Quantification of MUC5AC protein in human tears.

PURPOSE: MUC5AC has been identified as a major secretory mucin of conjunctival goblet cells and precorneal tear film. However, no method has been reported to quantify MUC5AC protein in human tears. The objective of this study was to establish a method to measure the amount of MUC5AC in human tears and to correlate the amount of MUC5AC with age, gender, and dry eye diseases. METHODS: A goat antibody was raised to synthetic peptides corresponding to nonglycosylated epitopes of human MUC5AC mucin. This antibody and a horseradish peroxidase-coupled second antibody were used to develop a quantitative immunoassay to measure the MUC5AC concentration of tear samples collected on Schirmer strips. Porcine stomach mucin was used as a standard for the assay. The chemiluminescent MUC5AC signal was digitized and quantified. Tear samples from 19 healthy volunteers and 31 clinically diagnosed dry eye patients were analyzed. RESULTS: MUC5AC concentration in human tears ranged from undetectable to more than 200 microg/mL porcine stomach mucin equivalent. In the healthy population, low, moderate, and high concentrations were found in the tear samples from younger and older persons and from both men and women. The mean MUC5AC content in tears was lower in the dry eye patients than in the age- and gender-matched healthy individuals. CONCLUSIONS: A method was established to quantify MUC5AC in human tear samples obtained on Schirmer strips. There was no correlation between the amount of MUC5AC and age or gender in the healthy population. Dry eye disease patients, however, typically showed reduced concentrations of soluble MUC5AC in the tear film.

Quantification of MUC2 and MUC5AC transcripts in human conjunctiva.

PURPOSE: Transcripts of mucins 1, 4, and 5AC have been identified in human conjunctival tissue. Of these, only MUC5AC has been localized to goblet cells. MUC2 is a goblet cell mucin originally identified in the intestinal mucosa. The presence of MUC2 transcripts and levels of MUC2 and MUC5AC transcripts in normal human conjunctiva, as determined by quantitative polymerase chain reaction (PCR), is reported. METHODS: RNA from conjunctivae of six donors (3 men, 3 women, 44 to 69 years, all white) was isolated and subjected to competitive reverse transcription-PCR. Internal standards, which are dsDNA molecules with ends complementary to a given primer pair but containing nonhomologous central sequences, were prepared for each gene assayed. Titration of a constant amount of cDNA against serial dilutions of the internal standard allowed quantification of the template cDNA. MUC2 and MUC5AC levels were compared to levels of the "housekeeping" gene, beta2-microglobulin (beta2M). The identity of PCR products was confirmed by sequencing. RESULTS: In the six individual samples tested, beta2M mRNA is expressed, on average, at approximately 10(-20) moles per sample (1 microg RNA) or approximately 63.5x10(4) molecules. The mRNA encoding MUC5AC, a relatively abundant ocular mucin, exists at levels 10-fold lower than beta2M. In contrast to previous reports of MUC2 mRNA being absent at the ocular surface, these results show that MUC2 transcripts are present and expressed at levels 5900-fold lower than for MUC5AC. Apparently, MUC2 transcripts exist on the order of only approximately 100 to 1000 molecules/microg of RNA in the analyzed samples. CONCLUSIONS: MUC2 transcripts are present in human conjunctival tissue and their abundance is much lower than that of MUC5AC. This is the first application of competitive PCR to the quantitative analysis of mucin expression in human ocular tissue. The sensitivity of this method allows the detection of MUC2 transcripts that were not detected by Northern blot analysis or in situ hybridization in previous studies. It also makes possible the comparison of relative levels of expression for ocular mucins.

MUC5AC mucin is a component of the human precorneal tear film.

