The Effect of Resveratrol on Protecting Corneal Epithelial Cells from Cytotoxicity Caused by Moxifloxacin and Benzalkonium Chloride

Abstract

Purpose.: Moxifloxacin (MOX), a fourth generation fluoroquinolone (FQ), has a wide antibacterial spectrum, but may show cytotoxicity characterized by high productions of reactive oxygen species (ROS). This study investigated the protective role of a common antioxidant agent, resveratrol (trans-3,5,4′-trihydroxystilbene), against the cytotoxicity caused by MOX.

Methods.: Experiments were performed with a human corneal epithelial cell line (HCECs; ATCC-CRL-11515). Another commonly used FQ, levofloxacin (LEV), and the most commonly used preservatives, benzalkonium chloride (BAC), were also used for comparison with MOX. Cell viability and morphologic changes after treatment were evaluated with trypan blue exclusion assay, propidium iodine/annexin V-FITC staining, and flow cytometry. Chemiluminescence immunoassay was used for ROS quantification. MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] assay, wound healing assay, and intracellular detections of oxidative stress were performed to evaluate the effects of resveratrol.

Results.: The MOX group, similar to the BAC group, showed significant cell shrinkage and death compared with the LEV group. High ROS production in HCECs of MOX group was observed both by chemiluminescence immunoassay and intracellular images. Within the observations of MTS [3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium] assay, live cell images, and wound healing process in vitro, the cytotoxic effects of the MOX and BAC groups were opposed by resveratrol. Human corneal epithelial cells pretreated with resveratrol demonstrated better cell viability and healing rate in the early stage.

Conclusions.: The protective effects of antioxidant agents indicate that MOX, similar to BAC, causes oxidative stress–related cell damage. The results also inspired us to think about a “supplementary regimen” to increase safety and decrease the adverse effect in the treatment of corneal infections.

Introduction

Bacterial keratitis is a common ocular infection and a leading cause of ocular morbidity and blindness worldwide. Fluoroquinolones (FQs), derive from nonfluorinated nalidixic acid, are antibacterial agents, which are widely used in the treatment of ocular infectious diseases, including bacterial keratitis. Fluoroquinolones offer a broad-spectrum activity and act by inhibiting two bacterial enzymes, bacterial topoisomerase IV, and DNA gyrase, which are specific to bacteria.¹ Moxifloxacin (MOX) is a fourth generation FQ, which has a wide-antibacterial spectrum, demonstrating increased activity against gram-positive cocci, and also retains the potency of third generation agents against gram-negative pathogens.^2–4 Therefore, MOX is commonly used in the management of ocular infections and, recently, as prophylactic therapy before or after ophthalmic surgery.^3–6 

Although FQs are generally considered as safe antimicrobial agents, results from previous studies have indicated that FQs may be toxic to corneal epithelial cells and could delay the wound healing process.^7–11 In clinical practice, studies have reported increased incidence of corneal perforations following FQ treatment,^12,13 and severe sterile corneal ulcers following the topical application of MOX eyedrops.¹⁴ In addition to the antibiotics itself, a further concern regarding the toxicity of eye drops is the toxicity of its preservatives, benzalkonium chloride (BAC), which is the most commonly added preservative in commercial ophthalmic FQ solutions. Earlier studies established its cytotoxic effects on corneal and conjunctival cells,^9,15,16 and further research has indicated that BAC provides the main source of cytotoxicity in commercial ophthalmic solutions.^9,17 Our previous study investigated the cytotoxicity of different FQs on cultured human corneal epithelial cells (HCECs), observing that preservatives play a big role on the cytotoxicity observed with FQ eyedrops.¹⁷ However, as shown in our study, some FQs still showed prominent cytotoxicity toward corneal epithelial cells without BAC, including MOX. 

It is well known that excessive reactive oxygen species (ROS) may lead to cell damage and cell death.¹⁸ In different cell types, ROS may work as the mitotic stimulator, cellular senescence inducer, cell death mediator, or even join the early stages of cell apoptosis.^19–24 Several studies have pointed out that systemic FQ administration causes oxidative stress on tendon cells, subsequent tendinitis, and even tendon rupture.^25–29 Therefore, it is reasonable to infer that ROS may also play a role in corneal cytotoxicity with topical FQ usage. Two common antioxidant agents, Resveratrol (trans-3,5,4′-trihydroxystilbene) and N-Acetylcysteine (NAC), are used in ocular diseases. Resveratrol, a major polyphenol found in red wine, has been proven to be a good ROS scavenger to both singlet oxygen or superoxide anions.^30–33 As for the eye diseases, some studies also found its protective effects for AMD.³⁴ N-Acetylcysteine (NAC), a kind of amino acid structured as acetylated L-cysteine, is the most abundant antioxidant in human cells.³⁵ NAC is crucial for the formation of glutathione (GSH) and glutathione peroxidase, which is essential for clearing free radicals.^35–37 Some studies have also shown that it can slow down the progression of senile cataract.^38,39 In the present study, we used these two antioxidants to evaluate their protective effects on corneal epithelial cells in terms of the influence of ROS under FQs. 

