PURPOSE. Amniotic membrane (AM)-reconstructed conjunctival surfaces recover the normal epi-thelial phenotype with a significantly higher cell density than the control. The present study wasundertaken to examine how AM modulates rabbit conjunctival epithelial cell differentiation.
METHODS. Rabbit conjunctival epithelial cells (RCEs) were cultured on the basement membrane sideof dispase-pretreated AM, with or without seeding rabbit conjunctival fibroblasts (RCFs) on thestromal side. After 7 to 12 days, half of the cultures were raised to the air-liquid interface, and theremainder stayed submerged. A small group of air-lifted cultures containing RCFs was treated withretinoic acid. After 1, 2, and 4 weeks, cultures were terminated and processed for immunostainingwith antibodies directed against distinct types of mucins (SMC and AM3), glycocalyx (AMEM2),keratin K3 (AE5), and K12 (AK2). Additionally, western blot analysis was performed for K3 keratinexpression. Ultrastructural changes were evaluated by transmission electron microscopy.
RESULTS. In general, RCEs grown on AM were uniformly small, negative to AE5 and AK2 antibodies,and positive to AMEM2 and ASPG1 antibodies. Epithelial stratification and cell polarity withprominent microvilli, tight junctions, and hemidesmosomes were more pronounced in air-liftedcultures. RCEs cocultured with RCFs showed scattered AM3-positive goblet cells, which were notincreased by retinoic acid.
CONCLUSIONS. RCES cultured on AM primarily exhibit a nongoblet conjunctival epithelial phenotype.Epithelial stratification and cell polarity, features essential for epithelial differentiation, are pro-moted by air-lifting. This culture model will be useful for studying how growth and differentiationof conjunctival epithelial cells can be modulated further. (Invest Ophthalmol Vis Sci. 1999;40:878-886)
Epithelia covering the corneal and the conjunctival sur-faces are distinctly different. Although both are nonke-ratinized, stratified epithelia, the conjunctival but not
the corneal epithelium contains goblet cells. Their distinctepithelial phenotypes can be demonstrated by the expressionof different keratin pairs and types of mucin and glycocalyx.For example, the corneal epithelium expresses the keratin K3and K12 pair, which is distinctly different from the conjuncti-val epithelium, and keratins expressed by the corneal andconjunctival epithelia are distinctly different from those in thekeratinized skin epidermis.1'4 Conjunctival goblet cells secrete
From the 'Ocular Surface and Tear Center, Department of Oph-thalmology, Bascom Palmer Eye Institute, Miami; and the 2Departmentof Cell Biology and Anatomy, University of Miami School of Medicine,Florida.
Supported in part by Public Health Service Research GrantEY06819 from the U. S. Department of Health and Human Services,National Eye Institute, National Institutes of Health, Bethesda, Mary-land; an unrestricted grant from Research to Prevent Blindness, NewYork, New York; and a research fellowship grant (Me 1623/1-1) fromthe Deutsche Forschungsgemeinschaft (DM), Bonn, Germany.
Presented in part at the annual meeting of the Association forResearch in Vision and Ophthalmology, Fort Lauderdale, Florida, May1998.
Submitted for publication September 16, 1998; revised December3, 1998; accepted December 17, 1998.
Proprietary interest category: SCGT has filed a patent for themethod of preparation and clinical uses of amniotic membrane.
Reprint requests: Scheffer C. G. Tseng, Bascom Palmer Eye Insti-tute, William L. McKnight Vision Research Center, 1638 NW 10thAvenue, Miami, FL 33136.
a gel-forming mucin, MUC5A/C.5 Conjunctival nongoblet epi-thelial cells and corneal epithelial cells express two types ofmembranous mucins, MUC16 and sialomucin complex (SMC),7
and conjunctival nongoblet cells express MUC4 mucin.5 Inaddition, a set of glycoproteins constituting the glycocalyx hasalso been described for ocular surface epithelia.8"10 The loss ofgoblet cells,1 x change of keratin expression,3 and loss of mu-cin/glycocalyx expression12 are invariably found in squamousmetaplasia, a hallmark of different forms of dry eye and ocularsurface disorders. Therefore, it is important to understand howepithelial phenotypes of mucin expression and nonkeratiniza-tion are modulated to endow the ocular surface with sufficientmoisture and maintain ocular surface integrity.
