB~ NaB H4 treatment specifically labeled the sialoglyco- FIGURE 3. Neuraminidase treatment of cell surface sialogly- proteins. In the unbound fraction, only a small por- coproteins of rabbit corneal epithelium in culture. Mem- tion of GP12 was detected. Again, GP12 eluted in the brane sialoglycoproteins of migrating epithelial cell cultures were radiolabeled with NalCXi/NaB'^FU treatment as de- unbound fraction migrated faster than that eluted in scribed in the Methods section and were then treated with the bound fraction. neuraminidase or with enzyme buffer alone for 1 hour. After Polyacrylamide Gel Electrophoresis of enzyme treatment, cell membranes were solubilized and 3 were electrophoresed in 10% gels using 37 ng of protein in Galactose Oxidase/NaB H4 Labeled Migrating each lane. Note that, with the exception of a portion of and Nonmigrating Corneal Epithelial Cells GP12 (arrow), no other components were resistant to en- To identify the major membrane glycoproteins of rab- zyme treatment. bit corneal epithelium that contain oligosaccharides with terminal galactose/N-acetylgalactosamine residues and to determine whether alterations in these 3 H represents volatile radioactivity. It is known that glycoproteins occur during cell migration, we ana- during the reduction with borohydride, volatile radio- lyzed electrophoretic patterns of CHAPS-solubilized activity representing tritiated hydrogen and water is 3 material of galactose oxidase/NaB H4-treated M- and generated, and the amount of such radioactivity pres- N cells. The most dramatic difference between the ent after the reaction varies depending on the lot of electrophoretic patterns of M and N cells was that 3 2327 NaB H4 crystal used in the reaction. GP1, which appeared as an intense band in the N cells, was barely detectable in the M cells (Fig. 5) in all Effect of Neuraminidase Treatment on three preparations. Furthermore, three components Radiolabeled Sialoglycoproteins of Corneal (GP12, 28 kd; GPU, 32 kd; and GP10, 42 kd) were Epithelium present in a higher amount in M cells than in N cells. 3 To ensure that the NaIO4/NaB H4 treatment specifi- Although these three components were present in a cally labeled the cell surface sialic acid residues, the higher amount in M cells than in N cells in all the radiolabeled cells were treated with neuraminidase or preparations analyzed, their relative levels varied

Downloaded from iovs.arvojournals.org on 09/24/2021 360 Investigative Ophthalmology & Visual Science, February 1995, Vol. 36, No. 2 N M N M D O 200- 00 Origin — 200- CC 46- 97- 5 o 69- UJ 30- O X U 46- BPB- LU CB FG CC 30 D a O N M LU _J o 22

BPB —GP12 WGA FIGURE 5. Polyacrylamide gel electrophoresis in sodium dodecyl sulfate of corneal epithelial cell surface glycoproteins Chromatography containing terminal galactose/N-acetylgalactosamine residues. Migrating (M) and nonmigrating (N) cells in culture FIGURE 4. Fluorography patterns of corneal epithelial cell were treated with galactose oxidase/NaB3H and the radiola- surface siajoglycoproteins eluted from the WGA-agarose col- 4 beled glycoproteins were solubilized in 0.5% CHAPS con- umn. Membrane sialoglycoproteins of •migrating epithelial 3 taining protease inhibitors as described in the Methods sec- cell cultures were radiolabeled with NaIO4/NaB H4 and were then chromatographed on a WGA-agarose column. tion. Electrophoresis was performed in 10% gels using 75 Unbound fraction was eluted with 0.2% CHAPS in 50 mM fig of protein in each sample. Protein components were Tris/HCl, pH 7.8, 0.15 M NaCl, and the bound fraction was visualized by Coomassie blue staining (CB, top left), and the eluted with the same buffer containing 0.3M N-acetylglucos- same gel was subsequently processed for fluorography to amine. Samples applied to each lane originated from 330 visualize radioactive components (FG, top right).Th e molec- jig of the original cell protein. Note that with the exception ular weight scale is based on the electrophoretic mobility of of a portion of GP12 (arrow), no components were seen in reduced standard proteins. The glycoproteins GP1 through the WGA unbound fraction. WGA = wheat germ agglutinin; GP12 are designated in order of increasing mobility on the CHAPS = 3-[(3-Cholamidopropyl) dimethylammonio]-l- gel. Note that GPJ. is present in a higher amount in N cells propanesulfonate. than in M cells, arid GP10, GPU, and GP12 are present in a higher amount in M cells than in N cells. To visualize GP10 to GP13 more clearly, some gels were overexposed (bottom). BPB = bromophenol blue; CHAPS = 3-[(3-Chol- amidopropyl) dimethylammonio]-l-propanesulfon'ate.

