Biochemistry and Molecular Biology

Surface Chemistry Study of the Interactions of Benzalkonium Chloride with Films of Meibum, Corneal Cells Lipids, and Whole Tears
Georgi A. Georgiev,1 Norihiko Yokoi,2 Krassimir Koev,3 Elena Kutsarova,1 Slavyana Ivanova,1 Alexander Kyumurkov,4 Albena Jordanova,5 Rumen Krastev,6,7 and Zdravko Lalchev1
PURPOSE. To perform a surface chemistry study of the interactions between benzalkonium chloride (BAC), a common preservative used in ophthalmic formulations, and tear lm (TF) constituents. METHODS. The interactions between BAC and human tears, meibum, and rabbit corneal cell lipid extracts at the airwater interface were examined in vitro during controlled compression expansion of the lm area by a Langmuir surface balance, surface potential measurements, and pendant drop-axisymmetric drop shape analysis (PD-ADSA). Surface pressurearea isotherms and isocycles were used to assess the samples lateral elasticity and capability of compressing and spreading during dynamic area changes. Lipid lm morphology was monitored by Brewster angle microscopy. The viability of BAC-treated Statens Seruminstitut rabbit cornea (SIRC) cell cultures was also examined. The BAC concentration was kept within the clinical range of 0.001% to 0.02%. RESULTS. In the Langmuir balance and PD-ADSA experiments, the interactions between BAC and lipids or tears resulted in (1) impaired lipid spread and formation of discontinuous nonuniform surface layers, (2) increased surface pressurearea hysteresis during compression and expansion, and (3) displacement of the lipids by BAC from the surface. A decrease (50%) in SIRC cell viability was observed. The effects occurred within seconds after BAC exposure, and their magnitude increased with BAC concentration. CONCLUSIONS. The surface chemistry approach used in this study provided molecular-scale insights into the detrimental effect of BAC on TF, which well explain the TF instability and corneal epithelial barrier dysfunction after exposure to BAC in the in vivo human eye. (Invest Ophthalmol Vis Sci. 2011;52: 4645 4654) DOI:10.1167/iovs.10-6271 he cationic surfactant benzalkonium chloride (BAC) is a commonly used preservative, as it is one of the few ingredients that fullls the European Pharmacopoeia challenge test for inhibition of bacterial growth during storage of ophthalmic formulations.1 However, in vivo studies have repeatedly reported that BAC decreases tear lm (TF) noninvasive breakup time and induces dry eye symptoms.2 4 The molecular scale mechanisms of these adverse effects remain unclear. The purpose of the present study was to gain insight into how the interactions with BAC modify the surface properties of individual TF compounds (meibomian lipids) and of whole human tears. In vitro, the samples were spread to form lm over the airwater interface of a Langmuir trough or of a pendant drop. The subphase below the lms consisted of physiological saline solution: with or without BAC applied in the clinical concentration range of 0.001% to 0.02%. The lms were subjected to quasi-equilibrium and dynamic area compression and expansion (Figs. 1A, 1B), and the dependence of the lm surface pressure on area was measured by Langmuir surface balance and by pendant drop-axisymmetric drop shape analysis (PD-ADSA). The surface potentials and the morphology of the Langmuir lms were monitored with a surface potential (V) sensor and Brewster angle microscopy (BAM), respectively. As a detrimental effect of BAC on the corneal surface integrity has been reported frequently,57 Langmuir surface balance experiments were also performed, to study the interactions of BAC with the membrane lipid extract of Statens Seruminstitut rabbit cornea (SIRC) cells. The lipid extract lms mimic the corneal epithelium membranes outer half,8 which is oriented toward the aqueous tear and is exposed to interaction with eyedrop constituents. The surface chemistry data were compared with the changes of SIRC cell viability and capability of forming a conuent cellular monolayer after treatment with BAC. The in vitro surface chemistry results correlated with the recently published4 in vivo ndings on impaired spreading of the tear lm lipid layer (TFLL) and on the barrier dysfunction of the corneal epithelium after treatment of human eyes with BACpreserved timolol.

From the 1Departments of Biochemistry and 4Cellular Biology, Faculty of Biology, University of Soa, Soa, Bulgaria; the 2Department of Ophthalmology, Kyoto Prefectural University of Medicine, Kyoto, Japan; the 3Department of Ophthalmology, University Hospital Aleksandrovska, Soa, Bulgaria; the 5Institute of Biophysics, Bulgarian Academy of Science, Soa, Bulgaria; the 6Natural and Medical Sciences Institute at the University of Tu bingen, Tu bingen, Germany; and the 7 Max Planck Institute of Colloids and Interfaces, Potsdam, Germany. Supported by Contract DO 02-280/2008 from the Bulgarian Ministry of Education and Science. Submitted for publication July 25, 2010; revised January 29 and March 27, 2011; accepted March 28, 2011. Disclosure: G.A. Georgiev, None; N. Yokoi, None; K. Koev, None; E. Kutsarova, None; S. Ivanova, None; A. Kyumurkov, None; A. Jordanova, None; R. Krastev, None; Z. Lalchev, None Corresponding author: Georgi A. Georgiev, Department of Biochemistry, Faculty of Biology, University of Soa, Soa, Bulgaria; ggeorg@biofac.uni-soa.bg.
Investigative Ophthalmology & Visual Science, June 2011, Vol. 52, No. 7 Copyright 2011 The Association for Research in Vision and Ophthalmology, Inc.

MATERIALS

AND

METHODS

Collection of Human Meibum and Whole Tears

The samples for the study were collected from healthy volunteers among the regular laboratory members: three women (23 40 years old) and one man (31 years old). Their surface properties were identical with the ones of meibomian and tear samples obtained from 20

4645

4646

Georgiev et al.

