Chitosan Nanoparticles as a Potential Drug Delivery System for the Ocular Surface: Toxicity, Uptake Mechanism and In Vivo Tolerance

Abstract

purpose. To study the in vitro and in vivo interaction of chitosan nanoparticles (CSNPs), a new particulate drug carrier, with epithelial cells on the ocular surface.

methods. CSNPs labeled with fluorescein isothiocyanate–bovine serum albumin were produced by ionotropic gelation. Human conjunctival epithelial cells (IOBA-NHC) were exposed for 15, 30, 60, and 120 minutes to three different CSNP concentrations. Immediately after treatment and after a 24-hour recovery period in culture medium, cell survival, and viability were measured. The association of CSNPs with IOBA-NHC cells was investigated by confocal microscopy. The influence of temperature and the effect of metabolic inhibition were studied by fluorometry. The in vivo uptake and acute tolerance of the ocular surface to CSNPs were evaluated in rabbits.

results. Cell survival and viability of CSNP-exposed cells were equivalent to that of the control. Uptake of CSNPs was continuous for the 2-hour duration of these experiments and was temperature dependent. Metabolic inhibition by sodium azide had no effect on CSNP uptake. The rabbit ocular surface showed no signs of inflammation or alteration after CSNP exposure compared with the control. Fluorescence microscopy of rabbit eyeball and lid sections confirmed in vivo uptake by conjunctival and corneal epithelia.

conclusions. CSNPs were internalized by IOBA-NHC cells by an active transport mechanism that did not compromise cell viability. Moreover, these nanoparticles were well tolerated by the ocular surface tissues. These facts add further support for the potential use of these colloidal systems to delivery drugs to the ocular surface.

Ocular drug delivery is extraordinarily hampered by the defensive barriers of the eye. Both corneal and conjunctival epithelia are sealed with intercellular tight junctional complexes that limit the entrance of potentially harmful substances, or for that matter, helpful substances such as therapeutic drugs. In addition, the mucous tear film entraps debris, particles, and drugs and removes them through the lacrimal drainage system. These protective physiological mechanisms often limit the efficacy of eye disease treatments. To compensate for this loss, drugs must be applied frequently, which can generate problems of patient compliance. Physicochemical stability and limited transport across epithelia are additional problems for drug delivery, especially in the case of large biomolecules such as peptides, proteins, and genes. Thus, the efficacy of a topically administered drug delivery system depends on interaction with the ocular mucosa, protection from degradation, and facilitation of delivery to the ocular tissues. 

Previous studies showed that both poly(alkylcyanoacrylate) ^{1 2} and poly-ε-caprolactone ^{3 4 5} colloidal systems are able to improve the intraocular penetration of drugs. However, the duration of these systems at the ocular surface is limited to a few hours. In contrast, mucoadhesive polymers can increase the residence time of drugs on the ocular surface. ⁶ One of these, the cationic polymer chitosan, possesses some favorable biological characteristics such as biodegradability, nontoxicity, and biocompatibility ^{7 8 9} that make it a promising candidate for ocular drug delivery. ¹⁰  

Our group has developed different chitosan-based nanocarriers. These nanocarriers enhance the transport of drugs across the nasal ^{11 12} and intestinal ¹³ mucosal barriers, and they exhibit very promising behavior as topical ocular drug delivery systems. ^{14 15} Chitosan-coated nanocarriers, composed of nanocapsules with an oily core and a chitosan coating, have a great affinity for corneal and conjunctival epithelial cells. ¹⁶ Similarly, nanoparticles made solely of chitosan exhibit a facilitated interaction with the cornea and conjunctiva where they remain for more than 24 hours. ¹⁵ The surface localization of cyclosporine-loaded nanoparticles was responsible for high and sustained levels of this drug delivered to the corneal and conjunctival epithelia. ¹⁴ Based on these observations, we investigated the mechanism of interaction of newly developed, improved chitosan nanoparticles (CSNPs) with conjunctival epithelial cells and evaluated the in vivo tolerance of this new nanoparticulate drug delivery system. 

Materials and Methods

Chitosan in the form of hydrochloride salt (Protasan CL 110) was purchased from Pronova Biopolymer AS (Oslo, Norway). Pentasodium tripolyphosphate (TPP), fluorescein isothiocyanate-bovine serum albumin (FITC-BSA), benzalkonium chloride (BAC), and tetramethylrhodamine-isothiocyanate (TRITC)-phalloidin were obtained from Sigma-Aldrich (St. Louis, MO). Culture plates were obtained from Nunc (Roskilde, Denmark). DMEM/F12 culture medium and other cell culture reagents were from Invitrogen-Gibco (Inchinnan, UK). Propidium iodide (PI) was obtained from Invitrogen (Leiden, The Netherlands). Fluorescence anti-fade mounting medium (Vectashield) was from Vector Laboratories (Burlingame, CA). Xylazine (Rompun) was from Bayer AG (Leverkusen, Germany), ketamine (Ketolar) from Pfizer Ireland Pharmaceuticals (County Dublin, Ireland) and topical anesthetic from Alcon (Barcelona, Spain). Polyether sulfone paper filters (0.20-μm pore size, 13 mm in diameter; Supor 200) were obtained from Gelman Laboratory (Ann Arbor, MI). All other reagents were from Sigma-Aldrich, unless otherwise specified. 

Preparation of CSNPs

CSNPs were prepared by ionotropic gelation of chitosan with TPP according to the procedure developed by Calvo et al. ¹⁷ Briefly, 1.2 mL of 0.5 mg/mL TPP aqueous solution were added to 3 mL of 1 mg/mL chitosan and stirred at room temperature. The CSNPs were labeled by adding 0.725 mg/mL FITC-BSA to the TPP aqueous solution. The nanoparticles formed spontaneously and were then concentrated by centrifugation at 11,000 g in a glycerol bed for 45 minutes. The supernatants were discarded, and CSNPs were resuspended in purified water for further characterization. 