Downloaded from iovs.arvojournals.org on 09/24/2021 Corneal Epithelial Cell Surface Glycoproteins 361

among different preparations so that in one prepara- mined whether GP1 is a novel glycoprotein identified tion they were present in a moderately (Fig. 5, top) for the first time or whether it represents a known higher amount and in two preparations they were component. Furthermore, although we have assumed present in a significantly higher amount in M cells that the 240-kd glycoprotein labeled by both NaIO4/ 3 3 than in N cells (Fig. 5, bottom). NaB H4 and galactose oxidase/NaB H4 treatment is the same component, the possibility remains that both techniques have labeled different components of simi- DISCUSSION lar molecular weight. Although positive identification To understand the role of corneal epithelial cell sur- of GP1 must be made, it is unlikely to be BPAG1, face glycoproteins in cell migration and cell-cell and a known 230-kd glycoprotein found along the basal cell-matrix interactions, the present study focuses on membrane of corneal epithelium, because it could identifying major surface glycoproteins of migrating not be precipitated by human autoimmune bullous and nonmigrating corneal epithelium using two well- pemphigold antiserum that cross-reacts with the rabbit characterized cell surface carbohydrate labeling tech- antigen (Mohan et al, unpublished observations). Fur- niques. That the mild periodate treatment followed thermore, recent studies suggest that BPAG1 may be 3 2fl 30 by reduction with NaB H4 specifically labeled the cell a transmembrane glycoprotein; ' die intracellular surface sialic acid residues was demonstrated by the domain of many transmembrane glycoproteins forms fact that most components labeled by the technique a complex with molecules of cytoskeleton, and such were susceptible to neuraminidase treatment and re- complexes are not usually extractable, in buffers con- acted with WGA, a lectin known to bind specifically taining nonionic detergents (see below). Nor is GP1 to sialic acid and N-acetylglucosamine residues.24 A likely to be a subunit of fibronectin because it did not small portion of GP12 that was resistant to neuramini- bind to a gelatin-Sepharose (Pharmacia, Piscataway, dase and did not react with WGA may represent a NJ) column (Ahmed and Panjwani, unpublished ob- 28 modified form of sialic acid. The most commonly servations) . found sialic acid in higher animals is N-acetylneur- Another major finding of the present study is the aminic acid, but various modifications, such as hydrox- identification of several cell surface glycoproteins ylation of the N-acetyl group (N-glycolylneuraminic found in elevated amounts in migrating epithelium acid) and acetylation of N-acetyl- and N-glycolylneur- compared to nonmigrating epithelium. Regardless of aminic acid, are also found, though to a lesser de- the labeling technique used, GP-12 (28 kd) was found gree.28 The modified sialic acids differ with respect to in higher amounts in migrating epithelium than in their susceptibility to neuraminidases from different nonmigrating epithelium. In addition, a 21-kd sialo- sources.28 Heterogeneity within GP12 was also evident glycoprotein (GP13) was detected only in migrating by the observation that its neuraminidase-resistant epidielium, and two glycoproteins (GP10, 42 kd; fraction migrated faster on electrophoresis gels than GP11, 32 kd) containing oligosaccharides with termi- the fraction that was susceptible to the enzyme. nal galactose / N-acetylgalactosamine residues were The most significant finding of the present study detected in a higher amount in migrating than in is the identification of a 240-kd membrane glycopro- nonmigrating epithelium. Because GP12 and GP13 tein (GP1) that appeared as an intense band in fluo- were synthesized in elevated amounts by migrating rographs of nonmigrating epithelium but was barely epithelium even when the cells were cultured in the detectable in the extracts of migrating epithelium. presence of 5-flurouracil to inhibit cell mitosis (data 3 This glycoprotein was labeled by NaIO4/NaB H4, as not shown), they are likely to be related to cell migra- 3 well as galactose oxidase/NaB H4 labeling techniques, tion rather than cell mitosis. The glycoproteins, which indicating that it contains oligosaccharides with termi- were detected in a higher amount in migrating com- nal sialic acid and galactose/N-acetylgalactosamine pared to nonmigrating epithelium, may play a role in residues. Because GP1 is not detected in significant cell migration and cell-cell or cell-matrix interac- amounts until the cells reach confluency and it disap- tions. Many studies have shown that in sparse cell cul- pears again when confluent cells are induced to mi- tures, as well as in cells plated in low Ca++ culture grate by subculture of primary cells, it appears likely media to prevent the formation of cell-cell junction, that the synthesis of this glycoprotein is induced upon membrane glycoproteins, especially those that medi- establishment of cell-cell contact and that there is a ate cell-cell and cell-matrix interactions, are evenly downregulation of this component during cell migra- distributed over the entire cell surface and, as die cells tion. Downregulation of GP1 may be of significance reach confluency and become tightly packed, there is in events that mediate breakage of cell-cell contact a redistribution of these glycoproteins such that they in response to injury, an event prerequisite for the are localized at the site of cell-cell or cell-matrix initiation of reepithelialization. It remains to be deter- attachment.31"35 Many transmembrane glycoproteins