IOVS, June 2011, Vol. 52, No. 7

FIGURE 1. Simplied depiction of a Langmuir trough (A) and of PD-ADSA (B). Both setups perform controlled area compression expansion cycling of the lipidwhole tear sample (by the movable barriers of the trough or by suction in and out of the drop) and enable measuring the dependence of surface pressure () on the lm area (A) during cycling. (C) Arrows: the direction of the area changes. The compression and expansion of (A) isotherms can show signicant hysteresis (C, loop 1) or can closely overlap (C, loop 2).

(12 women and 8 men) healthy individuals (data not shown). None of the volunteers used glasses or contact lenses and did not apply eye cosmetics before the sample collection. All volunteers gave written informed consent. The collection procedures were in accordance with the Declaration of Helsinki and were approved by the Soa University Ethics Committee. Human meibomian lipids were expressed from eyelids by applying pressure with two opposing cotton buds.9 The meibum was collected from the lid margin with a platinum spatula daily for a week, to minimize the effect of day-to-day intraindividual variations. The samples were collected, weighed, and dissolved in chloroform to a unied stock solution with a concentration of 1 mg lipid/mL. Before the experiments, the obtained donor-specic lipid solutions were kept at 80C. The composition of meibum was examined by thin-layer chromatography and revealed the presence mainly of wax and sterol esters, in agreement with a previous report.10 Tears from both eyes of each volunteer were collected with 25-L capillary tubes, pooled, and stored in capsules (Eppendorf, Fremont, CA), as previously described,11 for up to 5 minutes before the PD-ADSA experiments. The tears were collected and tested at least twice from the each person, to account for intraindividual variation.

and fast dynamic compression expansion isocycling. In the quasiequilibrium experiments, the compression rate was 11 mm/min, which is known to be sufciently slow to allow for the reorientation and relaxation of the lipid molecules at the surface.13 To examine the performance of meibum in making rapid area changes, dynamic isocycling was performed with the maximum possible barrier rate (140 mm/min) at which there was no leakage of the lm. Ten consecutive cycles were performed with each sample. Normally, between the rst and third cycles, the (A) loops achieved a stationary shape, and the (A) isocycles were then obtained and analyzed. For certain experiments, the lm surface potential V was registered by using a vibrating plate with a sensor (Spot; Kibron).13,14 All isotherms were repeated at least three times; the difference between the repetitions was up to 2%. The corneal temperature usually ranges between 30C and 33C15,16 and at normal to cold weather conditions TF temperature can be lower due to heat exchange. The presented experiments were performed at 28C. The effects of BAC on the samples surface properties were reproducible at 37C. The morphology of the lms was observed by BAM17 (MicroBAM2; Nima Technology Ltd., Coventry, UK). For experiments on the membrane lipid extract of SIRC cells the protocols were identical with those for the human extract.

Isolation of the Membrane Fraction of SIRC Cells

The plasma membrane fraction was obtained by the procedure of Evans with modications.12 Control experiments tested the effect of BAC (Sigma-Aldrich, St. Louis, MO) on lms by synthetic lipids (Avanti Polar Lipids, Alabaster, AL), such as dipalmitoylphosphatidylcholine (DPPC) and dipalmitoylphosphatidylglycerol (DPPG).

PD-ADSA of Dynamic Isocycles of Whole Human Tears

The experiments were performed with a contact angle meter (CAM 101; KSV, Helsinki, Finland) in pendant-drop mode.18,19 The instrument was equipped with a humidity- and temperature-controlled chamber set at 28C. Whole human tears (volume 30 L) were used to form drops (25 mm2 initial area) at the end of a stainless steel needle attached to a motor-driven syringe (Hamilton, Reno, NV). A dynamic compression-expansion cycle of the drop was completed at a constant rate within 30 seconds. The drop shape was recorded with a video camera. The images were processed and tted to the Gauss-Laplace equation with the built-in software, to obtain the tears area and the surface pressure. The accuracy of surface pressure measurements was 0.5 mN/m. Ten consecutive cycles were performed with each sample. The (A) loops achieved stationary shape between the rst and third cycles, and the (A) isocycles were obtained.

Langmuir Surface Balance Experiments

Surface pressure ()area (A) isotherms were measured with a computer-controlled Langmuir surface balance (Kibron, Helsinki, Finland, equipped with a Trough XL; area 225 cm2, volume 40 mL), by the Wilhelmy wire probe method. Human meibum dissolved in chloroform was spread (43 L of 1 mg/mL) over the airsaline solution interface with a microsyringe (Hamilton, Reno, NV). An acrylic cover was put over the trough to protect against dust contamination and to suppress subphase evaporation. After 15 minutes was allowed for chloroform evaporation, lm area compression was started with two symmetrically moving barriers. For the experiments in which the effect of BAC on meibum lm was tested, before the compression, up to 250 L BAC solution was injected into the saline solution below the lipid lm to the designated preservative concentration in the trough subphase. Two types of measurements were performed: slow quasi-equilibrium compression

Analysis of the Quasi-equilibrium Compression Isotherms

The TFs reciprocal compressibility , Cs1, at a given surface pressure was calculated from (A) quasi-equilibrium compression isotherm using the equation

IOVS, June 2011, Vol. 52, No. 7

C S 1 A

Surface Chemistry of BAC Interaction with TF

4647

d dA

(1)
T

where A is the area at the indicated . The higher the value of Cs1, the lower the lateral elasticity of the surface lms and vice versa.20

Analysis of the Dynamic CompressionExpansion Isocycles

The surface pressurearea curves obtained at compression and expansion of the lm surface can strongly differ (Fig. 1C, loop 1) or closely overlap (Fig. 1C, loop 2) in their course. The discussed effects can be quantitatively evaluated by the integration of (A) expansion compression curves and calculation of the isotherm reversibility (Rv):

conuence, before and after BAC treatment, was monitored, and images were captured regularly with an inverted microscope (Eclipse TS100-F; Nikon, Tokyo, Japan), with a photo documentation system, at 20 magnication. The changes in cell viability due to treatment with BAC were evaluated by MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide) colorimetric assay.25

