Whole-Eye Perfusion Model for Screening of the Ocular Formulations via Confocal Laser Scanning Microscopy

Abstract

Various physiological, anatomical barriers make ocular drug delivery very challenging. Hence, better in vitro screening models are needed for rapid screening of the formulations. In this study, a simple whole-eye perfusion model was designed and its application was explored for screening targeted formulation across the full-thickness cornea using confocal laser scanning microscopy. PEG-cholecalciferol-based integrin targeted coumarin-6 micelles (TC6M) and non-targeted coumarin-6 micelles (NTC6M) were developed by solvent diffusion evaporation technique. The formulations NTC6M and TC6M had particles size 23.5 ± 5 nm and 28.5 ± 6 nm respectively and osmolality of 294–300 mOsml/Kg. The whole-eye perfusion model was developed using porcine eye. TC6M and NTC6M were instilled on the excised porcine eyes as well as in the eyes of NZW rabbits. Corneas were excised from the experimental eyes; coumarin-6 penetration across the corneas was analyzed using confocal microscope. Coumarin-6-loaded micelles had particle size below 50 nm. NTC6M formulations showed penetration to the deeper layers up to 500 μm porcine eyes and up to 50 μm in rabbit corneas. However, TC6M formulations exhibited superior retention, as higher fluorescent intensities were observed in upper layers up to 50 μm depth in the porcine eye and 20 μm depth in rabbit eye. Hence, applicability of whole-eye perfusion model in preliminary screening of the formulations was successfully demonstrated. Whole-eye perfusion model when combined with confocal microscopy has potential to be used as an efficient tool for rapid screening and optimization of various ophthalmic formulations.

INTRODUCTION

Human eye has unique anatomy and compartment organization which makes it a multiplex organ. It has physiological barriers such as tear flow, reflex blinking, and nasolachrymal drainage. Besides physiological barriers, it is comprised of two types of anatomical barriers (a) static and (b) dynamic. While static barrier involves the outermost layer epithelium, stroma, and tight junctions of blood aqueous barrier, dynamic barrier comprises of lymph flow, conjunctiva blood, and tear drainage. These physiological and anatomical protective barriers are difficult to circumvent, making ocular drug delivery more challenging (1).

Topical drug application is the most frequently used route for treatment of various anterior eye diseases. However, only 5% of the topical formulation is bioavailable owing to various barriers in the cornea (2). Also, it is quite challenging to monitor the amount of formulation actually penetrating into the tissues from the site of application. Even though various techniques have been implied in monitoring the disease states like, e.g., transmission electron microscope (TEM), scanning electron microscope (SEM), multiphoton laser scanning microscopy, and confocal scanning, very few tools are available to analyze the penetration of the formulation across the site of application. Confocal microscopy has been used as an adjunct tool in therapeutic monitoring and diagnosis of dry eye disease (DED), Sjogren’s syndrome, and corneal nerve diseases (3–5). Various in vitro and ex vivo studies have been performed for trans-corneal penetration and absorption using Franz diffusion cell, Ussing chamber, and perfusion models. Majumdar et al. and Ahuja et al. studied trans-corneal permeation of ketorolac tromethamine eye drops and diclofenac from oil drops respectively across the excised goat corneas using Franz diffusion chamber (6,7). KM Hamalainen et al. used an Ussing chamber to quantitatively characterize paracellular routes in rabbit ocular tissue such as cornea, conjunctiva, and sclera (8). Olsen et al. successfully demonstrated permeability of sclera to various hydro-philic molecules, i.e., dexamethasone, methotrexate, and insulin using an Ussing apparatus (9). Horizontal diffusion assemblies/Ussing chambers are better tools while studying flux, diffusion rate, and concentration of ions across the tissues as molecules are free to travel in either direction. Perfusion models on the other hand involve improvement of diffusion chambers to accommodate for perfusion lines mimicking circulation of aqueous humors and precision pumps for tear fluids (10–12). Pescina et al. worked on an ex vivo model using porcine eye and evaluated permeation of various drugs with different physicochemical properties. They also highlighted the role of charge on the molecule in the permeability (13). Besides, whole-eye models have also been reported for trans-corneal penetration studies. Whole-eye models have been considered better as they provide closer resemblance to in vivo conditions and eye anatomy along with accountability to trans-scleral penetration. Various whole-eye models have been reported so far, e.g., Morrison et al. used cryopreserved whole bovine eye and demonstrated the mechanism of increased permeability of riboflavin when administered along with cyclodextrin and in presence of calcium sequestering agents such as ethylenediaminetetracetic acid (EDTA) and ethylene glycol-bis (β-aminoethyl ether)-N,N,N′,N-tetraacetic acid (EGTA) (14,15). Kompella et al. reported use of whole bovine eye for studying nanoparticles disposition across cornea and conjunctiva (16). While whole-eye models have been implied for penetration studies, whole-eye perfusion models have not been discussed so far. To the best of our understanding, this is the first report where a simple, cost-effective whole-eye perfusion model has been established, used in conjugation with confocal microscopy, and explored for rapid screening of ophthalmic formulations based on nanoparticles uptake in various layers of full-thickness cornea. In the current study, we developed whole-eye perfusion model using porcine eye, evaluated corneal penetration of coumarin-6, and compared the ex vivo data with in vivo studies done in rabbits. To conclude, we demonstrated the application of whole-eye perfusion model in the screening of ophthalmic formulations.