Downloaded from iovs.arvojournals.org on 09/24/2021 362 Investigative Ophthalmology & Visual Science, February 1995, Vol. 36, No. 2

are extractable in nonionic detergents from the sparse Key Words cell cultures in the absence of cell-cell contact; upon corneal epithelium, wound healing, cell migration, plasma reaching confluency and the establishment of cell- membranes, glycoproteins cell contact, the same glycoproteins form a complex with the molecules of cytoskeleton and become insolu- Acknowledgment 32 34 ble in nonionic detergents. " For example, in the The authors thank Dr. Nikhat Ahmed for performing the presence of extensive cell-cell contact, approximately cell viability study and for help with the estimation of molec- 35% of uvomorulin, a glycoprotein found at the cell- ular weight of GP1 using cross-linked hemocyanin standards cell junction, could be extracted from the MDCK cells, (210 kd, trimer; 280 kd, tetramer). They also thank Drs. whereas in the absence of extensive cell-cell contact, Zheng Zhao and Royce Mohan for helpful discussions. >90% of this glycoprotein is extracted in buffers con- References taining nonionic detergents.34 Properties of different plasma membrane glycoproteins vary with respect to 1. Hanna C. Proliferation and migration of epithelial their extractability in buffers containing nonionic de- cells. AmJOphthalmol. 1966; 61:55-62. 2. Pfister RR. The healing of corneal epithelial abrasions tergents. Unlike E-cadherin and a number of other in the rabbit: A scanning electron microscope study. plasma membrane glycoproteins that are easily ex- Invest Ophthalviol. 1975; 14:648-661. tracted from the membranes of sparse cell cultures in 3. Cavanagh HD, Pihlaja D, Thoft RA, Dohlman CH. buffers containing nonionic detergents, desmocollins, The pathogenesis and treatment of persistent epithe- the membrane glycoproteins found in desmosomes, lia] defects. TrAmAcadOphthalOtol. 1976;81:OP-754- are insoluble in the nonionic detergents regardless of 769. whether the cell-cell contacts have been induced.36 4. Cavanagh HD, Colley A, Pihlaja DJ. Persistent corneal Whether the plasma membrane glycoproteins of epithelial defects. Int Ophthalmol Clin. 1979; 197-206. 5. Gipson IK, Anderson RA. Effect of lectins on migra- rabbit corneal epithelium identified in this study using tion of the corneal epithelium. Invest Ophthalmol Vis cells grown in culture are direcdy relevant to the cells Sti. 1980; 19:341-349. in vivo remains to be established. In this study, only 6. Jorna A, Raizman M, Panjwani N. The role of corneal primary cell cultures were used, and dierefore there fucosylated glycoconjugates in corneal epithelial cell is good probability of preserving many of the in vivo migration. ARVO Abstracts. Invest Ophthalmol Vis