As a first step toward these objectives, it is necessary toestablish an in vitro culture system to facilitate goblet celldifferentiation. Studies have shown that growth and differen-tiation of conjunctival epithelial cells can be modulated byvitamin A,13 by such matrix components as collagen gel andmatrigel,14 and by inclusion of conjunctival fibroblasts in anorganotypic culture.15
Amniotic membrane (AM), with its thick basement mem-brane and avascular stromal matrix, has recently been usedsuccessfully for ocular surface reconstruction in a variety ofocular surface disorders.16"23 Impression cytology has shownthat the reconstructed conjunctival surface fully recovers itsnormal conjunctival epithelial phenotype with an average2-fold increase in epithelial cell density and a 10-fold increasein goblet cell density, compared with densities of those cells incontrol samples.24 We thus wondered whether AM may be an
IOVS, April 1999, Vol. 40, No. 5 Conjunctival Epithelium on Amnlotic Membrane 879
FIGURE 1. Experimental design with AM. (A) A side view showinghow the AM was fastened to a culture plate insert with a nonabsorb-able suture; (B) a view from above the insert with the AM placed in a24-well tissue culture plate; (C) illustration of the setup of the in vitroreconstruction system for culturing conjunctival epithelial cells on thebasement membrane side and their own fibroblasts on the stromal sideof the AM.
ideal matrix substrate to establish an in vitro culture system.Herein, we report how AM was used as a natural substrate tomodulate the epithelial phenotype of rabbit conjunctival epi-thelial cells (RCEs), with or without cocultured rabbit conjunc-tiva] fibroblasts (RCFs).
MATERIALS AND METHODS
Animals
All procedures were performed according to the ARVO State-ment for the Use of Animals in Ophthalmic and Vision Re-search. New Zealand White rabbits of both sexes aged be-tween 4 and 6 months were used in all experiments. Beforeeuthanasia with an intravenous overdose of sodium pentobar-bital, they received an intramuscular injection of 50 mg xyla-zine hydrochloride and 50 mg ketamine hydrochloride.
Materials
Dulbecco1 s modified Eagle's medium (DMEM), HEPES-buffer,trypsin-EDTA, amphotericin B, and fetal bovine serum (FBS)were purchased from Gibco (Grand Island, NY). Dispase II andfluorescein isothiocyanate (FITQ- conjugated and affinity-puri-fied goat anti-mouse IgM antibody were obtained from Boehr-inger Mannheim (Indianapolis, IN). The IgG monoclonal anti-body AE5, recognizing the 64-kDa basic keratin K3 waspurchased from ICN (Costa Mesa, CA). The mouse monoclonalIgG antibody 15H10 directed against ASPGl of SMC was a kindgift from Kermit Carraway, University of Miami, Florida. TheFITC-conjugated goat anti-mouse IgG and IgM antibodies ad-
FIGURH 2. Phase contrast images of the in vitro reconstruction system and of air-lifted cultures with RCEs and RCFs 18 days after air-lifting. (A)One clay after seeding RCFs on the stromal side of the AM; (B) low magnification showing RCFs on the stromal side and RCEs on the basementmembrane side at day 4. Note the initial formation of small cohesive epithelial islands on the basement membrane of the AM. The same area ofthe culture is shown at high magnification with different focus on RCFs (C, black arroivbead and on RCEs D, white star). In (E) low-magnificationand (F) high-magnification photos, epithelial cells after 18 days of air-lifting are small, round, and compactly organized, liar, (A, C, D, F) 60 ju.ni;(B, E) 300 jmi.
S80 Meller and Tseng IOVS, April 1999, Vol. 40, No. 5
FIGURE 3- Semithin sections of air-lifted and submerged 50-day-old RCEs cocultured with RCFs (black stars'). Air-lifting was performed for 34 days.(A) Epithelial stratification and cell polarity were more pronounced in air-lifted cultures with small, round, and cuboidal basal cells (blackarrowheads). (B) Submerged cultures showed less stratification and cell polarity. Bar, 25 Jim.
sorbed with human serum proteins, gentamicin, hydrocorti-sone, mitomycin C, dimethyl sulfoxide, cholera toxin, andinsulin-transferrin-sodium selenite media supplement werefrom Sigma (St. Louis, MO). The mouse monoclonal antibodiesAK2 to K12 keratin,4'25 AM3 to conjunctival goblet cells,26 andAMEM2 to mucosal epithelial membrane-associated glycoca-lyx12 were of the IgM class and were developed in our labora-tory. The tissue culture plastic plates (24-well) were fromBecton Dickinson (Lincoln Park, NJ). Culture plate inserts usedfor the three-dimensional reconstruction culture system werefrom Millipore (Bedford, MA).
Isolation and Cultivation of Rabbit ConjunctivalEpithelial Cells
Immediately after rabbits were euthanatized, the entire sheetof conjunctiva was removed 1 mm from the glandular edge ofthe tarsal plate and 5 mm from the limbus, and the excessiveTenon's tissue was trimmed off. This sheet was rinsed threetimes with DMEM containing 50 jug/ml gentamicin and 1.25jxg/ml amphotericin B, placed on a sterile paraffin sheet, anddigested with dispase II (1.2 U/ml in Mg2+- and Ca2+-freeHanks' balanced salt solution) for 1 to 2 hours at 37°C inhumidified 5% CO2.
The loosened epithelial aggregates were dispersed fromthe surface by gentle pipetting several times with the originalmedium containing dispase II. Loosened RCEs were separatedinto single cells by a second digestion with 0.1% trypsin and
0.02% EDTA in HBSS for 5 minutes. The enzymatic reactionwas blocked with DMEM containing 10% FBS; single RCEswere obtained by slowly passing the cell-containing mediumthrough a 23-gauge needle 7 to 10 times and were collected bycentrifugation at 800^ for 4 minutes.
