Evaluation of moxifloxacin-induced cytotoxicity on human corneal endothelial cells

Moxifloxacin hydrochloride (MXF) is widely used for the prevention of bacterial endophthalmitis after intraocular surgeries. However, the safety issue of intracameral injection of MXF for human corneal endothelial cells (HCECs) is still debatable. In this study, we investigated concentration-dependent cytotoxicity (0.05–1 mg/ml) of MXF for immortalized HCECs (B4G12 cell) and the underlying mechanism. Reactive oxygen generation (ROS) and cell viability after MXF exposure was measured. Flow cytometric analysis and TUNEL assay was used to detect apoptotic HCECs after MXF exposure. Ultrastructure of damaged HCECs by MXF was imaged by transmission electron microscope. Western blot analysis and caspase 2, 3 and 8 analysis were used to reveal the underlying mechanism of MXF induced damage in HCECs. We found that MXF induced dose-dependent cytotoxicity in HCECs. MXF exposure increased ROS generation and induced autophagy in HCECs. Increased LDH release represented the cellular membrane damage by MXF. In addition, caspases activation, Bax/Bcl-xL-dependent apoptosis pathway and apoptosis inducing factor nuclear translocation were all involved in MXF induced HCECs’ damage, especially after exposure to high dose of MXF (0.5 and 1.0 mg/ml). These findings suggest that MXF toxicity on HCECs should be thoroughly considered by ophthalmologists when intracameral injection of MXF is planned.

www.nature.com/scientificreports/ apoptotic and necrotic cell populations, compared with low concentrations of MXF. Because the totally destroyed HCECs were eliminated during the staining and flow cytometric analysis procedure, some discrepancy was inevitable between cell viability data and flow cytometric data. For example, over 50% of cell death was observed in CCK analysis ( Fig. 1), while less than 30% of apoptotic and necrotic cell population in flow cytometric data after 48 h of 0.5 mg/ml of MXF exposure (Fig. 5). Induction of apoptosis of corneal endothelial cells by MXF was further verified in ex vivo porcine corneas. Incubation of corneas with MXF (0.5 and 1.0 mg mL) for 24 h resulted in apoptotic death of corneal endothelium (Fig. 6).

Molecular evidences of MXF induced immortalized HCECs apoptosis.
The activation of caspases 2, 3, and 8 pathways was investigated in immortalized HCECs after 48 h of exposure to MXF. Exposure to MXF resulted in significant dose dependent increases of caspase 2, 3, and 8 activation (Fig. 7). These findings indicate caspases pathway is involved in MXF induced HCECs apoptosis. AIF is normally distributed in mitochondria in the resting state and translocates to the nucleus under stress to induce apoptosis. AIF is one of the key molecules involved in a caspase-independent apoptosis pathway 14 . Although the total protein expression level of AIF was not affected, increased AIF nuclear translocation compared with control was observed after MXF exposure (Fig. 8). In addition, pro-apoptotic Bax expression increased with decreased anti-apoptotic Bcl-xL protein expression, which indicated the activation of Bax/Bcl-xL-dependent apoptosis pathway after MXF exposure ( Fig. 9).

Discussion
In this study, we found that MXF can induce immortalized HCECs death in a dose-dependent manner. The major cell damage occurred via apoptosis, combined with necrotic cell death ( Supplementary Fig. 4). We verified that MXF exposure increased ROS and autophagy significantly in immortalized HCECs. The induced apoptosis by MXF involved both caspase dependent and caspase independent pathways. We found increased LDH release from immortalized HCECs after high concentrations of MXF exposure. This finding is consistent with the previous studies 12, 13 . Haruki et al. reported that MXF over 0.5 mg/ml caused HCEC membrane damage and death 13 . Increased oxidative stress after MXF exposure, found in our study, is also consistent with the previous studies. It is known that increased ROS can induce autophagy and eventual cell death when autophagy failed to overcome ROS 15 . Akal et al. reported increased oxidant content in cornea and increased caspase 3 and 8 staining in rat corneal tissue after MXF 0.05 mg/0.01 ml intraocular injection 12 . We also observed increased apoptosis of HCECs after MXF exposure. However, interestingly, our study found that MXF-induced apoptosis in HCECs was both caspase dependent and independent, which differs from the previous animal study. This difference may be due to differences in the regenerative capacity of endothelial cells. Although we used immortalized human cornea endothelial cells in the current experiment, normal HCECs do not regenerate in vivo, while corneal endothelial cells of rabbits and rats can regenerate after various forms of damage 16,17 .