RESULTS
Adsorption of BAC at the Pure AirWater Interface and at Interface with Predeposited Meibomian Film
The adsorption of BAC at the pure airwater interface was estimated, to determine the critical micelle concentration (CMC) of the preservative (Fig. 2A). The CMC corresponded to 0.01% BAC, at which a plateau of at 39 mN/m was realized. The obtained value concurred with the one determined in another study.7 Thus, the range of clinically used BAC concentrations (0.001% 0.02%) spans more than two concentration regions: below and above the CMC of the preservative. The adsorption of BAC to the airwater interface covered with predeposited meibum lm with initial surface pressure 0 was also measured (Fig. 2B). After BAC was injected into the subphase, if the preservative penetrated the surface lm, it was registered as an increase in surface pressure (). The pene-

R v 100

Af

dA

Al

expansion

Af

(2)

dA

Al

compression

For lms of hydrophobic molecules like the meibomian lipids, the magnitude of the hysteresis is determined by the kinetics with which the compressed surface lms restore their structure and spread at the interface during area expansion.21 Highly reversible lms (Fig. 1C, loop 2) restore their structure and spread at the interface with a rate commensurate to that of area expansion (achieved by the barriers of the Langmuir trough or by the dilation of the pendant drop). Lower reversibility of the lms (Fig. 1C, loop 1) reects slower reorganization and spreading of the surface lms compared with the rate of area expansion.

Evaluation of the Adsorption of BAC at the Pure AirWater Interface and at the Interface with Predeposited Meibomian Film
A multiwell plate trough (Kibron) consisting of 15 individual circular wells, 500 L volume each, was used in these experiments.22,23 For measuring the adsorption of BAC at the pure airwater interface, a preservative solution with the desired concentrations was loaded into the wells, and the equilibrium surface pressures were recorded. For probing the adsorption of BAC to the surface with predeposited meibomian lm, lipids dissolved in chloroform were spread at the airwater interface and 5 minutes were allowed, for the chloroform to evaporate. Various amounts of meibum were deposited so that different initial surface pressure (0) values were achieved. Then 3 L of BAC solution were injected into the subphase to give 0.001% bulk concentration of preservative. The equilibrium increase in the surface pressure () after the injection of BAC was measured. The increments were plotted versus 0, to measure the critical surface pressure CR, the surface pressure at and above which BAC will no longer be capable of penetrating the lipid lm.8

Cultivation of SIRC Cells, Treatment with BAC, and Measurements of Cell Culture Conuence and of Cell Viability
SIRC cells were cultured in Dulbeccos modied Eagles medium (DMEM), supplemented with 10% fetal bovine serum, until 95% to 100% conuence was reached. The conuent cellular monolayers consisted of 5.8 105 cells (6 104 cells/cm2). Then, the medium was removed, and the cells were incubated with 0.001%, 0.005%, 0.01%, and 0.02% BAC in phosphate-buffered saline for up to 5 minutes, as described previously.24 The BAC solution was then swept out, and the cells were washed and reincubated in DMEM. The SIRC cultures FIGURE 2. Adsorption of BAC at the pure airwater interface (A) or at the interface covered with predeposited lipid lm with surface pressure 0 (B). Top: before the plateau of the surface pressure is reached, BAC molecules exist as monomers in the subphase, whereas once stops changing, the interface is saturated with BAC and the preservative molecules form micelles (lled circle) in the subphase. The results at the bottom were obtained at 0.001% BAC. At CR, BAC monomers cannot penetrate the lipid lm and remain in the subphase. The slope is calculated by least-squares linear t (r 0.978).

4648

Georgiev et al.

IOVS, June 2011, Vol. 52, No. 7 aggregates that contain some areas of lower lipid amounts, resembling black stripes. When the surface pressure was increased above 10 to 11 mN/m (i.e., the value at which inection in CS1 was observed), the black stripes closed, and the lipid lm became more dense and uniform. A similar correlation between the inection of the (CS1) dependence and the closing of the black stripes is already reported in bovine meibum lms.29 When the lm was subjected to dynamic compression and expansion (Fig. 3B), the shape of the compression (A) isotherm remained identical with the one observed at the quasiequilibrium compression rate. Both (A) trends, at compression and at expansion, closely overlapped, as shown in other studies,27,28 and as reected by the very high value of the isotherm reversibility, Rv, of 98%. In the next set of experiments, BAC was injected in concentrations from 0.001% (below CMC) to 0.02% (above CMC) into the subphase below the meibum lms that were expanded to their initial area and to 0 surface pressure. We found that the effects of 0.001% and 0.002% BAC and the effects of 0.01% and 0.02% on the surface properties of the tested samples (meibum, corneal cells lipid extracts, and whole tears) were identical. The results described were chosen to illustrate the impact of the preservative in the two regions: (1) below the CMC of BAC and (2) at the CMC of BAC (as the preservative effects above CMC are identical with the ones at CMC). For the entire concentration range, the preservative rapidly adsorbed to the surface lm, which was registered as an increase in and in the surface potential immediately after the injection of BAC into the subphase (Figs. 4A, 4B). Notable is the difference in the kinetics of and V changes after injection of BAC in concentrations below CMC (Fig. 4A). It can be seen that up to 400 seconds after the inclusion of BAC, increased steeply to 33 mN/m, whereas V rose smoothly from 180 to 250 mV. After this moment slowly (for the next 900 seconds) decreased up to a plateau value of 23 mN/m, whereas V continued to increase up to 520 mV. The opposite trends in the transients of and V point to the complex structural reorganization taking place at the interface. When BAC was injected into the subphase at (and above) CMC, both and V rose rapidly and simultaneously and soon reached stable plateau values (Fig. 4B).

