Quaternary Ammoniums and Other Preservatives’ Contribution in Oxidative Stress and Apoptosis on Chang Conjunctival Cells

Abstract

purpose. To investigate some of the toxicity mechanisms of 10 preservatives currently used in ophthalmic solutions in vitro.

methods. A continuous human conjunctival cell line was treated with different concentrations of various preservatives for 15 minutes and for 15 minutes followed by 24 hours of cell recovery: three benzalkonium chlorides (BACs) with different hydrocarbon chain length, benzododecinium bromide (BOB), cetrimide (Cet), phenylmercuric nitrate (PM), thimerosal (thi), methyl parahydroxybenzoate (MPHB), chlorobutanol (clb), and EDTA. An inhibition study was then conducted using a 1-hour vitamin E pretreatment followed by a 15-minute BAC treatment. Membrane integrity was assessed using a neutral red test and chromatin condensation with a Hoechst 33342 test. Reactive oxygen species were measured using dichlorofluorescein diacetate test for H₂O₂ production and hydroethidine test for O₂ ^.− production. These tests were performed using microplate cold light cytofluorometry. Cell size and DNA content were also analyzed using flow cytometry. Confocal microscopy was used to explore morphologic changes.

results. A significant decrease of membrane integrity with chromatin condensation was observed with all the quaternary ammoniums tested at concentrations of 0.005% and higher. The effect was amplified after 24 hours of cell recovery. The other preservatives tested did not decrease membrane integrity. H₂O₂ production was observed with all the preservatives, whereas O₂ ^.− production was significantly higher with the quaternary ammoniums at 0.005% and 0.01%, compared with the other preservatives. Flow cytometry results confirmed the cytotoxicity observed with cold light cytofluorometry.

conclusions. The quaternary ammoniums tested (BAC, BOB, and Cet) were the most cytotoxic preservatives in the current model. An apoptotic mechanism appeared to be present at low concentrations of quaternary ammoniums, whereas a necrotic process appeared at higher concentrations. Superoxide anions may play an important role in tissue damage induced by preservatives in ocular surface disorders.

Preservatives are used in most ophthalmic preparations, including eye drops and contact lens solutions. Although topically administered medications are increasingly used with apparent safety and good tolerance, there is growing evidence that long-term use of topical drugs can induce changes in the ocular surface and may often produce damage to conjunctival and corneal epithelial cells. There have been several reports of the toxic effects of prolonged topical treatments, partly due to the preservatives associated in the formulation of such treatments. ^{1 2 3} In the eye, preservative turnover is very slow, and quaternary ammonium molecules can be retained in ocular tissues up to 7 days. ⁴ The lipophilic nature of some preservatives causes them to bind to the ocular tissues immediately after topical application. Previous studies by Burstein ³ have shown that topically applied benzalkonium chloride (BAC), the most commonly used preservative in ophthalmic solutions, can cause morphologic disruption of the corneal epithelium at high concentrations. ³ In addition, there is evidence that clinical concentrations of BAC may change the ionic resistance of the cornea by intercalating into cellular membranes, which results in increased permeability. ⁵ Three types of mechanisms have been described: detergent effects causing loss of tear film stability, toxic effects to the corneal and conjunctival epithelia, and immunoallergic reactions. ^{2 6 7} Furthermore, repeated doses of preserved eye drops can lead to a cumulative effect, because the preservatives are in prolonged contact with the epithelium. Several studies have confirmed the participation of preservatives in induction of ocular surface inflammation, ^{8 9} allergy, ⁶ fibrosis, ¹⁰ and dry eye syndrome. ^{11 12} Preservatives are also suspected of strongly increasing the risk of failure of trabeculectomy in glaucoma. ^{13 14 15 16}  

In vitro models have been developed to predict the cytotoxic potential of preservatives. These models were essentially based on corneal epithelial cells ^{17 18} or on other epithelial systems with characteristics similar to those of the superficial layer of the corneal epithelium (Madin–Darby canine kidney cells). ¹⁹ The human continuous conjunctival cell line has also been useful for ocular toxicological studies. ^{20 21 22} We recently showed that BAC is a strong proapoptotic agent in Chang’s conjunctival cells. ²³  

The purpose of this study was to investigate, by flow cytometry and microplate cold light cytofluorometry, the cytotoxicity of 10 of the most common preservatives used in ophthalmic solutions. Because stimulation of reactive oxygen species (ROS) constitutes one of the mechanisms of cytotoxicity, we investigated ROS production induced by preservatives in a well-adapted cellular model for in vitro cytotoxicity. Moreover, the protective effect of vitamin E against the cytotoxicity of preservatives was explored. The new technique of microplate cold light cytofluorometry for cytotoxicity assays has been validated in Chang conjunctival cells in previous studies ^{23 24 25} and allows the use of numerous fluorescent probes directly on living cells. Their specificity, sensitivity and standardization ensure that they are well-adapted to cellular heterogeneity and comply with the requirements of cellular pharmacotoxicology screening procedures. Thus, this study could be better performed physiologically on live rather than dead cells, because labile markers can be significantly affected by the use of extraction techniques. Therefore, we analyzed, in the Chang’s human continuous conjunctival cell line, the immediate and delayed actions of different concentrations of preservatives on membrane integrity, cell size, DNA condensation, and ROS production. Variations of labile markers (ROS) were therefore instantaneously detected, thereby providing reliable data. To our knowledge, this is the first report to describe the relation between oxidative stress and apoptosis after preservatives treatments and to compare in a similar biologic way the cytotoxicity of 10 different preservatives. Results thus obtained on the human conjunctival cell line may contribute to a better understanding of preservative cytotoxicity. 

