e u r o p e a n j o u r n a l o f p h a r m a c e u t i c a l s c i e n c e s 3 4 ( 2 0 0 8 ) 263–273

avai lab le at www.sc iencedi rec t .com

journa l homepage: www.e lsev ier .com/ locate /e jps

In vitro modulation of preservative toxicity: High molecularweight hyaluronan decreases apoptosis and oxidative stressinduced by benzalkonium chloride

Thierry Pauloina,∗, Melody Dutota, Jean-Michel Warneta,b,c, Patrice Rata,b,c

a Laboratoire de Toxicologie, Faculte des Sciences Pharmaceutiques et Biologiques, Universite Paris Descartes, 75006 Paris, Franceb INSERM UMRS 872, Institut Biomedical des Cordeliers, 75006 Paris, Francec Centre Hospitalier National d’Ophtalmologie des Quinze-Vingts, 75012 Paris, France

a r t i c l e i n f o

Article history:

Received 29 February 2008

Received in revised form

9 April 2008

Accepted 24 April 2008

Published on line 1 May 2008

Keywords:

Hyaluronan

Preservative

Apoptosis

Oxidative stress

P2X7 cell death receptor

Benzalkonium chloride

a b s t r a c t

Objective: Benzalkonium chloride (BAK) is one of the most often used preservative in phar-

maceutical products and it is known to induce toxic effects. Hyaluronan (HA), a linear

biopolymer, is involved in several biological processes. The aim of this work is to in vitro

investigate if HA is able to decrease BAK toxicity.

Methods: Two human epithelial cell lines were treated with different incubation time pro-

tocol with BAK and three different molecular weights HA (HA 20 kDa, HA 100 kDa and HA

1000 kDa, 0.2%, w/v). Flow cytometry, fluorescence microscopy, microplate cytofluorometry

and confocal microscopy were performed to evaluate expression of CD44 receptor, cell via-

bility, oxidative stress, mitochondrial mass, chromatin condensation, plasma-membrane

permeability, DNA fragmentation and cytoskeleton morphology.

Results: The three HAs studied induce neither oxidative stress nor apoptosis. HA 1000 kDa

significantly decreases oxidative stress, apoptosis and necrosis induced by BAK. Experi-

ments with HA 20 kDa or HA 100 kDa did not show the same effects. For instance, the more

molecular weight decreases, the more protection decreases. Moreover, we suggest that HA

interacts with cell plasma-membrane and inhibits cell death receptors.

Conclusion: High molecular weight HA (1000 kDa, 0.2%) is an effective protective agent against

BAK.

© 2008 Elsevier B.V. All rights reserved.

1

Pcmgit

in most of the eyedrops that are stored for extended peri-

0d

. Introduction

reservatives such as benzalkonium chloride (BAK) are a cru-ial ingredient in aqueous products because they preventicrobial infection. For example, many people using eyedrops

et the tip of the bottle in contact with their eyes when theynstill a drop. This contact allows cutaneous bacteria to con-aminate medicated eyedrops. When the eyedrops are used

∗ Corresponding author. Tel.: +33 1 53 73 98 65.E-mail address: thierry.pauloin@gmail.com (T. Pauloin).

928-0987/$ – see front matter © 2008 Elsevier B.V. All rights reserved.oi:10.1016/j.ejps.2008.04.006

again, bacteria might be instilled in the eyes along with theeyedrop. There are different types of eyedrop preservatives,some more powerful than others. BAK, which belongs to thequaternary ammonium group, is a potent preservative used

ods of time. The mechanism of its microbiocidal action isthought to be due to the disruption of the plasma membrane,which increases cellular permeability, and induces cell death.

u t i c

264 e u r o p e a n j o u r n a l o f p h a r m a c e

BAK is active against bacteria, fungi and protozoa and is cur-rently used in concentrations ranging from 0.001% to 0.1%.To summarize, BAK is one of the most often used preserva-tives because of its efficiency, its stability and its low cost.The downside of BAK is that several toxic effects have beenobserved. Studies have shown that BAK is likely to damage thehuman ocular surface and impair the local tolerance to eye-drops (Tripathi et al., 1992; Eleftheriadis et al., 2002; Baudouinet al., 1999). BAK is also used in aqueous nasal formulations.It has been proven to induce the impairment of mucociliaryclearance (Batts et al., 1990; Boek et al., 1999; Arnitz et al., 2006).Some morphological changes in the nasal respiratory mucosahave been observed (Berg et al., 1997; Cureoglu et al., 2002;Lebe et al., 2004). BAK, also added to shampoo, is suspected toinduce contact dermatitis (Oiso et al., 2005).

Preservatives prevent microbial proliferation and there-fore must be used at effective concentrations, although toxic.Indeed, microbial proliferation and its consequences couldbe worse than the toxicity of the preservative. Our goal wasto find a molecule able to protect cells from preservativetoxicity.

Hyaluronan (HA) is a linear non-sulfated polysaccharidechain composed of �-1,4-glucuronic acid alternated with �-1,3-N-acetylglucosamine (Weissman and Meyer, 1954), whichbelongs to the glycosaminoglycan family. Disaccharides arehydrophilic molecules that confer to HA an excellent water-holding capacity. Indeed, HA can incorporate a volume ofwater more than 1000 times its initial volume, constitutinga viscous polymer. HA is found extensively in the extracel-lular matrix of all tissues and is abundant in the vitreousbody of the eye, in the synovial fluid of articular joints andthe intercellular space of the epidermis. HA is synthetisedas large 1000 kDa disaccharide chains and is then progres-sively degraded in the matrix to give intermediate size HA(∼10 to ∼100 kDa) or small fragments (<1 kDa). HA regulatescell movement and the transport of extracellular components(Laurent and Fraser, 1992; Toole, 2004). Following tissue injury,native HA is depolymerised and HA fragments accumulate,inducing the expression of inflammatory genes (Horton et al.,1998, 1999; McKee et al., 1996). During the healing process,HA fragments levels increase which promotes inflammatorycell migration and proliferation (Chen and Abatangelo, 1999).Native HA does not induce inflammatory or proliferative genes(Noble, 2002). Thereby, HA is a remarkable biopolymer thatappears to have an amazing array of biological functionsdepending on the number of disaccharides.