Materials and Methods

Cell Culture

A human corneal epithelial cell line was obtained from American Type Culture Collection (ATCC-CRL-11515; Manassas, VA, USA). The cells were maintained in Keratinocyte-Serum Free Medium (GIBCO-BRL 17005-042; GIBCO Laboratories, Grand Island, NY, USA) supplemented with 0.05 mg/mL bovine pituitary extract, 5 ng/mL human recombinant epidermal growth factor, 0.005 mg/mL insulin, 500 ng/mL hydrocortisone, 100 IU/mL penicillin, 100 μg/mL streptomycin, and 0.125 μg/mL amphotericin B. The cells were cultured at 37°C in a moist atmosphere with 5% carbon dioxide. The culture medium was changed every 2 days. In the present study, CRL-11515 HECEs within seven passages after purchase were cultured in a Keratinocyte-Serum Free medium. In every section of the experiments, we used cells of the same passage among groups to ensure the comparability. 

Preparation of Test Solutions

The testing materials including three standard powders of levofloxacin (LEV; Daiichi Pharmaceutical Co., Tokyo, Japan), MOX (Sigma-Aldrich Corp., St. Louis, MO, USA), and BAC (Sigma-Aldrich Corp.), were dissolved in distilled water to a stock concentration of 1.5%, 1.5%, and 0.2%, respectively. Levofloxacin and MOX stock solutions (pH 7.4) were diluted with a culture medium to 0.5% before cell treatment, and a stock solution of 0.2% BAC (pH 7.3) was further diluted to 0.001% with culture medium prior to use. The culture medium supplemented with the previously mentioned serum and antibiotics served as the control. The antibiotics did not contain preservatives or pharmaceutical excipients. It could, therefore, be assumed that the observed effects were attributable to the test solutions. 

The testing antioxidant materials included Resveratrol (Sigma-Aldrich, Corp.) and NAC (Sigma-Aldrich, Corp.). Resveratrol was initially dissolved in dimethylsulfoxide (DMSO) at stock concentration of 10 mM as blocking concentration and diluted with a culture medium to 5 μM during cell treatment. N-Acetylcysteine was dissolved in distilled water at a stock concentration of 100 mM, and was diluted with a culture medium to 10 μM before cell treatment. 

Trypan Blue Staining and Trypan Blue Exclusion Assay

To evaluate the cell viability and morphologic changes after exposure to test solutions, trypan blue staining and trypan blue exclusion assay were used, HCECs were seeded in six-well culture plates (2 × 10⁵ cells/well) and incubated in culture medium for 48 hours until reaching confluence. Then we removed the culture medium, and added 1 mL test materials. The cultures were incubated at 37°C in a moist atmosphere with 5% carbon dioxide for 1 hour. With solutions withdrawn, the cells were gently washed with PBS and stained with 0.2% trypan blue (Sigma-Aldrich Corp.). Phase-contrast microscopy (Eclipse TS100; Nikon, Tokyo, Japan) was then used to record morphologic change of cells. 

To assess the proportion of viable cells after treatments with trypan blue exclusion assay, the treated cells were washed with PBS and trypsinized using 1 mL trypsin at 37°C in a moist atmosphere with 5% carbon dioxide for 10 minutes. Trypsinization was halted by adding the same amount of culture medium. The cell mixture was then pipetted and transferred to a 15-mL tube and centrifuged at 1500g for 5 minutes. After discarding the supernatant, cells were stained using 0.2% trypan blue. The numbers of stain-positive (dead) and stain-negative (viable) cells in each culture were counted using a hemocytometer chamber. The mean percentage of stain-negative cells in each independent experiment was compared with that in the control group (n = 9). 

In some groups, cultured HCECs were initially exposed to the control or a test solution on six-well plates for 60 minutes, then solutions were removed, and the cells were incubated in fresh culture medium for a 24-hour recovery period. The cells were then stained with 0.2% trypan blue (Sigma-Aldrich Corp.) and examined using phase-contrast microscopy to observe the reversibility of cytotoxicity. 