MATERIALS AND METHODS

Materials

Poly(ethyleneglycol) methyl ether (average Mwt~2000) (mPEG2000), methylenechloride, N,N-dimethylpyridin-4-amine (DMAP), cholecalciferol, calcium chloride (CaCl₂), acetone, N,N-diethylethanamine (TEA), glutaric anhydride, dichloromethane (DCM), potassium chloride (KCl), N,N′-dicyclohexylcarbodiimide (DCC), methanol, ethyl ester (ethyl acetate), sucrose, sodium bicarbonate (NaHCO₃), sodium chloride (NaCl), and petroleum ether were obtained from Sigma-Aldrich, (St. Louis, MO). 1, 2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[maleimide (polyethylene glycol)-2000 conjugated to Arg(R)-Gly(G)-Asp(D) (RGD) tripeptide motif (DSPE-PEG-RGD), 2 K was purchased from Biochempeg (Watertown, MA). 1, 2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[methoxy(polyethylene glycol)-2000] (mPEG-2000-DSPE) was procured from Lipoid (Germany). 3-(2-Benzothiazolyl)-N, N-diethylumbelliferylamine, 3-(2-Benzothiazolyl)-7-(diethylamino) coumarin (coumarin-6) was obtained from Santa Cruz Biotechnology.

Synthesis of PEG-Cholecalciferol Conjugate

We synthesized a conjugate of mPEG2000 and cholecalciferol by using 2 steps; procedure and details of the reactions have been reported previously (17). In short, the first step of synthesis involved reaction of cholecalciferol and glutaric anhydride (1:5 mol ratio) in the presence of TEA. Reaction involved refluxing for 3 days in dark, followed by removal of solvent (DCM) and purification of residue to obtain cholecalciferol glutarate. Confirmation of the product was ascertained by using IR and ¹HNMR spectroscopy. Ring opening of glutaric anhydride and esterification with cholecalciferol was confirmed by appearance of stretching bands at 1735 cm⁻¹, corresponding to the carbonyl (C = O) group of ester and at 1710 cm⁻¹ due to free carboxylic acid at end. Similarly, in ¹HNMR, esterification of H3 was confirmed by downfield shift from δ 3.9 ppm to δ 5.0 ppm and appearance of steroid skeleton shift within range of δ 0.5 ppm to δ 2.9 ppm. The next step involved the reaction of cholecalciferol glutarate with mPEG2000 (1:1 mol ratio) using DCC (coupling reagent) and DMAP (catalyst). The final reaction product PEG-cholecalciferol conjugate (PEGCCF) was purified and confirmed by FTIR (Thermo Scientific Nicolet iS5 FT-IR instrument; Waltham, MA, USA) and NMR spectroscopy (Varian Mercury 300 MHz) (17).