In this present study, we investigate in vitro the ability ofHA to reduce preservative toxicity in two human epithelialmodels. Moreover, we suspect that HA molecular weight is animportant factor in cell protection. Thereby, our experimentswere made using a native HA (1000 kDa) and two differentintermediate size HA (20 and 100 kDa).

2. Materials and methods

2.1. Cell culture

Wong Kilbourne derivative of Chang conjunctival epithelialcell line (WKD, ECACC 93120839) was cultured in Dulbecco’s

a l s c i e n c e s 3 4 ( 2 0 0 8 ) 263–273

minimum essential medium (Eurobio, Les Ulis, France) supple-mented with 10% foetal bovine serum (Eurobio), 1% glutamine(Eurobio), 1% penicillin and streptomycin (Eurobio). Humancorneal epithelial cell line (Araki-Sasaki et al., 1995) (HCE,Riken cell bank RCB 1384) was cultured in Dulbecco’s min-imum essential medium mixed with Ham’s F12 medium(50–50) (Eurobio) supplemented with 10% foetal bovine serum(Eurobio), 1% glutamine (Eurobio), 1% penicillin and strepto-mycin (Eurobio). Cultures were maintained at 37 ◦C, 5% CO2

in a humidified incubator. Normal culture development wasassessed daily by inverted microscopy. Confluent cultureswere removed by trypsin incubation (Eurobio) and cells werecounted.

2.2. Cell incubation

All of the microplates wells are exactly seeded with the samenumber of cells (200 �l of a 9 × 104 cell/ml solution) with bothcell lines. When cells reached approximately 80% of con-fluence, the culture medium was removed and cells werepreincubated 20 min with 0.2% (w/v) HA 1000 kDa (Soliance,Pomacle, France), 0.2% (w/v) HA 100 kDa (Soliance) or 0.2%(w/v) HA 20 kDa (Soliance) dissolved in PBS (Eurobio) and thenincubated 20 min with different concentration of BAK (Sigma,St. Louis, Missouri, USA). PBS was used as control. Molecularweight and purity of each HA was controlled and certified bySoliance.

2.3. Lysosomal membrane integrity

Neutral red (Fluka, Buchs, Switzerland) uptake assay is acell viability assay, based on the ability of viable cells toincorporate neutral red, which is a weak cationic dye thatreadily penetrates cell membranes by non-ionic diffusion,accumulating in lysosomes, where it binds with anionicsites in the lysosomal matrix. Lysosomal membrane integrityis closely correlated with cell viability and it is evaluatedwith neutral red fluorescence (excitation, 535 nm; emission,600 nm). In accordance with the validated protocol of Boren-freund and Puerner, 200 �l per well of neutral red solution(50 �g/ml) containing culture medium were added to livingcells (Borenfreund and Puerner, 1985). After a 3-h incubationtime at 37 ◦C in moist atmosphere with 5% CO2, the wells werewashed with PBS to remove any remaining unincorporateddye. The dye was then released from the cells using 200 �lof lysis solution per well (1% acetic acid, 50% ethanol and49% dH2O). The plate was agitated on a microplate shaker for20 min and then fluorescence was measured using microplatefluorometry (Safire, Tecan, Lyon, France).

2.4. Intracellular redox status

Alamar Blue assay uses resazurine (Sigma), a visible blue fluo-rogene probe, which is reduced to a red fluorescent compound(resorufin) by cellular redox enzymes (de Fries and Mitsuhashi,1995; Larson et al., 1997). After treatment, the medium was

removed and 200 �l per well of resazurine solution (0.1 mg/ml)containing culture medium supplemented with 2.5% foetalbovine were added to living cells. After a 6-h incubation time at37 ◦C in moist atmosphere with 5% CO2, resorufin fluoromet-

t i c a

ru

2

RfliolIl(flctHla4flbdHdsclitHcset

2

MONmw3ttwac

2

Ycfl1afas

e u r o p e a n j o u r n a l o f p h a r m a c e u

ic signal (excitation, 535 nm; emission, 600 nm) was measuredsing microplate fluorometry.

.5. Reactive oxygen species production

eactive oxygen species were evaluated using two differentuorometric methods. H2DCF-DA (Invitrogen, Carlsbad, Cal-

fornia, USA) probe is a standard probe to detect reactivexygen species in cells, especially hydroperoxide. It is color-ess and non-fluorescent until cleavage of the diacetate group.t is non-polar compound which is hydrolyzed by intracel-ular esterases to become a non-fluorescent polar derivativeH2DCF), which is oxidized rapidly to give the highly greenuorescent 2′,7′-dichlorofluorescein in the presence of intra-ellular reactive oxygen species (Gomes et al., 2005). Afterreatment, the medium was removed and 200 �l per well of

2DCF-DA solution (0.1 mg/ml) containing PBS were added toiving cells. After a 20-min incubation time at 37 ◦C in moisttmosphere with 5% CO2, fluorometric signal (excitation,90 nm; emission, 535 nm) was measured using microplateuorometry. Intracellular superoxide levels were measuredy quantifying the fluorescence of the oxidation product ofihydroethidium (HE). Oxidation of HE by superoxide convertsE, which exhibits weak blue fluorescence, to an ethidiumerivative that exhibits peak fluorescence in the rhodaminepectrum (excitation, 480 nm; emission, 586 nm). HE, which isell permeant, enters the cell and, after oxidation, accumu-ates in the nucleus, where it binds DNA with a small shiftn its emission spectrum to 567 nm (Zhao et al., 2003). Afterreatment, the medium was removed and 200 �l per well ofE solution (1.58 �g/ml) containing PBS were added to livingells. After a 20-min incubation time at 37 ◦C in moist atmo-phere with 5% CO2, fluorometric signal (excitation, 480 nm;mission, 567 nm) was measured using microplate fluorome-ry.

.6. Mitochondrial mass

itochondrial mass was evaluated using the NonylAcridinerange probe (NAO, Molecular Probes, PoortGebouw, Theetherlands), which stains lipids found exclusively in theitochondrial inner membrane. The microplate is incubatedith the dye solution (10 �M in culture medium) for 30 min at

7 ◦C. After a 1 h recovery period, wells were washed with PBSo remove any remaining unincorporated dye. The dye washen released from the cells using 200 �l of lysis solution perell (1% acetic acid, 50% ethanol and 49% dH2O). The plate was

gitated on a microplate shaker for 20 min and then fluores-ence was measured (excitation, 490 nm; emission, 530 nm).