Preparation of Rabbit Conjunctival Fibroblasts
The remaining stromal conjunctival tissue was dissected intoexplants of approximately 2 X 2 X 2 mm3, placed on 100-mmtissue dishes, and covered overnight with a drop of FBS aloneor DMEM containing 10% FBS, the antibiotics listed earlier, and1 M HEPES buffer. One day later, 10 ml of the same mediumwas added, and explant cultures were incubated at 37°C in95% humidity and 5% CO2. The medium was changed every 3to 4 days. After cells reached subconfluence, RCFs were sub-cultured with 0.1% to 0.25% trypsin, 0.02% EDTA in HBSS at37°C with a 1:3 to 4 split for several passages.
Three-Dimensional Cultures Using AmnioticMembrane
The method of preparation of preserved human AM has beenreported.17 In the present study the tissue samples were kindlyprovided by Bio-Tissue (Miami, FL). In preliminary experimentsit was noted that attachment and growth of RCEs seededdirectly on the basement membrane side of the AM wereimpeded by the devitalized amniotic epithelium. To resolvethis problem, the AM was pretreated with dispase (1.2 U/ml in
IOVS, April 1999, Vol. 40, No. 5 Conjunctiva! Epithelium on Amniotic Membrane 881
FIGURE 4. Differences in microvilli of apical cell membrane. Trans-mission electron microscopy shows that apical microvilli (arroivheaclsin A and B, and stars in C) were more pronounced in air-lifted cultures(A) and (B) than in submerged sutures (C). Bars, (A) I fim; (B, C) 0.1
Mg2+- and Ca2+-free HBSS) for 15 to 30 minutes, followed bygentle scraping with a rubber policeman. With a nonabsorb-able suture, AM of approximately 1,5 X 1,5 cm2 was suturedonto a culture plate insert with the basement membrane facingup and was placed in a 24-well tissue culture plastic plate (Fig.1). RCEs were seeded at a density of 1.0 to 1.6 x 105 cells/culture insert on the basement membrane of the AM. In asubgroup RCFs were seeded at a density of 1.0 X 105 cells/culture insert on the stromal side of the AM 24 hours beforeseeding with RCEs. Cells were then cultured in a medium ofequal volume of HEPES-buffered DMEM containing bicarbonateand Ham's F12 supplemented with 0.5% dimethyl sulfoxide, 2ng/ml mouse epidermal growth factor, 5 /xg/ml insulin, 5jug/ml transferrin, 5 ng/ml selenium, 0.5 /Lig/ml hydrocorti-sone, 30 ng/ml cholera toxin A subunit, 5% FBS, 50 jLLg/mlgentamicin, and 1.25 jLtg/ml amphotericin B. The cultures wereincubated at 37°C in 5% CO2 and 95% air, and the medium waschanged every 2 to 3 days. After RCEs reached subconfluence,half of the cultures were air-lifted by lowering the air-liquidinterface to the level of the cultured epithelial cells. Aftercultures were air-lifted, the medium was changed daily. In asubgroup, retinoic acid prepared at a final concentration of10~H M in dimethylsulfoxide under amber light was applied,with the same cautious light condition used each time themedium was changed.
ImmunostainingSpecimens from submerged and air-lifted cultures were re-moved from the culture insert, and frozen 6-/xm-thick sectionswere prepared, using ornithine carbamoyltransferase as theembedding medium (Tissue-Tek, Sakura FineTEK, Torrance,CA). Sections were stained with hematoxylin-eosin and peri-odic acid-Schiff s reagent and subjected to immunofluores-cence studies. All sections were preincubated with goat serum(1:1000) to prevent nonspecific staining. After rinsing twice for5 minutes with PBS, sections were incubated with mousemonoclonal AM3 (1:300), AMEM2 (1:300), AE5 (1:100), AK2(1:500), and ASGP1 (1:400) for sialomucin complex. After twowashings with PBS for 5 minutes, they were incubated with anFITC-conjugated secondary antibody (goat anti-mouse IgG at1:100 for ASPG1 and AE5; goat anti-mouse IgM at 1:100 forAK2, AM3, and AMEM2). After two additional washings withPBS, sections were mounted with an antifade solution andanalyzed with a fluorescence microscope (Axiophot; CarlZeiss, Oberkochen, Germany).
Sodium Dodecyl Sulfate-Polyacrylamide GelElectrophoresis and Immunoblot AnalysisAfter 4 weeks, air-lifted and submerged epithelial cultures,with or without cocultured RCFs, were rinsed with PBS for 3minutes and homogenized in 1 ml 0.5% Triton in 50 mMTris/HCI (pH 7.4). The insoluble precipitate collected by cen-trifugation was extracted with 0.5 ml 9 M urea in 50 mMTris/HCI (pH 7.4), containing a mixture of protease inhibitors.The same volume of the urea-soluble fraction was mixed with4% sodium dodecyl sulfate (SDS) and 0.6 M dithiothreitol. Thesamples were subjected to 7% SDS-polyacrylamide gel electro-phoresis and immunoblotted with the monoclonal antibodyAE5. Normal rabbit corneal epithelium and rabbit conjunctiva!epithelium cultured in air-lifted organotypic cultures4 express-ing K3 keratin were used as positive control samples. Amnioticmembrane without cultured cells was used as a negative con-trol.
