Molecular Vision 2004; 10:555-565 <http://www.molvis.org/molvis/v10/a68/> Received 13 November 2003 | Accepted 2 August 2004 | Published 18 August 2004

Download Reprint

Delivery of antiangiogenic and antioxidant drugs of ophthalmic interest through a nanoporous inorganic filter

Kathleen E. Orosz,¹ Somil Gupta,^1,2 Marla Hassink,¹ Mohamed Abdel-Rahman,¹ Leni Moldovan,³ Frederick H. Davidorf,¹ Nicanor I. Moldovan^1,2,3
 
(The first two authors contributed equally to this publication)
 
 

¹Division of Ophthalmology, ²Biomedical Engineering Center, and ³Davis Heart & Lung Research Institute, Ohio State University, Columbus, OH

Correspondence to: Nicanor I. Moldovan, Davis Heart and Lung Research Institute, Departments of Internal Medicine and Ophthalmology, The Ohio State University Medical Center, Room 305A, 473 West 12th Avenue, Columbus, OH, 43210; Phone: (614) 247-7801; FAX: (614) 293-5614; email: moldovan-1@medctr.osu.edu

Abstract

Purpose: We propose a novel method of administration of antiangiogenic and antioxidant drugs, with potential clinical application in the treatment of proliferative diabetic retinopathy (PDR) and age-related macular degeneration (AMD). We suggest the encapsulation of drugs in implantable sustained release devices, limited by membranes with pores in the tens of nanometers diameter range, which display a slower, quasi-linear release kinetics, and a better selectivity than other membranes. In this paper we explored the feasibility of this approach by testing in vitro several key elements of the nanofilter system: diffusion of drugs of interest, efficacy in producing desirable effects on cells, and biocompatibility of used material with some of the cells encountered in the ocular cavity.

Methods: We used an aluminum oxide filter (Anopore^TM) with pores of 20 nm as a limiting medium for the administration of drugs. First, we induced an oxidative stress in human retinal endothelial cells (HREC) by treating them with hydrogen peroxide diffused across the filter, in the absence or in the presence of catalase. HREC attached to the culture plate, or emerging as angiogenic sprouts from aggregates embedded in collagen gels, were also exposed to vitamin C or to endostatin delivered across the nanoporous filter. Direct exposure of the cells to the agents served as positive controls. Growth of cells on the filter was considered an indication for biocompatibility.

Results: Catalase diffused across the nanoporous membrane counteracted the cytotoxic effect of hydrogen peroxide on HREC. We also found that vitamin C, acting directly or after diffusion across the filter, up to concentrations physiologically present in the eye, was a concentration dependent modulator of HREC's ability to survive and sprout. Additionally, we confirmed the ability of endostatin to block the growth of HREC either attached or sprouting from cell aggregates, after diffusion across the Anopore^TM nanofilter.

Conclusions: The drug delivery method based on the administration of angiostatic and antioxidant agents across the inorganic aluminum oxide nanoporous filter passed the key in vitro tests for diffusibility and biocompatibility, opening the way for medical applications.

Introduction

Sustained release of ophthalmic drugs is a rapidly developing field, suitable for treatment of conditions where systemic therapy may be accompanied by serious side effects, and where the repeated intravitreal drug administration carries significant risks. Drug delivery to the choroid, retina, and vitreous is difficult because of the obstacle represented by the blood-retina barrier and the extraocular epithelia. Only a fraction of the drug administered orally or by subcutaneous or intramuscular routes reaches the retina, requiring large and often toxic doses to be therapeutically effective [1]. This limitation is aggravated by the fact that drug levels should be sustained for extended periods of time at the target site. Moreover, because many drugs undergo significant hepatic degradation, their use requires frequent administration and/or high doses. Therefore, one possible approach to improve retinal drug delivery is to facilitate localized delivery to the posterior segment of the eye by using novel drug delivery methods, such as sustained release methods. Applications of this technology to ophthalmology have also received significant attention, notably in the form of intravitreal gancyclovir device [2] or of encapsulated cell derived cilliary neurotrophic factor [3].