.7. Plasma-membrane permeability

O-PRO-1 (Invitrogen), a DNA probe, is usually used to dis-riminate cells dying by apoptosis versus necrosis with bothow cytometry and fluorescence microscopy (Idziorek et al.,995; Choucroun et al., 2001). This fluorescent probe evalu-

tes membrane permeability modifications that early appearor the apoptotic process. YO-PRO-1 specifically binds to DNA,nd its fluorescence can be detected (excitation, 491 nm; emis-ion, 509 nm). Cellular membrane is not permeable to the

l s c i e n c e s 3 4 ( 2 0 0 8 ) 263–273 265

probe, and its DNA binding shows the opening of specificmembrane pores that appear for apoptosis. Fluorescence wasperformed after 10 min of incubation using microplate fluo-rometry. They were then observed by inverted fluorescencemicroscopy (DMIRB; Leica, Heidelberg, Germany) and pho-tographed (Coolpix 5000; Nikon, Tokyo, Japan).

2.8. Chromatin condensation assessment

Hoechst 33342 (Invitrogen) was used to evaluate chromatincondensation in cells with propidium iodide (Invitrogen) todiscriminate necrotic cells. The UV fluorescent probe Hoechst33342 (excitation, 360 nm; emission, 450 nm) enters living cellsand apoptotic cells, whereas propidium iodide enters necroticcells much faster than Hoechst 33342 (Belloc et al., 1994). Thecells were exposed for 20 min to a Hoechst 33342 solution at aconcentration of 10 �g/ml containing 1 �l of propidium iodide(1 mg/ml). They were then observed by inverted fluorescencemicroscopy (DMIRB; Leica) and photographed (Coolpix 5000;Nikon).

2.9. DNA fragmentation evaluation

In apoptotic cells, DNA is progressively fragmented and canbe quantified. The cells were seeded in 6-well plates andkept at 37 ◦C for 24 h. After the different incubations and a24-h recovery time in culture medium supplemented with2.5% foetal bovine, cells were gently detached and collectedafter 10 min of incubation in 0.5 mM EDTA. Then, they werewashed in PBS and suspended in 1 ml of PBS. Cells are fixedwith a 1% PFA solution for 24 h. Finally, cells are treated 5 minwith saponin (Sigma), washed with PBS and incubated for3 min with 50 �g/ml of propidium iodide (Invitrogen) beforeflow cytometry analysis (FC500, Beckman Coulter, Minneapo-lis, Minnesota, USA).

2.10. Expression of CD44 receptor

Culture cells were re-suspended at 2.5 × 105 cells in 30 �l ofculture medium and incubated for 45 min at 4 ◦C with CD44monoclonal antibodies (Sigma). After three washes in coldphosphate-buffered saline supplemented with 0.5% of BSAcells were incubated 30 min at 4 ◦C with a goat anti-mouseimmunoglobulin-FITC (Dakocytomation, Glostrup, Denmark)After three washes in cold phosphate-buffered saline sup-plemented with 0.5% of BSA, cells were fixed with 1%paraformaldehyde in PBS. Cellular debris were eliminatedfrom the analysis using a gate on forward and side scatter.For each sample 5 × 103 cells were analyzed (FC500, BeckmanCoulter, Minneapolis, Minnesota, USA).

2.11. Inverted immunofluorescent analysis of actin

For immunofluorescent examination of actin expression, cellswere seeded on coverslips and fixed with 4% paraformalde-hyde in PBS. After being washed twice with PBS, cells were

permeabilized with 0.1% Triton X-100 and stained, firstly,with Alexa-conjugated phalloidin (Invitrogen) for 30 min, andsecondly stained with TOTO (Invitrogen) for 10 min. TheAlexa-phalloıdin-stained actin and TOTO-stained nuclei were

266 e u r o p e a n j o u r n a l o f p h a r m a c e u t i c a l s c i e n c e s 3 4 ( 2 0 0 8 ) 263–273

flowD44

with negative control, in order to have the more relevantassays.

Preincubation for 20 min with HA 1000 kDa before a 20-min incubation time with BAK on both epithelial cell lines

Fig. 2 – Analysis of HAs cell biocompatibility. Lysosomalmembrane integrity was evaluated with the neutral reduptake assay and after a 24-h incubation time with PBS(control), HA 20 kDa 0.2%, HA 100 kDa 0.2% or HA 1000 kDa0.2% on HCE (corneal epithelial cell line) and WKD

Fig. 1 – Expression of the CD44 receptor was analyzed usingline (A) and the conjunctival epithelial cell line (B) express C

observed under inverted fluorescence microscope (DMIRB;Leica) and photographed.

2.12. Statistical analysis

Results were obtained in fluorescence units and wereexpressed as a percentage of the control. Each drug con-centration was tested in 12 wells, and each experiment wasperformed in triplicate. The mean values for each concen-tration were analyzed by one-way ANOVA followed by theDunnett test (Sigma Stat 2.0; SSPS, Chicago, Illinois, USA) andthe level of significance was fixed at 0.05.

3. Results

3.1. Corneal and conjunctival cell lines express theCD44 receptor

CD44 is the major receptor for hyaluronan. We analyzedthe expression of the CD44 receptor using flow cytometry.Results indicated that the corneal epithelial cell line (Fig. 1A;CD44 mean fluorescence: 158.5 ± 13, control mean fluores-cence: 4.8 ± 20) and the conjunctival epithelial cell line (Fig. 1B;CD44 mean fluorescence: 30.3 ± 13, control mean fluores-cence: 3.7 ± 12) express CD44 receptor. Expression of CD44 washigher in the corneal epithelial cell line than in the conjunc-tival epithelial cell line.

3.2. No significant toxic effect was observed with thethree studied HAs

Cell viability was evaluated using neutral red uptake assay(Fig. 2). Cytotoxicity of HA 20 kDa (0.2%, w/v), HA 100 kDa (0.2%,

w/v) and HA 1000 kDa (0.2%, w/v) were investigated on thecorneal epithelial cell line and the conjunctival epithelial cellline. Even after a 24-h incubation time, no significant modifi-cation of lysosomal membrane integrity was observed.

cytometry. Results indicated that the corneal epithelial cellreceptor.