tration of BAC into the meibum lm decreased with the increase in 0. This result is easy to explain, because a higher value of 0 means a higher surface concentration of lipid and less free area available for the incorporation of BAC molecules. By the linear extrapolation of (0) dependence, the socalled critical surface pressure CR was obtained: the minimum surface pressure of the predeposited lipid lm at which and above which BAC is not able to enter the interfacial lm.8 It can be seen that CR was identical with the surface pressure of BAC at CMC (39 mN/m). The results presented in Figure 2B are from experiments with the meibum of one person. The value of CR 39 0.7 mN/m was reproduced also with meibum from each of the other three volunteers and with mixed meibum (an equiweight mixture of the samples collected from each volunteer) which suggests reproducibility of CR despite individual differences. Synonymously, CR exceeding 35 mN/m was registered for synthetic lipids such as DPPC or corneal cell lipid extract (data not shown). These values were signicantly higher than the surface pressure of the tears in an open eye11,26 (30 mN/m) and of the cellular biomembranes8 (30 33 mN/m). Therefore, it can be expected that under physiological conditions, BAC penetrates and alters the integrity of the TFLL and of the membranes of the corneal epithelial cells.

Compression Isotherms and Dynamic Isocycles of Human Meibum and Tears, with and without BAC
The surface pressurearea compression isotherms of human meibum across the subphase of saline solution are shown in Figure 3A. The shape of the isotherms was similar to the ones reported by other authors.27,28 The (A) isotherm was converted by equation 1 to the dependence of the reciprocal compressibility (CS1) on (Fig. 3A, bold line). Across the entire scale, CS1 remained below 100 mN/m, and the relation between CS1 and was not linear. The (CS1) trend showed the lowest (30 mN/m) CS1 at of 11 mN/m, indicating that structural changes were taking place in the lipid lm at the given . BAM (Figs. 3C, 3D) revealed that, even at low surface pressures (between 0.5 and 10 mN/m), meibum formed continuous but rough lms, which is characteristic of lipid multilayers. The lms were composed of bright lipid

FIGURE 3. Langmuir surface balance studies of human meibum lms over pure saline solution subphase. (A) Quasi-equilibrium (A) isotherm processed by equation 1 to derive the presented (CS1) dependence. (B) Dynamic (A) isocycle with an isotherm Rv of 98%. (CF) BAM images (1700 1700 m) of meibum lm representative of the designated range. The dark, less-reective black stripes tended to close at 10 mN/m.

IOVS, June 2011, Vol. 52, No. 7

Surface Chemistry of BAC Interaction with TF

4649

FIGURE 4. Kinetics of surface pressure and surface potential changes after injection of BAC at concentrations below CMC (A; 0.001% BAC) and at CMC (B; 0.01% BAC), in the subphase of meibum lms expanded to the initial area.

of their dipole moments, which in turn will increase V. At further compression to 20% of the initial area (Fig. 5B), the lipid condensates consolidated, but the surface layer texture remained heterogeneous and did not recover the continuous morphology of the lipid lm in absence of preservative (Figs. 3E, 3F). At 0.01% preservative (Figs. 5C, 5D) for the entire range of surface pressures, the lms looked dark and uidlike (i.e., the surface was enriched with BAC molecules), and some isolated bright lipid islands were observable only at a high degree of compression. The ability of BAC to alter the surface pressure and the structure of meibomian lipid lms agrees with in vitro ndings31,32 showing perturbations of the lipid lms at 0.001% of BAC and with in vivo observations2,4 of the instability of TF in human eyes (after treatment with BAC-preserved timolol) and rabbit eyes (after instillation of 0.001% BAC). The effects of BAC on whole tears isocycles measured by PD-ADSA were similar to those discussed above (Fig. 6). Although the results of tears of only one person are shown, experiments using the samples of all volunteers revealed similar trends. Pure tear (A) isocycles displayed high reversibility (Rv 98%). The pure tears surface pressure at the initial area, 32 to 33 mN/m (i.e., surface tension, 40.9 39.9 mN/m), agreed very well with the surface tension (43.6 2.7 mN/m) of uncompressed tears previously reported.11,26 BAC readily adsorbed at the airtear interface, which is registered as the increase in tear surface pressure in the initial area. Below CMC (0.001% 0.005% preservative) the isotherms reversibility decreased (Rv 80% compared with Rv 98% for pure tears), whereas at and above CMC (0.01% 0.02% preservative), the reversibility increased but the isotherm shape became almost at and practically coincided with the (A) isocycles of the pendant drop, by pure BAC solution.

The dynamic isocycling of the mixed BACmeibum lms revealed (A) loops different from the ones of the pure meibum (Fig. 5). Below CMC of the preservative, the compression and expansion (A) trends showed increased hysteresis and decreased isotherm reversibility (Rv 80% at 0.001% BAC and Rv 75 at 0.005% BAC), compared with the pure lipid lms (Rv 98%). At (and above) CMC (0.01% BAC), the isotherm reversibility recovered (Rv 98%) but the shape of (A) dependence was almost at in contrast to the steep isotherm observed when only lipid was present at the surface (Fig. 3A). Flat isotherms are characteristic of water-soluble surfactants, whereas for water-insoluble substances (like lipids), a steady increase of when the area is compressed is to be expected.30 The alterations in the lm texture caused by the injection of BAC in concentrations below CMC were visualized by BAM (Figs. 5AD). It can be seen that at preservative concentrations below CMC (0.001% and 0.005% BAC; Fig. 5A) when compressed to 80 40% of the initial area, the mixed lm was discontinuous and nonuniform, in contrast to the morphology displayed by the lipid alone (Figs. 3C, 3D). The two surface lm patterns separated at the interface: (1) condensed lipid-like aggregates, with appearance different from the pattern characteristic of the meibum alone and (2) uidlike, dark, noncontrast regions. These nonreective, dark regions represent areas enriched with BAC molecules, which, owing to their high solubility in water, do not form visible aggregates at the surface. The formation of the two types of regions (the reective bright islands and dark liquid areas) may explain the opposite trends of and V between 300 and 1300 seconds on Figure 4A. Although the restructuring of the surface lm may result in the gradual decrease of the surface pressure, the condensation of the lipid molecules should increase the normal component