The main difference between the previous studies and our current study is a more in-depth investigation of the possible mechanisms of MXF induced HCECs apoptosis. As shown previously, the increase of a cleaved  www.nature.com/scientificreports/ form of caspase 3 (the common pathway caspase) indicates that both intrinsic and extrinsic caspase pathways are involved in MXF-induced HCECs apoptosis. The activation of caspase 8 indicates the activation of extrinsic pathway of apoptosis which is mainly mediated by the signal through the receptor on the cell membrane 15,18 . However, the activation of caspase 2 indicates the apoptosis pathway induced by the increased cellular stress such as oxidative stress, ER stress or DNA damage 15,18,19 . The finding of the activation caspase 2, 3 and 8 can be an evidence of the involvement of both intrinsic and extrinsic caspase pathways in MXF-induced HCECs' damage. The intrinsic apoptotic pathway may also be activated by the cells if the extrinsic signal is not sufficient to execute apoptosis 18 . In addition, we found that the Bax/Bcl-xL pathways were also involved in MXF-induced HCECs apoptosis. Upregulation of the pro-apoptotic Bcl-2 family, such as Bax, and downregulation of the anti-apoptotic Bcl-2 family, such as Bcl-xL leads to caspase independent apoptosis. [20][21][22][23] Increased nuclear translocation of AIF can be another evidence of caspase independent apoptosis induced by high dose of MXF exposure. Our findings are similar to the previous study investigating HCEC toxicity of another ophthalmic antibiotic, norfloxacin 24 . MXF is one of the most widely used broad spectrum topical ophthalmic antibiotics. MXF interferes with bacterial DNA gyrase and topoisomerase IV. The inhibition of either enzymes results in bacterial death. Commercial topical MXF (Vigamox) can reach anterior chamber concentrations between 1.55 and 2.28 µg/ml after 2-3 days of topical application (usually four times a day) 25 . Considering that the MIC range of MXF of most fluoroquinolone sensitive pathogens is between 0.25 and 2.5 µg/ml 26 , the topical application of MXF should be effective for the control of anterior chamber infection.
In addition, due to easy access to commercialized topical MXF formulation, some surgeons have been injecting small amounts of MXF (Vigamox) into the anterior chamber (intracameral) at the end of the cataract surgery to prevent bacterial endophthalmitis 27 , the effectiveness of which in preventing postoperative bacterial infection by MXF intracameral injection has been repeatedly reported 4,28,29 . The routine intracameral injection of MXF reduced postoperative bacterial endophthalmitis rate 3.5 to 4.0 times in a large case series (N = 617,453 and N = 116,714) 29,30 . Additionally, thorough anterior chamber irrigation, using MXF solution, has been suggested as an effective preventive action for postoperative bacterial endophthalmitis 31 .