Propidium Iodine/Annexin V-FITC Staining and Flow Cytometry for Detection of Apoptosis/Necrosis

To further clarify the mechanism of cytotoxicity, propidium iodine/annexin V-FITC (PI/annexin V) staining was used to identify apoptotic and necrotic cells, respectively. Human corneal epithelial cells in six-well plates were exposed to the control or a test solution for 1 hour. The cells were gently washed with PBS and then washed in annexin V-FITC solution (BD Biosciences, San Jose, CA, USA), containing propidium iodine (Sigma-Aldrich Corp.), and Hoechst 33258 (Sigma-Aldrich Corp.). After incubating the cells in a dark room for 15 minutes, cultured HCECs were washed with PBS, trypsinized by dissociation buffer (GIBCO) and resuspended in blocking solution (Ca²⁺, Mg²⁺-free HBSS containing 2% goat serum). Then the cell morphology and signal intensity was analyzed on a Becton Dickinson FACScanR flow cytometer (Franklin Lakes, NJ, USA). Forward-scattered light (FSC) and side-scattered light (SSC) were recorded for cell volume and morphology. Signal intensity of PI/annexin V were calculated for evaluation of apoptosis/necrosis. 

Production of Reactive Oxygen Species as Detected Using Chemiluminescence

Immunoassay.

After incubation with the control or a test solution in 3-cm culture plates for 1 hour, the solutions were collected. A 0.2-mL solution was placed in a dark chamber of the luminol-enhanced and lucigenin-enhanced chemiluminescence analyzing system (CLD-FS1; Tohoku Electronic Industrial Co., Sendai, Japan) to measure the production of ROS and superoxide anions. After an initial 50 seconds, to determine the background level, 0.5 mL 0.015 mM luminol (Sigma-Aldrich Corp.) or 0.5 mL 0.005 mM lucigenin (Sigma-Aldrich Corp.) was added to react with ROS or superoxide anions, respectively. Chemiluminescence was continuously measured for 180 seconds. 

Evaluation of Antioxidant Effects on Cell Viability Using MTS Assay

For viability studies, cells were seeded in 96-well plates (2 × 10⁴ cells/well) and incubated for 24 hours. Once a confluent cell layer was obtained, the media were removed and 100 μL test materials were added in various concentrations. The MTS [3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium] assay (CellTiter 96 Aqueous One Solution Cell Proliferation Assay; Promega, Madison, WI, USA) was performed at five incubation time points (30 minutes, and 1, 4, and 8 hours). The test materials were removed and 20 mL MTS reagent along with 100 mL culture medium (without serum and antibiotics) were added to each well. Human corneal epithelial cells were further incubated at 37°C for 3 hours. Absorbance measurements were read at 490 nm using a fluorescence absorbance 96-well plate reader afterwards. Negative control groups consisted of cells in media (without serum and antibiotics) without the test materials added (pH = 7.3). Groups of media, pH adjusted to 5.0 (without serum, antibiotics or test materials added), were processed identically and incubated alongside the treated groups. Three independent experiments (n = 4 per group) were conducted. Cell viability was expressed as a percentage of the negative control group. To distinguish the toxicity from water and ensure the adequacy of nutrient for cell growth, we compared HCECs cultured in media without test materials (negative control) and those in distilled water (vehicle control). To further explore the effect of antioxidant agents, we added 100 μL antioxidant agents as pretreat medication for 60 minutes, removed the media, and then added 100 μL test materials as previously stated. 

Evaluation of Antioxidant Agents on the Bactericidal Activity of Antibiotics

The antimicrobial activity of MOX and LEV with/without 5 μM resveratrol against standard strain of Staphylococcus aureus ATCC 29213, Escherichia coli ATCC 25922, Serratia marcescens ATCC 8100, and Pseudomonas aeruginosa ATCC 27853 was determined by broth microdilution method. Serial double dilutions of standard powder of MOX and LEV were prepared at a concentration range of 0.06 to 32 μg/mL according to Clinical and Laboratory Standards Institute (CLSI) recommendations. The bacteria were inoculated into cation-adjusted Müller-Hinton (M-H) broth to reach a final concentration of 10⁵ colony forming units (CFU)/ml. The inoculated microdilution trays were sealed in plastic bags and incubated at 35°C in ambient air for 22 hours. The minimum inhibitory concentration (MIC) was defined as the lowest concentration of drug that inhibited visible growth. 