Interactions of BAC with Membrane Lipid Extract of SIRC Cells and with Cell Cultures
The inuence of BAC (below and above CMC) on the surface properties is identical with the data discussed above (Fig. 7): (1) decrease in Rv and formation of heterogeneous lms at 0.001% to 0.005% BAC and (2) at isotherms and dark uidlike lms at and above CMC. BAC strongly decreased the viability of SIRC cells and their capability of forming conuent cellular monolayers (Fig. 8). The damage in the cellular layer appeared rapidly, within 5 to 10 minutes after treatment with BAC, despite the brief exposure of the cells to the preservative. The SIRC cell line was validated by the European Centre for Validation of Alternative Methods (1990; Invittox protocol 40; Invittox Database for In Vitro Toxicology; http://www.invittox.com) for use in ophthalmic cytotoxicity studies of corneal cells, and the SIRC cells were found to properly mimic the properties of the corneal epithelium in pharmacotoxicologic studies,3336 including the damage due to exposure to BAC.35 Although SIRC cells are commonly referred to as an epithelial cell line,3338 they actually may be of broblast origin.39 Thus, it is most appropriate to describe them simply as a corneal cell line.

DISCUSSION
Surface Behavior of Pure Human Meibum and Whole Tears
Our BAM images and compressibility analysis (Fig. 3) agree with the literature data40 that meibum spontaneously forms a multilayer (instead of a monolayer) lm when spread at the airwater interface. Normally, lipid monolayers at low surface pressure have high lateral elasticity (low CS1 values) and a

4650

Georgiev et al.

IOVS, June 2011, Vol. 52, No. 7

FIGURE 5. Top: Dynamic (A) isocycles of meibum lms after injection into the subphase of the denoted concentrations of BAC. The isotherm Rv is shown for each isocycle. (AD) Representative BAM images (1700 m 1700 m) are shown of meibum lms in the presence of BAC at a low to intermediate (80% 40% of the initial area) and a high (20% of the initial area) degree of area compression.

liquid-expansionlike phase that does not signicantly modify the Brewster angle of the surface. Only at further compression when high surface pressures are achieved are condensed lipid phases with low lateral elasticity (CS1 100 mN/m) observed. In contrast, human meibum, even at low (0.510 mN/m), forms a rough, continuous lm at the airwater interface, becoming more uniform when increases and preserves

high lateral elasticity (CS1 100 mN/m) for the whole range. The Langmuir lms of meibum displayed nonlinear (CS1) dependence, with a minimum CS1 (at 11 mN/m) and subsequent reversal of the slope toward higher CS1 valuesa trend typical of multicomponent viscoelastic lipid multilayers.30 The structural rearrangements at this point (closure of

FIGURE 6. Dynamic (A) isocycles of whole human tears, with or without BAC (at the designated concentrations). The isotherm Rv is shown for each isocycle.

IOVS, June 2011, Vol. 52, No. 7

Surface Chemistry of BAC Interaction with TF

4651

FIGURE 7. Top: Dynamic (A) isocycles (top) of SIRC membrane lipid extract (SIRCm), with or without BAC injected into the subphase. BAC concentrations were varied from below CMC (0.005% BAC) to equal to or more than the CMC (0.01% 0.02% BAC) of the preservative. Bottom: typical BAM images (1700 1700 m) at 30% of the initial area of the SIRCm lms, pure and with the denoted BAC concentrations in the subphase.

the black stripes and formation of a more continuous and thicker lm) may be due to the squeezing out of lm-forming molecules from the interface during compression,41 which then forms a multilayered structure, as observed in a synthetic meibum replica.42 The analysis of TFLL kinetics of spreading performed by the Voigt model43 and by the spring-sloppy pistons model44 synonymously reveal the viscoelastic behavior of the TF lipid layer.

The multilayer structure may well be responsible for the characteristic surface properties (low CS1 and high Rv) of meibomian lipids and whole tears. Triglyceride cholesterol ester mixtures (i.e., compositions partially similar to meibum) are reported45 to form highly compressible and reversible multilayers. The very high isotherm reversibility, Rv, of 98% of meibum and tears indicates rapid spread and recovery of the lm structure during dynamic area changes.

FIGURE 8. Top: dependence of the number of viable cells in SIRC cellular monolayers on time in untreated cultures and cultures treated with the designated concentrations of BAC. Bottom: conuent monolayer images (620 620 m) characteristic of untreated SIRC cells, which became discontinuous and patchy within 5 minutes after treatment with the denoted BAC concentrations; the effect of BAC concentration below CMC (0.001%, 0.002%, and 0.005%) were almost identical and therefore only the image at 0.005% preservative is presented.

4652

Georgiev et al.