The most common concentrations of MXF, used for intracameral injection or irrigation, are between 0.25 and 0.5 mg/ml 2,3 . Commercial MXF eyedrops, such as Vigamox, are preservative free and contain 5 mg of MXF  26 . Considering aqueous humor circulation system, the direct comparison between intracameral injection and 24 h culture medium exposure is impossible. However, in the current study, we found that even lower concentrations, such as 0.05 mg/ml of MXF, can induce significant ROS generation and death of HCECs as demonstrated in cell viability assay. Although the rapid clearance of moxifloxacin injected into the anterior chamber is highly expected in normal condition, the cataract surgeon should not overlook the reduced aqueous outflow, often found immediately after surgery due to retained viscoelastic device in the anterior chamber. Partial obstruction of aqueous outflow can decrease the clearance of MXF after intracameral injection and increase the exposure time of HCECs to MXF. Our study has several limitations. Firstly, we used the HCEC cell line B4G12. Unfortunately, it is well known that the primary culture of HCECs has limited proliferation capacity. Ideally, primary-cultured HCECs from several different donors might have been better suited to our study. However, the heterogeneous population of HCECs can also cause donor-specific effects, such as age and topography-related (central vs. peripheral origin) differences 32 . For these reasons, we alternatively manipulated B4G12 instead of primary cultured HCECs in this study for the stable experiments and results because B4G12 expresses HCEC-specific 9.3. E-antigen, occludin and ZO-1 protein 33 , and shows differentiated HCECs properties in its morphology and function [34][35][36] . Secondly, the lack of an in vivo toxicity experiment is another drawback. Finding observed in vitro is not always repeatable in vivo because of the many confounding factors. However, HCECs toxicity study is impossible to be conducted on human eyes, because HCEC damage is irreversible and can lead to blindness. Unfortunately, convenient animal eye models, as for rabbits for example, differ from humans, in that HCECs have in vivo regeneration capability. Finally, the increased pro-apoptotic pathways were only verified after relatively high dose of MXF exposure. Therefore, the underlying mechanisms of low doses of MXF induced HCECs are still unclear. Future experiment using inhibitors or activators may enhance the evidence of the related signal pathways found in the current study.
In summary, we demonstrated a dose-dependent toxicity of MXF for immortalized HCECs. Caspases activation, Bax/Bcl-xL-dependent apoptosis pathway and AIF activation were all involved in the mechanisms of MXF toxicity for HCECs. Our results suggest that intracameral injection or irrigation of MXF should be tapered to the lowest effective concentration to avoid irreversible damage of HCECs. In addition, careful and regular postoperative monitoring of corneal endothelium is necessary after intracameral injection of MXF. Lactate dehydrogenase (LDH) assay. Cellular membrane damage was measured using an LDH cytotoxicity detection kit (Takara Bio Inc., Shiga, Japan). Briefly, HCECs were cultured at 1 × 10 4 cells/well in a 96-well plate and incubated for 24 h. Following the adherence of cells, various concentrations (0, 0.05, 0.125, 0.25, 0.5, 1.0 mg/ml) of MXF were applied to cells for 24 h, 48 h, and 72 h respectively. The wells with vehicle only and the wells with 1% triton X-100 addition were used as the negative and positive controls, respectively. Following the incubation of cells, all the supernatants were transferred to a new 96-well plate, the reaction mixture was added, and they were all incubated for 20 min at room temperature. Absorbance was measured at 490 nm.  The membranes were then incubated with peroxidase-conjugated secondary antibodies for 1 h at room temperature. Blots were developed using an enhanced chemiluminescence kit (catalog number: RPN2232; GE healthcare, Buckinghamshire, UK) and visualized using a Fujifilm Image Reader LAS-3000 (Fujifilm, Tokyo, Japan). Each experiment was repeated at least three times, and the densitometric analysis was performed using Multi Gauge V3.0 (Fujifilm Life Science, Tokyo, Japan).  www.nature.com/scientificreports/ solution was added to each trypsinized cells at a volume per volume ratio of 1:30. Following 1 h incubation at 37 °C, cells were washed and the signal was detected fluorescence intensity at Ex = 488 nm/Em = 530 nm using fluorescence microplate reader. For caspase 3 assay, cells were re-suspended in lysis buffer on ice for 10 min. After the lysis step, cytosolic extracts were collected by centrifugation and mixed with Dithiothreitol (DTT) contained reaction buffer and DEVD-chromophore p-nitroaniline (p-NA) substrates. DEVD is an amino acid sequence (Asp-Glu-Val-Asp) cleaved by caspase 3. The detection buffer mixed with cellular supernatants was incubated at 37 °C for 90 min and the absorbance was measured at OD 400-405 nm. For caspase 8 assay, each trypsinized cells were aliquoted, 300 µL, into microtubes. FITC-IETD-FMK was added into each tube and incubated for 1 h at 37 °C incubator (5% CO 2 ). Following incubation, FITC-labeled, activated caspase 8 in HCECs was detected by fluorescence plate reader at Ex = 485 nm/Em = 535 nm.