Preparation of Integrin Targeted Coumarin-6 Micelles and Non-targeted Coumarin-6 Micelles

Coumarin-6-loaded nanoparticles were prepared using previously reported solvent diffusion evaporation technique with slight modifications (18). Briefly, coumarin-6 solution in acetone (0.1% w/v) was added dropwise to aqueous sucrose solution (10% w/v) which also contained PEGCCF (0.83% w/v), mPEG-2000-DSPE (0.4% w/v), and DSPE-PEG-RGD, 2 K (0.04% w/v). The aqueous to organic phase ratio was 3: 1. The dispersion was stirred at 600 rpm at room temperature for 12–14 h facilitating complete removal of acetone along with self-assembly of targeted coumarin-6 micelles (TC6M). The TC6M thus obtained were filtered through corning a 0.22-μm PES syringe filter (VWR). Non-targeted coumarin-6 micelles (NTC6M) were prepared similarly except using mPEG-2000-DSPE instead of DSPE-PEG-RGD, 2 K. Coumarin-6 was added to the formulations to aid the observations via confocal microscopy and demonstrate the application of whole-eye perfusion model.

Characterization of TC6M and NTC6M.

Coumarin-6 concentration in both TC6M and NTC6M was determined by fluorescence spectrometry (Tecan plate reader) using λ_exc 450 nm and λ_emi 505 nm. Coumarin-6 standards were prepared in DMSO. The poly-dispersity index (PDI), particle size, and zeta potential of the nanoparticles were obtained by Nicomp 380 ZLS Particle sizer (Port Richley, FL). The encapsulation efficiency of TC6M and NTC6M was determined as a ratio of actual amount of dye encapsulated in micelles divided by total amount of dye initially used for encapsulation.

Whole-Eye Perfusion Model

Fresh porcine eyes were obtained from Jhonston’s meat market Monticello, FL (slaughterhouse). Perfusion experiments were performed within 1–2 h of the animal sacrifice. Porcine eyes (n = 3) with their anterior portion exposed to the surface were placed on propylene hollow circular stands supported by cotton at the base and placed in glass petriplates. Petriplates were placed in water bath maintained at 37°C. Petriplates were filled with keratinocyte serum free medium (KSFM with pituitary extract and EGF). The perfusion was performed by cannulating the anterior portion of the eyes (especially anterior chamber) with 27G needle, flushing, and inflating with KSFM, followed by constant medium flow rate of 1–2 μL/min (maintained using peristaltic pumps) during the study (Fig. 2a). Coumarin-6-loaded TC6M and NTC6M drops (50 μl) were instilled on the corneas with the help of single-channel pipette (Eppendorf, NY, USA). To maintain moisture, 50 μl of simulated tear fluid (STF) containing KCl 0.14% w/v, NaCl 0.68% w/v, CaCl2 0.008% w/v, and NaHCO3 0.22% w/v maintained at 37°C was administered on the corneas every 15 min. Study was continued for 2 h. After 2 h, perfusion was stopped; eyeballs were washed 3 times with DPBS, and corneas were excised. Confocal laser scanning microscopy (Nikon Eclipse TE2000-U) with z-stacking (10 μm intervals) was performed on the corneas and images were analyzed using NIS-Elements viewer (4.20).

Fig. 2.

Whole-eye perfusion model using porcine eyes. a Schematic showing perfusion in porcine eye. The anterior portion of the eyes was cannulated using a 27G needle. KSF media was perfused through corneas at the rate of 1–2 μl/min. STF instilled to mimic tear flow. b Complete setup of whole-eye perfusion model maintained at 37°C. Perfusion was facilitated with peristaltic pumps

In vivo Perfusion Study

Female New Zealand White (NZW) rabbits (age 70 days) (n = 3) were purchased from Charles River Lab., (Wilmington, MA). TC6M drops 50 μl were instilled in one eye (right eye) and NTC6M were instilled in the other eye (left eye) of the rabbits. After 12 h, rabbits were sacrificed and eyes were enucleated. Corneas were isolated, washed with DPBS, and observed under confocal laser scanning microscope and images were captured.

STATISTICAL ANALYSIS

The experiments were performed in triplicate and data was presented as mean ± standard deviation. Statistical analysis was carried out using ANOVA (Bonferroni’s multiple comparison tests) using GraphPad Prism 6, (GraphPad, LaJolla, CA). A value of p < 0.05 was considered statistically significant difference.