Physicochemical Characterization of CSNPs

Mean particle size and size distribution of the nanoparticles were determined by photon correlation spectroscopy (Zetasizer 3000-HS; Malvern Instrument, Malvern, UK). The effect of different temperatures and time in culture medium on the mean size of particles was also determined. 

The zeta potential of nanoparticles was studied by using laser Doppler anemometry with a Zetasizer 3000-HS (Malvern Instruments). This technique allows a fast and high-resolution determination of the surface charge of colloidal particles by measuring their velocity in an electrical field. Samples were diluted with 0.1 mM KCl and placed in an electrophoretic cell where a potential of ±150 mV was established. The zeta potential was calculated from the mean electrophoretic mobility values with Smoluchowski’s equation. ¹⁸ Each CSNP batch was analyzed in triplicate. 

The amount of FITC-BSA encapsulated into the CSNPs was determined by spectrophotometry (495 nm, UV-1603; Shimadzu, Germany) and calculated as the difference between the FITC-BSA used to prepare the CSNPs and that present in the supernatant after centrifugation. In the same way, the release of FITC-BSA from CSNPs was checked after a 4-hour incubation in phosphate-buffered saline (PBS) at 37°C. 

IOBA-NHC Cell Line and Culture Conditions

The IOBA-NHC cell line derived from normal human conjunctival epithelium ¹⁹ was cultured in DMEM/F12 supplemented with 2 ng/mL human epidermal growth factor, 0.1 μg/mL cholera toxin, 1 μg/mL bovine pancreatic insulin, 10% fetal bovine serum, 0.5 μg/mL hydrocortisone, 50 U/mL penicillin, 50 mg/mL streptomycin, and 2.5 μg/mL amphotericin B. Cells in passages 69 to 79 with a viability equal to or higher than 98% were used. 

In Vitro Cell Survival after Exposure of IOBA-NHC Cells to CSNPs

IOBA-NHC cell survival after exposure to 0.25, 0.5, or 1 mg/mL CSNPs diluted in PBS was determined both immediately and after a 24-hour recovery period in culture medium. Cells were seeded onto two six-well plates (2 × 10⁵ cells/well). When confluence was reached after approximately 24 hours, cells were washed two times in supplement-free DMEM/F-12 and they were exposed to CSNPs. Each plate included the three masked CSNP suspensions prepared in PBS at three different concentrations, and the following controls: FITC-BSA solution in PBS, culture medium (negative control), PBS (negative control), and 0.005% BAC in culture medium (positive control of cell damage). After incubation for 15, 30, 60, or 120 minutes, plates were washed three times with PBS (pH 5.0). One plate was immediately trypsinized with 0.25% trypsin with 1 mM EDTA in Ca²⁺ and Mg²⁺-free Hanks’ balanced salt solution. The other plate was maintained for an additional 24-hour period in supplemented DMEM/F-12, after which it was processed as described above. Recovered cells were counted in a hemocytometer and the viability measured with the Trypan blue dye exclusion test. Cell survival was expressed as the percentage of cells recovered after treatment compared to the control cells exposed to culture medium. Experiments were performed in quadruplicate (n = 4), and treatments were always performed in duplicate. 

Interaction of CSNPs with IOBA-NHC Cells

Immunofluorescence Assays.

To evaluate the ability of CSNPs to cross the plasma membrane and enter the cell, confluent IOBA-NHC cells were exposed to CSNPs and examined under a confocal laser scanning microscope (model LSM310; Carl Zeiss Meditec, Jena, Germany). Cells were seeded onto glass coverslips in 24-well plates (3 × 10⁵ cells/well) and grown to confluence. Before adding CSNPs, the cells were washed for 1 hour in supplement-free DMEM/F-12 culture medium and then exposed to one of the three CSNP concentrations for 15, 30, 60, and 120 minutes. 

After incubation, cells were washed three times with 0.27% glucose in PBS (pH 5.0), and once with cold PBS (pH 7.4), then fixed in ice cold methanol for 10 minutes and counterstained with PI (1:12,000) to identify nuclei. In some experiments, after fixation the cells were permeabilized for 30 minutes in PBS (pH 7.4), containing 0.27% glucose and 0.2% Tween-20, and actin filaments were stained with TRITC-phalloidin (1:200). The preparations were viewed under the confocal microscope equipped with a krypton-argon laser that excited FITC at 488 nm and PI and TRITC at 543-nm. Detection was with a band-pass emission barrier filter. Except where noted, images were collected with the 63× objective without zoom, at 512 × 512 pixels per inch. The field of view was 193 μm wide when using the 63× objective and 226 μm wide when using the 40× objective. To identify the localization of CSNPs within the cell, we captured six serial optical sections every 0.5 μm along the z-axis, from the top to the bottom of the cell layer. A scale bar was superimposed on the digital images, which were then prepared for publication (Photoshop 4.0; Adobe Systems, Mountain View, CA). Red fluorescent images of nuclei were merged with the corresponding green fluorescent images of CSNPs at the same z-axis value. Controls included FITC-BSA, PBS, and culture-medium–treated cells. All treatments were assayed in duplicate in three independent experiments (n = 3) and three independent fields were studied and photographed for each slide. 