Evaluation of Antioxidant Effects on Wound Healing Process

Human corneal epithelial cells were cultured on 24-well culture plates as described above. When reaching confluence, the HCECs were wounded by Ti:Sapphire laser to create a 500-μm wide wound. The damaged cells were washed out with PBS and then incubated with culture medium, LEV (0.05%), MOX (0.025%, 0.05%, and 0.1%), or BAC (0.001%) (with/without 5 μM resveratrol or 10 μM NAC). The wound healing process of each group was observed at 8 hours, and 1, 2, 3, and 4 days. The extent of healing was determined by the ratio of the differences between the zero hour and the remaining wound areas at different time points. 

Detection of the Dynamic Change of Oxidative Stress in Live Cells

To observe the antioxidant effects and the dynamic change of oxidative stress in live cells, we use the fluorogenic probes of oxidative stress (CellROX; Life Technologies Corporation, Carlsbad, CA, USA) for this experiment. After HECEs treated with test solutions (with or without the ROS scavenger–resveratrol) for 1 hour, the cells were stained with 5 μM fluorogenic probes for 30 minutes. Then the agents were washed out and replaced with the initial test solutions. The intracellular fluorescent intensity was observed with fluorescence microscopy and pictures were taking 2 hours after the initial treatment. Ten fields were calculated in every plate and three repeat experiments were performed. The calculations of signal intensity and subsequent analysis were done with MATLAB (The MathWorks, Inc., Natick, MA, USA). 

Statistical Analysis

A computerized database was established to facilitate data management and statistical analysis. Subsequent data analysis was carried out using Microsoft Excel 2007 (Microsoft Corporation, Redmond, WA, USA), SPSS 12.0 (SPSS, Inc., Chicago, IL, USA), and SAS 9.1.3 (SAS Institute, Inc., Cary, NC, USA) software. Two-sided P ≤ 0.05 was considered statistically significant. Two-sample t-tests were used for comparing mean values of continuous variables between two groups. Fisher's exact tests or χ² tests were used to examine the associations between categorical variables. The comparisons among the three tested materials were accomplished by using a one-way ANOVA. 

Results

Quantified Evaluation of Cytotoxicity in Nonpreservative Fluoroquinolones

In the control groups, HCECs exposed to the culture medium remained confluent and maintained a normal shape. However, cells incubated in LEV, MOX, and BAC showed time-dependent morphologic changes with cell detachment. Trypan blue staining revealed that significantly higher populations of cells were stained positive in the MOX and BAC groups compared with those in the LEV and control groups. Figure 1A showed the trypan blue staining of HCECs exposed to test solutions (MOX/BAC/LEV) for 60 minutes. With flow cytometry, Figure 1B revealed a proportion of cells in the MOX and BAC groups with a decrease in size, which is consistent with cell shrinkage associated with apoptosis. To further confirm the phenomenon, trypan blue exclusion assay was performed and results are shown later in Figure 1F. The MOX and BAC groups displayed significant cytotoxic effects within 5 minutes of exposure (P = 0.032, P = 0.004, respectively, compared with the control group). The cytotoxicity was further increased with prolonged exposure until 3 hours (double asterisks in the figure stand for P < 0.001 compared with control group). The number of viable cells in the LEV group, however, did not significantly differ from the control group until 1 hour of exposure (t-test, P = 0.982, P = 0.989, P = 0.019, P = 0.017, at 5 and 30 minutes, and 1 and 3 hours compared with the control). 

Figure 1

 

Cytotoxicity of MOX, LEV, and BAC after exposure of test solutions for 60 minutes. (A) Trypan blue staining of HCECs of each group. (B) Flow cytometry for cell sorting of each group. (C, D) Quantifications of the PI/annexin V double stain intensity by flow cytometry. (E) The results of the 24-hour recovery test with trypan blue staining. (F) Trypan blue exclusion assay to quantify the cell viability change in each group. (G) The dot-blot chart depicted the difference between the control and MOX group. Much more HCECs exposed to MOX heading for apoptosis than cells of the control group. (H–I) The statistical analysis for annexin V and PI staining among groups after 1-hour exposure. (Both * and **, P < 0.05 compared with the control group.)

Figures 1C and 1D showed the quantifications of the PI/annexin V staining, and the statistical analysis among groups was shown later in Figures 1H and 1I. After 1-hour exposure, the MOX group showed significantly high annexin V and PI staining (*P < 0.05 in both staining compared with the control group), representing the active status of apoptosis. The BAC group showed high PI staining (*P < 0.05) but low annexin V staining. (**P < 0.05) No significant signal change was found in cells of LEV group. We further depicted the difference between control and MOX group in Figure 1G. With the dot-blot chart, we can clearly see much more HCECs exposed to MOX heading for apoptosis than cells of the control group. 