Measurement of reactive oxygen species (ROS
Annexin V FITC/PI staining for quantification of apoptosis. The late stages of cell death, resulting from either apoptosis or necrosis processes, were detected using a FITC Annexin V apoptosis detection kit II (BD Biosciences, San Jose, CA, USA), according to the manufacturer's instructions. Briefly, MXF-treated HCECs were washed with cold DPBS and then re-suspended using 1 × binding buffer (5 × 10 5 cells/mL); 10 μg of purified Terminal deoxynucleotidyl transferase (TUNEL) assay. For the detection of fragmented DNA, due to apoptosis at the cellular level in HCECs, TUNEL assay was performed using the APO-BrdU TUNEL assay kit (catalog number: A23210: Molecular Probes), according to the manufacturer's protocol. HCECs were seeded at a density of 6 × 10 4 cells per milliliter and grown on Nunc Lab-Tek II chamber slides (Thermo Fisher Scientific; Waltham, MA, USA) and 0, 0.125, 0.25, and 0.5 mg/mL of MXF were treated for 24 h. The cells were fixed with 3.7% paraformaldehyde for 10 min at room temperature, and permeabilization was carried out using 0.1% triton x-100 for 5 min at room temperature. Following the washing of steps with DPBS, the cells were blocked using 1% BSA (Sigma-Aldrich) in DPBS for 30 min at RT. The chamber slides were incubated overnight at 4 °C, which were labeled using TdT enzyme and anti-BrdU mixture solution. www.nature.com/scientificreports/ Gatan, Warrendale, PA, USA). The length of the electron micrograph was measured using the GMS software (Gatan).
Ex vivo porcine corneal endothelial toxicity. Fresh pig eyes were purchased from the local slaughter house. The eyes were enucleated immediately after the death and transferred within 6 h with ice. Corneal button was harvested and cut in quadrants with a sharp surgical blade. Each segment was stained with 0.005% trypan blue mixed with minimum essential medium (MEM) for 5 min. Corneal endothelial cell viability was assessed by examining the blue-stained area under an inverted-phase contrast microscope. After the baseline viability assessment, the corneal segments were incubated with the exposure to various concentrations of MXF at 37 °C in a 5% CO 2 and 95% air-humidified atmosphere for 24 h. The tissue culture medium was serum-free MEM containing L-glutamine (2 mM), NaHCO3 (20 g/L), penicillin (100 IE/mL), and streptomycin (0.1 mg/ mL). After incubation, the corneal segments were fixed in 4% paraformaldehyde and treated in 30% sucrose. The corneal tissues were embedded in OCT compound, and then cut into 15-μm-thick horizontal sections with a cryostat. For evaluation the MXF-induced corneal apoptosis, we carried out TUNEL staining (In Situ Cell Death Detection Kit, Fluorescein; Roche Applied Science, Mannheim, Germany). According to the commercial protocol, the tissue sections were permeabilized with 0.1% Triton X-100 on ice for 2 min. and then incubated with the TdT enzyme at 37 °C for 1 h, and washed with PBS (5 min, shaking, three times). For filamentous actin staining, TRITC-conjugated phalloidin (1 µg/mL; Sigma-Aldrich) was added for 20 min and rinsed three times with PBS (5 min per rinse). The counterstaining of the cellular nucleus was performed with DAPI. The slides were examined with an LSM800 confocal laser scanning microscope, and the digital images were analyzed with ZEN lite 2012 software.
Statistical analysis. The data were presented as mean ± standard error, and the statistical significance was determined by ANOVA and Dunnett's multiple comparison test. P values of less than 0.05 were regarded as significant, according to GraphPad Prism Ver. 5.01 (GraphPad Software Inc., La Jolla, CA, USA).

Data availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.