Materials and Methods

Conjunctival Cell Line

Wong–Kilbourne-derived human conjunctival epithelial cells, an established cell line (Wong–Kilbourne derivative of Chang conjunctiva, clone 1-5c-4, American Type Culture Collection [ATCC] certified cell line [CCL], 20.2), were cultured under standard conditions (humidified atmosphere of 5% CO₂ at 37°C) in Dulbecco’s minimum essential medium (DMEM; Eurobio, Les Ulis, France) supplemented with 10% fetal bovine serum (Dominique Dutscher, Brumath, France), 1% glutamine (Eurobio), 0.1% ampicillin (Panpharma, Fougères, France), and 2% kanamycin (Bristol Myers Squibb, Paris, France). Cells from passages 6 through 17 (after ATCC initial passage 65) were used in all experiments. Normal culture development was assessed daily by phase-contrast microscopy. Confluent cultures were removed by gentle trypsin incubation, and cells were counted. They were then seeded into 96-well culture plates for microtitration analysis (5,000 cells per well; [Nunc, Roskild, Denmark] and plated in 15-cm² flasks (Nunc) for flow cytometric analyses. Cultures were kept at 37°C for 24 hours. After subconfluence was attained (culture surface covering nearly 70%), cells were exposed to the different formulations. Because this cell line spontaneously undergoes apoptosis at 100% confluence, ²⁵ 70% of confluence was thus chosen to avoid any artifact in membrane integrity assays. 

Preservative Treatments

Five different preparations of quaternary ammonium molecules were examined. Three formulations of BAC were tested: (1) C14 benzalkonium chloride, alkyl dimethylbenzylammonium chloride (100% 14-carbon alkyl; BAC₁₀₀); (2) C14/C12 benzalkonium chloride (58% 14-carbon alkyl; 32% 12-carbon alkyl; BAC₅₈); and (3) C14/C12 benzalkonium chloride (32% 14-carbon alkyl; 58% 12-carbon alkyl; BAC₃₂). Their antiseptic action is effective at low concentrations and is due to the hydrocarbon chain length. The maximum activity is obtained with the C14 molecules and the minimum with the C8 and C18 preservatives. ²⁶ In eye drops, the precise composition of BAC is almost never known, because a mixture of C12 and C14 chains is mostly used. We therefore decided to compare the eventual difference in toxic effects of various preparations of BAC that have different antiseptic activities. 

Benzododecinium bromide (BOB) and cetrimide (Cet) were also tested. The five quaternary ammoniums were each tested at concentrations of 0.00001%, 0.0001%, 0.001%, 0.005%, and 0.01%, the concentration used in most eye drops being 0.01%. 

Five other preservatives were tested at five concentrations: (1) phenylmercuric nitrate (PM) at concentrations ranging between 0.000001% and 0.001%, its usual concentration being 0.001%; (2) thimerosal (thi) at concentrations ranging between 0.000004% and 0.004%, usually used at 0.004%; (3) methyl parahydroxybenzoate (MPHB) at concentrations ranging between 0.00003% and 0.03%, its usual concentration being 0.03%; (4) chlorobutanol (clb) at concentrations ranging between 0.00005% and 0.05%, its usual concentration being 0.5%, because high concentrations of this drug are required to produce antimicrobial effects, and we could not obtain this concentration because other excipients are needed to make clb soluble; and (5) EDTA at concentrations ranging between 0.00001% and 0.01%, the most common concentration used being 0.01%. 

An inhibition study was performed using a 1-hour vitamin E pretreatment followed by a 15-minute BAC 0.001% treatment. 

All preservatives and vitamin E were provided by Transphyto, Clermont-Ferrand, France. All dilutions were realized in culture medium. The complete culture medium was used as a negative control. 

Durations of cell treatments with preservatives were chosen as a compromise between in vitro and in vivo data currently available on preservatives and in line with our previous work using the same conjunctival cell line. In vitro, it has been demonstrated that a 100-second application of 0.007% BAC produces lysis of 50% of conjunctival cells. ²⁷ A 1-hour application of 0.0013% to 0.007% BAC solution on epithelial corneal cells also produces a 50% decrease in membrane integrity. ²⁸ In vivo, in corneal and conjunctival tissues, BAC has a half-life of 20 hours for the epithelium and 11 hours for the total conjunctiva. ⁴  

In the present study two incubation times were therefore applied to control and treated cells: 15 minutes of treatment and 15 minutes followed by 24 hours of cell recovery in normal culture medium, as performed in our previous studies. ^{23 24 25} The 24-hour cell recovery period was also tested as a way of approaching the clinical conditions in which the conjunctival tissue may recover after eye drop instillation. 