Transmission Electron MicroscopySelected specimens were fixed in 2% glutaraldehyde and pro-cessed for conventional transmission electron microscopy.Samples were rinsed in 0.1 M phosphate buffer (pH 7.3),postfixed in 1% osmium tetroxide, and embedded in Epon.Ultrathin sections were cut and conventionally stained withuranyl acetate and lead citrate and examined with an electronmicroscope (model 420; Philips, Eindhoven, The Netherlands).Semithin sections were stained with 1% methylene blue, 1%azure II, and 1% borax.
RESULTS
Epithelial MorphologyRCFs attached to the stromal side of the AM within 4 to 24hours and exhibited typical morphology (Fig. 2A). Most RCEsalso attached to the basement membrane side within 24 hours,predominantly as single cells. After 3 to 4 days, a cohesive cellsheet was formed by small and uniform epithelial cells (Fig,2B). At high magnification, RCEs and RCFs could be distin-guished by their characteristic cell shapes (Figs. 2C, 2D, re-spectively, when viewed at the same site on different focal
882 Meller and Tseng IOVS, April 1999, Vol. 40, No. 5
FIGURE 5- Differences in intercellular contacts. Transmission electron microscopy shows that intercel-lular contacts among epithelial cells in the stratified layers were eiilianced in air-lifted cultures (A), (B)compared with those in submerged cultures (C), which showed wide intercellular spaces. Intercellulardigitation (arrows) and desmosome formation {arrowheads) contributed to a cohesive, compactlyorganized epithelial sheet in air-lifted cultures. Bars, (A) 0.1 /am; (B, C) 1 fxm.
planes). RCFs were spindle-shaped and were found on thestromal side, whereas RCEs were small and round and werelocated on the basement membrane side.
Under all conditions tested, confluence of epithelial cellswas usually observed within 7 to 14 days, when air-lifting wasintroduced. Air-lifted cultures showed a higher proliferationrate than did submerged cultures, especially when retinoicacid was added (not shown). Stratified cell layers developedprimarily from small epithelial cell islands noted in the earlystage in air-lifted cultures. Submerged cultures with or withoutfibroblasts were comparably less proliferative and did not re-veal obvious stratification under phase contrast microscopy.The morphology of RCEs did not change significantly in 18days (Figs. 2E, 2F) or even up to 41 days (not shown) afterair-lifting. Most of the RCEs remained uniformly small andshowed little signs of senescence (i.e., intracytoplasmic vacu-olation or desquamation) for up to 60 days (the longest timetested) of culturing in all groups.
In general, RCEs were stratified into more cell layers whenair-lifted compared with stratification in those that remainedsubmerged. Furthermore, air-lifted cultures with RCFs exhib-ited more stratification than those without RCFs. A represen-tative sample of air-lifted and submerged cultures with RCFsafter 50 days of culturing is shown in Figure 3. In these samplesair-lifting was performed for 34 days after 16 days of sub-merged culturing. Besides the visualization of more epithelialcell layers, it was noteworthy that basal cells of the air-liftedculture were more cuboidal than those of the submergedcultures.
Transmission electron microscopy showed that air-liftedcultures showed more prominent cell polarity in formation ofmicrovilli on the apical cell membrane, the extent of intercel-lular interdigitations, and the density of intracellular desmo-somes and hemidesmosomes (Figs. 4, 5, 6). Microvilli coatedwith glycocalyx materials were more pronounced in air-liftedcultures (Figs. 4A, 4B) than were those found in submergedcultures (Fig. 4C). Intercellular contacts among epithelial cellsin the stratified layers were enhanced in air-lifted cultures (Figs.5A, 5B) compared with contacts seen in submerged cultures(Fig. 5C), which showed wide intercellular spaces. This wasattributed to an increase in intercellular digitation and desmo-some formation that resulted in a cohesive and compactlyorganized epithelial sheet in air-lifted cultures. Adjacent to thebasement membrane of the AM, the air-lifted culture showedmore hemidesmosomes (Fig. 6A, 6B) than did the submergedcultures (Fig. 6C, 6D). Furthermore, in air-lifted cultures, basalepithelial cells were typically cuboidal to round shaped, andcontained a large number of mitochondria and endoplasmicreticulum (Fig. 6B). Suprabasal cells also had a high content ofthese organelles (Fig. 5A). Intermediate to superficial cells hada denser amount of cytoplasmic filaments than did basal cells(cf. Figs. 5A and 6A). Superficial epithelial cells became flat-tened and less electron dense, and in some areas became roundto oval shaped with multiple vesicles (Fig. 3A). The latter celltype was not found in the submerged cultures. In air-lifted andsubmerged cultures, no goblet cells with characteristic ultra-structural features were found.