PURPOSE: Mucins are important structural and functional components of the precorneal tear film, yet little is known of their composition and synthesis. The mRNAs of MUC1, MUC4, and MUC5AC have previously been identified in human conjunctiva. Of these, only MUC5AC mRNA appears to be associated with goblet cells. The purpose of the this study was to quantify MUC5AC transcript levels, to identify MUC5AC protein in conjunctiva, tears, and goblet cells and to determine whether this mucin is secreted in response to the calcium ionophore ionomycin. METHODS: MUC5AC mRNA from normal human conjunctiva was identified, quantified, and compared with beta2-microglobulin levels using a competitive reverse transcription-polymerase chain reaction (RT-PCR) method. An antibody to a MUC5AC peptide was used to localize this mucin in conjunctival sections by immunohistochemistry. Anti-MUC5AC antiserum was used to label western blot analysis of conjunctiva and tears. Conjunctival tissues were incubated with ionomycin, and secreted mucins were detected with Helix pomatia agglutinin conjugated to horseradish peroxidase and with anti-MUC5AC antiserum. RESULTS: MUC5AC and beta2-microglobulin transcripts were expressed at a ratio of approximately 1:500. Immunochemical labeling showed that MUC5AC was localized in conjunctival goblet cells and at the apical surface of the conjunctival epithelium. MUC5AC protein was present in conjunctiva and in the tear film. Ionomycin stimulation of conjunctival secretion resulted in a fourfold increase in total mucin secretion and in a corresponding increase in secreted MUC5AC. CONCLUSIONS: MUC5AC is synthesized by goblet cells of the normal human conjunctiva, and this mucin is a component of conjunctival secretions and of normal human tears.

Regulation of ocular mucin secretion by P2Y2 nucleotide receptors in rabbit and human conjunctiva.

The effects of adenine analogues on secretion of high molecular weight, mucin-like glycoproteins (mucins) by conjunctival goblet cells were investigated using isolated rabbit and human conjunctiva. Mucin secretion was assayed using a quantitative dot-blot assay of Helix pomatia agglutinin-horseradish peroxidase binding to mucins absorbed to nitrocellulose filters. In rabbit conjunctiva, exogenous ATP (10(-7)-10(-3) m) induced a concentration-dependent, four-fold increase in mucin secretion that reached a plateau 15 min after drug addition. The rank order of potency of agonists was UTP>=ATPgammaS>ATP>ITP>ADP>>AMP. Adenosine, alpha,beta-methylene- ATP and beta,gamma-methylene-ATP were ineffective in stimulating mucin release. The response to ATP was unmodified by the putative P2 purinergic antagonists suramin or reactive blue (5x10(-5) m). In human conjunctiva, ATP and UTP were nearly equipotent in stimulating mucin secretion with an EC50 congruent with5x10(-6) m. These findings demonstrate that rabbit and human conjunctival cells contain functional P2Y2 (formerly designated as P2U) nucleotide receptors that govern mucin secretion. These receptors may provide useful pharmacological targets for therapeutic modulation of tear film mucins in dry-eye disorders and/or corneal wound healing.

PGE2 synthesis and response pathways in cultured corneal endothelial cells: the effects of in vitro aging.

PURPOSE: The purpose of these studies is to develop an in vitro model of corneal endothelial aging and to investigate age-related changes in morphology, mitosis, prostaglandin synthesis and prostaglandin response pathways. METHODS: First-passage rabbit corneal endothelial cells were grown in vitro for up to 30 days after subculture. PGE2 synthesis was measured by radioimmunoassay. EP2 receptors were evaluated, by determination of PGE2 stimulated by flow cytometry and by bromodeoxyuridine (BrdU) incorporation in subconfluent, confluent and injured cultures. RESULTS: Rabbit corneal endothelial cells become less dense and more irregular in shape as they age in culture, thus resembling their in vivo counterparts. PGE2 synthesis and response decrease with culture age. Injury results in enhanced PGE2 synthesis in both younger and older cultures. In younger cultures, injury also results in mitosis of cells at the wound margin, and this response is greatly diminished in older cultures. CONCLUSIONS: The morphologic and mitotic changes seen in rabbit corneal endothelial cultures in vitro resemble those seen as a consequence of aging in humans and rabbits. Prostaglandin synthesis and response pathways are modified as a result of aging and may play a role in the autocrine regulation of wound repair, especially in younger cells.

Corneal endothelial repair. Regulation of prostaglandin E2 synthesis.