3.3. HA 1000 kDa significantly reduces BAKcytotoxicity

Cell viability was evaluated using two different and inde-pendent assays. Neutral red uptake assay (Fig. 3A) evaluatedlysosomal membrane integrity and Alamar Blue assay (Fig. 3B)evaluated intracellular redox status.

A 20-min incubation time with BAK 0.002%, with thecorneal epithelial cell line, and BAK 0.005%, with the con-junctival epithelial cell line, induced a significant decrease inlysosomal membrane integrity (−53% and −67% with HCE andWKD, respectively) and in intracellular redox status (−61% and−56% with HCE and WKD, respectively). These BAK concentra-tions were chosen to have a mid cytotoxic effect compared

(conjunctival epithelial cell line). *p < 0.05, **p < 0.01, and***p < 0.001 compared with PBS control. Results areexpressed as percentages of control values. No significanttoxic effect was observed even after a 24-h incubation timewith the three studied HAs.

e u r o p e a n j o u r n a l o f p h a r m a c e u t i c a l s c i e n c e s 3 4 ( 2 0 0 8 ) 263–273 267

Fig. 3 – Analysis of cell viability. Lysosomal membrane integrity was evaluated with the neutral red uptake assay andintracellular redox status was evaluated with the Alamar Blue assay after a 20-min preincubation time with PBS (white), HA20 kDa 0.2% (light-grey), HA 100 kDa 0.2% (dark-grey) or HA 1000 kDa 0.2% (black) followed by a 20-min incubation time withBAK 0.002% (HCE: corneal epithelial cell line) or BAK 0.005% (WKD: conjunctival epithelial cell line). *p < 0.05, **p < 0.01, and* d asp e in

sarc1bsrfr

fnr

3

Opedw

iHt

Hwdwre(

a

**p < 0.001 compared with BAK control. Results are expressereincubated with HA 1000 kDa showed a significant increas

ignificantly improved lysosomal membrane integrity (+27%nd +14% with HCE and WKD, respectively) and intracellularedox status (+23% and +27% with HCE and WKD, respectively)ompared with BAK control cells. A preincubation with HA00 kDa induced an increase in intracellular redox status inoth epithelial cell lines but did not significantly modify lyso-omal membrane integrity. Besides, intracellular redox statusesults obtained with HA 1000 kDa are significantly differentrom HA 100 kDa (p < 0.001 and p < 0.01, with HCE and WKD,espectively).

Corneal and conjunctival epithelial cell lines preincubatedor 20 min with HA 20 kDa did not show significant differenceeither in lysosomal membrane integrity nor in intracellularedox status compared with BAK control cells.

.4. HA 1000 kDa decreases BAK oxidative stress

xidative stress was evaluated using two different and inde-endent assays. Hydroperoxide production (Fig. 4A) wasvaluated using the fluorogen 2′,7′-dichlorodihydrofluoresceiniacetate (H2DCFDA) probe and superoxide anion productionas evaluated using the dihydroethidium probe.

A 20-min incubation time with BAK induced an importantncrease in hydroperoxide production (+179% and +94% withCE and WKD, respectively) and in superoxide anion produc-

ion (+271% and +387% with HCE and WKD, respectively).Preincubation of both cell lines for 20 min with HA 20 kDa,

A 100 kDa or HA 1000 kDa before a 20-min incubation timeith BAK induced a significant decrease in hydroperoxide pro-uction compared with BAK control cells. All results obtainedith HA 1000 kDa (−95% and −70% with HCE and WKD,

espectively), on both cell lines, were significantly differ-

nt from results obtained with HA 20 kDa or HA 100 kDap < 0.001).

A 20-min preincubation time with HA 1000 kDa induceddecrease in superoxide anion production on corneal and

percentages of control values (dotted line). Cellscell viability on both epithelial cell lines.

conjunctival epithelial cell lines (−120% and −291% with HCEand WKD, respectively) compared with BAK control cells. Apreincubation with HA 100 kDa induced a decrease in super-oxide anion production only on the conjunctival epithelial cellline. Corneal and conjunctival epithelial cells preincubatedfor 20 min with HA 20 kDa showed no significant differencein superoxide anion production compared with BAK controlcells.

3.5. HA 20 kDa, HA 100 kDa and HA 1000 kDadecreases BAK induced mitochondrial mass

Mitochondrial mass was evaluated using the NonylAcridineOrange assay (Fig. 4B).

A 20-min incubation time with BAK induced an importantincrease in mitochondrial mass (+163% and 185% with HCEand WKD, respectively).

Preincubation of both cell lines for 20 min with HA 20 kDa,HA 100 kDa or HA 1000 kDa before a 20-min incubation timewith BAK induced a significant decrease in mitochondrialmass, compared with BAK control cells. Results obtained withHA 1000 kDa (−105% and −140% with HCE and WKD, respec-tively) were significantly different from results obtained withHA 20 kDa or HA 100 kDa (p < 0.001).

3.6. HA 1000 kDa decreases chromatin condensationand plasma-membrane permeability

Apoptosis was evaluated using two different and inde-pendent assays. Chromatin condensation evaluation wasperformed using Hoechst 33342 fluorescence (Fig. 5A). Plasma-membrane permeability was evaluated using YO-PRO-1 probe

(Fig. 5B).