In addition, we tested the recoverability of cell damages. After a 60-minute treatment of these agents, HCECs were put into the culture setting for 24 hours to evaluate their recoverability. Figure 1E shows the results of the recovery test with trypan blue staining. Strikingly, an irrecoverable cell damage was observed in the BAC group, a partial recovery in the MOX group, and a complete recovery in the LEV and control groups. 

ROS Productions of HECES After Drug Exposure and Effects of Antioxidant Agents on Cell Viability

To further explore the possible causes of cytotoxicity in different groups, we evaluated the ROS production in these cells. In Figure 2A, with chemiluminescence immunoassay, cells of the MOX group showed statistically more ROS production than those in the LEV and control groups (P < 0.001, both in total ROS production and superoxide anion production). The trend was similar in the BAC group. However, after 60-minute exposure to drugs, the MOX group showed an even higher ROS production than that in the BAC group (P = 0.001). The LEV group revealed low ROS production and showed no statistical difference with the control group (P = 0.73 in total ROS production and P = 0.75 in superoxide anion production). These results also showed that the amount of ROS production was proportional to the cytotoxicity among these groups. 

Figure 2

 

(A) Chemiluminescence immunoassay for quantification of the ROS production in groups. Left: Total ROS production. Right: Superoxide anion production. (B–D) Temporal observation of cell viability and the effects of antioxidant agents by MTS assay. (Taking the control group as 100% at every particular period of time.)

In Figure 2B, with the help of MTS assay, we observed a temporal change of cell viability among groups from 30 minutes, and 1, 4, and until 8 hours. In Figure 2B, all three groups showed a gradually decreased number of viable cells compared to the control group (we took the control group as 100% at particular periods of time). Correspondingly, the LEV group showed a higher percentage of viable cells compared with both the MOX and BAC groups at several time points (P < 0.05 in 30 minutes and 1 hour, P < 0.001 in 4 hours). However, the difference disappeared after the cell was exposed for 8 hours. Cell viability in the MOX group was similar to that in BAC group, only with borderline significance in the 1 hour group (P = 0.07, higher percentage of cell viability in the MOX group than in the BAC group). 

In viewing of the ROS production induced by BAC and MOX, we tested the effect of antioxidants on cell death induced by BAC or MOX. Figure 2C showed the comparisons among the BAC group, the BAC pretreated with resveratrol group, and the BAC pretreated with NAC group. Human corneal epithelial cells pretreated with resveratrol showed a better cell viability in the early stages (P = 0.007 at 30 minutes and P = 0.03 in 1 hour, compared with that in the BAC group without pretreatment). However, the difference disappeared 1 hour later. Pretreatment with NAC did not provide protective effects at all time points. Figure 2D illustrates the comparisons among the MOX group, MOX pretreated with resveratrol group, and MOX pretreated with NAC group. Human corneal epithelial cells pretreated with resveratrol showed a better cell viability in 30 minutes (P = 0.01). However, the difference disappeared later. Pretreatment with NAC also did not provide protective effects at all time points. Thus, resveratrol, rather than NAC, provides a transient protection of HCECs from damages induced by BAC and MOX. 

Evaluation of Antioxidant Agents on the Bactericidal Activity of Antibiotics

The MIC of LEV and MOX against four tested bacteria (S. aureus ATCC 29213, E. coli ATCC 25922, S. marcescens ATCC 8100, and P. aeruginosa ATCC 27853) was 0.25/0.06, 0.06/0.06, 0.5/0.5, and 2/4 (μg/mL), respectively. The MIC of LEV and MOX against the four tested bacteria was within the CLSI MIC quality control ranges that confirmed the accuracy of the test. The MIC was not changed after adding 5 μM resveratrol, indicating that resveratrol did not influence the bactericidal activity of LEV and MOX. 

Effects of Antioxidant Agents on Wound Healing Process

To further simulate clinical conditions, wound healing rate was evaluated in cultured HCECs. With the initial 500-μm wide wound, we observed wound healing rates among groups from 8 hours, and 1, 2 , 3, and until 4 days. In Figure 3A, all three groups (LEV, MOX, and BAC) demonstrated a gradually decreased percentage of uncovered area similar to a corneal wound healing process. However, compared with the control group, healing rate was delayed in all three groups before day 3 (P < 0.05 in all subgroup comparisons). In this aspect, the BAC group showed the slowest wound healing rate compared with that in the LEV and MOX groups. Nevertheless, HCECs in all three groups could achieve complete healing at day 4. 