Fluorometry Study

To determine whether CSNP uptake by IOBA-NHC cells was energy dependent, confluent IOBA-NHC cells were exposed to 100 μL of 0.5 mg/mL CSNPs (50 μg/well) for 15, 30, 60, and 120 minutes either at 37°C in the presence or the absence of a metabolic inhibitor, 100 mM sodium azide, or at 4°C. This sodium azide dose was chosen, as previous results ^{13 20} (data not shown) had shown that it was not harmful for cells at the assayed times. CSNP content inside cells was determined by fluorometry in cell lysates after FITC-CSNP exposure, as previously described by Russell-Jones et al. ²¹ with some modifications. After treatment, cells where washed three times with PBS (pH 5.0) and once with 5 mM EDTA (pH 5.0), to remove nanoparticles not taken up by the cells, including those attached to the cell membrane and to the plastic of the well. The cells where then frozen for 30 minutes, thawed, and disrupted with lysis buffer (pH 8.0), containing 2% SDS and 50 mM EDTA. Alternatively, after CSNP removal and cell washing, culture medium was added, and cells were maintained for an additional 24 hours, either at 37°C or 4°C. After this period, the culture medium was collected, and the remaining cells were lysed as just described. 

Cells exposed to the vehicle (PBS; pH 7.4), served as a control for intrinsic fluorescence, both in the presence and the absence of 100 mM sodium azide. Wells without cells, but incubated with CSNPs and processed as just described, served as the control for residual fluorescence after washing. Fluorescence in cell lysates and cell culture medium was immediately measured in a fluorometer (Fluoroskan II; Labsystems, Cheshire, UK) at 458 nm (excitation) and 538 nm (emission), and the data were analyzed on computer (DeltaSoft II software; V 3.62 FL; BioMetallics, Inc., Princeton, NJ). 

The amount of CSNPs taken up by the cells was calculated by interpolating the fluorescence intensity obtained in each cell lysate in a standard curve of known CSNP concentrations in lysis buffer, individually prepared for each single experiment. CSNP uptake was expressed as a percentage of the amount of CSNPs added to each well (50 μg). 

When standards were diluted in culture medium, the fluorescence was quenched, probably by interference with components of the medium. Normal levels of fluorescence were reached when the standards were prepared in culture medium diluted 1:7 with lysis buffer. Therefore, all culture medium samples were diluted in this fashion to avoid the interference of medium components in fluorescence emission. A standard curve was prepared in lysis buffer using different dilutions of CSNPs in culture medium, and the fluorescence was measured. Treatments were always performed in duplicate in at least three independent experiments (n = 3). Cellular uptake experiments at 37°C and at 4°C were run simultaneously with the same batches of cells and CSNPs. Similarly, experiments comparing uptake with and without sodium azide were run simultaneously with the same batches of cells and CSNPs. 

In Vivo Tolerance Assay

To test ocular tolerance to CSNPs, an in vivo local, acute tolerance assay was performed. Experiments were approved by the IOBA Research Committee and conformed to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Five female albino New Zealand rabbits weighing 2.0 to 2.5 kg were used. Animals received 30 μL of the 0.5 mg/mL CSNP formulation in the right eye every 30 minutes for 6 hours. The contralateral left eye was used as the control and received no treatment. A sixth rabbit was used as a sham-treated control. Three, 6, and 24 hours after the first instillation, animal discomfort and clinical signs in conjunctiva, cornea, and lids (Table 1)were macroscopically evaluated by a trained ophthalmologist, according to a modification of the scoring system established in the Organization for Economic Cooperation and Development (OECD) guidelines (1987) for ocular irritation testing. ²²  

The animals were euthanatized by air embolism after being deeply anesthetized by intramuscular administration of an overdose of anesthetic and paralyzing mixture of xylazine (20 mg/kg) and ketamine (200 mg/kg). Eyeball and lid tissues were removed, fixed in Davison fixative solution, and embedded in paraffin for a pathology study. Eyeball and lid sections (6 μm) were evaluated in a masked fashion according to the following criteria: alteration of the corneal, limbal, or conjunctival epithelia, edema in lid tissues, presence of inflammatory eosinophils, neutrophils, mast cells or lymphocytes, and any other abnormality. Animals wore Elizabethan collars during the study to avoid licking and scratching their eyes and face. 

In addition, conjunctival impression cytology (CIC) specimens were collected from both control and treated eyes in all animals 6 days before the experiment and immediately after sign evaluation 24 hours after the first instillation. Conjunctival cells were harvested using polyethersulfone paper filters. The filter paper was cut into eight pieces, and one piece was used for each eye of one animal. The filter was applied to the superior bulbar conjunctiva after topical anesthesia instillation, carefully pressed down, and subsequently removed with adherent cells. Filters were then placed in 96% alcohol and stained according to our modification of Tseng’s PAS-Papanicolau procedure. ²³ After staining, CICs were evaluated in a masked fashion according to Nelson’s classification. ²⁴ Briefly, this classification system evaluates the morphology of conjunctival ocular surface and the degree of squamous metaplasia by using a grade from 0 to 3 where grades 0 and 1 were for normal and grades 2 and 3 were for abnormal morphology. Grades were assigned for nuclear morphology, nuclear/cytoplasmic ratio, metachromasia, and goblet cell density. 

In Vivo CSNP Uptake

In vivo uptake of CSNPs was evaluated by fluorescence microscopy. Paraffin-embedded eyeballs and eyelids from the six rabbits used for the in vivo tolerance assay were used. Paraffin was removed from the sections with xylene-alcohol, mounted with fluorescence antifade mounting medium and examined under an epifluorescence microscope (DM L; Leica, Wetzlar, Germany). Both right (treated) and left (control) eyes from each animal were evaluated in duplicate. Sections from the sham control rabbit were examined as well. 