A 20-min incubation time with BAK induced an importantincrease in chromatin condensation (+301% and +411% withHCE and WKD, respectively) and a huge plasma-membrane

268 e u r o p e a n j o u r n a l o f p h a r m a c e u t i c a l s c i e n c e s 3 4 ( 2 0 0 8 ) 263–273

Fig. 4 – Oxidative stress evaluation. (A) Hydroperoxide production was evaluated with the DCFH-DA assay and superoxideanion production was evaluated with the hydroethidium assay after a 20-min preincubation time with PBS, HA 20 kDa 0.2%,HA 100 kDa 0.2% or HA 1000 kDa 0.2% followed by a 20-min incubation time with BAK 0.002% (HCE: corneal epithelial cellline) or BAK 0.005% (WKD: conjunctival epithelial cell line). *p < 0.05, **p < 0.01, and ***p < 0.001 compared with BAK control.Results are expressed as percentages of control values (dotted line). Cells preincubated with HA 1000 kDa showed asignificant decrease in oxidative stress on both epithelial cell lines. (B) Mitochondrial mass was evaluated using theNonylAcridine Orange assay after a 20-min preincubation time with PBS, HA 20 kDa 0.2%, HA 100 kDa 0.2% or HA 1000 kDa0.2% followed by a 20-min incubation time with BAK 0.002% (HCE: corneal epithelial cell line) or BAK 0.005% (WKD:conjunctival epithelial cell line). *p < 0.05, **p < 0.01, and ***p < 0.001 compared with BAK control. Results are expressed as

d wial ce

percentages of control values (dotted line). Cells preincubatesignificant decrease in mitochondrial mass on both epitheli

permeability (+2424% and +2428% with HCE and WKD, respec-tively).

A 20-min preincubation time with HA 100 kDa and HA1000 kDa before a 20-min incubation time with BAK induceda significant decrease in chromatin condensation, comparedwith BAK control cells. Results obtained with HA 1000 kDa(−178% and −252% with HCE and WKD, respectively) weresignificantly different from results obtained with HA 100 kDa(p < 0.001). Results obtained on corneal and conjunctivalepithelial cell lines with HA 20 kDa were not significantly dif-ferent from results obtained with BAK control.

Preincubation of the corneal epithelial cell line for 20 min

with HA 20 kDa, HA 100 kDa or HA 1000 kDa before a 20-minincubation time with BAK induced a significant decrease inplasma-membrane permeability compared with BAK controlcells. Results obtained with HA 1000 kDa (−1952%) were signif-

th HA 20 kDa, HA 100 kDa and HA 1000 kDa showed all lines.

icantly different from results obtained with HA 20 kDa or HA100 kDa (p < 0.001). The conjunctival epithelial cell line prein-cubated with HA 1000 kDa showed a significant decrease inplasma-membrane permeability (−1492%) whereas preincu-bation with HA 20 kDa and HA 100 kDa had no significanteffect.

3.7. HA 1000 kDa decreases BAK-induced DNAfragmentation

Flow cytometric analysis measuring propidium iodide stain-ing of DNA (Fig. 6) demonstrated that after a 24-h recovery

time, a very low dose of BAK (0.001%) even induced a sig-nificant increase in DNA fragmentation on the conjunctivalepithelial cell line (mean fluorescence 22.5%, p < 0.001) com-pared with control cells (mean fluorescence 5.2%). After 24 h

e u r o p e a n j o u r n a l o f p h a r m a c e u t i c a l s c i e n c e s 3 4 ( 2 0 0 8 ) 263–273 269

Fig. 5 – Apoptosis evaluation. (A) Chromatin condensation (Hoechst 33342) and (B) plasmic-membrane permeability(YO-PRO-1) were evaluated after a 20-min preincubation time with PBS, HA 20 kDa 0.2%, HA 100 kDa 0.2% or HA 1000 kDa0.2% followed by a 20-min incubation time with BAK 0.002% (HCE: corneal epithelial cell line) or BAK 0.005% (WKD:conjunctival epithelial cell line). *p < 0.05, **p < 0.01, and ***p < 0.001 compared with BAK control. Results are expressed aspercentages of control values (dotted line). Cells preincubated with HA 1000 kDa showed a significant decrease in chromatinc th ep

oscciDptwc

3

TwBbaa

ondensation and in plasmic-membrane permeability on bo

f incubation with HA 20 kDa, HA 100 kDa or HA 1000 kDa, noignificant modification of DNA fragmentation was observedompared with the control (data not shown). Cells prein-ubated with HA 100 kDa or HA 1000 kDa before a 20-minncubation time with BAK showed a significant decrease inNA fragmentation (mean fluorescence HA 1000 kDa 9.4%< 0.001, HA 100 kDa 13.6% p < 0.001) compared with BAK con-

rol cells. Results with HA 20 kDa (mean fluorescence 21.1%)ere not significantly different from results obtained with BAK

ontrol.

.8. HA 1000 kDa protects F-actin cytoskeleton

he corneal epithelial cell line was preincubated for 20 minith HA 1000 kDa (0.2%), and then incubated for 20 min with

AK 0.002%. As shown in Fig. 7B compared to Fig. 7A, incu-ation with BAK induced a disruption of F-actin cytoskeletonrrangement and altered cell morphology. Indeed, there wasdecrease in the number of elongated bundles and the ring

ithelial cell lines.

like structure disappeared. A preincubation for 20 min withHA 1000 kDa before a 20-min incubation time (Fig. 7C) withBAK obviously protected actin structure against BAK-inducedcytoskeleton disorganization.

4. Discussion

Among all organs exposed to toxic products, the eye is one ofthe most sensitive. Cornea and conjunctiva are the main con-stituents of the ocular surface. In this study, solutions weretested during short incubation times (20 min) to reproducewhat can happen in vivo. Results showed that benzalkoniumchloride, the preservative currently used in most ophthalmicproducts, induced an important toxicity on human epithelial

corneal and conjunctival cell lines. Indeed, we observed in ourstudy an important cell viability decrease (confirmed by twoindependent assays) associated with huge plasma-membranepermeability, important chromatin condensation, disruption

270 e u r o p e a n j o u r n a l o f p h a r m a c e u t i c a l s c i e n c e s 3 4 ( 2 0 0 8 ) 263–273

Fig. 6 – DNA fragmentation. Conjunctival epithelial cells were preincubated with PBS, HA 20 kDa, HA 100 kDa or HA1000 kDa and incubated 20 min with PBS or BAK 0.001%. After a 24-h recovery time, DNA fragmentation was evaluated. (A)Preincubation with PBS and incubation with PBS. (B) Preincubation with PBS and incubation with BAK. (C) Preincubationwith HA 1000 kDa and incubation with BAK. (D) Preincubation with HA 100 kDa and incubation with BAK. (E) Preincubation

d 20DNA

with HA 20 kDa and incubation with BAK. Cells preincubateincubation time with BAK showed a significant decrease in

of F-actin cytoskeleton arrangement and DNA fragmentation.Moreover, we demonstrated that BAK induced an oxidativestress with overproduction of hydroperoxide and superoxideanion, as well as an increase in the mitochondrial mass.