IOVS, April 1999, Vol. 40, No. 5 Conjunctival Epithelium on Amniotic Membrane 883
f W* *v**£ *
J.J
FtGURE 6. Differences in formation of hemidesmosomes. Transmission electron microscopy shows thathemidesmosomes (arrows) were more pronounced in air-lifted cultures (A, B) than in submerged cultures(C, D). Bars, (A, C) 0.1 /xm; (B, D) 1 fim.
Epithelial Phenotype
To verily the epithelial phenotype, we used a panel of mono-clonal antibodies to keratins, mucins, and glycocalyx glyco-proteins in immunofluorescence staining. To verify their anti-genic epitopes, we first stained normal rabbit ocular surfaceepithelia in vivo. As reported,27 AE5, which recognizes K3keratin, stained the full thickness of the corneal epithelium(Tig. 7B), but not the conjunctival epithelium (Fig. 7A). AE5antibody did not stain the RCEs cultured on the AM, no matterwhether air-lifting was conducted or whether RCFs were in-cluded (Tig. 8B). This result was further confirmed by westernblot analysis (Fig. 8A). Normal in vivo corneal epithelium (la-beled as CO in Fig. 8A) and RCEs grown in air-lifted organo-typic cultures with RCFs (Fig. 8A, OT and A+F+) werestrongly positive for 64-kDa keratin 3, and submerged organo-typic cultures were weakly positive for keratin 3 (labeled asFig. 8A, OT and A—F+). This finding was consistent with thosepreviously reported/' In contrast, no AM cultures, with orwithout air-lifting and RCFs, expressed keratin 3 (Fig. 8A).Immunostaining for K12 keratin by AK2 was negative in thenormal conjunctival epithelium in vivo and in AM cultures (notshown).
ASGP1 recognizes SMC, a newly found transmembranemucin expressed by corneal epithelium and conjunctival non-goblet epithelial cells in vivo in rats.7 This antibody stainednongoblet epithelial cells and the membranous portion of thegoblet celts of the rabbit conjunctival epithelium (Fig. 7C) andmost of the RCEs grown on AM (Fig. 8C). AMEM2 recognizes
the mucin-like glycoconjugates found in the glycocalyx of allwet mucosal epithelia, including ocular surface epithelia.10
AMEM2 stained nongoblet epithelial cells of the rabbit conjunc-tiva (Fig. 7D) and stained most epithelial cells of RCEs grownon AM (Fig. 8D), a pattern resembling that of ASGP1 (Fig. 8C).AM3 recognizes conjunctival goblet cell-secreted mucins andstains rabbit conjunctival goblet cells in vivo (Fig. 7E). On crosssections AM3 stained no cells (not shown) but detected a fewscattered goblet cells grown on AM when prepared as flat-mounts (Figs. 8E, 8F). Air-lifting or the addition of fibroblasts orretinoic acid did not significantly affect the latter observation(not shown). Collectively, these results indicate that the result-ant phenotype of RCEs grown on AM retained conjunctivalorigin and was predominantly that of the nongoblet epithelialcell. Goblet cell differentiation was not frequently observed.
DISCUSSION
The major finding reported is that RCEs grown on AM maintaina conjunctival nongoblet epithelial phenotype. Epithelial dif-ferentiation and cell polarity are promoted by air-lifting withcocultured RCFs. We think that such a culture system will beuseful for exploring how goblet cell differentiation can bepromoted further. New information thus derived will be valu-able for understanding the mechanism by which squamousmetaplasia develops in various ocular surface and tear disor-ders.
884 Meller and Tseng IOVS, April 1999, Vol. 40, No. 5
FIGURE 7. Fmmunostaining using monoclonal antibodies AE5, ASGP1,AMEM2, and AM3 in the normal rabbit ocular surface. AE5 did not stainthe conjunctival epithelium (A) but stained the full-thickness cornealepithelium (B). ASGP1 stained the superficial epithelial cells of non-goblet epithelial cells and the membranous portion of some gobletcells (C). AMEM2 stained the superficial cell layers of all nongobletepithelial cells (D). AM3 stained selectively all goblet cells (E). Bar, (A,B, C, D, E) 100 ju,m.
It has been recognized that the basement membrane fa-cilitates migration of epithelial cells,28 reinforces adhesion ofbasal epithelial cells,29'30 promotes epithelial differentia-tion,- '̂~3'' and prevents epithelial apoptosis.3536 Collectively,these effects may explain why we observed that RCEs culturedon AM were uniformly small and round and could be main-tained for up to 50 days in vitro without notable signs ofsenescence (Fig. 2). This finding is consistent with the clinicaldata from impression cytology that AM-reconstructed conjunc-tival surfaces contain uniformly small basal epithelial cells withtwice the cell density of age- and sex-matched normal controlsamples.24 We are currently testing the hypothesis that the AMcan preserve epithelial progenitor cells and prolong their lifespan.