RESULTS AND DISCUSSION

Characterization of PEGCCF

Cholecalciferol has been reported to exhibit immunomodulatory effect leading to reduction in ocular inflammation (19). However, being hydrophobic in nature, it needs a carrier system for delivery inside the ocular cells. Hence, it was conjugated with mPEG-2000 to obtain PEGCCF which can then perform dual functions; one is self-assembly into nanomicellar delivery system and second was to act as carrier for any desirable therapeutic agent.

However, in the current formulation, instead of drug, only coumarin-6 was loaded to facilitate observation via confocal microscopy and prove the applicability of whole-eye perfusion model. The final reaction product PEGCCF was purified and confirmed by presence of C-O-C stretch (ether-dialkyl) at frequency 1104 cm⁻¹ in FTIR and the appearance of protons due to PEG group at δ 3.5 ppm to δ 3.6 ppm in NMR spectroscopy (17). The residual content of DCM in the end product was below 100 ppm, hence did not exceed the permissible level of 600 ppm (as per ICH guidelines for class 2 solvents).

Characterization of TC6M and NTC6M

The conventional solvent diffusion evaporation technique was used to prepare micelles. The particle size of NTC6M and TC6M was 23.5 ± 5 nm and 28.5 ± 6 nm (Fig. 1a) with zeta potential of 2.36 ± 0.30 mV and − 6.77 ± 2.41 mV and PDI 0.38 ± 0.06 and 0.44 ± 0.03 respectively. Both the formulations had low particle size ensuring better permeability. Even though both the micellar formulations had low zeta potential values (absolute value), they were sterically stabilized by presence of PEG moieties on the surface. It has been reported that PEG moieties prevent aggregation, create stearic barrier, and thus provide stability (20). Further, low PDI values of the micelles reflect homogeneity of the dispersion along with high stability. The encapsulation efficiency and drug loading of NTC6M and TC6M was 5.59 ± 0.81% w/w; 55.9 ± 8.1 μg/ml and 6.12 ± 0.21%w/w; and 61.2 ± 2.1 μg/ml. The reason for low encapsulation efficiency might be high solubility of coumarin-6 in organic solvent used, low concentration of surfactant, or method of preparation. The formulations were transparent and bright yellow in color (Fig. 1b). Yellow color of the formulations was due to hydrophobic fluorescent dye coumarin-6, which absorbs and emits light in regions of visible spectrum.

Fig. 1.

Characterization of coumarin-6 loaded micelles. a) Particle size distribution of NTC6M and TC6M. b) Appearance of NTC6M and TC6M. The formulations were transparent and bright yellow in color

Whole-Eye Perfusion Model

Perfusion models are needed for development, testing, and screening of ophthalmic formulations. Generally, penetration studies are performed by ex vivo or in vitro preparation of the corneal tissues and placing the tissues either in Franz diffusion, Ussing chamber, and perfusion/diffusion chambers. Various modifications have been reported in the Franz diffusion such as glass well (21), fluid circulation in donor compartment (22), and Ussing chamber, e.g., controlled STF by reduced volume of insert attached to infusion pump (23). Thiel et al. designed polycarbonate and steel perfusion model for drug penetration and toxicity studies. Although the perfusion devise was capable of assessing epithelium and endothelium throughout the experiments, but it was limited by its complex chamber design. Dutescu et al. demonstrated penetration of cyclosporine across rabbit corneas placed on artificial chambers receiving constant supply of aqueous humor (11). Although diffusion models have successfully demonstrated consistent measurement of drug transport across the cornea, however, the dimension and volume of anterior chambers could not simulate real-time conditions (24). Hence, whole-eye models are needed to mimic eye anatomy and in vivo conditions.