Reactive oxygen species overproductions may result fromalterations of the mitochondrial respiratory chain that can nolonger reduce oxygen, which is the final electron acceptor.

This overproduction results in the opening of the mitochon-drial pores, which induces stretching of mitochondrial crestsand formation of megamitochondria (Matsuhashi et al., 1997;Karbowski et al., 1999). Superoxide-generating systems have

Fig. 7 – F-actin detection on the corneal epithelial cell line usingincubation for 20 min with BAK 0.002% and (C) preincubation witBAK 0.002%. A decrease in cell size associated with cell disorganwith HA 1000 kDa showed few morphologic changes.

min with HA 100 kDa or HA 1000 kDa before a 20-minfragmentation compared with BAK control cells.

been demonstrated to be cytotoxic by degrading DNA and pro-moting peroxidation of membrane lipids (Freeman and Crapo,1982). It is not clear whether reactive oxygen species are a partof the signal transduction cascade triggered by various induc-ers of apoptosis, or whether they are generated in a parallelpathway that can independently trigger apoptosis. Resultsshowed three different characteristics of apoptotic cells: per-

meability of the plasma membrane, chromatin condensationand DNA fragmentation.

In accordance with previous in vitro studies (Debbasch etal., 1999, 2001; De Saint Jean et al., 2002), we demonstrated

Alexa-phalloıdin staining. (A) Negative control, (B)h HA 1000 kDa 0.2% followed by incubation for 20 min withization was observed with BAK, whereas cells preincubated

t i c a

tpObewohodfaB

(o

dwwaiHcBpva(swi(ssWWp1vpwomwrw

imsrbwlp

tme

r

e u r o p e a n j o u r n a l o f p h a r m a c e u

hat BAK is a toxic agent. Nevertheless, BAK is necessary torevent microbial development in pharmaceutical products.ur idea was to investigate in vitro whether HA, a well-knowniopolymer, is able to reduce BAK toxicity. HA possesses inter-sting physicochemical properties. This viscous biopolymerith negative charges can neutralize the toxic cationic chargef the BAK quaternary ammonium. Besides, HA is rich inydroxyl functions, which can potentially absorb reactivexygen species. Nevertheless, the biological functions of HAepend on its number of disaccharides. In this study, weocused on the relation between HA molecular weight and itsbility to improve the safety of aqueous products containingAK.

Firstly, we assessed the safety of the three HAs we studiedHA 1000 kDa, HA 100 kDa and HA 20 kDa). No cytotoxicity wasbserved even after a 24-h incubation time.

Then, we evaluated HA protection against BAK with threeifferent protocols. Following the first protocol, HA and BAKere incubated at the same time. In the second protocol, BAKas incubated before HA. Results with both these protocols

nd with both epithelial cell lines did not show any signif-cant protective effect of HA. Following the third protocol,A was incubated before BAK. Results showed that a prein-ubation with HA 1000 kDa alone was able to reduce all theAK-induced toxic effects we previously observed. HA cyto-rotective effect was similar on both epithelial cell lines. Celliability was significantly higher than that observed with BAKlone: there was an increase in lysosomal membrane integrityHCE: +49%, WKD: +42%) and an increase in intracellular redoxtatus (HCE: +59%, WKD: +61%). Oxidative stress was reduced,ith significant decreases in overproduction of hydroperox-

de (HCE: −36%, WKD: −34%) as well as of superoxide anionHCE: −32%, WKD: −60%). Last but not least, apoptosis wasignificantly lower than that observed with BAK. We observedignificant decreases in chromatin condensation (HCE: −44%,

KD: −49%) and in membrane permeability (HCE: −77%,KD: −60%). These results were confirmed using two inde-

endent assays on both cell lines. Results showed that HA000 kDa decreased DNA fragmentation (−58%) of conjuncti-al cells and ultrastructural studies showed that HA 1000 kDarotected corneal cells. Results with HA 100 kDa and HA 20 kDaere different from those obtained with HA 1000 kDa: webserved that the more HA molecular weight decreases, theore cytoprotection decreases. Consequently, the moleculareight of this linear molecule is a fundamental factor. These

esults lead to the conclusion that only the high moleculareight HA (1000 kDa) has a significant cytoprotective effect.

HA specifically interacts with several cell surface receptors,ncluding the cluster determinant 44 (CD44) considered as the

ain HA cell-surface receptor (Tammi et al., 1998). We demon-trated that both epithelial cell lines express this membraneeceptor, suggesting that HA could strongly bind to cell mem-ranes. These results were in accordance with clinical studies,hich showed that HA coats the corneal epithelium for at

east 1 h (Polack, 1982). We suggest that HA 1000 kDa coats androtects cell membranes interacting with CD44 receptor.

BAK induced a strong plasma-membrane permeabiliza-ion. A YO-PRO-1 probe was used to evaluate plasma-

embrane permeabilization. YO-PRO-1 is also used tovaluate P2X7 cell death receptor activity. This receptor, a

l s c i e n c e s 3 4 ( 2 0 0 8 ) 263–273 271

member of the ionotropic P2X receptor family, is activatedby extracellular ATP (Surprenant et al., 1996; Rassendren etal., 1997), which can induce the formation of cytolytic pores.Cytolytic pores are permeable to large molecules such as flu-orescent dyes like YO-PRO-1 (Viginio et al., 1999) leading toapoptosis through different pathways (Budagian et al., 2003;Donnelly-Roberts et al., 2004; Wang et al., 2003). In a pre-vious study, we observed that BAK, like other quaternaryammonium preservatives, was able to induce oxidative stress,apoptosis and P2X7 receptor activation. We suggested that themodulation of P2X7 receptor activation could be an interest-ing tool to limit preservative toxicity (Dutot et al., 2008). In thisstudy, we observed that HA 1000 kDa highly decreased BAK-induced P2X7 receptor activation (HCE: −77%, WKD: −60%)and significantly increased cell viability.