Statistical Analysis

Data are expressed as the mean ± SEM. The statistical significance of the differences between CSNP and the control solutions at each time point was analyzed by multivariate analysis of variance (ANOVA) with the Scheffé test as a post hoc test (SAS statistical package 8.01; SAS Institute, Cary, NC) or with Student’s t-test. Data from the in vivo tolerance assay were analyzed by the Fisher exact test. Differences were considered to be significant when P < 0.05. 

Results

Characterization of CSNPs

As determined by photon-correlation spectroscopy and laser Doppler anemometry, the average size of the CSNPs were 289 ± 13 nm, and the average zeta potential was +35.9 ± 1.4 mV, consistent with the cationic charge of chitosan. The encapsulation efficiency of FITC-BSA was 92.3% ± 0.4%. Furthermore, less than 1% of FITC-BSA encapsulated was released after 4 hours of incubation in PBS, indicating it was strongly attached to CSNPs. Therefore, the detected fluorescence corresponded to the presence of CSNPs. 

The stability of CSNP size at different temperatures and times was also determined by photon correlation spectroscopy (Table 2) . At 37°C, CSNPs aggregated after a 1-hour incubation in culture medium. Partial aggregation occurred after 2 hours at room temperature or 24 hours at 4°C. Therefore, we investigated cell survival after exposure to CSNPs for a maximum of 2 hours to determine the effect on cells before and after they aggregated. 

Cell Survival after Exposure to CSNPs

Most of the immediate cell survival values, as determined by counting recovered cells after exposure to CSNPs, were above 70% at all times and concentrations tested with the exception of the 0.25-mg/mL concentration, which resulted in cell survival of 47% after 15 minutes of exposure (Fig. 1A) . Cell survival immediately after CSNP exposure for 30 minutes was significantly higher (P < 0.05) than after exposure for 15 or 60 minutes, as determined by the Scheffé post hoc test. Cell survival after 0.5- and 1 mg/mL CSNP exposure for 30 minutes was at its maximum (103% and 108%, respectively). These high cell survival rates were also observed when cells were maintained in culture medium for an additional 24-hour period after exposure to CSNPs (Fig. 1B) . The only exception was for cells incubated in 1 mg/mL CSNP, where the 24-hour survival was of 45% for the 2-hour treatment. Cell survival rates after the 24-hour recovery period of PBS and CSNP-treated cells were significantly lower than those of the control culture medium–exposed cells (P < 0.05). There were no significant differences in cell survival in the immediate or 24-hour recovery period among the different concentrations of CSNPs tested or the PBS-treated cells. The survival of BAC-exposed cells was significantly lower when determined immediately (Fig. 1A)and after the 24-hour recovery period (Fig. 1B) . The great majority of the recovered cells were living, as shown by exclusion of Trypan blue. The viability of recovered cells was always higher than 92% when measured immediately after CSNP exposure and higher than 86% in cells maintained for 24 hours in culture medium after CSNP exposure (Figs. 1C 1D) . No significant differences in cell viability were observed among CSNP concentrations or incubation times, or when compared with culture medium controls. As expected, viability of BAC-treated cells was significantly lower of that of controls, both immediately after treatment and after the 24-hour recovery period. 

Interaction of CSNPs with IOBA-NHC Cells

The interaction of CSNPs with IOBA-NHC cells was studied by confocal microscopy. Control cells exposed to culture medium showed no green fluorescence (Figs. 2A 2B) . Cells exposed to the FITC-BSA solution had an evenly distributed green fluorescence in the cytoplasm (Fig. 2C) . In contrast, CSNPs appeared as discreet nanometer-sized, fluorescent dots inside the cells (Fig. 2D) . This confirmed that the staining was not due to the release of fluorescent dye. No morphologic alterations in cells exposed to CSNPs were observed compared with control cells. 

Based on confocal microscopy images, we determined that the thickness of fixed IOBA-NHC cells was approximately 3 μm. Thus, by taking six serial images at 0.5-μm intervals along the z-axis, from the top to the bottom of the cell monolayer, we confirmed the intracellular presence of the CSNPs (Fig. 3) . The extent of cell-associated CSNPs increased as the cytoplasmic area increased progressively along the z-axis. 

The presence of CSNPs in the intracellular compartment was observed at each time interval and for each concentration (Fig. 4) . The internalization of CSNPs was observed as soon as 15 minutes after incubation, and more nanoparticles were present inside the cells after 30 minutes. Uptake was concentration dependent, with the greatest number of fluorescent dots found with 0.5 and 1 mg/mL CSNPs, and fewer for 0.25 mg/mL. Aggregates were seen with longer incubation times, especially after 2 hours (Fig. 4) . This effect was more pronounced when higher CSNP concentrations were used. 

To detect a possible alteration in the cytoskeleton after cell exposure to CSNPs, TRITC-phalloidin was used in some experiments to counterstain actin filaments. No differences were observed in actin filament distribution in CSNP-exposed cells compared with the control (data not shown). 

Effect of Metabolic Inhibition on CSNP Uptake

The CSNP content inside the cells was quantitatively measured by fluorometry as previously described by Russell-Jones et al. ²¹ The quantity of cell-associated fluorescence reflects the amount of CSNPs internalized and, to a minor extent, that which is bound to the cell surface after washing with 5 mM EDTA (pH 5.0). The 0.5-mg/mL CSNP concentration was chosen for this study for several reasons: (1) This concentration showed less in vitro aggregation than 1 mg/mL; (2) in vitro cell recovery rates were in general better than or equal to that for the 0.25 mg/mL formulation; and (3) confocal microscopy images showed that cellular uptake was higher and faster than that for 0.25 mg/mL. 