Increased epithelial stratification and cell polarity wasobserved when RCEs cultured on AM were further exposed tothe air-liquid interface, an effect termed air-lifting (Fig. 3). Thisfinding resembles that noted earlier in an organotypic culture,
in which a native type I collagen gel impregnated with RCFswas used.4 Although the exact mechanism by which air-liftingtriggers epithelial stratification remains unknown, a stark con-trast in the resultant epithelial phenotype is noted between AMcultures and reported organotypic cultures. Stratified RCEs inair-lifted AM cultures exhibited a genuine conjunctival pheno-type with negative AE5 staining and absence of K3 keratinexpression by western blot analysis (Fig. 8) and negative AK2staining (not shown). In contrast, as reported earlier,4 stratified
CO OT OT AM AM AM AM AM
A+F+ A-F+ A+F+ A-F+ A+F- A-F- no
71 Kd-
FIGURE 8. Immunoblotting for AE5 and immunostaining for AE5,ASGP1, AMEM2, and AM3 in RCEs grown on AM (A). Normal cornealepithelium in vivo (CO) and organotypic cultures (OT) with RCFs andair-lifting (A+F+) and OT with RCFs but without air-lifting (A-F+)were used as positive control samples. All expressed K3 (64 kDa)keratin. In contrast, all AM cultures with or without air-lifting (A+, A—)or with or without RCFs (F+, F—) did not express K3 keratin. Amnioticmembrane alone (no, last lane) was used as a negative control sample.AE5 was negative (B). ASGP1 (C) and AMEM2 (D) stained nearly allepithelial cells of frozen sections intensely. AM3 stained some scat-tered goblet cells in a flatmount preparation (E, F). Bar, (B, C, D, E)100 /xm; (F) 25 jam.
IOVS, April 1999, Vol. 40, No. 5 Conjunctival Epithelium on Amniotic Membrane 885
RCEs in air-lifted organotypic cultures showed positive AE5staining and K3 keratin expression, and submerged organo-typic RCEs cultures showed weakly positive keratin expres-sion, findings consistent with those in our previous report.4 NoAM cultures expressed K3 keratin (Fig. 8) or K12 keratin (notshown).
Although the K3 and K12 keratin pair has been consideredthe cornea-type differentiation marker,2'27 our reported datapoint out that K12 keratin is more specific than K3 keratin inthis regard.4 K4 and K13 keratins, which are thought to bespecific for nonkeratinized stratified epithelia, are also ex-pressed in the conjunctival epithelium more abundantly thanin the corneal epithelium.37'38 They were not investigated inthe present study. Taken together, these results indicate thatAM cultures permit RCEs to express more conjunctival epithe-lial phenotype than do organotypic cultures. This differencemay be attributable to the unique property of the basementmembrane of the AM. The basement membrane has beendescribed by Kurpakus et al.33 to influence the phenotype andcytokeratin expression of ocular surface epithelia.
We also observed that air-lifting further promoted expres-sion of such ultrastructural features as microvilli of the apicalmembrane, intercellular junctions, and density of desmosomesand hemidesmosomes (Figs. 4, 5, 6). Because the resultantepithelial phenotype responded positively ASGP1 and AMEM2antibodies (Fig. 8), markers for nongoblet epithelial cells, wethink that such morphologic changes of epithelial stratificationand polarity are therefore correlated with the biochemicalphenotype of the nongoblet conjunctival epithelial phenotype.Future studies are needed to determine which component orcomponents in the basement membrane are responsible forcontrolling the genetic expression of such an epithelial phe-notype.
In the currently defined culture conditions, goblet celldifferentiation marked by positive AM3 staining was occasion-ally found (Fig. 8). It has been recognized that in vitro differ-entiation of goblet cells requires stringent culture conditions indifferent epithelial tissues (for review see reference 15). Solu-ble factors such as calcium and thyroxine promote epithelialdifferentiation in general.39"0 Vitamin A13 and factors increas-ing cyclic adenosine monophosphate production, such as chol-era toxin, are known to induce an increased secretion ofmucin.4' In contrast, hydrocortisone has been shown to inhibitthe differentiation of goblet cells in duodenal explants.40 In thepresent study, addition of retinoic acid did not stimulate gobletcell differentiation.
Besides soluble factors, the extracellular matrix may playan important role. When the AM is used as a natural substrate,cultured tracheal epithelial cells predominantly express ciliawithout goblet cells.42 This result, resembling that described inthe present study, indicates that the basement membrane alonemay not be sufficient to induce goblet cell differentiation.Because AM-reconstructed conjunctival surfaces actually showa 10-fold increase in goblet cell density compared with densityin the control sample, determined by impression cytology,24
we wonder whether additional mesenchymal elements areneeded to promote goblet cell differentiation. This view issuggested by studies of tracheal epithelial cells, which losegoblet cell differentiation when cultured on plastic dishes butreexpress goblet cells when reseeded on the denuded anddevitalized tracheal lumen.43'44 This finding strongly suggeststhat stromal fibroblasts may play an important role in modulat-
ing goblet cell differentiation. Our preliminary data indicatethat RCFs added to the same side of the basement membranepromoted goblet cell differentiation much more than whenadded to the stromal side (Meller and Tseng, unpublishedresults). Therefore, future studies will be directed towardstudying whether the thick basement membrane of the AMmay have impeded the epithelial-mesenchymal interactionneeded to support goblet cell differentiation.