Sti. characteristics of the cells. In a recent study, we dem- 1993;34:1553. onstrated that a glycolipid, paragloboside,12'13 is syn- 7. Gipson IK, Riddle CV, Kiorpes TC, Spurr SJ. Lectin thesized in a higher amount in migrating epithelium binding to cell surfaces: Comparisons between normal than in nonmigrating epithelium regardless of and migrating corneal epithelium. Dev Biol. 1983; whether the epithelia are prepared using primary cell 96:337-345. cultures, organ cultures, or in vivo techniques. Be- 8. McLaughlin BJ, Barlar EK, Donaldson DJ. Wheat germ agglutinin and concanavalin A binding during epithe- cause of difficulties in obtaining sufficient migrating lial wound healing in the cornea. Curr Eye Res. 1986; and nonmigrating epithelium from corneas in vivo 5:601-609. and in organ culture, we think diat further character- 9. Zieske JD, Higashijima SC, Gipson IK. Con A- and ization of plasma membrane glycoproteins identified WGA-binding glycoproteins of stationary and migra- in the present study using cells grown in culture will tory corneal epithelium. Invest Ophthalmol Vis Sti. provide tools to study die role of plasma membrane 1986;27:1205-1210. glycoproteins of corneal epithelium in an in vivo 10. Zieske JD, Higashijima SC, Spurr-Michaud SJ, Gipson model of corneal wound healing. IK. Biosynthetic responses of the rabbit cornea to a keratectomy wound. Invest Ophthalmol Vis Sti. 1987; In summary, the present study has identified a 28:1668-1677. number plasma membrane glycoproteins of corneal 11. Panjwani N, Michalopoulos G, Song J, Zaidi TS, Yo- epidielium and has demonstrated that during cell mi- geeswaran G, BaumJ. Neutral glycolipids of migrating gration, the level of one cell surface glycoprotein is and nonmigrating rabbit corneal epithelium in organ markedly reduced whereas the levels of four cell sur- and cell culture. Invest Ophthalmol Vis Sti. 1990; face glycoproteins are elevated. Isolation and charac- 31:689-695. terization of these glycoproteins is under way in our 12. Zhao Z, Jungalwala F, BaumJ, Panjwani N. Identifica- laboratory. Given their strategic location on the cell tion of paragloboside-related glycolipids of rabbit cor- surface, the glycoproteins identified in this study neal epithelium: Their potential role in cell migration. ARVO Abstracts. Invest Ophthalmol Vis Sa. could serve as mediators of important biologic pro- 1992; 33:891. cesses, including cell migration and cell-cell and cell- 13. Ahmad S,Jungalwala F, BaumJ, Panjwani N. Potential matrix interactions, both of which are important role of a corneal glycolipid paragloboside in corneal events in wound healing. epithelial cell migration: An immunohistochemical

Downloaded from iovs.arvojournals.org on 09/24/2021 Corneal Epithelial Cell Surface Glycoproteins 363