Different controls were included to assure that the fluorescence measurements corresponded to cell-associated CSNPs. Control IOBA-NHC cells exposed to PBS did not show intrinsic fluorescence (data not shown). Cell-free wells exposed to CSNPs in PBS showed a slight residual fluorescence, which was much lower than that measured in CSNP-exposed cells and was constant for all time points studied. Fluorescence in wells after cell lysis buffer removal was negligible, assuring that all fluorescence content was collected in the lysis buffer. 

Effect of Temperature.

CSNP uptake at 37°C increased steadily up to 2 hours of incubation (Fig. 5A) . Whereas cellular integrity was not affected by incubation at 4°C, the uptake was significantly reduced at 30-minute and longer exposures. Fluorescence values of cells that were maintained for an additional 24-hour recovery period in culture medium were significantly reduced at all time-points compared with those measured immediately after CSNP treatment (Fig. 5B) . There were no significant differences between those values obtained in cells maintained 24 hours at 37°C compared with 4°C (Fig. 5B) . Cell-free wells incubated with CSNPs had a similar pattern of fluorescence at 24 hours (Fig. 5C) . To determine whether cell-associated fluorescence was lost to the culture medium, the media were collected after the 24-hour recovery period, and the fluorescence was measured. The fluorescence, expressed as micrograms of CSNPs, was lower in supernatants for cells kept at 4°C compared with this kept at 37°C (Fig. 6A) , although the differences were significant only at 15 and 120 minutes. This effect was also observed in cell-free wells, although the differences were not significant at any time (Fig. 6B) . 

Effect of Azide.

CSNP uptake at 37°C in the presence of the metabolic inhibitor sodium azide (100 mM) was not significantly different from the control up to 2 hours (Fig. 7) . When cell-free wells were incubated with CSNPs with and without azide, similar levels of fluorescence were obtained (data not shown), thus ruling out a direct effect of sodium azide on fluorescence emission. 

In Vivo Tolerance Assay

Six days before the experiments began, the animals showed no clinical signs and ocular surface structures were normal. The mean clinical macroscopic signs score was 0. The CIC Nelson classification grade was 0 to 1 for both eyes, based on the normal appearance of the conjunctival epithelium with respect to goblet and nongoblet epithelial cells. There were abundant goblet cells and scattered presence of polymorphonuclear cells. Twenty-four hours after the first instillation of 0.5 mg/mL CSNPs, another CIC sample was taken from both eyes. Similar grades (0–1) were recorded, and no differences were observed between control and treated eyes (Figs. 8A 8B) . Rabbits showed no signs of discomfort at 24 hours. However, treated eyes had a clinical macroscopic sign score of 1, because of a minimal mucus-like discharge at 6 hours after the first instillation. The discharge may have been caused by the ability of CSNPs to aggregate and by the deficiency daily hygiene caused by the Elizabethan collars worn to prevent licking and scratching the eyes and face. 

Pathology of the rabbit eyeballs and lids confirmed the presence of normal ocular surface structures in both control and treated eyes. Conjunctival, limbal, and corneal epithelia displayed a normal number of cell layers composed of cells with appropriate morphology (Figs. 8C 8D 8E 8F) . Goblet cells remained abundant and filled with secretory product. Conjunctiva-associated lymphoid tissue was identified in all conjunctiva, and no differences in size or location were observed in CSNP-exposed eyes compared with control eyes. In addition, there were scattered polymorphonuclear cells in the conjunctival stroma, and no tissue edema was present in the cornea, conjunctiva, or lids. No differences were observed between CSNP-treated and control eyes. 

In Vivo CSNP Uptake

As expected, no fluorescence signal was detected in sections of sham control rabbit eyes (Figs. 9A 9D) . Sections from CSNP-treated eyes revealed fluorescence localized throughout the nonnuclear cytoplasm of corneal and conjunctival epithelial cells (Figs. 9B 9E) . Corneal epithelium from CSNP-treated eyes was uniformly fluorescent (Fig. 9B) . In some parts of the bulbar conjunctival epithelium, the fluorescence was localized in the apical membrane of epithelial cells. In the palpebral conjunctiva, it was evenly distributed along the entire cell membrane, as well as the cytoplasm (Fig. 9E) . Fluorescence was similarly detected in goblet cells. Occasionally, sections of contralateral control corneas and conjunctivas also emitted weak fluorescence signals that were much less intense than that of the treated eye (Figs. 9C 9F) . 

Discussion

In this work, we investigated a new kind of colloidal bioadhesive, CSNPs, developed by our group. Our results demonstrated that CSNPs are promising candidates for ocular surface drug delivery as they presented negligible toxicity, were able to enter conjunctival epithelial cells in vitro, and were well tolerated in vivo. Drug delivery systems for the ocular surface must overcome important physical barriers to reach their target cells. Different colloidal systems have been developed to solve this problem (for a review, see Ref. ⁶ ). Among them, chitosan-based systems are acknowledged as more suitable for the ocular pathway, based on the favorable biological characteristics of chitosan. ^{7 8 9} Several studies have demonstrated that nanoparticles are transported across epithelia more readily than are microparticles. ^{25 26} CSNPs can be easily prepared under mild conditions, and they can incorporate macromolecular bioactive compounds. ¹⁷ This propriety is extremely useful for drugs, proteins, genes, or hydrophobic molecules that are poorly transported across epithelia. 

A previously developed chitosan-based nanoparticle system successfully delivered cyclosporine A (CyA) to the ocular mucosa. ^{14 15} Animals treated with CyA-loaded chitosan nanoparticles had significantly higher corneal and conjunctival drug levels than did those treated with suspensions of CyA in chitosan aqueous solutions. ¹⁴ However, in those experiments, the kind of nanoparticles used could not be separated from the remaining chitosan in solution. Thus, the chitosan-induced effects on tight junctions could not be distinguished from effects exerted by nanoparticles alone. Another drawback of that preparation was that the FITC was covalently bound to the particles. ¹⁵ That attachment changed the conditions and behavior of the chitosan polymer and therefore changed the conditions for preparation of the nanoparticles. 