Experimental Procedures

Experiments were performed using microplate cold light fluorometry, which allows fluorometric detection (280–870 nm) with high sensitivity (picograms to femtograms per milliliter) and specificity. Fluorometry was performed with a microplate cytofluorometer ²⁹ (Fluorolite 1000; Dynex; Cergy Pontoise, France). According to the recommendations of the European Centre for the Validation of Alternative Methods (ECVAM), three cellular markers were evaluated: cellular viability, cellular proliferation, and cellular metabolism with ROS production. ³⁰  

To complete these results, cell size and DNA content were also analyzed by flow cytometry. All flow cytometric measurements were performed on a commercially available flow cytometer (EPICS XL; Beckman Coulter, Miami, FL) equipped with an argon laser emitting at 488 nm, using software provided by the manufacturer (EPICS XL system II; Beckman Coulter) for data analysis. 

This new and original technique allows direct use of numerous fluorescent probes directly on living cells and allows analysis of 96 wells in less than 1 minute. Furthermore, each cell sample can be considered sufficiently similar to the other samples. ²⁹ All fluorescent probes were added to live cells, in mostly physiological conditions, because this method allows detection of the fluorescent signal directly in the microplate cytofluorometer. 

Four different tests were used according to previously validated methods in a Chang’s cell line ^{20 22} and other cell systems. ^{31 32} Briefly, membrane integrity, closely correlated with cellular viability, was evaluated with neutral red (Fluka, Ronkonkoma, NY) using fluorometric detection (excitation, 535 nm; emission, 600 nm). Neutral red was used at 50 μg/ml. In accordance with the validated protocol of Borenfreund and Puerner, ³³ 200 μl per well of medium containing neutral red was added to living cells, and the microplates were incubated for 3 hours at 37°C in atmosphere with 5% CO₂. The neutral red fluorescence was measured as previously described. ²⁹  

H₂O₂ was detected with the 2′,7′-dichlorofluorescein diacetate (DCFH-DA; Molecular Probes, Eugene, OR) dye added to live cells before any treatment, as previously described. ³² This probe is a nonfluorescent cell-permanent compound currently used in flow cytometry that we adapted to microplate cytometry. Once inside the cell, it is cleaved by endogenous esterases and can no longer pass out of the cell. The de-esterified product becomes the fluorescent compound 2′,7′-dichlorofluorescein on oxidation by ROS. The fluorescent signal detected (excitation, 490 nm; emission, 535 nm) has been demonstrated to be proportional to ROS production. ^{31 32}  

O₂ ^.− was detected using hydroethidine (Molecular Probes). It was oxidized to the fluorescent ethidium cation by O₂ ^.−, allowing the cation to bind to nuclear DNA with an extensive fluorescent enhancement. ³⁴ The probe was used on cells at 5 μM after 10 minutes (excitation, 485 nm; emission, 600 nm). 

Hoechst 33342 (Molecular Probes) is a specific UV fluorescent probe (excitation, 360 nm; emission, 450 nm). It specifically reacts with the DNA, at adenine and thymine levels, by intercalation after 30 minutes. ^{35 36} This probe was used on cells at a final concentration of 10 μg/ml. One microliter of propidium iodide (Sigma, St Louis, MO) at 0.5 mg/ml was added to the Hoechst 33342 solution to control necrosis of cells. In all experiments, the background fluorescence was determined on wells without cells but containing the dye solution and was deduced from all control and treated wells. 

Microplate cold light cytofluorometry results were obtained in fluorescence units and were expressed as a percentage of the control. Wells containing cells with complete culture medium but without any treatment were used as the control. Each drug concentration was tested in six wells, and each experiment was performed in triplicate. Statistical comparisons were performed using an analysis of variance (ANOVA) test to compare the five quaternary ammoniums. The Mann–Whitney test and the z correlation test at a 0.05 level of significance were also performed (Statview IV for Windows; Abacus, Berkeley, CA). 

Alteration of cell size after preservative treatments was confirmed with flow cytometric analysis of forward scatter on a linear mode performed 15 minutes after the treatment. Cells were trypsinized, washed with cold phosphate-buffered saline (PBS), and analyzed for size. At least 3000 cells were analyzed per sample. 

After 15 minutes of treatment, cells were trypsinized, washed with cold PBS, and fixed 10 minutes with 95% ethanol in PBS at −20°C. Samples were washed with cold PBS, stained with propidium iodide at room temperature for 20 minutes, and analyzed on the flow cytometer. The sub-G1 region was determined by a gate defined in the controls in the whole-cell population, as described previously. ^{20 37}  

In parallel, standard immunofluorescence was performed to assess morphologic patterns of cells. Cells were cultured on slides and treated with preservatives for 15 minutes. They were washed with PBS and fixed as previously described. Phalloidin (Alexa 488; 200 units/ml, Molecular Probes) was then added to explore for F-actin. After 30 minutes of incubation, cells were washed in PBS. Propidium iodide was added to mark cell nuclei before examination with a confocal epifluorescence microscope (E800 PCM 2000; Nikon, Tokyo, Japan). 