Nanotechnology offers the possibility of control and manipulation of matter at molecular level. Its applications in the field of ophthalmology already surfaced [4-6]. In this context, sieving of molecules through pores with diameters comparable with the size of molecules is of considerable practical interest. This phenomenon found applications in various domains, such as immunoisolation [7,8].

Nanofiltration would be particularly promising for the development of implantable or trans-dermal drug delivery systems with controlled rate of release [9]. However, the basic science governing the mechanism of diffusion through nanopores is far from being understood. Because there is no general theory available to explain this type of diffusion [10], and because the transport of each molecular species across a given porous filter has unique features, an empirical investigation of the process is required. The thermodynamic, continuous type of models do not apply to the diffusion through nanopores, therefore either numerical simulations [11,12] or other approaches (as used here) are required. Moreover, when it comes to transport of molecules of therapeutic importance, additional analyses are needed to study the biological effects of the drug in the particular conditions of administration and on given cell types, as well as the biocompatibility of the drug delivery system.

These issues were addressed in the current paper with reference to compounds potentially useful for the treatment of angiogenesis dependent, proliferative diseases such PDR and AMD. We consider that this novel drug delivery approach may target both the initiation steps, as well as the vasoproliferative components of these diseases. Nanoporous alumina capsules were recently presented in literature regarding their technology of fabrication and diffusion properties [13]. However, despite the fact that the field is ripe for biological extensions, no such applications were described yet. We suggest that such capsules can be implanted into the eye, similar to the procedure for in situ treatment of cytomegalovirus retinitis [14].

As a proof of concept for the novel drug delivery system based on diffusion through pores with diameters in the nanometer range, we chose to test the effect on endothelial cells in vitro. We used three types of molecules with various mechanisms of action and acting in different phases of the proliferative disease development: (a) catalase, the enzyme limiting the concentrations of hydrogen peroxide (H₂O₂), a compound known to be present during the initiating phase of many ocular diseases resulting from free radical insults [15]; (b) ascorbic acid, a substance with complex action [16,17], which interferes both with the formation of free radicals and with the angiogenic process per se [18]; (c) endostatin, a potent anti-angiogenic molecule with potential applications in the treatment of angiogenesis dependent diseases [19,20]. As a model of reactive oxygen species, we chose H₂O₂, normally found in the aqueous humor [21] and suspected to be a key player in the propagation of free radical induced damage, particularly in PDR [22].

Since oxidative stress is increased in many ocular diseases [23], antioxidant therapy with vitamin C has been suggested [24,25]. Orally administered vitamin C and vitamin E have been shown to decrease nitric oxide levels, one of the substrates for forming the peroxynitrite (along with superoxide), and an oxidant that causes tissue injury [26]. However, systemic administration of ascorbic acid for managing eye conditions may not be feasible because of the very high local concentrations of ascorbate in the eye. This was shown to be one of the main components of the aqueous humor (1.16 mM), with a molarity that is 25 times higher than in the plasma [27]. Diabetes reduces the available vitamin C pool in the vitreous [28]. The current understanding of the pathogenesis of PDR is that hyperglycemia induces a pro-oxidant state in the cells of the retinal microvasculature, altering their redox metabolism [29,30]. However, besides its function in regulating the redox system, vitamin C may be a potential modulator of angiogenesis, either through its impact on collagen synthesis [31] or, when present in high concentrations as in the vitreous, through a direct cytotoxic effect [18].

Angiogenesis in PDR is stimulated not only by increased ocular levels of VEGF [32], but also by the decrease in concentrations of the endogenous inhibitors, such as vitamin C, TGF-beta [33] or endostatin [34]. A similar scenario may apply during the development of AMD. Therefore, administration of endostatin in this setting is expected to have significant anti-angiogenic effects.

Our data demonstrate the feasibility of the nanofilter based approach to the administration of drugs by passive diffusion across an aluminum oxide nanofilter, in amounts sufficient to show a meaningful biological effect. As targets of these compounds we chose the human retinal endothelial cells (HREC), either directly attached to the culture dishes, or in a model of angiogenesis. This is the first step in the process of designing of a drug delivery capsule with pores within nanometer size range.