We present the next generation of chitosan nanoparticles with improved characteristics. CSNPs were prepared using the chitosan salt Protasan. They were loaded with FITC-BSA, and they were isolated and purified from the remaining chitosan solution by centrifugation. These CSNPs have been shown to be a good drug delivery system for intestinal ¹³ and nasal ^{11 12 27} mucosa. Administration of drug-loaded chitosan-based nanoparticles improved both insulin ¹¹ and tetanus toxoid ^{12 27} transport across the nasal epithelium in mice. With those results in mind, we used two approaches, in vitro and in vivo, to test these CSNPs in ocular tissues. 

In our in vitro experiments, we used the IOBA-NHC cell line to determine whether this new CSNP system was noxious for the conjunctival epithelium. Three different CSNP concentrations were assayed to find the optimal one and to determine the localization of CSNPs within the cell. Based on cell survival and viability values, the 0.5 mg/mL concentration was the best because it rendered maximal cell survival. Based on confocal microscopy, this concentration also allowed the clearest identification of CSNPs inside cells and showed less aggregation than the most concentrated formulation tested, 1 mg/mL. 

Cell survival in cultures exposed to CSNPs was very high, especially after 30 minutes of incubation. The significantly higher cell recovery levels at 30 minutes compared with 15 and 60 minutes may be related to the nanoparticle aggregation observed after 1 hour of incubation at 37°C. Large aggregates on the cell surface may cause a transient, pseudotoxic effect. The relatively low immediate cell survival obtained after 15 minutes of exposure may be related to a transient, artifactual membrane weakening due to the cell manipulation methodology. Fifteen minutes may not be enough time for cells to recover from the initial cell manipulation and/or the effect of PBS in which the CSNPs are suspended. 

Cell survival 24 hours after CSNP exposure was also high at all tested concentrations and exposure times, except the 2-hour exposure to 1 mg/mL CSNPs. Although survival was high, levels were significantly lower than that of control cells exposed to culture medium alone. The difference was probably because of the effect of PBS on cells, since similar survival rates were recorded immediately and after 24 hours in culture medium for cells exposed to PBS and all CSNP concentrations. Thus, no inherent toxicity can be attributed to CSNPs, per se. This was further confirmed by the viability of recovered cells, which were approximately 90% when measured immediately after CSNP exposure (>92%) and after the 24-hour recovery period (>86%). Even though PBS may cause transient damage to cells, it is the best medium for the CSNPs, because it maintains their physicochemical characteristics and is readily adjusted for pH and osmolarity. 

Confocal microscopy demonstrated that CSNPs crossed the plasma membrane without causing any apparent alteration in cytoskeleton. We obtained serial optical sections every 0.5 μm in the z-axis within the cell thickness, to confirm the exact location of the fluorescence in cells. We are certain that the observed fluorescence corresponded to intact CSNPs and not to the released label based on our previous physicochemical characterization studies. Those experiments showed that FITC-BSA was strongly attached to CSNPs, as less than 1% of FITC-BSA was released after 4 hours of incubation. Furthermore, the fluorescence pattern was punctiform bright dots that were clearly different from the homogenous appearance observed in FITC-BSA-exposed cells. This fact allowed us to assign the bright-dotted fluorescence to CSNPs. The CSNP content inside the cells increased with concentration and exposure time. 

These results agree with those in our previous studies ¹⁵ in which intracellular chitosan-based nanoparticles were detected in cross-sections of rabbit conjunctival epithelium after in vivo administration. However, in contrast to the uneven distribution of nanoparticles in the rabbit conjunctival epithelium, a uniform distribution was present in the IOBA-NHC cells after CSNP exposure. This difference may be due to the improved nanoparticles used in the present study. Also, the IOBA-NHC cell line is a homogeneous epithelial cell line, ¹⁹ whereas in the conjunctiva, different cell types are present. In fact, confocal microscopy images of cross-sections of corneal epithelium from chitosan-based nanoparticle-treated rabbits were drastically different from those of the conjunctiva. ¹⁵ This suggests that uptake mechanism depends on the cell type and/or tissue. Internalization of other types of nanoparticles in conjunctival epithelium has also been described using conjunctival primary culture models. ²⁸  

The uptake of CSNPs at 37°C was continuous, at least up to the 2-hour time-point studied. The percentage of CSNPs taken up, ∼38% at 1 hour, was much higher than those previously reported for MTX-E12 and Caco-2 cells in similar studies: 13% and 7.8%, respectively. ¹³ For the MTX-E12 cells, the largest fraction of nanoparticles was bound to mucus. ¹³ The mucoadhesion mechanism of chitosan is due to an ionic interaction between the positively charged amino groups of chitosan and the negatively charged sialic acid residues in mucus. ⁸ IOBA-NHC cells express MUC1, -2, and -4 mucin RNA. ¹⁹ Both Muc1 and -4 proteins are membrane-spanning mucins, whereas Muc2 protein is a gel-forming type, ²⁹ and each contains many sialic residues. Therefore, electrostatic interaction of the CSNPs with the negatively charged glycocalyx of the IOBA-NHC cells may be responsible for the high levels of uptake compared with others cell lines. 