Results

Cellular Viability and Membrane Integrity Evaluation

Membrane integrity significantly decreased after 15 minutes of treatment with all the quaternary ammoniums tested (Fig. 1A) . This toxicity appeared at concentrations of 0.005% and 0.01%, and membrane integrity decreased between 20% and 36% of the control value (P < 0.001 compared with control for all preservatives). After 24 hours of cell recovery (Fig. 1B) , significant cellular damage was found at 0.001% and above. The same decrease of membrane integrity was observed with BAC₁₀₀, BAC₅₈, and BAC₃₂, ranging between 70% with 0.001% BAC to 30% with 0.01% BAC. With BOB, this decrease varied from 56% at 0.001% to 32% at 0.01%, and with Cet, it varied from 57% at 0.001% to 31% at 0.01%. No significant difference was found when the five quaternary ammoniums were compared using ANOVA. 

Clb, EDTA, organomercurials (thi and PM), and MPHB did not decrease membrane integrity after 15 minutes or after 24 hours of cell recovery for all the concentrations tested (data not shown). In Table 1 we present the mean values obtained with all the concentrations tested for Clb , EDTA, organomercurials, and MPHB (no significant differences between the five concentrations tested) and the mean values obtained with 0.005% and 0.01% BAC₁₀₀, BAC₅₈, BAC₃₂, BOB, and Cet. 

H₂O₂ Production with DCFH-DA Test

Significant ROS production was observed with all the quaternary ammoniums tested, even at lowest concentrations such as 0.00001% (Fig. 2) . Maximum production was observed at 0.001% for BAC₁₀₀ (mean fluorescence, 169% of the control), BAC₅₈ (mean fluorescence, 220% of the control), BAC₃₂ (mean fluorescence, 176% of the control), and Cet (mean fluorescence, 235% of the control). At higher concentrations, H₂O₂ production decreased, possibly because of the cytotoxicity demonstrated by cell viability analysis. The maximum observed with BOB appeared at 0.01% (mean fluorescence, 255% of the control). Clb, EDTA, organomercurials, and MPHB also showed significant H₂O₂ production, but the maximal production was observed with thi (Fig. 3) . 

O₂ ^.− Production with Hydroethidine Test

Quaternary ammoniums induced an O₂ ^.− synthesis at 0.005% and 0.01% with no difference among the three BAC formulations, BOB, and Cet, according to ANOVA. The maximum was observed for all the quaternary ammoniums at 0.01%: 172% with BAC₁₀₀, 201% with BAC₅₈, 172% with BAC₃₂, 172% with BOB, and 159% with Cet (Fig. 4) . Significant O₂ ^.− production was also observed with thi at 0.004% (mean fluorescence, 124%, P < 0.001 compared with control) and clb at 0.05% (mean fluorescence, 126%, P < 0.001 compared with control), but it was significantly less than the production observed with quaternary ammoniums (P = 0.0004 with thi compared with quaternary ammoniums; P = 0.005 with clb compared with quaternary ammoniums; Fig. 5 ). 

DNA Condensation Evaluation

All quaternary ammoniums and clb molecules induced a concentration-dependent increase of the fluorescence ratio after 15 minutes of treatment, although all the quaternary ammoniums tested did not produce the same intensity of fluorescence compared with control (Figs. 6 A 6B ). Mean fluorescence increased from 122% to 547% with BAC₁₀₀, from 129% to 452% with BAC₅₈, from 100% to 361% with BAC₃₂, from 109% to 367% with BOB, from 119% to 360% with Cet, and from 150% to 260% with clb. Chromatin condensation observed with all the quaternary ammoniums tested at 0.005% and 0.01% was significantly higher than that with clb (mean fluorescence, 370% with quaternary ammoniums, 162% with clb; P < 0.0001 for all concentrations tested). As shown in Figure 7 with Hoechst staining, cells treated with quaternary ammoniums showed, in a concentration-dependent manner, chromatin condensation and fragmentation typical of apoptosis, when compared with control cells. We present in Table 2 the mean values obtained with all the concentrations tested for organomercurials, MPHB, and EDTA (no significant difference among the five concentrations tested and among the four preservatives, according to ANOVA). The chromatin condensation observed was significantly less than that observed with the quaternary ammoniums molecules (mean fluorescence, 119% with EDTA and MPHB and 128% with organomercurials versus 370% with quaternary ammoniums, with P = 0.0022 when EDTA or MPHB were compared with quaternary ammoniums and P = 0.0002 when organomercurials were compared with quaternary ammoniums). 