Methods

Cells and materials

HREC were obtained as a cell line from Applied Cell Biology Research Institute through Cell Systems (Kirkland, WA). HREC were cultured in CSC Complete Medium also from Cell Systems. For experimentation, the cells were grown in a basal medium with or without other additives. The basal medium consisted of Medium 199 (Gibco, Grand Island, NY) supplemented with 10% fetal bovine serum (FBS, Gemini Bioproducts, Calabasas, CA) and 1% penicillin/streptomycin/amphotericin (PSA, Gibco). The additives were specified below for each experimental condition.

Porous aluminum oxide filters (Anopore^TM) with pores of 20 nm diameter were obtained from Nalge Nunc International (Rochester, NY) as ready to use tissue culture inserts. These inserts were of two sizes: one to fit in 6 well plates (used for the study of endostatin diffusion), and ones fitting 24 well plates (for the assessment of the diffused catalase and of Vitamin C).

The in vitro angiogenesis model

HREC aggregates were formed by a hanging droplet method [35]. Each droplet consisted of 1x10⁴ cells in a 40 μl suspension of CSC Complete Medium. Five droplets were suspended from a parafilm lined lid of a 60 mm x 15 mm culture dish and allowed to aggregate for three days at 37 °C, 5% CO₂. The dish also contained 4 ml of Dulbecco's Modified Eagles Medium (DMEM, Gibco) for humidity. After three days, the medium was removed from the aggregate, a plastic ring (plastic coverslip with hole punched from Nalge Nunc International) was placed around it, and it was embedded in 60 μl of 2.5 mg/ml type I collagen (Becton Dickinson, Franklin Lakes, NJ) as described in Vernon and Sage [35] (Figure 1A). The collagen discs either were placed alone in wells containing growth medium, or below Anopore^TM filter inserts (Nunc 25 mm Tissue Culture Inserts, Nalge Nunc International, Naperville, IL), from which the tested drugs could diffuse to the aggregates (Figure 1B). Alternatively, HREC were occasionally grown either directly on the bottom of the wells, or on the filter inside of the Anopore^TM insert and exposed to the drugs diffusing across the filter from the outside.

After incubation in 37 °C, 5% CO₂ for 4-7 days, the sprouting from aggregates was quantified using the SigmaScan Pro software (Version 5.0.0; Sigma, St. Louis, MO), by determining the cumulative length of the cell outgrowths. The sprouts lengths were drawn manually on the computer screen, guided by the actual digital image. Additives included VEGF and bFGF (R & D Systems, Minneapolis, MN), endostatin (Calbiochem, San Diego, CA), H₂O₂, and catalase from Aspergillus niger (both from Sigma). Catalase has a molecular weight (MW) of 345 kDa and a radius of 5.83 nm [36]. For endostatin, the hydrodinamic radius (estimated by comparison to fluorescein dextran of similar MW, 20 kDa [13]), is 5.56 nm. For ascorbic acid (MW=176.1 Da), we calculated based on a published relationship between molecular mass and molecular radius [37], a value of about 0.3 nm.

Assessment of the protective effect of filter administered catalase on H₂O₂ induced cytotoxicity in HREC

The cell monolayer model was used to test the effects of filter diffused catalase on endothelial cell survival in the presence of H₂O₂. In this model, HREC were grown in culture medium overnight on glass coverslips (Fisher Scientific, Pittsburgh, PA) placed at the bottom of a 24 well plate (Becton Dickinson). After one day, filter inserts were added and the cells were exposed to the experimental conditions. Control wells contained basal medium on both sides of the filter. To determine the effect of H₂O₂ alone on cell survival, 100 μM H₂O₂ in basal medium was added to the lower well. To counteract the effect of H₂O₂, 200 U/ml catalase in basal medium was added to the upper well lined by the filter, and medium containing 100 μM H₂O₂ within the lower well. The cells were observed for 1-2 days by phase contrast microscopy to determine cell survival. Similar culture conditions were used in the HREC aggregate model to demonstrate the protective effects of trans-filter catalase on angiogenesis in H₂O₂ induced oxidative stress. The experimental conditions were the same as the cell monolayer model, except the aggregates were exposed to 1 mM H₂O₂ and 400 U/ml catalase.