study. ARVO Abstracts. Invest Ophthalmol Vis Sci. surface galactose and galactosamine in glycolipid and 1992; 33:892. glycoprotein of human erythrocytes. / Biol Chem. 14. Stepp MA, Spurr-Michaud S, Tisdale A, Elwell J, Gip- 1973;248:4311-4317. son IK. Alpha6Beta4 integrin heterodimer is a compo- 26. Laemmli UK. Cleavage of structural proteins during nent of hemidesmosomes. Proc Natt Acad Sci USA. the assembly of the head of bacteriophage T4. Nature. 1990;87:8970-8974. 1970; 227:680-685. 15. Grushkin-Lerner LS, Trinkaus-Randall V. Localization 27. Baumann H, Doyle D. Turnover of plasma membrane of integrin and syndecan in vivo in a corneal epithelial glycoproteins and glycolipids of hepatoma tissue cul- abrasion and keratectomy. Curr Eye Res. 1991; 10:75- ture cells. JBiol Chem. 1978; 253:4408-4418. 85. 28. Varki A. Diversity in the sialic acids. Glycobiology. 16. Tervo K, Tervo T, van Setten GB. Virtanen I. Integrins 1992; 2:25-40. in human corneal epithelium. Cornea. 1991; 10:461- 29. Sawamura D, Li K, Chu ML, UittoJ. Human bullous 465. pemphigoid antigen (BPAG1): Amino acid sequences 17. Stepp MA, Spurr-Michaud S, Gipson IK. Integrins in deduced from cloned cDNAs predict biologically im- the wounded and unwounded stratified squamous epi- portant peptide segments and protein domains. J Biol thelium of the cornea. Invest Ophthalmol Vis Sci. Chem. 1991;266:17784-1779O. 1993;34:1829-1844. 30. Ikeda S, Yaguchi H, Ogawa H. Expression of extracel- 18. Cowen SJ, Anhalt GJ, Wicha MS, Sugar A, Labib RS, lular and intracellular bullous pemphigoid antigens Diaz LA. Distribution of pemphigus and pemphigoid at the dermal-epibolic junction in organ culture of antigens, laminin, and type IV collagen in corneal human skin. /Invest Dermatol. 1990;94:336-340. epithelium. Invest Ophthalmol Vis Sci. 1981;21:879- 31. Herzlinger DA, Ojakian GK. Studies on the develop- 882. ment and maintenance of epithelial cell surface polar- 19. Fujikawa LS, Foster CS, Gipson IK, Colvin RB. Base- ity with monoclonal antibodies. / Cell Biol. 1984; ment membrane components in healing rabbit cor- 98:1777-1787. neal epithelial wounds: Immunofluorescence and ul- 32. Ojakian GK, Schwimmer R. The polarized distribution trastructural studies./ Cell Biol. 1984;98:128-138. of an apical cell surface glycoprotein is maintained 20. Takahashi M, Fujimoto T, Honda Y, Ogawa K. Distri- by interactions with the cytoskeleton of Madin-Darby bution change of fodrin in the wound healing process canine kidney cells./ Cell Biol. 1988; 107:2377-2387. of the corneal epithelium. Invest Ophthalmol Vis Sci. 33. Shore EM, Nelson WJ. Biosynthesis of the cell adhe- 1992; 33:280-285. sion molecule uvomorulin (E-Cadherin) in Madin- 21. Jumblatt M, Neufeld A. Beta-adrenergic and seroto- Darby canine kidney epithelial cells. / Biol Chem. nergic responsiveness of rabbit corneal epithelial cells 1991;266:19672-19680. in culture. Invest Ophthalmol Vis Sci. 1983;24:1139- 34. Nelson WJ, Shore EM, Wang AZ, Hammerton RW. 1143. Identification of a membrane-cytoskeletal complex 22. Gahmberg CG, Andersson LC. Selective radioactive containing the cell adhesion molecule uvomorulin (E- labeling of cell surface sialoglycoproteins by perio- Cadherin), ankyrin, and fodrin in Madin-Darby ca- date-tritiated borohydride. / Biol Chem. 1977; 252: nine kidney epithelial cells. / Cell Biol. 1990; 110:349- 5888-5894. 357. 23. Panjwani N, Baum J. Rabbit corneal endothelial cell 35. Herz RE, Ojakian GK. Differential targeting of an epi- surface glycoproteins. Invest Ophthalmol Vis Sci. 1985; thelial plasma membrane glycoprotein in polarized 26:450-456. Madin-Darby canine kidney cells. / Biol Chem. 24. Bhavanandan VP, Katlic AW. The interaction of wheat 1989;264:4605-4612. germ agglutinin with sialoglycoproteins: The role of 36. Penn EJ, Hobson C, Rees DaMA. The assembly of sialic acid. / Biol Chem. 1979;254:4000-4008. the major desmosome glycoproteins of Madin-Darby 25. Gahmberg CG, Hakomori S-I. External labeling of cell canine kidney cells. FEBS Lett. 1989;247:13-16.

Downloaded from iovs.arvojournals.org on 09/24/2021