After 24 hours of cell recovery (Fig. 6C) , the quaternary ammoniums still induced a marked increase of Hoechst fluorescence at 0.001% and higher, ranging from 165% with BAC₁₀₀ to 415% with BAC₃₂. No significant difference was observed between the five quaternary ammoniums. Quaternary ammoniums at 0.005% and 0.01% induced a significant chromatin condensation compared with EDTA, organomercurials, and MPHB (mean fluorescence, 299% with quaternary ammoniums at all concentrations of 0.005% and 0.01%, 121% with EDTA, 118% with organomercurials, 117% with MPHB, P < 0.0001 for all values). The chromatin condensation observed with clb was less significant (mean fluorescence, 241%, P < 0.05) compared with quaternary ammonium molecules (Table 2) . 

Cell Size Analysis

Concentration-dependent toxicity was confirmed by flow cytometry by the alteration of cell size after quaternary ammonium treatment (Fig. 8) . Cells treated with 0.01% BAC₅₈ had a 49% reduction of cell size in comparison with untreated cells. No difference was observed among the five quaternary ammoniums. The other preservatives tested did not show any alteration of cell volume (data shown only for MPHB and three concentrations of BAC₅₈). 

DNA Content

We measured sub-G1 cell population after 15 minutes of treatment with the different preservatives. Normal untreated cells showed a 17%± 5% sub-G1 population. The population of sub-G1 cells was 42% ± 3%, 54% ± 7%, and 74% ± 9% for 0.0001%, 0.001%, and 0.01% quaternary ammonium treatments, respectively (Fig. 9) . The other preservatives tested did not show any significant increase in the apoptotic cell population (sub-G1 population varied from 17% to 23%; data shown only for MPHB and three concentrations of BAC58). 

Morphologic Changes

A concentration-dependent cell retraction was observed after treatment with quaternary ammoniums molecules, whereas no morphologic change was observed with the other preservatives tested (Fig. 10) . 

Correlation between the Different Tests Performed on Preservatives

Membrane integrity and chromatin condensation showed significant negative correlation (r = −0.863, P < 0.0001; Fig. 11 ). Poor but significant correlation was found between H₂O₂ production and membrane integrity or chromatin condensation (H₂O₂ production versus membrane integrity: r = 0.417, P = 0.0003; H₂O₂ production versus chromatin condensation: r = −0.269, P = 0.0104). However, this correlation was only due to the two highest concentrations of quaternary ammoniums tested. A significant negative correlation was shown with membrane integrity and O₂ ^.− production (r =− 0.551, P < 0.0001). The results obtained with the neutral red probe after 15 minutes of treatment were confirmed after 24 hours of cell recovery (r = 0.770, P < 0.0001). At the same time, chromatin condensation was well correlated with the O₂ ^.− production (r = 0.738, P < 0.0001) and was associated with the decrease of membrane integrity after 24 hours of cell recovery (r = −0.697, P < 0.0001). H₂O₂ and O₂ ^.− productions, however, were not correlated. 

Inhibition Study

No cytotoxicity was observed after a 1-hour vitamin E treatment (Fig. 12) . Membrane integrity and chromatin condensation were not altered, compared with complete culture medium alone. No ROS production was detected. After a 1-hour vitamin E pretreatment followed by a 15-minute BAC 0.001% treatment, there was no alteration of membrane integrity (mean fluorescence, 88% with BAC 0.001% versus 120% with a vitamin E pretreatment followed by a BAC treatment; P < 0.001 compared with BAC). Significant decreases in chromatin condensation, H₂O₂, and O₂ ^.− production were observed compared with BAC. However, after a vitamin E pretreatment followed by a BAC treatment, H₂O₂ production was increased (mean fluorescence, 146% with vitamin E with BAC versus 105% with BAC alone; P < 0.001 compared with BAC), whereas no significant difference was observed when O₂ ^.− productions were compared. 

Discussion

Cationic agents are used in pharmaceutical preparations for antimicrobial preservation because of their ability to lyse microbial cellular membranes. The quaternary ammonium cationic surfactants (BAC, BOB, Cet) we tested have detergent-like properties that may cause cell damage by emulsification of the cell wall lipids. Numerous clinical and biologic side effects of surfactant preservatives have been described, such as ocular irritation (with lacrimation, hyperemia, photophobia, and edema), punctate keratitis, gray corneal epithelial haze, pseudomembrane formation, decreased corneal epithelial microvilli, and cytotoxicity to the corneal epithelial cells. ^{1 2 5} In addition, cell permeability caused by quaternary ammoniums is potentiated when EDTA is used. BAC disrupts the bacterial external membrane, and EDTA disorganizes the cell envelope. ³⁸  

Organomercurials include thi, phenylmercuric nitrate or acetate, and mercuric oxide. They bind to the cell membrane protein sulfhydryl groups, causing an increase in cellular permeability. Adverse ocular side effects due to these preservatives are rare. The most striking side effect is mercurial deposits in various ocular tissues. These compounds have also induced allergic reactions in sensitized persons; erythema, edema, and hyperemia of the eyelids; or conjunctivitis. Thi, however, has been shown to produce cytotoxicity for corneal epithelial cells. ^{39 40 41 42 43 44}  