PURPOSE: The authors have investigated the hypothesis that prostaglandin E2 (PGE2) synthesis is regulated during corneal endothelial wound healing. Previous studies have shown that PGE2 is an important mediator of endothelial mitosis, migration, and differentiation. METHODS: Biosynthesis of PGE2 was investigated in a wound closure model of the cultured rabbit corneal endothelium and in cultures treated with experimental agents. Prostaglandin E2 synthesis was measured by enzyme-linked immunosorbent assay. Pharmacologic experiments were designed to evaluate the contributions of protein kinases, phospholipase A2, and cyclooxygenase to endogenous PGE2 synthesis. RESULTS: Prostaglandin E2 synthesis is increased markedly in response to injury and is proportional to the extent of wounding. Biosynthesis of PGE2 returns to basal levels concurrently with recovery of the injury. Synthesis is dependent on the activities of protein kinase C (PKC), phospholipase A (PLA), and cyclooxygenase. Two forms of cyclooxygenase are present in corneal endothelial cells, and pharmacologic studies indicate that the activity of the COX 2 contributes to injury-dependent PGE2 synthesis. CONCLUSIONS: Prostaglandin E2 synthesis is increased in injured corneal endothelial cells. This synthesis is dependent on the coordinated regulation of PKC, PLA, and cyclooxygenase. Prostaglandin E2 synthesis presents an attractive target for pharmacologic manipulation of endothelial recovery from injury.

EP2-receptor stimulated cyclic AMP synthesis in cultured human non-pigmented ciliary epithelium.

Prostaglandins are locally produced mediators of physiologic function which act through specific receptors to influence cellular response pathways. We have conducted a series of experiments designed to obtain a preliminary pharmacologic characterization of the prostaglandin E2 receptor subtype mediating increased synthesis of adenosine 3',5'-cyclic monophosphate (cyclic AMP) in an SV40 transformed line of human non-pigmented ciliary epithelium (ODMC2). Human non-pigmented epithelial (NPE) cells were grown to confluency, preincubated with indomethacin (to prevent endogenous PGE2 synthesis) and with isobutyl-methyl-xanthine (to prevent breakdown of cAMP), challenged with PGE2 or receptor-selective prostanoids and extracted. Prostanoid-stimulated cAMP synthesis was determined using a protein binding assay and normalized to cellular protein. Prostanoid agonists known to be relatively selective for the EP2 receptor subtype stimulated cAMP synthesis by human NPE cells. EP1 and EP3, IP, FP and DP selective prostanoids had negligible effects on stimulation of cAMP synthesis. The results of these studies lead us to conclude that human NPE cells express receptors of the EP2 subtype which mediate PGE2 stimulated synthesis of cyclic AMP.

Autocrine regulation of corneal endothelium by prostaglandin E2.

PURPOSE: Previous studies have shown that prostaglandin E2 is synthesized by rabbit corneal endothelial cells in culture and that PGE2 acts to increase synthesis of adenosine 3'-5'-monophosphate (cyclic AMP) and to inhibit endothelial migration in response to experimental wounds. The present study was undertaken to identify endogenously produced prostanoids, to evaluate the effect of PGE2 on corneal endothelial cell cycle parameters, to examine PGE2 receptor effector coupling, and to examine the distribution of cyclooxygenase, a key enzyme regulating PGE2 synthesis, in cultured rabbit corneal endothelial cells. METHODS: Prostaglandin biosynthesis was evaluated by thin layer chromatographic analysis of metabolically labeled arachidonic acid products and by radioimmunoassay of PGE2. Corneal endothelial mitosis was examined by flow cytometric analysis of cycling cells and by direct cell counts. PGE2 receptor-mediated cyclic AMP synthesis was quantified by a protein binding assay, and cyclooxygenase was localized by immunohistochemistry. RESULTS: PGE2 is the major prostanoid produced by rabbit corneal endothelial cells, and its synthesis is enhanced by mitogenic stimulation and is blocked by the cyclooxygenase inhibitor, indomethacin. Inhibition of PGE2 synthesis results in a larger percentage of cycling cells and agonists selective for EP2 receptors reverse indomethacin-induced increases in cell density. Inhibition of endogenous PGE2 synthesis also results in upregulation of receptor-stimulated cyclic AMP synthesis. Cyclooxygenase is present in confluent endothelial cells and is particularly abundant in subconfluent cultures and in cells migrating to close an experimental wound. CONCLUSION: The synthesis of PGE2 is regulated in corneal endothelial cells, and this autocoid acts to inhibit endothelial mitosis and activity of prostaglandin receptors of the EP2 subtype.

Endothelin does not affect experimentally induced corneal neovascularization.

Endothelin, a potent vasoconstrictor, was found to be ineffective in the treatment of experimentally induced corneal neovascularization. Endothelin was administered topically, subconjunctivally, and intraluminally in serial concentrations ranging from 0.0005 to 5.0 micrograms/mL in New Zealand white rabbits without effect. Electron microscopy of the neovascular cornea revealed the vessels consisted only of endothelin and pericytes. Hence, the vessels were not responsive to endothelin because they lacked contractile smooth muscle.