HA 1000 kDa is a very large biopolymer that coats cell mem-brane via strong links with CD44 receptors. We suggested thatit physically masks at the same time P2X7 cell death receptor.Therefore, HA could prevent P2X7 receptors to be activated bytheir ligand, in this case extracellular ATP released by necroticcells. This hypothesis is very promising for two reasons: onthe one hand, HA physically protects cell membranes and onthe other hand, it inactivates plasma membrane cell deathreceptors and acts as a specific antiapoptotic agent.

In conclusion, our present study showed that HA 1000 kDa,HA 100 kDa and HA 20 kDa do not have any toxic effect on twohuman epithelial cell lines. HA 1000 kDa significantly reducesall the BAK-induced cytotoxic effects we observed. We sug-gested that HA 1000 kDa forms a cytoprotective coat on cellmembrane. This cytoprotective coat prevents BAK cytotoxic-ity and could isolate cell death receptor, such as P2X7, fromits potential ligands. Biopolymer HA 1000 kDa, which featuresantioxidant and antiapoptotic effects, has promising thera-peutical topic applications in diverse inflammatory or toxicsyndromes, notably in ophthalmology and in dermatology.For example, high molecular weight hyaluronan could be pro-posed to the millions of patients who daily use preservedocular medications (antiglaucomatous eyedrops, contact lenscare solutions,. . .) and risk impairment of their ocular surface.However, caution must be used when extrapolating these invitro findings to the clinical setting. Further in vivo studies ofHA protective effects are needed before a final conclusion canbe reached.

e f e r e n c e s

Araki-Sasaki, K., Ohashi, Y., Sasabe, T., Hayashi, K., Watanabe, H.,Tano, Y., Handa, H., 1995. An SV40-immortalized humancorneal epithelial cell line and its characterization. Invest.Ophthalmol. Vis. Sci. 36, 614–621.

Arnitz, R., Ott, H.W., Gstottner, M., Nagl, M., Scholtz, A.W., Neher,A., 2006. A novel N-chlorotaurine-corticosteroid combinationas a preservative-free local disinfectant: influence on theciliary beat frequency in vitro. Acta Otolaryngol. 126, 291–294.

Batts, A.H., Marriott, C., Martin, G.P., Wood, C.F., Bond, S.W., 1990.

The effect of some preservatives used in nasal preparationson the mucus and ciliary components of mucociliaryclearance. J. Pharm. Pharmacol. 42, 145–151.

Baudouin, C., Pisella, P.J., Fillacier, K., Goldschild, M., Becquet, F.,De Saint, J.M., Bechetoille, A., 1999. Ocular surface

u t i c

272 e u r o p e a n j o u r n a l o f p h a r m a c e

inflammatory changes induced by topical antiglaucomadrugs: human and animal studies. Ophthalmology 106,556–563.

Belloc, F., Dumain, P., Boisseau, M.R., Jalloustre, C., Reiffers, J.,Bernard, P., Lacombe, F., 1994. A flow cytometric method usingHoechst 33342 and propidium iodide for simultaneous cellcycle analysis and apoptosis determination in unfixed cells.Cytometry 17, 59–65.

Berg, O.H., Lie, K., Steinsvag, S.K., 1997. The effects of topicalnasal steroids on rat respiratory mucosa in vivo, with specialreference to benzalkonium chloride. Allergy 52, 627–632.

Boek, W.M., Romeijn, S.G., Graamans, K., Verhoef, J.C., Merkus,F.W., Huizing, E.H., 1999. Validation of animal experiments onciliary function in vitro. II. The influence of absorptionenhancers, preservatives and physiologic saline. ActaOtolaryngol. 119, 98–101.

Borenfreund, E., Puerner, J.A., 1985. Toxicity determined in vitroby morphological alterations and neutral red absorption.Toxicol. Lett. 24, 119–124.

Budagian, V., Bulanova, E., Brovko, L., Orinska, Z., Fayad, R., Paus,R., Bulfone-Paus, S., 2003. Signaling through P2X7 receptor inhuman T cells involved p56lck, MAP kinases, andtranscription factor AP-1 and NF-kappa B. J. Biol. Chem. 278,1549–1560.

Chen, W.Y., Abatangelo, G., 1999. Functions of hyaluronan inwound repair. Wound Repair Regen. 7, 79–89.

Choucroun, P., Gillet, D., Dorange, G., Sawicki, B., Dewitte, J.D.,2001. Comet assay and early apoptosis. Mutat. Res. 478, 89–96.

Cureoglu, S., Akkus, M., Osma, U., Yaldiz, M., Oktay, F., Can, B.,Guven, C., Tekin, M., Meric, F., 2002. The effect ofbenzalkonium chloride on rabbit nasal mucosa in vivo: anelectron microscopy study. Eur. Arch. Otorhinolaryngol. 259,362–364.

de Fries, R., Mitsuhashi, M., 1995. Quantification of mitogeninduced human lymphocyte proliferation: comparison ofAlamar Blue assay to 3H-thymidine incorporation assay. J.Clin. Lab. Anal. 9, 89–95.

De Saint Jean, M., Debbasch, C., Brignole, F., Warnet, J.M.,Baudouin, C., 2002. Relationship between in vitro toxicity ofbenzalkonium chloride and preservative-induced dry eye.Adv. Exp. Med. Biol. 506, 697–702.

Debbasch, C., De Saint Jean, M., Pisella, P.J., Rat, P., Warnet, J.M.,Baudouin, C., 1999. Quaternary ammonium cytotoxicity in ahuman conjunctival cell line. J. Fr. Ophtalmol. 22, 950–958.

Debbasch, C., Pisella, P.J., De Saint, J.M., Rat, P., Warnet, J.M.,Baudouin, C., 2001. Mitochondrial activity and glutathioneinjury in apoptosis induced by unpreserved and preservedbeta-blockers on Chang conjunctival cells. Invest.Ophthalmol. Vis. Sci. 42, 2525–2533.

Donnelly-Roberts, D.L., Namovic, M.T., Faltynek, C.R., Jarvis, M.F.,2004. Mitogen-activated protein kinase and caspase signalingpathways are required for P2X7 receptor (P2X7R)-induced poreformation in human THP-1 cells. J. Pharmacol. Exp. Ther. 308,1053–1061.

Dutot, M., Warnet, J.M., Baudouin, C., Rat, P., 2008. Cytotoxicity ofcontact lens multipurpose solutions: role of oxidative stress,mitochondrial activity and P2X7 cell death receptoractivation. Eur. J. Pharm. Sci. 33, 138–145.