Clb is mainly known to induce cytotoxicity in corneal epithelial cells. In previous studies, it was reported that occasional use (twice daily up to 12 days) of a clb-preserved artificial tear resulted in only modest exfoliation of corneal epithelial cells (i.e., up to a maximum of 14%). ⁴⁵ These changes were reversible, and it was therefore suggested that the eye could adapt to repeated use of these preserved artificial tears. ^{46 47}  

Our results revealed new aspects of preservative toxicity. After a 15-minute application of all the quaternary ammoniums tested, membrane integrity of treated cells was altered, whereas there was no variation of membrane integrity with the other preservatives. The difference was significant among all the quaternary ammoniums tested at 0.005% and 0.01% and among the other preservatives, even at the highest concentrations (P = 0.002 compared with quaternary ammoniums). After a 0.01% quaternary ammonium treatment, cells showed characteristics of immediate abundant lysis, with membrane debris and low cell size on flow cytometric analysis graphs. The effects of all the quaternary ammoniums tested were progressive. There was a 75% decrease of membrane integrity after a treatment with quaternary ammoniums at 0.005% and 0.01% that persisted after 24 hours. No significant difference was observed among the five quaternary ammoniums analyzed, and the three BAC tested were similar despite supposed differences in their antimicrobial activities. In contrast, the other preservatives tested showed no alteration of membrane integrity, even at high concentrations and even after 24 hours of cell recovery. 

However, ROS evaluation showed that H₂O₂ production increased with the noncytotoxic preservatives and with quaternary ammonium concentrations less than or equal to 0.001%, even though there was no alteration of cell viability, whereas H₂O₂ production was decreased with quaternary ammoniums at high concentrations (0.005% and 0.01%), possibly because of the decrease of cell viability observed for these two concentrations. Concerning O₂ ^.−, there was a marked increase in production with all quaternary ammoniums tested at 0.005% and 0.01%, whereas a nonsignificant increase was observed at the lowest concentrations. O₂ ^.−, but not H₂O₂, could therefore play a role in the decrease of membrane integrity, because these two parameters were well correlated. Furthermore, O₂ ^.− was also associated with chromatin condensation—these two parameters being very significantly correlated—suggesting that O₂ ^.− may induce apoptosis or at least participate in epithelial cell degeneration. In a cardiomyocyte model, ROS clearly induced apoptosis. ⁴⁸  

Superoxide-generating systems have been demonstrated to be cytotoxic for cultured cells, to degrade polysaccharides and DNA, to promote peroxidation of membrane lipids, to alter vascular permeability, and to potentiate inflammation. ⁴⁹ In addition, these free radicals may play a significant role in the generation of chemotactic factors and in augmentation of the inflammatory response by inactivation of normally available serum antiproteases that are known to neutralize the effects of leukocytic proteases. ⁵⁰ It was also recently shown that topical application of a free radical scavenger (ascorbic acid) decreased oxygen radical tissue damage after excimer keratectomy and reduced the acute inflammatory reaction efficiently. ⁵¹ An inhibition study using a 1-hour vitamin E pretreatment followed by a 15-minute BAC 0.001% treatment was conducted in our model. Chromatin condensation and O₂ ^.− production were significantly decreased compared with BAC alone. The same levels of fluorescence were found after vitamin E treatment or vitamin E associated with BAC treatment. H₂O₂ production was observed, whereas there was no alteration of membrane integrity. Reactive oxygen metabolites thus appear to play an important role in cytotoxic effects and in the amplification of the inflammatory process. 

A slight O₂ ^.− production was also shown with clb and thi at the highest concentrations tested, without any consequence for membrane integrity, even though an increase of chromatin condensation was observed with clb. 

Only a few studies were performed to evaluate the extent of oxygen radical damage to the ocular surface. Presence of oxygen free radicals has been demonstrated in the tear fluid of patients with dry eye syndrome ⁵² or in vivo after excimer laser corneal surgery. ^{53 54 55 56} In a rabbit model, presence of lipid peroxidation was demonstrated in the superficial corneal stroma after excimer surgery. The lipid peroxidation was hypothesized to be from oxygen free radicals generated by the infiltrating polymorphonuclear cells at the site of tissue damage. ⁵⁶ Furthermore, Shimmura et al. ⁵³ clearly identified the specific species of radicals formed (OH.) by the excimer laser and the cytotoxic effects on keratocytes in a contamination-free culture. They showed that hydroxyl radicals may be partly responsible for stromal fibroblast cell apoptosis after excimer laser treatment. In our study, O₂ ^.− production was thus observed when membrane integrity decreased (after 15 minutes and also after 24 hours of cell recovery). 