Figure 3

 

Effects of antioxidant agents on wound healing rates. (A) All three treated groups demonstrated a slower wound healing process than the control group (*P < 0.05 compared with all three treated groups). (B) Percentage of delayed healing in comparison between the MOX group and MOX pretreated with resveratrol, as well as between the BAC group and BAC pretreated with resveratrol. In both MOX and BAC groups, HECEs pretreated with resveratrol showed a better healing rate in early stages (single asterisk in the MOX group and double asterisk in the BAC group, P < 0.05). (C) Multiple comparisons between the MOX group and the MOX pretreated with resveratrol group. In the 0.025% and 0.05% MOX groups, HECEs pretreated with resveratrol showed an increased healing rate at the early stages (P <0.05 in 0.025% MOX group, day 1, and 0.05% MOX group, days 1 and 2). (D–G) Showing the pictures of fluorescent gradients in HCECs of the control, LEV, MOX, and MOX with resveratrol groups. (H–I) Demonstrating the statistical analysis about intracellular accumulation of oxidative stress after exposed to FQs with/without resveratrol.

We further investigated pretreatments with the antioxidants of 5 μM/mL resveratrol or 10 μM NAC, 30 minutes in the HCECs. Figure 3B showed the comparisons between the MOX group and MOX pretreated with resveratrol, as well as between the BAC group and BAC pretreated with resveratrol. Human corneal epithelial cells pretreated with resveratrol evidenced a better healing rate in early stages, both in the MOX and BAC groups (single asterisk in the MOX group and double asterisk in the BAC group; Fig. 3B). However, the difference disappeared with longer observation. Pretreatment with NAC did not provide protective effects at all time points. Pretreatment with resveratrol showed no benefits in the LEV group. 

We further studied the cytotoxicity of MOX at different concentrations. Figure 3C described the multiple comparisons between the MOX group and the MOX pretreated with resveratrol group. Three different concentrations of MOX exposure (0.025%, 0.05%, and 0.1% of MOX) were used. In the 0.025% and 0.05% MOX group, HCECs pretreated with resveratrol showed an increased healing rate at the early stages (P < 0.05 in 0.025% MOX group, day 1, and 0.05% MOX group, days 1 and 2). However, in the 0.1% MOX group, there was a significantly delayed wound healing before day 3 and resveratrol did not increase the wound healing rate. Resveratrol seemed to have more effects on or upon HECEs exposed within lower-concentration MOX. 

The Dynamic Change of Oxidative Stress in Live Cells

The accumulation of oxidative stress and the effect of ROS scavenger–resveratrol in cellular level were shown in Figures 3D through 3G and analyzed in Figures 3H and 3I. With the fluorogenic probes, we recorded that ROS accumulated prominently in HCECs of MOX group (P < 0.001), and resveratrol effectively reduced the oxidative stress (P = 0.01). It is noteworthy that, despite the usage of resveratrol, cells of MOX group still produced more ROS compared with control group (P = 0.01). Though there was 9% increase of ROS accumulation in cells of LEV group (P = 0.02), it could not be reduced by resveratrol. 

Discussion

The epithelium, being the main barrier of the cornea, is crucial for the normal physiology of cornea. In infectious keratitis, the healing of epithelial defects is an important indicator of treatment. However, in severe corneal infection, topical antibiotic drops are usually administered in a high frequency, leading to higher concentrations in the ocular surface. The cytotoxicity of drops is believed to be dose-dependent. Therefore, its cytotoxicity could be more prominent and notable. 

It is well known that preservatives lead to considerable cytotoxicity and commercial drops often blame their cytotoxicity to their preservatives. Benzalkonium chloride is a frequently added preservative in ophthalmic solutions and its cytotoxicity for corneal and conjunctival epithelium is widely accepted.^40,41 However, in our previous study, different fluoroquinolones (FQs) presented intrinsic cytotoxicities to the corneal epithelium.¹⁷ In the present study, therefore, we take the commonly used preservative BAC as the negative control group, to make a more extended comparison. Among the clinically used FQs, both LEV and MOX ophthalmic solutions are self-preserved without containing preservatives. To clarify their (own) potential cytotoxicity may provide valuable information for safety concern in daily practice. 