Whole-eye models have been investigated for studying penetration and absorption of proteins and peptides (16), colloidal systems (25), and silica nanoparticles (14) across the cornea. The whole-eye models reported so far lack arrangements for any perfusion lines. Therefore, in the current study, we designed a simple whole-eye perfusion model (Fig. 2b), where fresh whole porcine eyes were perfused with KSF medium in the anterior segment. The perfusion with flow rate of 1–2 μl simulated the aqueous humor. Aqueous humor is a clear fluid continuously released at rate of 1.5–3.0 μl/min by ciliary bodies; circulates in the anterior segment to provide nutrition to avascular tissues, e.g., cornea, lens, and trabecular meshwork; and exits from trabecular meshwork (26). We chose porcine eye as they are easily available, robust, and large. Moreover, their size, weight, and volume of aqueous and vitreous humor resemble human eye and have also been reported as a good model for human eye in the literature (24,27). STF drops (50 μl) were instilled on the corneas every 15 min to mimic tear flow. The basal tear flow and reflex tear flow in normal human eye has been reported as 1.18 ± 0.36 μl/min and 5.71 ± 5.86 μl/min respectively (28). The perfusion studies were continued for only 2 h and corneas were excised from the globe thereafter. It has been reported that epithelial layers remain intact for about 2 h and start to loosen at after 3 h in isolated corneas. Also, due to disorganization of superficial layers of epithelium and hydration of stromal layers, the corneal tissues become opaque after some time (13,29). Using confocal microscopy, micelles uptake in various layers of full-thickness cornea was determined in terms of fluorescence intensity at various depths of whole cornea. Deep penetration of up to 500 μm was observed in porcine cornea treated with NTC6M (Fig. 3a) validating that smaller particle size made it easier for the formulation to penetrate. Further, higher fluorescence intensity was observed in porcine corneas treated with TC6M (Fig. 3b) as compared to NTC6M suggesting that integrin targeting improved the retention. Ken-ichi Ondo et al. demonstrated the presence integrins β1 in porcine epithelium as well as isolated porcine epithelial cells by immunofluorescence (30). Yu-Hua Weng et al. reported overexpression of integrins β1 in human corneal epithelial cells (HCEC) and rabbit corneal epithelial cells (RCEC) and also illustrated the binding of integrin with RGD present in RGD functionalized targeted nanomicelles (31). Integrins are integral membrane associated glycoproteins which help in attachment to glycoproteins such as laminins, collagens, and fibronection and vitronectins (32). β1, β2, and αv are three major groups of integrins. The alpha chain αv can bind and form heterodimer with various β chains including β1, β3, β5, β6, and β8. Usually, extracellular domain of integrins such as αv, α5β1, and αIIbβ3 can recognize ligands/glycoproteins, fibronectin, laminin, collagens etc., present in extracellular matrix via sequence of three amino acids, i.e., arginine-glycine-aspartic acid (RGD) (33,34). Irene Sanchez et al. has reported that overall thickness of porcine cornea by ultrasound pachymetry was 1013 μm (in ex vivo) and 666 μm in live animal. Thickness of epithelium is about 80 ± 25 μm and stroma is about 900 μm (35). The higher fluorescent intensity was observed in TC6M only in upper layers of cornea up to 50 μm (p < 0.01) (Fig. 4a) due to presence of majority of integrins in the epithelium of the porcine cornea. A sharp decrease in fluorescent intensity after 50 μm can be explained by very low concentration or absence of integrins in deeper layers of porcine cornea (Fig. 4b). However, presence of fluorescence in the deeper layers of NTC6M treated corneas was due to absence of RGD, thus lack of integrin targeting in those formulations (Fig. 4b). Similar results were observed in in vivo perfusion studies performed in the NZW rabbits (Fig. 6b). Various integrins have been reported in the epithelium of rabbit cornea, e.g., α6β4 located in hemidesmosomes of adhesion complexes in basal epithelium, α6 and β1 in basal membrane, and basal epithelial cells (36). In rabbits, average corneal thickness has been reported to be between 300 μm and 400 μm with epithelial thickness of about 47 ± 3 μm (37). As per confocal microscopy data, some fluorescence was observed up to depth of 300 μm in NTC6M-treated eyes (Fig. 5a). However, in TC6M-treated eyes, the highest intensity of fluorescence was observed at upper layers with very low fluorescent intensity in deeper layers (Fig. 5b). These results are supported by studies done by Yu-Hua Weng et al. revealing better uptake of RGD functionalized targeted nanomicelles of flurbiprofen as compared to flurbiprofen micelles in HCEC and RCEC (31). Presence of integrins in upper layers could be a reason for better binding only in upper layer which led to the highest fluorescent intensity at about 20 μm depth in TC6M-treated corneas (Fig. 6a, b) (p < 0.05). As compared to porcine cornea, intensity of fluorescence was lower in the layers of rabbit cornea which might be due to continuous tear flow of about 7.58 ± 2.3 μl/min (38) along with reflex blinking of 10–12 times per minute (39) and nasolachrymal drainage. Even though there were differences in the fluorescent intensities in two models, it can be clearly concluded that targeted formulation provided better retention in upper layers as compared to non-targeted formulation. Thus, it can be said that the data obtained from the whole-eye perfusion model very much reflected what was being observed during in vivo conditions. Hence, whole-eye perfusion model can be used for initial screening of the formulations depending upon the desired depth of penetration thereby reducing the number of formulation to be tested on animals, which in turn may reduce total number of animals being used in studies. Nevertheless, there were several limitations in the current study, i.e., quantification at each depth was relevant to upper layers (which was chosen based on showing maximum fluorescent intensity), applicability of reported data only to given experimental condition which may or may not extricated to other nano-formulations or dyes, able to determine drug penetration through trans-corneal route only and not through trans-scleral route (sclera being opaque cannot be used in the study), and variation in eye anatomy and physiology from species to species. The whole-eye perfusion model may have scope in evaluation of non-fluorophore drugs provided they are conjugated or tagged with fluorophore moieties. Further studies are warranted to confirm the applicability of model to determine drug permeation in aqueous humor, lens, and vitreous body. However, use of whole-eye perfusion model and monitoring fluorescence at various depths via confocal microscopy may be envisaged as an efficient tool for preliminary screening and optimization of the formulations.