Whatever the mechanisms involved, these alterations observed with quaternary ammoniums occurred together with morphologic modifications (cell size reduction, chromatin condensation, cytoskeleton retraction, increased sub-G1 population), all highly suggestive of the apoptotic process. According to ECVAM recommendations, cellular DNA has been evaluated using the Hoechst test. Hoechst 33342, a DNA fluorochrome, unlike propidium iodide, is not excluded by live or apoptotic cells. It has been observed that short exposure of cells to low concentrations of Hoechst 33342 leads to strong labeling of apoptotic cells. ⁵⁷ Live cells, however, require much longer incubation with Hoechst to obtain a comparable fluorescence intensity. Supravital uptake of Hoechst combined with exclusion of propidium iodide (to identify necrotic and late apoptotic cells) has been proposed as an assay of apoptosis. ³⁵ In the present study, apoptosis was distinguished from necrosis using the Hoechst 33342 and the neutral red tests. Cells in apoptosis are characterized by decreased membrane integrity (detectable using the neutral red test) and by chromatin condensation, which increases Hoechst 33342 fluorescence. ^{57 58} The combination of these two parameters does not occur in normal cells or in cells undergoing necrosis. The quaternary ammoniums tested showed a concentration-dependent increase of chromatin condensation (P < 0.001 compared with the control at 0.005% and 0.01%) associated with a decrease of membrane integrity. This result is in accordance with previous studies conducted with quaternary ammoniums, in which different techniques were used on the same cell line. ²⁵ This increase of Hoechst 33342 fluorescence was also observed with clb at all concentrations, although there was no alteration of the cell cycle and of membrane integrity, only an increase in O₂ ^.− production at the higher concentration tested, whereas there was a nonsignificant increase of fluorescence with EDTA, organomercurials, and MPHB. A concentration-dependent apoptotic process was confirmed using flow cytometric analysis of DNA content with quaternary ammoniums but was not obtained with the other preservatives tested, even with clb. ⁵⁷  

We have recently shown that BAC may induce two distinct patterns of cell death: apoptosis and necrosis. Necrosis was found to be induced by high concentrations, whereas apoptosis appeared at lower concentrations with an expression of the apoptotic marker Apo 2.7 and typical apoptotic changes in DNA content. ^{21 25} A similar phenomenon may therefore occur with the three different BAC solutions, BOB, and Cet. 

Experimental models showed that BAC is, at least to a large extent, responsible for toxic and/or immunoinflammatory effects on ocular structures. ^{9 24} However, any extrapolation of information obtained from the Wong–Kilbourne cell line to the ocular surface must be made with caution. The effects observed in vitro may be concentration dependent. Unlike the in vitro situation, in vivo there is almost instantaneous dilution and continuous action of the lids. Therefore, the concentrations used in vitro cannot be obtained in vivo, and consequently the toxic effects observed in vitro may be less important in vivo. Furthermore, the preocular mucin and glycocalyx, which is normally present and protects the apical cell membrane in vivo may be absent in the Chang cells in culture, explaining a higher susceptibility to these preservatives at the concentrations tested. However, it was demonstrated that residual amounts of BAC could also be detected in the conjunctival epithelium 9 days after a single topical application. ⁴ Furthermore, the presence of delayed tear clearance may elevate the tear concentration of these preservatives to toxic levels. This may explain why medicamentosus is common in patients with delayed tear clearance. ⁵⁹ In addition, renewal of ocular surface epithelia can explain, at least in part, the less important toxicity observed in vivo, although it has been shown that preservatives induce corneal epithelial damage and limbal and conjunctival infiltration by immunocompetent cells. ⁶⁰  

Furthermore, interactions between inflammatory reactions and apoptosis are present in the ocular surface. ⁶¹ The present study also showed a clear correlation between apoptosis and superoxide anion production, which suggests that these reactive radical species may play an important role in the tissue damage that occurs in ocular surface disorders. It remains to be determined whether this free radical production constitutes the origin or the consequence of the cytotoxicity of quaternary ammoniums. 

Inflammation, free radical production, and apoptosis therefore seem to be closely related processes in ocular cells. In the near future, development of cytoprotective drugs, which could protect the ocular surface from drug-induced toxicity, is of great importance. Nevertheless, this study confirms that quaternary ammoniums, including the less extensively studied BOB and Cet, should be avoided as far as possible, especially in chronic ocular surface diseases such as glaucoma, dry eye, and allergy. The other preservatives tested were found to be less toxic and may thus be developed for prolonged use in eye drops. Some preservative-free ophthalmic solutions are also available in single-use doses or in multidoses. ^{62 63} In the future, they should be a successful substitute for the classic multidose preparations. 

 Submitted for publication August 17, 2000; revised October 27, 2000; accepted November 2, 2000.

 Commercial relationships policy: N.

 

Figure 1.

 

Membrane integrity evaluation with neutral red test after treatment with different concentrations of various quaternary ammoniums. (A) Fifteen minutes of treatment: There was a significant decrease of membrane integrity after 0.005% and 0.01% treatments. (B) Fifteen minutes of treatment followed by 24 hours of cell recovery: Significant cellular damage was found at concentrations of 0.001% and higher. (A, B) *P < 0.001 compared with control.

Table 1.