Cytotoxicity assays for vitamin C

To evaluate morphologically the cytotoxicity of vitamin C for retinal endothelial cells, HREC were grown on the bottom of 24 well culture plates overnight, and were then exposed to various concentrations of vitamin C (Sigma), buffered with 0.5 N NaOH to physiological pH (7.4) in basal tissue culture medium. After incubation overnight, the cultures were observed by phase contrast microscopy to determine cell survival. HREC aggregates were also exposed directly to 0.1-2.5 mM vitamin C in culture medium for the time of their growth, to determine the antiangiogenic capabilities of this compound. The sprouting from the aggregates was assessed both visually and quantitatively using the SigmaScan Pro program as described.

Alamar Blue assay (Trek Diagnostic Systems Inc., Westlake, OH) was used to assess HREC's metabolic activity, reflected simultaneously in inhibition of proliferation and cytotoxicity [18]. HREC's were seeded in a 96 well plate at a density of 7500 cells/well in DMEM/F12 supplemented with 10% FBS and 1% PSA. The cells were then grown for 24 h at 37 °C and 5% CO₂ to allow them to achieve confluence. After 24 h ascorbic acid was added to the wells in serial dilutions, and then incubated for further 24 h at 37°C and 5% CO₂. On day 3 the medium was aspirated and the cells were washed twice with PBS. A 10% solution of Alamar blue in culture medium was added to the wells and allowed to incubate for another 3 h. The plate was then read at 590 nm and calculations were done as per the manufacturer's instructions.

Furthermore, vitamin C was delivered to HREC aggregates through Anopore^TM membranes. To this end, the collagen embedded aggregates were placed in wells containing basal medium, and various concentrations up to 5.0 mM vitamin C were added inside the filter insert to diffuse toward the aggregates. The sprouting from the aggregates was assessed after 5 days both visually and quantitatively using the SigmaScan Pro program as described before.

Administration of endostatin across the nanoporous filter

First, we assessed by ELISA the diffusion across the Anopore^TM filter. To this end, 250 μl of 50 μg/ml of endostatin (Calbiochem) solution was added to the Nunc Anopore^TM inserts. The inserts were then placed in 2.5 ml of media in 6 well plates. 20 μl of sample was collected on each day from day 0 through day 7. The collected samples were then analyzed for the endostatin content using an ELISA kit (Calbiochem; catalog number QIA 65). When displaying the results, the concentrations were corrected for the volume effect of aliquots removal.

In order to test the effect of endostatin on HREC growth, we seeded these cells at an initial density of 2x10⁴ cells on the inner surface of Anopore^TM inserts. Then the inserts were placed in wells filled with either tissue culture medium, or this medium supplemented with 10 or 50 μg/ml of human recombinant endostatin. The cell incubations were then placed at 37 °C, in 5% CO₂. They were monitored daily to assess the viability of the cells, using phase contrast microscopy.

The HREC aggregate model was used as well to demonstrate the antiangiogenic effect of endostatin. Aggregates were placed in wells containing basal medium and Anopore^TM inserts were included in the wells. Solutions with initial concentrations of 0, 10, and 50 μg/ml endostatin were added inside the filter in order to let it diffuse to the HREC aggregates.

Results

In vitro sprouting from HREC aggregates

In order to set up a versatile angiogenesis model suitable for testing the effect of various drugs of ophthalmic interest on HREC in vitro, we adapted a technique initially described for bovine cells [35]. We found that HREC readily form aggregates in inverted droplets (Figure 2A). This relied on the ability of cells to form intercellular contacts in suspension and stabilizing the aggregate, unlike another tested cell type, the mouse aortic endothelial cells (data not shown). However, a robust sprouting in collagen gel was stimulated by the presence of a cocktail of angiogenic factors in the growth medium (Figure 2B). This sprouting could be completely blocked by adding in the medium, along with the growth factors, of an angiostatic compound, such as endostatin at 10 μM (Figure 2C).