Growth factors and corneal endothelial cells: I. Stimulation of bovine corneal endothelial cell DNA synthesis by defined growth factors.

Peptide growth factors and other physiological growth modifiers were evaluated for their ability to stimulate DNA synthesis in early passage cultures of bovine corneal endothelial cells (BCEC). Increasing concentrations of newborn bovine serum (0.5-10%) causes a progressive increase in DNA synthesis, which approached a plateau at 10% serum. Supplementing medium with 10% serum from different lots of newborn bovine serum or fetal bovine serum stimulated significantly different levels of DNA synthesis by BCEC. Addition of epidermal growth factor (EGF) (2 nM) to medium containing 10% newborn or fetal bovine serum further increased DNA synthesis. Dose-response curves for EGF, transforming growth factor-alpha, basic fibroblast growth factor (bFGF), and insulin-like growth factor I showed that each significantly stimulated high levels of DNA synthesis (200-700% increase) compared with BCEC cultured in serum-free medium. Vaccinia growth factor, insulin, and transforming growth factor-beta each significantly stimulated lower levels of DNA synthesis (30-200% increase), whereas nerve growth factor, multiplication stimulating activity, and platelet-derived growth factor all failed to significantly stimulate DNA synthesis above the level of serum-free medium. Other physiological growth modifiers were tested for their effects on DNA synthesis of BCEC. Transferrin and low levels of 3',5'-cyclic monophosphate (cAMP) stimulated very low levels of DNA synthesis (50% increase) whereas linoleic acid, high levels of selenium, or cAMP each inhibited DNA synthesis 25-75% below the level of BCEC cultured in serum-free medium. A series of eight formulations containing various combinations of EGF, FGF, insulin, transferrin, selenium, linoleic acid, retinoic acid, cAMP, heparin, and endothelial cell growth factor were tested for their mitogenic action on BCEC cultures. A formulation containing EGF, insulin, transferrin, selenium, and linoleic acid (EGF + ITSL) stimulated the highest level of DNA synthesis of BCEC, which was approximately 25% higher than the increase stimulated by addition of 10% newborn bovine serum. The formulation consisting of EGF + ITSL was also evaluated as a supplement to corneal storage media. Addition of EGF + ITSL to three corneal storage media (McCarey-Kaufman, K-Sol, CSM) significantly stimulated increases in cell numbers of approximately 50% above the unsupplemented corneal storage media. These results demonstrate that BCEC respond selectively to different defined peptide growth factors and physiological growth modifiers, and suggest that supplementation of corneal storage media with a defined formulation (EGF + ITSL) may enhance corneal endothelial cell density.

Growth factors and corneal endothelial cells: II. Characterization of epidermal growth factor receptor from bovine corneal endothelial cells.

Epidermal growth factor (EGF) is a potent mitogen for corneal endothelial cells and may play a role in endothelial wound healing. To further characterize the interaction of EGF with endothelial cells, we measured biochemical parameters of 125I-EGF binding to cultured bovine corneal endothelial cells (BCEC), determined the pattern of EGF-induced protein phosphorylation, and investigated the influence of retinoic acid (RA) and transforming growth factor beta (TGF-beta) on EGF-induced DNA synthesis and receptor levels. Binding of 125I-EGF to BCEC was dependent on time, reaching a plateau after approximately 2 h at 37 degrees C, was specific for EGF, and had high affinity (Kd = 100 pM) with approximately 21,000 receptors per cell. Cellular substrates for the EGF receptor kinase, which may function as initial second messengers for EGF, were detected by autoradiography of sodium dodecyl sulfate polyacrylamide gels of 32P-labeled BCEC proteins. EGF stimulated phosphorylation of 170, 37, 21 and 20-kDa proteins. Addition of 1 nM, 100 nM, and 10 microM RA to BCEC cultured in serum-free medium for 24 h progressively inhibited DNA synthesis by up to 80% compared with control cultures. However, when added in combination with 5 nM EGF, 1 nM and 100 nM RA synergistically stimulated DNA synthesis by up to 80% above the level of EGF stimulation without altering EGF receptor levels or binding affinity. Thus, short-term exposure of BCEC to RA potentiated EGF-stimulated DNA synthesis, most likely by acting at a postreceptor step.(ABSTRACT TRUNCATED AT 250 WORDS)

Growth factors and corneal endothelial cells: III. Stimulation of adult human corneal endothelial cell mitosis in vitro by defined mitogenic agents.