 

Membrane Integrity Evaluations

	15-Minute Treatment	15-Minute Treatment Followed by 24 Hours of Cell Recovery
Control (n = 60)	100	100
Quaternary ammoniums, 0.005 and 0.01% (n = 60)	24 ± 4.3*	28 ± 3.8*
Chlorobutanol (n = 60)	112 ± 4.3	97 ± 10.7
EDTA (n = 60)	107 ± 9.5	92 ± 16.3
Organomercurials (n = 120)	110 ± 16.3	91 ± 18.8
MPHB (n = 60)	113 ± 7.5	101 ± 11.3

 Data are expressed as mean percentages of control (±SD). Quaternary ammoniums: means of BAC, BOB, and Cet; organomercurials: means of thi and PM; clb, EDTA, organomercurials, MPHB: means of all concentrations tested.

*  P < 0.0001 compared with control and all other preservatives. No difference between all other preservative groups.

Figure 2.

 

H₂O₂ production evaluation after 15 minutes of treatment with quaternary ammoniums. *P < 0.001 compared with control.

Figure 3.

 

H₂O₂ production evaluation after 15 minutes of treatment with preservatives. *P < 0.001 compared with control. Bar shading: 1, 1/1000 dilution; 2, 1/100 dilution; 3, 1/10 dilution; 4, 1/2 dilution; and 5, the higher concentration tested for all preservatives: 0.004% for thi, 0.001% for PM, 0.03% for MPHB, 0.05% for Clb, and 0.01% for EDTA.

Figure 4.

 

O₂ ^.− production evaluation after 15 minutes of treatment with quaternary ammoniums.* P < 0.001 compared with control.

Figure 5.

 

O₂ ^.− production evaluation after 15 minutes of treatment with preservatives. *P < 0.001 compared with control. Key to shading is in Figure 3 .

Figure 6.

 

Chromatin condensation evaluation with Hoechst 33342 test after treatment with different concentrations of various preservatives. Fifteen minutes of treatment with (A) quaternary ammoniums: a significant concentration-dependent increase in fluorescence was observed, indicating an increase in chromatin condensation; (B) clb: a significant increase in fluorescence was found for all concentrations tested; and (C) quaternary ammoniums followed by 24 hours of cell recovery: chromatin condensation was still observed at 0.001% and higher concentrations. *P < 0.001 compared with control.

Figure 7.

 

Hoechst 33342 nuclear staining of the cultured cells. (A) Normal cell nuclei. Nuclei of cells treated with (B) 0.03% MPHB: no modifications compared with control; (C) 0.00001%, (D) 0.005%, and (E) 0.01% BAC showing a characteristic apoptotic peripheral condensation and fragmentation of chromatin, in a concentration-dependent manner. Magnification, ×40.

Table 2.

 

Chromatin Condensation Evaluations

	15-Minute Treatment	15-Minute Treatment Followed by 24 Hours of Cell Recovery
Control (n = 60)	100	100
Quaternary ammoniums, 0.005 and 0.01% (n = 60)	370 ± 18.2*	299 ± 19.3*
Clb (n = 60)	162 ± 16.3*	241 ± 20.7*
EDTA (n = 60)	119 ± 10.2	121 ± 10.4
Organomercurials (n = 120)	128 ± 18.1	118 ± 17.6
MPHB (n = 60)	119 ± 18.0	117 ± 17.7

 See Table 1 for explanation of data.

Figure 8.

 

Flow cytometric analysis of cell size after 15 minutes of treatment: (A) Control, (B) MPHB 0.03%, (C) BAC 10^{− 4}%, (D) BAC 10^{− 3}%, and (E) BAC 10^{− 2}%. FS, forward scatter.

Figure 9.

 

Flow cytometric analysis of DNA content 15 minutes after cell treatments: (A) Control (untreated cells) DNA. The number of cells is represented as a function of fluorescence (FL). The percentage of apoptotic cells (sub-G1 population) detected was (A) 20%, control; (B) 23%, MPHB 0.03%; (C) 46%, BAC 10^{− 4}%; (D) 53%, BAC 10^{− 3}%; and (E) 79%, BAC 10^{− 2}%.

Figure 10.

 

F-actin exploration using phalloidin. (A) Control cells and cells treated with (B) BAC 10^{− 4}%, (C) BAC 10^{− 2}%, and (D) Cet 10^{− 2}%. A concentration-dependent decrease of cell size associated with chromatin condensation and cell disorganization was observed with quaternary ammoniums tested at 10^{− 2}%. Magnification,× 1000.

Figure 11.

 

Correlations between the different tests performed. The results obtained with all the preservatives tested are presented. The different test compared are RN (neutral red): membrane integrity evaluation; H₂O₂: hydrogen peroxide production evaluation; Hoechst: chromatin condensation evaluation; and O₂ ^.−: peroxide anion production evaluation. The decrease of RN is correlated with an increase of O₂ ^.− production and with an increase of Hoechst fluorescence. H₂O₂ variations cannot be interpreted.

Figure 12.

 

Evaluation of membrane integrity, chromatin condensation, and ROS production after a 1-hour vitamin E treatment and a 1-hour vitamin E treatment followed by a 15-minute BAC 0.001% treatment.*** P < 0.0001; **P < 0.001;* P < 0.05 compared with BAC 0.001%.