MOX, a fourth generation FQ, has broad-spectrum activity and 3-fold higher concentration in the aqueous humor. It has been widely used for treating infectious keratitis and as a prophylactic regimen for endophthalmitis following cataract surgery.^42–44 However, the cytotoxicity of MOX has also caused concerns and several studies have demonstrated its toxicity. Matsumoto et al.⁴⁵ observed that MOX has greater inhibitory effects on corneal epithelial cell migration than gatifloxacin and LEV. Using an MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] assay, and by measuring the wound healing rate of injured cultured HCECs, Kim et al.⁴⁶ demonstrated that Vigamox (MOX) was more toxic than Cravit (LEV) in corneal epithelial cells at 0.25% concentration. Ayaki et al.⁴⁷ applied five preservative-free FQ on four different corneoconjunctival cell lines, ranking the FQs in order of cell toxicity as MOX equals gatifloxacinequals norfloxacin greater than LEV greater than tosufloxacin. In the first part of the present study, trypan blue exclusion assay confirmed that the MOX group (0.5% unpreserved MOX), which is similar to the BAC group (0.001% BAC), leads to significantly poorer cell viability than the control group and the LEV group (0.5% unpreserved LEV). Propidium iodine/Annexin V double staining and flow cytometry further demonstrated that HCECs of the MOX group underwent apoptosis while not being in the LEV group. Cells of the BAC group seemed to undergo late apoptosis or some necrosis in our observations. By morphologic observation, cell sorting by flow cytometry also revealed similar results, a large part of cells in MOX and BAC group showed a decrease in cell size, possibly representing cell shrinkage and death. These results were consistent with those from previous researches, and we provided more detailed observations in this study. We believed that the adverse effect of MOX is not caused by the primary pharmacological mechanism. This hypothesis is not only based on the fact that the antimicrobial mechanisms, which inhibit two bacterial enzymes, bacterial topoisomerase IV and DNA gyrase, are not designed for human cells, but also regarded as the prominent differences in cytotoxicity and its presentation between two most common used FQ agents, Lev and MOX. 

Cytotoxicity of preservatives has been reported relating to ROS production in many different studies. Dutot et al.⁹ observed that BAC could induce P2X7 cell death receptor activation associated with increased ROS and superoxide anion production and apoptosis. Chang et al.⁴⁸ found that benzyl alcohol, the preservative in triamcinolone acetonide, led to RPE cell death by apoptosis involving immediate production of ROS. It is reasonable to infer that ROS production may also be involved in the cell apoptosis of the BAC group. It is questionable whether the MOX group shared the same mechanism of cell apoptosis. Fluoroquinolones have been reported to induce tendinopathy due to oxidative stress.^26,28 In the present study, results from chemiluminescence immunoassay revealed significantly greater ROS and superoxide anion production by HCECs in both the MOX and BAC groups, compared with those in the LEV and control groups. The phenomenon further proved that the toxic mechanism of MOX may be similar to that of BAC. Oxidative stress should not be the common problems of all FQs because we found that LEV did not induce substantial ROS production. 

Oxidative stress is associated with several eye diseases such as cataract, AMD, and proliferative diabetic retinopathy (PDR).^49,50 Resveratrol, one of dietary polyphenols with an antioxidant effect, has been widely studied in terms of its therapeutic effect in eye disease, with most of discussion focusing on the protective effects toward RPE cells.^51–55 Some of these studies specifically focused on the protection from extrinsic harmful substance. Kubota et al.⁵³ reported that resveratrol prevents light-induced retinal degeneration via suppressing activator protein-1. They also found that resveratrol prevents ocular inflammation in endotoxin-induced uveitis by inhibiting oxidative damage and nuclear factor-kappa B activation.⁵⁴ Sheu et al.⁵⁵ reported that resveratrol protects HRPE cells from damage induced by acrolein, a compound found in cigarettes. Resveratrol has also been proven to have a neuroprotective effect on retinal ganglion cells (RGCs) in ischemic, glaucomatous or diabetic status.^56–58 As for the anterior segment of eye, Li et al.⁵⁹ observed that resveratrol exerts a protective effect against oxidative damage in lens epithelial cell cultures. Zheng et al.⁶⁰ found that the resveratrol protects human lens epithelial cells against H₂O₂-induced oxidative stress by increasing catalase, SOD-1, and HO-1 expression. These findings are thought to be effective in cataract prevention. In our study, HECEs pretreated with resveratrol indeed expressed better cell viability after exposure to MOX and BAC (Figs. 2C–E). However, the protective effects only achieved statistical significance within 1 hour of drug immersion. Fortunately, the in vivo drug concentration of the ocular surface is dynamic and medication would be pumped out by blinking and diluted by tear secretion. To further mimic clinical practice, we performed the wound healing assay with diluted medication (with 1:5, 1:10, 1:20 of commercial concentration) and detection of the dynamic change of oxidative stress in live cells to confirm the toxicity of the test solutions and the protective effect of antioxidant agents (Fig. 3). In wound healing assay, we observed that every test solution would lead to some delay in the wound healing process, including LEV. On the other hand, in fair concentration, all groups could achieve complete healing within 3 to 4 days. Resveratrol can provide some protective effect on groups of high ROS production (i.e., those exposed to MOX and BAC). Furthermore, according to our present study, pretreatment of resveratrol could provide more benefits to cells exposed to lower concentrations of MOX as seen in Figure 3C. With higher concentration (0.1%), delay of wound healing was prominent and resveratrol seemed ineffective. The intracellular accumulation of oxidative stress and the eliminating effect of resveratrol toward the MOX group further strengthened our idea that ROS scavenger may play roles in ameliorating cytotoxicity caused by MOX. 