Human corneal endothelial cells (HCEC) do not mitose extensively in vivo after damage to the endothelial layer. However, HCEC will divide in vitro if cultured under appropriate conditions. We measured the ability of various sera, plasma, growth factors, and nutritional substances to stimulate mitosis of HCEC during 5 days of organ culture after a central freeze injury to the endothelium. Supplementation of a chemically defined medium (CDM) with 20% fetal human serum (FHS) induced significantly higher numbers of mitotic figures or labeled nuclei of human or cat corneas compared with paired corneas cultured in CDM alone. Furthermore, addition of 20% FHS produced more labeled nuclei than did addition of 20% fetal bovine serum or 20% adult human serum. Dialyzed fetal human serum failed to stimulate mitosis, indicating that one or more components of fetal human serum with molecular weight less than 12,000 are essential for mitosis. Human plasma also failed to stimulate mitosis, but an extract of human platelets significantly stimulated high levels of nuclear labeling, suggesting that growth factors contained in platelet granules were responsible for serum-stimulated mitosis of HCEC. Addition of 100 nM epidermal growth factor (EGF) or 10 microM insulin to CDM supplemented with low levels of adult human serum (0.5%) stimulated significantly higher numbers of labeled nuclei compared with paired corneas cultured with 0.5% adult human serum. Supplementation of corneal storage media (K-Sol and CSM) with a mixture of chemically defined agents consisting of EGF, insulin, transferrin, selenium, linoleic acid, and albumin stimulated significantly higher numbers of labeled nuclei compared with paired corneas cultured in the unsupplemented corneal storage media.(ABSTRACT TRUNCATED AT 250 WORDS)

Prostaglandin E2 effects on corneal endothelial cyclic adenosine monophosphate synthesis and cell shape are mediated by a receptor of the EP2 subtype.

Corneal endothelial cells synthesize prostaglandin E2 (PGE2), and this synthesis is necessary for the maintenance of the normal polygonal shape of these cells. A series of experiments was done to examine the receptor-effector mechanism responsible for PGE2-mediated effects on cultured rabbit corneal endothelium. When challenged with exogenous PGE2, endothelial cells synthesized cyclic adenosine monophosphate (AMP) in a dose-dependent manner, and this synthesis was not antagonized by AH6809. The synthetic agonist 11-deoxy-PGE1, but not sulprostone, stimulated increased cyclic AMP synthesis. The pharmacologic profile of the endothelial PGE2 receptor is therefore consistent with that of an EP2 receptor linked to activation of adenylate cyclase. The prostaglandin agonists were also tested for their ability to prevent cellular elongation in response to indomethacin. The PGE2, 11-deoxy-PGE1, and 16,16-dimethyl PGE1 prevented elongation, but sulprostone and PGF2 alpha did not. The authors conclude that rabbit corneal endothelium in culture expresses a specific PG receptor of the EP2 subtype which is coupled to cyclic AMP synthesis and is involved in the regulation of cell shape.

Transforming growth factor alpha and its receptor in neural retina.

Transforming growth factor alpha (TGF-alpha) stimulates mitosis of many ectodermal cells but has not previously been studied for its role in neural tissues such as retina. We examined bovine retina for the presence of TGF-alpha mRNA, TGF-alpha protein and for the presence and location of the TGF-alpha/EGF receptor. Biochemical studies demonstrated a high level (770 fmol/mg protein) of specific, high affinity (Kd = 2 nM) TGF-alpha/EGF receptors in membrane homogenates of neural retina, but undetectable binding to homogenates of retinal pigment epithelium. Light microscopic autoradiograms of sections of neural retinal tissue incubated with 125I-EGF indicated that specific TGF-alpha/EGF receptors were present on one or more cell types of the retina with the exception of the outer segments of the photoreceptor cells. In addition, retinal cells appear to synthesize TGF-alpha since both mRNA for TGF-alpha and TGF-alpha protein (4.2 ng/mg protein) were detected in retinal extracts using cDNA hybridization and TGF-alpha RIA techniques. The role(s) of TGF-alpha and its receptor in retina is unknown, but it is possible that they interact via an autocrine/paracrine mechanism to influence retinal regeneration, proliferative retinopathies or neural transmission.

Elimination of hydroxypropyl methylcellulose from the anterior chamber of the rabbit.

