Human corneal epithelial cell primary cultures were obtained using leftover human donor peripheral corneas after transplantation (15). Under a tissue culture hood, Descemet's membrane, the endothelium, and the posterior stroma were removed under a dissecting microscope using forceps. The anterior cornea with intact epithelium was covered with 1.2 U/mL of Dispase II (Boehringer Mannheim, Mannheim, Germany) in calcium- and magnesium-free phosphate-buffered saline (PBS) and incubated at 37℃ in a humidified 5% CO₂ incubator for 1-hr. The epithelial cells were subsequently centrifuged at 1,000 revolutions per minute (rpm) for 5 min.

Primary cultures of corneal epithelial cells were prepared in 35-mm Petri tissue-culture dishes (Corning Incorporated, Corning, NY, USA), containing Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum, 20 ng/mL EGF, 100 U/mL penicillin, and 100 µg/mg streptomycin (all from Gibco BRL, Rockville, NY, USA). The cells were incubated at 37℃ under a humidified atmosphere of 95% air and 5% CO₂. The culture medium was changed every 2 or 3 days. Corneal epithelial cell growth commenced after 4-7 days and reached confluence within 21-28 days. Subsequently, the cells were enzymatically detached with 0.25% trypsin and 0.002% EDTA (Irvine Scientific, Santa Ana, CA, USA) at 37℃ for 10 min after washing once with Dulbecco's PBS (D-PBS) (Gibco BRL, Rockville, NY, USA). The suspended epithelial cells were centrifuged at 400 rpm for 10 min. The supernatant was removed and fresh medium was added. The number of cells in the cell suspension was counted in a hemacytometer and 5×10³ cells/well were plated in 96-well tissue culture plates. Second-passage human corneal epithelial cells were used in all experiments. The cells were incubated in 1 mL of culture medium at 37℃ (5% CO₂, 95% air) and were allowed to attach to the bottom of the well for 24-hr before 3 different NSAID eye solutions; diclofenac sodium (Ofenac®, Tajoon Pharmaceutical Ltd., Seoul, Korea), bromfenac sodium (Bronuck®, Senju Pharmaceutical Ltd., Osaka, Japan), pranoprofen (Pranoprofen®, Senju Pharmaceutical Ltd., Osaka, Japan) and the fluorometholone (FML®; Santen Inc., Osaka, Japan) as control were added. Since the effect of a drug on corneal epithelial cells can be underestimated when cells are grown too dense, the cells were cultured for approximately 4-5 days to ensure 80%-90% confluence.

To determine the proliferation rate of human corneal epithelial cells, the colorimetric tetrazolium salt 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyl tetrazolium bromide (MTT; Sigma, St. Louis, MO, USA) test was performed (11). The MTT assay is based on the generation of purple formazan from the methyl tetrazolium salt by the mitochondrial enzymes of viable cells. Cultured cells were seeded in 96-well culture plates at a concentration of 4×10³ cells/well and were allowed to form a monolayer within 24-hr. The cells were subsequently exposed to 150 µL of DMEM medium containing 1 of the 3 different NSAID eye solutions or the steroid drug 0.1% fluorometholone for 1-, 4-, 12-, and 24-hr. After drug exposure, the cells were washed twice with PBS and were incubated in culture medium for 24-hr, followed by the MTT assay. The balanced salt solution-treated group was used as a control. At the end of the incubation period, the MTT solution was carefully aspirated, with care taken not to disturb the purple formazan crystals formed at the bottom of each well. The formazan reaction product was dissolved by adding 150 µL dimethyl sulfoxide (Sigma, St. Louis, MO, USA) and the optical density of each well was measured using an automatic plate reader (Molecular Devices, Sunnyvale, CA, USA) at a 570- and 690-nm test and reference wavelength, respectively. In each experiment, 8 wells were used for each concentration. This procedure was repeated in triplicate (16).

Cellular metabolic activity was calculated using the means of the absorption rates at each exposure time and concentration. Cellular metabolic activity was calculated using the following formula: cellular metabolic activity (%)=absorption rate of each well/absorption rate of the control group×100. Data were analyzed using Kruskal-Wallis test and were considered statistically significant at P<0.05.