There were no previous reports regarding MOX and ROS that can be used for comparison. However, Debbasch's study on the conjunctival toxicity of 10 preservatives revealed that BAC caused apoptosis at low concentrations and necrosis at high concentrations.¹⁵ In our clinical practice, intensive application of topical antimicrobial drops was the best way to treat severe infectious keratitis. The extremely high frequency of medication would inevitably raise the drug concentration in the tear film. These patients with active infections commonly suffered from epithelial defect. Prompt re-epithelization is important for the health of ocular surface, and a similar phenomenon has also been mentioned in dry eye disease (Enriquez-De-Salamanca A, IOVS 2014;55:ARVO E-Abstract 1478; Abengózar-Vela A, et al. IOVS 2014;55:ARVO E-Abstract 3654). Abengózar-Vela et al. found that Quercetin and Resveratrol polyphenols may help to decrease TNF-α–related inflammatory pathways and damages of corneal epithelial cells. The effects can be observed both in vitro and in vivo. Therefore, our present study would inspire the future research on a “supplementary regimen” to increase safety and decrease adverse effects in the treatment course. 

The goal of our study was to clarify the safety of using FQs for infections of the ocular surface. We found that some FQs, especially MOX, tended to increase ROS production of human corneal epithelial cells, leading to subsequent cellular damage and apoptosis. A similar mechanism was also found in BAC, a common preservative. Resveratrol provided some protective effect, both in cell viability and in function concerning proliferation and wound healing, toward this kind of damage. Nevertheless, there are limitations in our study. We presented that an antioxidant with strong scavenger, resveratrol may effectively reduce the cytotoxicity caused by MOX. However, the molecular target of these cytotoxicities may need further study in pharmacology. 

In summary, MOX, as a potent fourth-generation FQ, has been widely accepted for clinical management of ocular infections and prophylactic therapy before or after ophthalmic surgery. Some recent studies also revealed that when used appropriately, MOX (similar to LEV) did not influence corneal wound healing.^61–63 On the contrary, through more detailed observation, we found the adverse effects of MOX toward corneal epithelial cells and wound healing. Thus, we proposed a new idea that antioxidant agents such as resveratrol may be useful in protecting ocular surface from cytotoxicity in topical antibiotics and preservatives. To the best of our knowledge, this is the first report discussing the protective effect of antioxidant agents toward intense topical antimicrobial FQs drops. These findings support the position that antioxidant supplement would be a possible solution to reduce the side effects of ocular medications. Further in vivo animal or clinical studies are mandatory to evaluate the clinical application of antioxidant supplement in treating microbial keratitis with MOX. 

Acknowledgments

The authors thank Yu-Chung Chuang, MD, MS, and Hsin-Fang Lo (Department of Internal Medicine, National Taiwan University Hospital, Taiwan) for the generous help in the evaluation of antioxidant agents on the bactericidal activity of antibiotics. They also thank Hsin-Yi Chen, PhD (Graduate Institute of Cancer Biology & Drug Discovery, College of Medical Science & Technology, Taipei Medical University, Taiwan), for the precious assistance in data interpretation. 

Supported by grants from the National Science Council, the Executive Yuan, Taiwan, NSC100-2314-B-002-059-MY3 (Taipei, Taiwan), and the National Center of Excellence for Clinical Trial and Research Grant DOH101-TD-B-111-001 (Taipei, Taiwan). 

Disclosure: T.-Y. Tsai, None; T.-C. Chen, None; I-J. Wang, None; C.-Y. Yeh, None; M.-J. Su, None; R.-H. Chen, None; T.-H. Tsai, None; F.-R. Hu, None 

Footnotes

 T-YT and T-CC contributed equally to the work presented here and should therefore be regarded as equivalent authors.