The elimination of hydroxypropyl methylcellulose (HPMC) of molecular weight 86,000 daltons (2%, viscosity 4,000 centipoises; 2.5%, 15,000 centipoises) and 120,000 daltons (2%, 15,000 centipoises) from the anterior chamber of rabbit eyes was determined colorimetrically and was nearly complete by 24 hours. The elimination rate was dependent on the viscosity of the HPMC injected into the eye. Elevation of the intraocular pressure by HPMC was detected three hours after injection, but the pressure was almost normal at six hours and was normal thereafter. This effect is independent of the molecular weight, the concentration, and the viscosity of the solutions used. Using an in vitro assay method, HPMC (86,000 daltons) at 2.5% concentration in tissue culture medium was innocuous to rabbit endothelial cells.

Pharmacological regulation of morphology and mitosis in cultured rabbit corneal endothelium.

Cultured rabbit corneal endothelial cells elongate when grown in the presence of epidermal growth factor (EGF) and indomethacin (INDO); whereas maintenance of the differentiated polygonal cell shape is apparently dependent upon endogenous synthesis of prostaglandin E2 (PGE2). In the current study, the authors demonstrate morphological changes in phenotypically altered cells and identify two intracellular pathways which interdependently regulate endothelial cells. Morphometric and mitotic analyses of cultures treated with a variety of pharmacological agents indicate that both protein kinases A- and C-dependent pathways regulate cell shape and cell division in corneal endothelial cells. Marked intracellular reorganization is associated with the morphological changes in the endothelial cells. When stained with rhodamine conjugated phallicidin, polygonal endothelial cells have circumferential bands of f-actin at their borders. EGF and/or INDO induce elongation and redistribution of f-actin into a diffuse cytoplasmic reticulum. Transmission electron microscopy demonstrates loss of several characteristic morphological markers for endothelial cells in response to pharmacologically induced elongation. The elongated cells lose intracellular junctions, apical/basal polarity and rough endoplasmic reticulum. These ultrastructural markers and circumferential f-actin bands are restored in cultures supplemented with exogenous PGE2. Modulation of these pathways in vivo may regulate cellular migration and mitosis during wound closure, stress, trauma and with age.

Corneal epithelial wound closure in tissue culture: an in vitro model of ocular irritancy.

The influence of 13 test agents on the ability of cultured rabbit corneal epithelial cells to migrate and re-cover a wound has been utilized to evaluate an in vitro model of ocular irritancy. Cells were grown under standard conditions and monitored for adequate cell density. Seven days after subculture, replicate wounds were produced and cultures were exposed to varying concentrations of a test agent in culture medium for 24 hr. Following exposure, cultures were fixed and stained to reveal remaining wound areas which were quantitated by computerized planimetry and compared to evaluate the deleterious effects of the test agents. The test ranks irritants in an order similar to that described in the literature for both in vivo and in vitro tests. This tissue culture model is conceptually simple, quantitative, and an alternative to the corneal component of whole animal testing for ocular irritancy.

Maintenance of corneal endothelial cell shape by prostaglandin E2: effects of EGF and indomethacin.

Confluent, cultured, rabbit corneal endothelial cells maintain a polygonal shape which is characteristic of these cells in vivo. When cultured in the presence of EGF (10 ng/ml) and/or indomethacin (1.0 microM), the endothelial cells have markedly different shapes at confluency. By morphometry, untreated cells are polygonal and have a maximum axis of 33 mu; EGF treatment causes a spindle-shaped elongation to 48 mu and indomethacin treatment causes a stellate-shaped elongation to 48 mu. There is a slight increase in cell density. When cells are cultured in the presence of both drugs, elongation is more pronounced to a fibroblastic appearing cell population, with maximum axes of 60 mu and more, but no additive increase in cell density. Continuity of cell borders is often lost. Corneal endothelial cells cultured in the presence of EGF, indomethacin, and PGE2 (0.5 microgram/ml) maintain their polygonal shape; PGF2 alpha is not effective at reversing the drugs' effects. Untreated and EGF-treated cells synthesize and release substantial quantities of PGE2 (2-4 ng/10(4) cells). Indomethacin completely inhibits PGE2 synthesis. It is concluded that PGE2 maintains the polygonal cell shape of the corneal endothelium in vitro and, perhaps, in vivo. The elongated forms of the cell may be related to migration and important in wound closure.