CSNP uptake was significantly reduced at 4°C, a condition that blocks active transport processes. However, metabolic inhibition of 100 mM sodium azide did not significantly affect it. These results partially agree with CSNP transport studies in Caco-2 cells, ¹³ in which the uptake was significantly reduced at 4°C and in the presence of sodium azide. Our results suggest that CSNPs enter IOBA-NHC cells by an active temperature-dependent mechanism, but no metabolic energy seems to be required, at least at the concentration and times studied. Conjunctiva and other tissues ³⁰ have a sodium azide-sensitive, polarized drug efflux pump (the P-gp) that actively removes drugs from the cells. As IOBA-NHC cells are derived from human conjunctiva, this pump could have been present. If so, inhibition by sodium azide would have caused increased accumulation of CSNPs by the cells. Because sodium azide had no effect, it seems unlikely that the pump exists, or if it does exist, it has no role in CSNP uptake. Further studies are needed to clarify the mechanisms involved in CSNP uptake. 

Cell-associated fluorescence after the 24-hour recovery period was significantly lower than that immediately after CSNP exposure, and there was no difference between cells maintained at 37°C and 4°C. Biochemical alteration or degradation of the CSNPs could account for this decrease in fluorescence. For instance, lysosomal enzymes and low pH could reduce the pH-sensitive fluorescence. The fluorescence loss may also be due to the release of CSNPs and/or FITC-BSA to culture medium. In fact, culture medium fluorescence was less in cells maintained at 4°C than at 37°C, although differences were significant at only 15 and 120 minutes of CSNP exposure. This temperature dependency suggests some kind of active transport mechanism for nanoparticles and/or degradation bioproducts of CSNP metabolism release from cells to medium. However, cell-associated fluorescence levels were not higher in cells maintained at 4°C than they were in those maintained 37°C. Fluorescence was also less in supernatants of control, cell-free wells maintained at 4°C than those maintained at 37°C, although not significant at any time. Some kind of physical effect, such as static quenching of the FITC fluorescence emission may occur. Further studies are necessary to clarify this point. 

Because of the absence of blinking or drainage mechanisms in vitro, we performed experiments to check the in vivo uptake and tolerance of CSNPs in rabbits. Fluorescence microscopy of eyeball and lids sections confirmed the in vivo uptake by conjunctival and corneal epithelia. Occasionally, sections from the contralateral eye emitted a weak fluorescence signal that was much less intense than that from the treated eye. Because animals wore Elizabethan collars to avoid licking and scratching the eyes and face, communication between the eyes through the lacrimal drainage system could be an explanation for this fact. 

Our in vivo studies also show that the CSNPs are well tolerated by ocular surface structures. Although rabbit and human ocular surfaces are not equivalent, a good correlation between rabbit and human eye irritation data exists when low volumes are used. ³¹ The absence of histologic alterations and abnormal inflammatory cells in cornea, conjunctiva, and lids was consistent with the lack of clinical signs on exposure to the CSNPs. These results correlate well with the in vitro study, where CSNPs had negligible toxic effect. Our previous studies in rabbit ^{14 15} showed that chitosan-based nanoparticles are promising vehicles for ocular drug delivery, mainly due to their ability to contact intimately the corneal and conjunctival surfaces, as well as for the ability to incorporate bioactive compounds. ¹⁷ Based on the results of the in vivo acute tolerance, the new type of CSNPs seem to be very promising as well, as no clinical or pathologic differences between the CSNPs and the control-treated eyes were present. 

In conclusion, we have shown that improved CSNPs readily penetrate conjunctival epithelial cells and are well-tolerated at the ocular surface of rabbits. These carriers hold promise as a drug delivery system for the ocular mucosa. Further studies are warranted to determine whether CSNPs can deliver drugs for the efficacious treatment of ocular surface diseases. 

 Presented in part at the annual meeting of the Association for Research in Vision and Ophthalmology, Fort Lauderdale, Florida, May 2003, and the 4th Annual International Symposium on Ocular Pharmacology and Pharmaceutics, Seville, Spain, February 28 to March 3, 2002.

 Supported by Spanish Ministry of Education and Science Grants FEDER-CICYT MAT2003-09713-02-01 and MAT 2004-04792-CO2-01.

 Submitted for publication April 22, 2005; revised October 7 and December 16, 2005; accepted February 24, 2006.

 Disclosure: A. Enríquez de Salamanca, None; Y. Diebold, None; M. Calonge, None; C. García-Vazquez, None; S. Callejo: None; A. Vila, Advancell, SL (E); M.J. Alonso (P)

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

Grading System of Macroscopically Evaluated Signs in the In Vivo Tolerance Study for the Colloidal Systems Tested

Grade	Discomfort	Cornea	Conjunctiva	Discharge	Lids
0	No reaction	No alterations	No alterations	No discharge	No swelling
1	Blinking	Mild opacity	Mild hyperemia/mild edema	Mild discharge without moistened hair	Mild swelling
2	Enhanced blinking/intense tearing/vocalizations	Intense opacity	Intense hyperemia/intense edema/hemorrhage	Intense discharge with moistened hair	Obvious swelling

Table 2.

 

Effect of Temperature on CSNP Size after Incubation in Cell Culture Medium

Temperature	Time (h)	Size (nm)
—	0	289 ± 13
37°C	0.5	359 ± 32
	1	>1000 (aggregates)
	2	>3000 (aggregates)
Room temperature	2	520 ± 37
4°C	24	659 ± 35

Figure 1.