The lactate dehydrogenase (LDH) assay measures leakage of the cytoplasmic enzyme LDH into the extracellular medium. The presence of LDH in the cell culture medium represents cell membrane damage. For this assay, 4×10³ corneal epithelial cells/mL were seeded in each well of 96-microtiter plates. Twenty-four hours after cell seeding, cells were exposed to 1 of the 3 different NSAID eye solutions or the fluorometholone. The LDH titer was assessed at 1-, 4-, 12-, and 24-hr after addition of the agent. After 24-hr, the supernatant was collected from each well. The cell monolayer was subsequently treated with a cell lysis solution for 30 min at room temperature and the cells and the lysates were collected. LDH activity was measured in both the supernatant and cell lysate fractions using CytoTox 96, a nonradioactive cytotoxicity assay kit (Promega, Madison, WI, USA), per the manufacturer's instructions. The absorbance was determined at 490 nm with a 96-well plate enzyme-linked immunosorbent assay (ELISA) reader (Moleculae Devices, Sunnyvale, CA, USA). LDH activity, which is proportional to color intensity, was expressed as optical density. This procedure was repeated in triplicate. The effect of the balanced salt solution-treated group was used as a control. To evaluate the significance of the differences in cytotoxicity between the treatments, the data were analyzed by the Kruskal-Wallis test and were considered statistically significant at P<0.05.

The composition differences between the NSAID and fluorometholone eye solutions were analyzed using a LX-20 chemistry analyzer (Beckman Coulter, Fullerton, CA, USA). The pH and the osmolarity of the eye solutions were measured using the Metrohm 780 (Metrohm, Zofingen, Switzerland) and Micro-Sample Osmometer (Fiske Associate, Norwood, MA, USA), respectively. The 3 NSAID eye solutions as well as the fluorometholone eye solution were tested 5 times. Data were reported as the means±standard deviation of all tests. The Kruskal-Wallis test was used for statistical analysis with P<0.05 considered statistically significant.

A scratch-wound assay using human corneal epithelial cells was used to determine whether NSAIDs or the steroid could promote wound closure. Human corneal epithelial cells (5×10³ cells/mL/well), which were grown to confluence in 24-well plates, were seeded at 500 µL/well in 6-well plates, and were subsequently incubated in DMEM under a 5% CO₂/95% air atmosphere at 37℃. The human corneal epithelial cells were wounded by scratching the surface of the culture layer with a 100-µL pipette tip. The scratched human corneal epithelial cells were washed with fresh medium to remove detached cells and the cells were subsequently incubated in the medium in the presence of the NSAIDs or fluorometholone for 1-, 24-, and 48-hr under 1/10 diluted eye solution tested. Cell migration was measured using an inverted phase-contrast light microscope. The balanced salt solution-treated group was used as a control.

To ensure that wounds of a similar-size area were compared, multiple positioning marks were made at the center of the denuded surface with a small needle. Eighteen hours after wounding, the monolayers were fixed and stained, and the migrated cells' edges were imaged. The size of wound gapping was measured by ROI manager tool in Image J software (US National Institutes of Health, Bethesda, MD, USA). To have better recognition by ROI manager, corneal image was preprocessed with versatile magic wand tool in Image J to enhance the contrast between acellular gap zone. After preprocessing, acellular area was automatically selected by ROI manager and the number of pixels in selected area was calculated. This procedure was performed 10 times for the statistical analysis. The Mann-Whitney U-test was used for statistical analysis with P<0.05 considered statistically significant.