Fig. 3.

Confocal microscopy 3D images of porcine corneas (636.5 μm × 636.5 μm × 500 μm) after ex vivo study. a) Penetration of NTC6M observed upto deeper layers. b) Penetration of TC6M depicting higher retention in the upper layers of epithelium

Fig. 4.

Distribution of coumarin-6 loaded TC6M and NTC6M in porcine corneas after performing ex vivo perfusion study for 2h. a) Uptake of NTC6M and TC6M in different corneal layers. Highest intensity of coumarin-6 was observed in the top most layers (0 μm–10 μm) of corneas. Decreased coumarin-6 intensity was observed at 50 μm depth. b) Graph of distribution of NTC6M and TC6M in porcine corneas. A higher intensity of coumarin-6 was observed in upper layers in TC6M treated corneas as compared to NTC6M. Coumarin-6 intensity gradually decreased in NTC6M treated corneas. Data shown as mean ± SD, *p < 0.05, **p < 0.01

Fig. 6.

Distribution of coumarin-6 loaded TC6M and NTC6M in rabbit corneas after performing in vivo study for 12 hours. a) Uptake of NTC6M and TC6M in different corneal layers. Highest intensity of coumarin-6 was observed in the top most layers (0 μm–10 μm) of corneas. Decreased coumarin-6 intensity was observed at 50 μm depth. b) Graph of distribution of NTC6M and TC6M in rabbit corneas. A higher intensity of coumarin-6 was observed in upper layers in TC6M treated corneas as compared to NTC6M. Coumarin-6 intensity gradually decreased in NTC6M treated corneas. Data shown as mean ± SD, *p < 0.05

Fig. 5.

Confocal microscopy 3D images of rabbit corneas (636.5 μm × 636.5 μm × 350 μm) after in vivo study. a) Penetration of NTC6M observed upto deeper layers. b) Penetration of TC6M depicting higher retention in the upper layers of epithelium

CONCLUSION

Owing to small surface area, anatomical, and physiological barriers, bioavailability of ocular topical formulation is very low. Hence, better in vitro screening models are warranted. The whole-eye perfusion model designed in the current study is very simple, cost effective, and may prove valuable in providing rapid screening of the several ophthalmic formulations and analyzing their penetration ability across various ocular tissues.