 

Immediate and 24-hour cell survival and viability after treatment with CSNPs. IOBA-NHC cells were exposed to 0.25, 0.5, or 1 mg/mL CSNPs for 15, 30, 60, or 120 minutes. Cell survival (A, B) and viability (C, D) of recovered cells were determined both immediately (A, C) and after a 24-hour recovery period (B, D) in culture medium. Cell survival was expressed as the percentage of cells recovered after treatment compared with recovery of cells maintained in culture medium alone (100%; data not shown). Immediate cell survival (A) after 30 minutes of CSNP exposure was significantly higher (P < 0.05) than that after 15 or 60 minutes, as shown by Scheffé post hoc statistical analysis. Cell survival of PBS- and CSNP-treated cells was significantly lower (P < 0.05) when compared with that of control cells treated with culture medium (data not shown). There were no significant differences in cell survival among the different concentrations, immediately (A) or after 24 hours (B). There were also no significant differences in the viability of recovered cells among CSNP concentrations or incubation times (C, D; n = 4). Data are expressed as the mean ± SEM. BAC, 0.005% benzalkonium chloride in culture medium; PBS, phosphate-buffered saline; FITC-BSA, fluorescein isothiocyanate bovine serum albumin solution in PBS.

Figure 2.

 

Confocal images of control IOBA-NHC cells. In the control cells, green fluorescence corresponds to CSNPs and not to the endogenous fluorescence or FITC-BSA released from the CSNPs. (A) Transmitted light micrograph of control cells exposed to culture medium. (B) No green fluorescence was detected in control cells exposed to the culture medium alone. (C) Cells exposed to FITC-BSA solution had an even distribution of fluorescence in the cytoplasm. (D) Cells exposed to CSNPs had nanometer-sized fluorescent dots inside them (n = 3). Magnification: (A–C) ×40; (D) ×63. Scale bar: (A–C) 50 μm; (D) 25 μm.

Figure 3.

 

Confocal serial images along the z-axis of IOBA-NHC cells. Cells were exposed for 30 minutes to 0.5 mg/mL CSNPs. This is a representative gallery of six serial micrographs showing the green fluorescence at 0.5-μm intervals along the z-axis, from the top to the bottom of the cell monolayer. Green images were merged with images of red PI-stained nuclei. Scale bar, 25 μm.

Figure 4.

 

Time-dependent CSNP uptake by IOBA-NHC cells. This is a representative series of cells exposed to 0.5 mg/mL CSNPs for 15, 30, and 120 minutes. For each exposure time, PI-stained and FITC-CSNP-loaded cells are shown separately, as well as with the merged images. CSNP presence was detectable at 15 minutes after exposure. Some aggregates were present in the longest incubation time (arrowhead, n = 3). Scale bar, 25 μm.

Figure 5.

 

Amount of CSNP uptake. CSNP uptake was quantified by measuring the fluorescent content in cell lysates, both immediately after exposure and after the 24-hour recovery period in culture medium. (A) Cell-associated fluorescence after CSNP exposure at 37°C or 4°C, expressed as a percentage of CSNP available for uptake, was significantly reduced at 4°C at 30, 60, and 120 minutes incubation time. (B) Micrograms of CSNPs in cells maintained for 24 hours after exposure at either 37° or 4°C was significantly lower than that measured immediately after CSNP exposure. (C) Micrograms of CSNPs in control, cell-free wells were much lower than cell-associated levels (B). Fluorescence in the control wells after 24 hours, at 37°C or 4°C, were lower than that measured immediately after CSNP exposure (n = 3). *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 6.

 

CSNPs in the culture medium after the 24-hour recovery period. (A) Culture medium-associated fluorescence, expressed as micrograms of CSNPs, was determined after 24 hours of culture at 4°C or 37°C. The medium CSNP content was generally lower for cells maintained at 4°C than at 37°C. The differences were significant for 15 and 120 minutes of exposure. (B) The CSNP content of the culture medium from control cell-free wells was much lower than that of wells with cells (in A). No significant differences were detected between values at 4°C and 37°C (n = 4). *P < 0.05, **P < 0.01.

Figure 7.

 

The effect of metabolic inhibition with 100 mM sodium azide on CSNP uptake. CSNP uptake was quantified as the fluorescence of cell lysates immediately after CSNP exposure at 37°C in the absence or the presence of sodium azide. The presence of sodium azide did not significantly inhibit CSNP uptake at any time point studied (n = 3).

Figure 8.

 

Ocular surface structures of CSNP-treated (OD) and control (OS) rabbit eyes. Rabbits were exposed to CSNPs for 24 hours. Representative conjunctival impression cytology (A, B) and conjunctival (C, D) and corneal (E, F) sections are shown. Conjunctival and corneal epithelia from both control and treated eyes displayed normal cell layers and morphology. No signs of tissue edema were observed in any structure studied after exposure to CSNPs compared with controls. Scale bar, 100 μm.

Figure 9.

 

CSNP in vivo uptake. Fluorescence microscopy of ocular surface structures of sham-treated (A, D), CSNP-treated (B, E), and contralateral control (C, F) rabbit eyes. Representative corneal (A–C) and conjunctival (D–F) sections are shown. No fluorescence was detected in sham control corneas (A) or conjunctivas (D). (B) Corneal epithelial cells of CSNP-treated rabbits were uniformly fluorescent. (B, inset): enlargement showing a detail of corneal epithelial fluorescence pattern. (E) Fluorescence in conjunctival epithelial cells was intense in apical cell membranes and positive along the basolateral cell membrane. (E, inset) Enlargement showing the basolateral membrane fluorescence staining in goblet and non–goblet cells. (C, F) Some fluorescence was detected in corneal and conjunctival epithelial cells from contralateral control eye (OS), although much less intense than in the treated (OD) eye. Scale bar (A–F) 50 μm; insets: 10 μm).

The authors thank to Alberto Sánz Cantalapiedra, MD, for fluorometry measurement assistance; Emiliano Becerra, MD, for eye and lid surgical extraction; Miguel Jarrín, MSc, for animal handling; Victoria Sáez for excellent technical assistance, and Agustin Mayo-Iscar, PhD, for statistical analysis.