IOVS, June 2011, Vol. 52, No. 7 enter the TFLL may be efciently screened from the rapid aqueous tear turnover and can cause long-term adverse effects. Such assumption correlates well with clinical ndings4a detrimental effect on TF stability observed even 30 minutes after instillation of BAC-preserved timolol (i.e., when the aqueous tear turnover should already be fully completed). BAM observations permitted correlation of the changes in (A) isocycles with the BAC-induced effects on the structure of the meibum lms. Polar penetrant molecules, like BAC, are known47 to spontaneously repel and separate in distinguished lm regions when brought in contact with hydrophobic lipids like meibum48 at the interface. This explains why penetration of BAC molecules (applied below CMC at 0.001% 0.005%), disturbs the integrity of the meibum. Heterogeneities (Figs. 5A, 5B) similar to the breakup of the TFLL observed in vivo,4 are formed in the resultant lm, which is composed of condensed meibum islands and of uid BAC-enriched regions. At and above CMC, the injected BAC adsorbs to the meibomian lm, not as individual monomers but as micelles that solubilize the lipids, and dark, uidlike BAC-enriched layers (Figs. 5C, 5D) are formed with very few lipid aggregates left on the surface. Concerning the rapid dilution of BAC in aqueous tears, the preservative effects at submicellar concentrations look more physiologically relevant. The interactions of BAC, below and above CMC, with meibum lms are illustrated in Figure 9. An essential prerequisite for stable TF to be observed is the presence of a continuous insoluble lipid multilayer as it enhances the formation of tangentially immobile surfaces that increase the TF resistance to thinning (the so-called Gibbs-Marangoni effect)49 and suppresses aqueous tear evaporation.50,51 Thus BAC-induced heterogeneity and solubilization of meibomian lm obstructs these key functions of the TFLL.

The data agree with in vivo studies,43 showing rapid TFLL spreading that closely follows the upward movement of the eyelid in healthy eyes. In contrast to multilayers, monolayers of disaturated phosphatidylcholines with acyl chains longer than C14 (the major phospholipids of the biomembranes) and of tissue lipid extracts (like pulmonary surfactant lipids) are much less reversible, and Rv can reach values as low as 45%.46

Surface Behavior of BAC-Treated Human Meibum and Whole Tears

In a detailed clinical study, Ishibashi et al.4 demonstrated a signicant decrease of TF noninvasive breakup time in eyes treated with BAC-preserved timolol, whereas TF in eyes exposed to timolol without BAC remained stable. This instability of TF produced by BAC is presumably at least partly due to the perturbation of the lipid layer by the preservative. BAC is a highly surface active, cationic compound that readily adsorbs at the pure airwater interface (Fig. 2A) and penetrates the preformed meibum lm (Fig. 2B) at surface pressures (39 40 mN/m) signicantly higher than the surface pressure of the tears in an open eye (30 mN/m).11,26 BAC exerted similar effects on the dynamic (A) isocycles of meibum (Fig. 5) and whole tears (Fig. 6). It penetrated the surface lm and decreased isotherm reversibility (below CMC) or displaced the meibumtear constituents from the interface, resulting in at (A) trends (above the preservatives CMC). Such plateaulike (A) isotherm is characteristic of adsorption layers formed from water-soluble surfactants and is identical with the isotherm obtained for pure BAC solutions. The lack of overlapping between the (A) isocycles of meibum/whole tears, with and without BAC, indicates that the ones that penetrated the surface lm strongly modify its structure, remain inside it, and do not desorb to the subphase, even at high degrees of compression. In vivo BAC molecules that manage to

FIGURE 9. A simplied presentation of the molecular mechanisms of the interactions between BAC and meibum. Top: continuous meibum multilayer composed of cholesteryl esters (sterol ring shown as eclipse), waxes, triglycerides, and minor fraction of polar lipids. When experiments are performed below the CMC of the preservative (bottom left), BAC molecules adsorbed like individual monomers to the meibomian lm. As BAC molecules are charged and polar whereas meibomian lipids are hydrophobic, both compounds spontaneously separate at the interface in uid BAC-enriched regions and thick lipid islands; the heterogeneous lm structure results in decreased isotherm reversibility (i.e., capability of spreading during area expansion) of the surface lm. When experiments are performed at and above CMC of the preservative (bottom right), BAC micelles (open circles) adsorb at the interface. The micelles solubilize most of the lipids (lled circles) and BAC molecules displace meibum from the interface.

IOVS, June 2011, Vol. 52, No. 7

Surface Chemistry of BAC Interaction with TF

4653

Effect of BAC on the Surface Behavior of Corneal Membrane Extracts and on SIRC Cell Viability and Capability of Forming Conuent Monolayers
In this study, BAC perturbed the integrity of membrane extract lms and decreased the viability and the conuence of SIRC cell cultures. BAM observations conrm that below the CMC, the surfactant induced heterogeneity in the lm structure, whereas at and above the CMC, it caused solubilization of the lms. These effects of BAC were also successfully reproduced when the lipid lm was composed of bovine meibum and a variety of synthetic lipids with long acyl chains: DPPC, DPPG, and a mixture of cholesterylmyristate and ethyl oleate (i.e., simplistic articial mimics of meibum; results not shown). This reproducibility indicates that the inuence of BAC on lipid lms is practically independent of the nature of the lipid and is determined by the intrinsic surfactant properties of BAC itself.