Acknowledgments
The authors thank Karl Meller from the Department of Cytology,Institute of Anatomy, Bochum, Germany, for preparing the specimenfor ultrastructural analysis and Kermit Carraway from the Departmentof Cell Biology and Anatomy, University of Miami, Florida, for provid-ing the monoclonal antibody to the sialomucin complex ASGP1.
References
1. Moll R, Franke WW, Schiller DL, Geiger B, Krepler R. The catalogof human cytokeratins. Patterns of expression in normal epithelia,tumors and cultured cells. Cell. 1982;31:ll-24.
2. Galvin S, Loomis C, Manabe M, Dhouailly D, Sun T-T. The majorpathways of keratinocyte differentiation as defined by keratinexpression, an overview. Adv Dermatol. 1989;4:277-299.
3. Tseng SCG, Hatchell D, Tierney N, Huang AJW, Sun T-T. Expres-sion of specific keratin markers by rabbit corneal, conjunctival,and esophageal epithelia during vitamin A deficiency. J Cell Biol.1984;99:2279-2286.
4. Chen WYW, Mui M-M, Kao W\V-Y, Liu C-Y, Tseng SCG. Con-junctival epithelial cells do not transdifferentiate in organotypiccultures: expression of K12 keratin is restricted to corneal epithe-lium. Curr Eye Res. 1994;13:765-778.
5. Inatomi T, Spurr-Michaud SJ, Tisdale AS, Zhan Q, Feldman ST,Gipson IK. Expression of secretory mucin genes by human con-junctival epithelia. Invest Ophthalmol Vis Sci. 1996;37:l684-1692.
6. Inatomi T, Spurr-Michaud SJ, Tisdale AS, Gipson IK. Human cor-neal and conjunctival epithelia express MUC1 mucin. Invest Oph-thalmol Vis Sci. 1995;36:1818-1827.
7. Price-Schiavi SA, Meller D, Jing X, Carvajal ME, Tseng SCG, Carr-away KL. Sialomucin complex at the rat ocular surface: a newmodel for ocular surface protection. Biochem J. 1998;335:457-463.
8. Jones DT, Monroy D, Pflugfelder SC. A novel method of tearcollection: comparison of glass capillary micropipettes with po-rous polyester rods. Cornea. 1997; 16:450-458.
9. Watanabe H, Fabricant M, Tisdale AS, Spurr-Michaud SJ, LindbergK, Gipson IK. Human corneal and conjunctival epithelia producea mucin-like glycoprotein for the apical surface. Invest Ophthal-mol Vis Sci. 1995;36:337-344.
10. Pflugfelder SC, Tseng SCG, Yoshino K, Monroy D, Felix C, Reis BL.Correlation of goblet cell density and mucosal epithelial mem-brane mucin expression with rose bengal staining in patients withocular irritation. Ophthalmology. 1997;104:223-235.
11. Tseng SCG. Staging of conjunctival squamous metaplasia by im-pression cytology. Ophthalmology. 1985;92:728-733-
12. Pflugfelder SC, Tseng SCG, Sanabria O, et al. Evaluation of subjec-tive assessments and objective diagnostic tests for diagnosing tear-film disorders known to cause ocular irritation. Cornea. 1998; 17:38-56.
13- Clark JN, Marchok AC. The effect of vitamin A on cellular differ-entiation and mucous glycoprotein synthesis in long-term rat tra-cheal organ cultures. Differentiation. 1979;l4:175-183.
14. Tsai RJF, Tseng SCG. Substrate modulation of cultured rabbitconjunctival epithelial cells. Invest Ophthalmol Vis Sci. 1988;29:1565-1576.
886 Meller and Tseng IOVS, April 1999, Vol. 40, No. 5
16. Kim JC, Tseng SCG. Transplantation of preserved human amnioticmembrane for surface reconstruction in severely damaged rabbitcorneas. Cornea. 1995;l4:473-484.
17. Lee S-H, Tseng SCG. Amniotic membrane transplantation for per-sistent epithelial defects with ulceration. Am JOphthalmol. 1997;123:303-312.
19. Shimazaki J, Yang H-Y, Tsubota K. Amniotic membrane transplan-tation for ocular surface reconstruction in patients with chemicaland thermal burns. Ophthalmology. 1997;104:2068-2076.
20. Shimazaki J, Shinozaki N, Tsubota K. Transplantation of amnioticmembrane and limbal autograft for patients with recurrent ptery-gium associated with symblepharon. BrJ Ophthalmol. 1998;82:235-240.