Changes in cellular morphology were investigated using transmission electron microscopy (TEM, JEOL 1200EX; Jeol Ltd., Tokyo, Japan) by a well-trained technician. Cells (5×10³ cells/mL/well) were grown to confluence in 24-well plates, were seeded at 500 µL/well in 6-well plates, and were subsequently incubated in DMEM under a 5% CO₂/95% air atmosphere at 37℃. After exposure to the 3 different NSAIDs and fluorometholone for 12- and 24-hr under 1/50 diluted eye solution tested, the cells were incubated at 37℃ for another 24-hr after being rinsed with D-PBS. The cells were fixed with 2% glutaraldehyde in 0.1 M of phosphate buffer (pH 7.4) for 12-hr and post-fixed with 0.1% osmium tetroxide for 2-hr. After rinsing with 0.1 M of phosphate buffer, the fixed cells were embedded in an Epon® 812 mixture (SPI Supplies/Structure Probe, West Chester, PA, USA) and ultrathin sections were cut, which were stained with uranyl acetate and lead citrate, and examined by TEM.

The study protocol was approved by the institutional review board (IRB) of Pusan National University Hospital (IRB No. E-2015067). Informed consent was waived by the IRB. All study conduct adhered to the tenets of the Declaration of Helsinki.

The MTT values among the NSAID eye solutions and the control at 1-hr of exposure were not significantly different. However, they insignificantly decreased with concentration and after 4-hr of exposure time. Especially, the cellular metabolic activity of diclofenac sodium or fluorometholone markedly decreased after 1/10-diluted eye solution for 12-hr of exposure compared to control (P=0.037, P=0.021), and was significantly maintained for 24-hr (P=0.032, P=0.018), compared with those of bromfenac sodium and pranoprofen (Fig. 1). However, there were not statistically significant difference of MTT value between diclofenac sodium and fluorometholone after 12-hr and 24-hr exposure, respectively (P=0.86, P=0.72).

Exposure to 1/100- and 1/50-diluted eye drops did not significantly affect the LDH titer at all exposure times measured compared to human corneal epithelial cell treated with BSS. Especially, the LDH titers at 1/10-diluted fluorometholone and diclofenac significantly increased after 12-hr exposure rather than other two NSAIDs (P=0.029, P=0.039). The LDH titers of diclofenac sodium markedly increased after up to 24-hr of exposure (P=0.041), but that of fluorometholone was the highest among eye solutions tested (P=0.024) (Fig. 2).

The preservative tested in the three NSAID and fluorometholone was 0.5% chlorbutanol in diclofenac sodium and 0.005% benzalkonium chloride (BAC) in the other 3 solutions. The Na⁺ concentration in diclofenac sodium was 9 mM/kg, lower than that of bromfenac sodium and pranoprofen. The concentration of Cl^- and K⁺ was similar in all eye drop solutions. The pH and osmolarity of diclofenac sodium, bromfenac sodium, and pranoprofen was 7.0, 8.0, and 7.7 and 301, 300, and 309 mOsm/kg, respectively (Table 1).

Using phase-contrast microscopy, human corneal epithelial cells treated with BSS were found to be densely distributed throughout the culture media with time. No treated cellular migration of control more migrated and covered to the wound gap region to 48-hr. With time, the diclofenac and fluorometholone-exposed cells did not migrate to the wound region compared with bromfenac and pranoprofen-treated cells. After 1/10 diluted eye solution for 48-hr exposure, the cell migration of diclofenac and fluorometholone significantly inhibited compared with that of control, respectively (P=0.001, P=0.002) (Fig. 3, Table 2). Compared with diclofenac and fluorometholone-treated cells, but there was no difference between control and two NSAIDs such as bromfenac and pranoprofen (P=0.328, P=0.486) (Fig. 3, Table 2).

The cellular morphology of control corneal epithelial cells was not changed, although loss of microvilli was seen in some cells. All 1/50-diluted NSAID eye solutions and the fluorometholone eye solution induced the cytoplasmic and nuclear changes of the corneal epithelial cells after 24-hr of exposure, such as loss of microvilli, a dilated rough endoplasmic reticulum, enlarged mitochondria, a disrupted cytoplasmic membrane, and nuclear condensation (Fig. 4). At 48-hr exposure to 1/50-diluted diclofenac sodium or fluorometholone, there was more severe cellular morphology of corneal epithelial cell rather than those of bromofenac sodium or pranoprofen. Especially, certain types of nuclear change, such as nuclear marginal condensation, were observed after 48-hr exposure to 1/50 diluted fluorometholone (Fig. 5).