References
1. Charnock C. Are multidose over-the-counter articial tears adequately preserved? Cornea. 2006;25:432 437. 2. Wilson WS, Duncan AJ, Jay JL. Effect of benzalkonium chloride on the stability of the precorneal tear lm in rabbit and man. Br J Ophthalmol. I975;59:667 669. 3. Debbasch C, Rat P, Warnet JM, De Saint Jean M, Baudouin C, Pisella PJ. Evaluation of the toxicity of benzalkonium chloride on the ocular surface. J Toxicol Cutan Ocul Toxicol. 2000;19:105 115. 4. Ishibashi T, Yokoi N, Kinoshita S. Comparison of the short-term effects on the human corneal surface of topical timolol maleate with and without benzalkonium chloride. J Glaucoma. 2003;12: 486 490. 5. Burstein NL. Preservative cytotoxic threshold for benzalkonium chloride and chlorhexidine digluconate in cat and rabbit corneas. Invest Ophthalmol Vis Sci. 1980;19:308 313. 6. Doughty MJ. Acute effects of chlorobutanol- or benzalkonium chloride-containing articial tears on the surface features of rabbit corneal epithelial cells. Optom Vis Sci. 1994;71:562572. 7. Deutschle T, Porkert U, Reiter R, Keck T, Riechelmann H. In vitro genotoxicity and cytotoxicity of benzalkonium chloride. Toxicol In Vitro. 2006;20:14721477. 8. Brockman H. Lipid monolayers: why use half a membrane to characterize protein-membrane interactions? Curr Opin Struct Biol. 1999;9:438 443. 9. Millar TJ, Tragoulias ST, Anderton PJ, et al. The surface activity of puried ocular mucin at the air-liquid interface and interactions with meibomian lipids. Cornea. 2006;25:91100. 10. Nagyova B, Tiffany JM. Components responsible for the surface tension of human tears. Curr Eye Res. 1999;19:4 11. 11. Tiffany JM, Winter N, Bliss G. Tear lm stability and tear surface tension. Curr Eye Res. 1989;8:507515. 12. Momchilova A, Markovska T. Phosphatidylethanolamine and phosphatidylcholine are sources of diacylglycerol in ras-transformed NIH 3T3 broblasts. Int J Biochem Cell Biol. 1999;31:311318. 13. Sabatini K, Mattila JP, Kinnunen PKJ. Interfacial behavior of cholesterol, ergosterol, and lanosterol in mixtures with DPPC and DMPC. Biophys J. 2008;95:2340 2355. 14. Brockman HL. Dipole potential of lipid membranes. Chem Phys Lipids. 1994;73:5779. 15. Shellock FG, Schatz CJ. Increased corneal temperature caused by MR imaging of the eye with a dedicated local coil. Radiology. 1992;185:697 699. 16. Morgan PB, Tullo AB, Efron N. Infrared thermography of the tear lm in dry eye. Eye. 1995;9:615 618. 17. Kaercher T, Ho nig D, Mo bius D. Brewster angle microscopy: a new method of visualizing the spreading of meibomian lipids. Int Ophthalmol. 1993;17:341348. 18. Kwok DY, Vollhardt D, Miller R, Li D, Neumann AW. Axisymmetric drop shape analysis as a lm balance. Colloids Surf A. 1994; 88:5158. 19. Li JB, Vollhardt D, Miller R, Weidmann G, Moehwald H. Isotherms of phospholipid monolayers measured by a pendant drop technique. Colloid Polym Sci. 1996;274:995999. 20. Ali S, Smaby JM, Brockman HL, Brown RE. Cholesterols interfacial interactions with galactosylceramides. Biochemistry. 1994;33:2900 2906. 21. Huehnerfuss H, Alpers W. Molecular aspects of the system water/ monomolecular surface lm and the occurrence of a new anomalous dispersion regime at 1.43 GHz. J Phys Chem B. 1983;87:5251 5258. 22. Alakoskela JM, Vitovic P, Kinnunen PK. Screening for the drugphospholipid interaction: correlation to phospholipidosis. Chem Med Chem. 2009;4:1224 1251. 23. Zhao H, Dubielecka PM, Soderlund T, Kinnunen PKJ. Interactions of adriamycin, cytochrome c, and serum albumin with lipid monolayers containing poly(ethylene glycol)-ceramide. Biophys J. 2002; 83:954 967. 24. Frentz M, Goss M, Reim M, Schrage NF. Repeated exposure to benzalkonium chloride in the Ex Vivo Eye Irritation Test (EVEIT):

General Comments on the Effects of BAC on the Surface Properties of Lipid Films: Possible Relation with BAC Adverse Effects In Vivo
We found that to partially suppress the capability of BAC to perturb the membrane integrity, it was necessary to decrease the preservative concentration by at least two orders (up to 105%), when compared with the lowest clinical concentration (0.001%). Key features of the BAC-induced alterations of lipid lm structure and surface properties in vitro are that:
BAC penetrates the lipid lms (meibum, tear surface lm, or cell biomembrane extract), modies their structure and properties, and stably resides in the lms during dynamic area changes. The BAC-induced effects occur rapidly (i.e., they are realized immediately after its injection).

The in vitro results provide a clue to the nature of the adverse effects of BAC in vivo. The concentration of BAC in aqueous tears decreases exponentially with each blink.52 However, it is highly probable that rapid changes in the integrity of the TFLL and of corneal cell membranes induced by BAC penetration take place in vivo. Once inserted in the lipid lms at the ocular surface, the preservative molecules will be excluded from the aqueous tear compartment and thus screened from the aqueous tear turnover. As the TFLL turnover extends for hours,53 the preservative molecules inserted in the lipid lm stay longer than those in the aqueous tears and can be responsible for the long-term adverse effects associated with BAC: TF instability and barrier dysfunction and damage of corneal epithelium.2,4,7 The preservative incorporation in the lipid layers after instillation of BAC-preserved eyedrops will be facilitated in dry eyes where the integrity of the TFLL and corneal cells membranes is compromised10,50 and the rate of aqueous tear turnover (i.e., the rate of preservative clearance) is frequently decreased.54 The detrimental effect of BAC on TF stability and corneal surface integrity may be caused, not only by direct interaction of the preservative with the lipid bilayer, but also by electrostatic aggregation between the cationic BAC molecules and the anionic polysaccharide moieties of the glycocalyx. Apart from straight interactions with TF compounds, BAC can exert toxicity by inducing inammation at the ocular surface.55 Once damaged, by one or by a combination of more mechanisms, the corneal surface loses its wettability, which in turn contributes to the destabilization of the TF.

Acknowledgments
The authors thank John Bush for reviewing the manuscript.

4654

Georgiev et al.