21. Tsubota K, Satake Y, Ohyama M, et al. Surgical reconstruction ofthe ocular surface in advanced ocular cicatricial pemphigoid andStevens-Johnson syndrome. Am] Ophthalmol. 1996; 122:38-52.
22. Tseng SCG, Prabhasawat P, Lee S-H. Amniotic membrane trans-plantation for conjunctiva] surface reconstruction. AmJ Ophthal-mol. 1997;124:765-774.
23. Tseng SCG, Prabhasawat P, Barton K, Gray T, Meller D. Amnioticmembrane transplantation with or without limbal allografts forcorneal surface reconstruction in patients with limbal stem celldeficiency. Arch Ophthalmol. 1998;ll6:431-44l.
24. Prabhasawat P, Tseng SCG. Impression cytology study of epithelialphenotype of ocular surface reconstructed by preserved humanamniotic membrane. Arch Ophthalmol. 1997;115:1360-1367.
25. Liu C-Y, Zhu G, Converse R, et al. Characterization and chromo-somal localization of the cornea-specific murine keratin geneKrtl.I2.J Biol Chem. 1994;260:24627-24636.
26. Huang AJW, Tseng SCG. Development of monoclonal antibodiesto rabbit ocular mucin. Invest Ophthalmol Vis Sci. 1987;28:l483-1497.
27. Schermer A, Galvin S, Sun T-T. Differentiation-related expressionof a major 64K corneal keratin in vivo and in culture suggestslimbal location of corneal epithelial stem cells. / Cell Biol. 1986;103:49-62.
28. Terranova VP, Lyall RM. Chemotaxis of human gingival epithelialcells to laminin. A mechanism for epithelial cell apical migration.J Periodontol. 1986;57:311 -317.
29. Khodadoust AA, Silverstein AM, Kenyon KR, Dowling JE. Adhesionof regenerating corneal epithelium. The role of basement mem-brane. AmJ Ophthalmol. 1968;65:339-348.
30. Sonnenberg A, Calafat J, Janssen H, et al. Integrin a6/j34 complexis located in hemidesmosomes, suggesting a major role in epider-
mal cell-basement membrane adhesion./ Cell Biol. 1991;113:907-917.
31. Guo M, Grinnell F. Basement membrane and human epidermaldifferentiation in vitro. / Invest Dermatol. 1989;93:372-378.
32. Streuli CH, Bailey N, Bissell MJ. Control of mammary epithelialdifferentiation: basement membrane induces tissue-specific geneexpression in the absence of cell-cell interaction and morpholog-ical polarity./ Cell Biol. 1991;115:1383-1395.
33. Kurpakus MA, Stock EL, Jones JCR. The role of the basementmembrane in differential expression of keratin proteins in epithe-lial cells. Dev Biol. 1992;150:243-255.
34. Barcellos-Hoff MH, Aggeler J, Ram TG, Bissell MJ. Functionaldifferentiation and alveolar morphogenesis of primary mammarycultures on reconstituted basement membrane. Development.1989;105:223-235.
35. Boudreau N, Sympson CJ, Werb Z, Bissell MJ. Suppression of ICEand apoptosis in mammary epithelial cells by extracellular matrix.Science. 1995;267:891-893.
36. Boudreau N, Werb Z, Bissell MJ. Suppression of apoptosis bybasement membrane requires three-dimensional tissue organiza-tion and withdrawal from the cell cycle. Proc Natl Acad Sci .1996;93:35OO-3513.
37. Kurpakus MA, Maniaci MT, Esco M. Expression of keratins K12,K4, and K14 during development of ocular surface epithelium.CurrEyeRes. 1994;13:805-8l4.
38. Krenzer KL, Freddo TF. Cytokeratin expression in normal humanbulbar conjunctiva obtained by impression cytology. Invest Oph-thalmol Vis Sci. 1997;38:142-152.
39. Black BL, Smith JE. Regulation of goblet cell differentiation bycalcium in embryonic chick intestine. FASEBJ. 1989;3:2653-2659.
40. Black BL, Moog F. Goblet cells in embryonic intestine: accelerateddifferentiation in culture. Science. 1977;197:368-370.
41. Virmani AK, Naziruddin B, Desai VC, Lowry JP, Graves DC, Sach-dev GP. Evidence for secretion of high molecular weight mucinsby canine tracheal epithelial cells in primary culture: effects ofselect secretagogues on mucin secretion. In Vitro Cell Dev Biol.1992;28A:120-127.
42. Noguchi Y, Uchida Y, Endo T, et al. The induction of cell differ-entiation and polarity of tracheal epithelium cultured on the am-niotic membrane. Biochem BiophysRes Commun. 1995;210:302-309.
43. Wu R, Martin WR, Robinson CB, et al. Expression of mucin syn-thesis and secretion in human tracheobronchial epithelial cellsgrown in culture. AmJ Respir Cell Mol Biol. 1990;3:467-478.
44. Inayama Y, Hook GE, Brody AR, et al. In vitro and in vivo growthand differentiation of clones of tracheal basal cells. Atn J Patbol.1989;134:539-549.