Eleftheriadis, H., Cheong, M., Sandeman, S., Syam, P.P., Brittain, P.,Klintworth, G.K., Lloyd, A., Liu, C., 2002. Corneal toxicitysecondary to inadvertent use of benzalkonium chloridepreserved viscoelastic material in cataract surgery. Br. J.Ophthalmol. 86, 299–305.

Freeman, B.A., Crapo, J.D., 1982. Biology of disease: free radicals

and tissue injury. Lab. Invest. 47, 412–426.

Gomes, A., Fernandes, E., Lima, J.L., 2005. Fluorescence probesused for detection of reactive species. J. Biochem. Biophys.Methods 397, 414–423.

a l s c i e n c e s 3 4 ( 2 0 0 8 ) 263–273

Horton, M.R., McKee, C.M., Bao, C., Liao, F., Farber, J.M.,Hodge-Dufour, J., Pure, E., Oliver, B.L., Wright, T.M., Noble,P.W., 1998. Hyaluronan fragments synergize withinterferon-gamma to induce the C-X-C chemokines mig andinterferon-inducible protein-10 in mouse macrophages. J. Biol.Chem. 273, 35088–35094.

Horton, M.R., Olman, M.A., Noble, P.W., 1999. Hyaluronanfragments induce plasminogen activator inhibitor-1 andinhibit urokinase activity in mouse alveolar macrophages: apotential mechanism for impaired fibrinolytic activity inacute lung injury. Chest 116, 17S.

Idziorek, T., Estaquier, J., De Bels, F., Ameisen, J.C., 1995. YOPRO-1permits cytofluorometric analysis of programmed cell death(apoptosis) without interfering with cell viability. J. Immunol.Methods 185, 249–258.

Karbowski, M., Kurono, C., Wozniak, M., Ostrowski, M., Tearnishi,M., Nishizawa, Y., Usukura, J., Soji, T., Wakabayashi, T., 1999.Free radical-induced megamitochondria formation andapoptosis. Free Radic. Biol. Med. 26, 396–409.

Larson, E.M., Doughman, D.J., Gregerson, D.S., Obritsch, W.F.,1997. A new, simple, nonradioactive, nontoxic in vitro assay tomonitor corneal endothelial cell viability. Invest. Ophthalmol.Vis. Sci. 38, 1929–1933.

Laurent, T.C., Fraser, J.R., 1992. Hyaluronan. FASEB J. 6, 2397–2404.

Lebe, E., Baka, M., Yavasoglu, A., Aktug, H., Ates, U., Uyanikgil, Y.,2004. Effects of preservatives in nasal formulations on themucosal integrity: an electron microscopic study.Pharmacology 72, 113–120.

Matsuhashi, T., Liu, X., Karbowski, M., Wozniak, M., Antosiewicz,J., Wakabayashi, T., 1997. Role of free radicals in themechanism of the hydrazine-induced formation ofmegamitochondria. Free Radic. Biol. Med. 23, 285–293.

McKee, C.M., Penno, M.B., Cowman, M., Burdick, M.D., Strieter,R.M., Bao, C., Noble, P.W., 1996. Hyaluronan (HA) fragmentsinduce chemokine gene expression in alveolar macrophages.The role of HA size and CD44. J. Clin. Invest. 98, 2403–2413.

Noble, P.W., 2002. Hyaluronan and its catabolic products in tissueinjury and repair. Matrix Biol. 21, 25–29.

Oiso, N., Fukai, K., Ishii, M., 2005. Irritant contact dermatitis frombenzalkonium chloride in shampoo. Contact Dermatitis52, 54.

Polack, F.M., 1982. Penetrating keratoplasty using MK storedcorneas and Na Hyaluronate (Healon). Trans. Am.Ophthalmol. Soc. 80, 248–261.

Rassendren, F., Buell, G.N., Viginio, C., Collo, G., North, R.A.,Surprenant, A., 1997. The permeabilizing ATP receptor, P2X7.Cloning and expression of a human cDNA. J. Biol. Chem. 272,5482–5486.

Surprenant, A., Rassendren, F., Kawashima, E., North, R.A., Buell,G.N., 1996. The cytolytic P2Z receptor for extracellular ATPidentified as a P2X receptor (P2X7). Science 272, 735–738.

Tammi, R., MacCallum, D., Hascall, V.C., Pienimaki, J.P., Hyttinen,M., Tammi, M., 1998. Hyaluronan bound to CD44 onkeratinocytes is displaced by hyaluronan decasaccharides andnot hexasaccharides. J. Biol. Chem. 273, 28878–28888.

Toole, B.P., 2004. Hyaluronan: from extracellular glue topericellular cue. Nat. Rev. Cancer 4, 528–539.

Tripathi, B.J., Tripathi, R.C., Kolli, S.P., 1992. Cytotoxicity ofophthalmic preservatives on human corneal epithelium. LensEye Toxic. Res. 9, 361–375.

Viginio, C., MacKenzie, A., North, R.A., Surprenant, A., 1999.Kinetics of cell lysis, dye uptake and permeability changes incells expressing the rat P2X7 receptor. J. Physiol. 519, 335–346.

Wang, C.M., Chand, Y.Y., Sun, S.H., 2003. Activation of P2X7purinoceptor-stimulated TGF-beta 1 mRNA expressioninvolves PKC/MAPK signalling pathway in rat brain-derivedtype-2 astrocyte cell line, RBA-2. Cell Signal. 15, 1129–1137.

t i c a

W

Z

e u r o p e a n j o u r n a l o f p h a r m a c e u

eissman, B., Meyer, K., 1954. The structure of hyalobiuronic acidand of hyaluronic acid from umbilical cord. J. Am. Chem. Soc.27, 1753–1757.

hao, H., Kalivendi, S., Zhang, H., Joseph, J., Nithipatikom, K.,Vasquez-Vivar, J., Kalyanaraman, B., 2003. Superoxide reacts

l s c i e n c e s 3 4 ( 2 0 0 8 ) 263–273 273

with hydroethidine but forms a fluorescent product that isdistinctly different from ethidium: potential implications inintracellular fluorescence detection of superoxide. Free Radic.Biol. Med. 34, 1359–1368.