IOVS, June 2011, Vol. 52, No. 7

39. Niederkorn JY, Meyer DR, Ubelaker JE, Martin JH. Ultrastructural and immunohistological characterization of the SIRC corneal cell line. In Vitro Cell Dev Biol. 1990;26:923930. 40. Svitova TF, Lin MC. Tear lipids interfacial rheology: effect of lysozyme and lens care solutions. Optom Vis Sci. 2010;87:10 20. 41. Graham DE, Phillips MC. Proteins at liquid interfaces: II. adsorption isotherms. J Colloid Interface Sci. 1979;70:415 426. 42. Petrov PG, Thompson JM, Abdul Rahman IB, et al. Two-dimensional order in mammalian pre-ocular tear lm. Exp Eye Res. 2007;84:1140 1146. 43. Yokoi N, Yamada H, Mizukusa Y, et al. Rheology of tear lm lipid layer spread in normal and aqueous tear decient dry eyes. Invest Ophthalmol Vis Sci. 2008;49:5319 5324. 44. King-Smith PE, Fink BA, Nichols JJ, Nichols KK, Braun RJ, McFadden GB. The contribution of lipid layer movement to tear lmthinning and breakup. Invest Ophthalmol Vis Sci. 2009;50(6):2747 2756. 45. Smaby JM, Brockman HL. Properties of cholesteryl oleate and triolein in mixed monolayers at the air-water interface. J Lipid Res. 1978;19:325331. 46. Blanco O, Pe rez-Gil J. Biochemical and pharmacological differences between preparations of exogenous natural surfactant used to treat Respiratory Distress Syndrome: Role of the different components in an efcient pulmonary surfactant. Eur J Pharmacol. 2007;568:115. 47. Mann EK, Lee LT, Henon S, Langevin D, Meunier J. Polymersurfactant lms at the air-water interface: 1. surface pressure, ellipsometry, and microscopic studies, Macromolecules. 1993;26: 70377045. 48. Butovich IA. The meibomian puzzle: combining pieces together. Prog Retin Eye Res. 2009;28(6):483 498. 49. Tadros TF. Applied Surfactants. Weinheim, Germany: Wiley-VCH Verlag GmbH & Co., 2005:73 84,259 284. 50. Bron AJ, Tiffany JM, Gouveia SM, Yokoi N, Voon LW. Functional aspects of the tear lm lipid layer. Exp Eye Res. 2004;78:347360. 51. Tiffany JM. The normal tear lm. Dev Ophthalmol. 2008;41:120. 52. Friedlaender MH, Breshears D, Amoozgar B, Sheardown H, Senchyna M. The dilution of benzalkonium chloride (BAC) in the tear Film. Adv Ther. 2006;23(6):835 841. 53. Mochizuki H, Yamada M, Hatou S, Tsubota K. Turnover rate of tear-lm lipid layer determined by uorophotometry. Br J Ophthalmol. 2009;93(11):15351538. 54. Yuan Y, Wang J, Chen Q, Tao A, Shen M, Shousha MA. Reduced tear meniscus dynamics in dry eye patients with aqueous tear deciency. Am J Ophthalmol. 2010;149:932938. 55. Baudouin C, Labbe A, Liang H, Pauly A, Brignole-Baudouin F. In Vitro preservatives in eyedrops: the good, the bad and the ugly. Prog Retin Eye Res. 2010;29:312334.

25.

26. 27.

28.

29.

30. 31.

32. 33.

34.

35.

36.

37.

38.

observation of isolated corneal damage and healing. Altern Lab Anim. 2008;36:2532. Mosmann T. Rapid colorimetric assay for cellular growth and survival-application to proliferation and cyto-toxicity assays. J Immunol Methods. 1983;65:55 63. Miller D. Measurement of the surface tension of tears. Arch Ophthalmol. 1969;82:368 371. Tragoulias ST, Anderton PJ, Dennis GR, Miano F, Millar TJ. Surface pressure measurements of human tears and individual tear lm components indicate that proteins are major contributors to the surface pressure. Cornea. 2005;24:189 200. Millar TJ, Mudgil P, Butovich IA, Palaniappan1 CK. Adsorption of human tear lipocalin to human meibomian lipid lms. Invest Ophthalmol Vis Sci. 2009;50:140 151. Georgiev GAs, Kutsarova E, Jordanova A, Krastev R, Lalchev Z. Interactions of meibomian gland secretion with polar lipids in Langmuir monolayers. Colloid Surf B Biointerfaces. 2010;78:317 327. Adamson AW, Gast AP. Physical Chemistry of Surfaces. 6th ed. New York, NY: John Wiley & Sons; 1997:101158, 537563. Holly FJ. Surface chemical evaluation of articial tears and their ingredients, II: interaction with a supercial lipid layer. Contact Lens. 1978;4:52 65. Zhao J, Wollmer P. Air pollutants and tear lm stability: a method for experimental evaluation. Clin Physiol. 200;121:282286. Goskonda VR, Khan MA, Hutak CM, Reddy IK. Permeability characteristics of novel mydriatic agents using an in vitro cell culture model that utilizes SIRC rabbit corneal cells. J Pharm Sci. 1999; 88:180 184. Goskonda VR, Hill RA, Khan MA, Reddy IK. Permeability of chemical delivery systems across rabbit corneal (SIRC) cell line and isolated corneas: a comparative study. Pharm Dev Technol. 2000; 5:409 416. Dutot M, Pouzau F, Larosche I, Brignole-Baudouin F, Warnet JM, Rat P. Fluoroquinolone eye drop-induced cytotoxicity: role of preservative in P2X7 cell death receptor activation and apoptosis. Invest Ophthalmol Vis Sci. 2006;47:28122819. Sakaguchi H, Ota N, Omori T. et al. Validation study of the Short Time Exposure (STE) test to assess the eye irritation potential of chemicals. Toxicol in Vitro. Published online February 1, 2011. Hutak CM, Kavanagh ME, Reddy IK, Barletta MA. Growth pattern of SIRC rabbit corneal cells in microwell inserts. J Toxicol Cutan Ocul Toxicol. 1997;16:145156. Tak RV, Pal D, Gao H, Dey S, Mitra AK. Transport of acyclovir ester prodrugs through rabbit cornea and SIRC-rabbit corneal epithelial cell line. J Pharm Sci. 2001;90:15051515.