Protective effects of trans-filter administered catalase on H₂O₂ induced cytotoxicity in HREC

We tested the ability of catalase, the antioxidant enzyme which processes H₂O₂, to cross the aluminum oxide filter and to protect the cells. To this end, we cultivated HREC in a medium supplemented with H₂O₂, in the presence or absence of an immersed Anopore^TM insert, containing catalase on the opposite side of the nanofilter. The control cells grew to form a monolayer (Figure 3A), but the cells exposed to H₂O₂ alone rounded up and died overnight (Figure 3B). The co-administration of catalase through the Anopore^TM filter blocked the effect of H₂O₂, and the cells survived to form a monolayer (Figure 3C).

This observation was replicated on the cell aggregate based angiogenesis model, illustrated in Figure 4. HREC aggregates maintained in supplemented basal medium showed a robust outgrowth after 6 days (Figure 4A). When H₂O₂ was introduced in this medium, there was no more sprouting emerging from the aggregates (Figure 4B). Again, the presence of Anopore^TM filter delivered catalase removed the oxidative stress, as seen by progression of the outgrowth in Figure 4C. Quantification of these effects is presented in Figure 4D.

Diffusion of vitamin C across the filter inhibits HREC sprouting

In order to confirm the reported direct angiostatic effects of ascorbate at doses close to those naturally found in human vitreous, we exposed cultivated HREC to vitamin C up to a concentration of 2.5 mM. We found that indeed vitamin C supplementation had a strong inhibitory effect (Figure 5), validated by the Alamar blue assay (Figure 5D).

Then we tested the ability of ascorbate to block the sprouting process in the aggregate angiogenesis model as well. We found that ascorbate, or its oxidation derivatives, could diffuse through the collagen gel and reach local concentrations high enough to induce blocking of the sprouting (Figure 6A), again abruptly at concentrations higher than 0.5 mM (Figure 6B). This inhibition was maintained with vitamin C administered across the Anopore^TM filter. The concentrations tested covered the range 0.025-5 mM. Above the concentration of 1 mM in the upper compartment, Vitamin C reduced very much the cumulative length of the outgrowth (Figure 7).

Trans-filter diffusion of endostatin blocks the growth and sprouting of HREC

We tested by ELISA the free diffusion of endostatin across the Anopore^TM aluminum oxide nanoporous filter, by collecting aliquots up to 7 days from the lower compartment of a bi-compartmental chamber similar to the one described in Figure 1B. The cumulative diffusion curve is presented in Figure 8, which shows the sustained passage of the drug across the filter, which did not plateaued off even after 7 days. At the end of one week, 812 ng out of the 12,500 ng initially placed in the upper compartment (which represented 6.49%), passed through the nanofilter in the second compartment.

Then, we examined the cell attachment and proliferation of HREC cultured on the bottom of Anopore^TM inserts (facing the upper compartment), after immersing them in tissue culture medium containing endostatin for more than a week. We found that already at 10 μg/ml endostatin present on the opposite side of the filter, after several hours the cells stopped to adhere, rounded up and detached (Figure 9). At the end of the incubation, the cells were almost completely removed, showing that the endostatin diffused across the nanofilter. Trans-filter diffused endostatin also efficiently stopped HREC sprouting from aggregates (Figure 10). After 6 days there was almost no sprout formation from any of the aggregates, as shown both microscopically (Figure 10A) and by the quantification of the assay (Figure 10B).

Discussion

Alumina films made by anodization are nanoporous and have highly uniform, straight and parallel pore channels [13]. They also have a high pore density (10⁹-10¹¹/cm²) and an almost monodisperse size distribution. These membranes can be made in few hours from aluminum metal, and they are relatively inexpensive. Aluminum oxide membranes are also commercially available as the bottom of tissue culture inserts (trade name: Anopore^TM). Nanoporous alumina capsules were recently described [13], but they were not tested yet in a biologically meaningful way. So far, the diffusion from this nanoporous capsules was characterized using fluorescein isothiocyanate and dextran conjugates of varying molecular weights, showing that molecular transport could be readily controlled by selection of capsule pore size. However, these results cannot be quantitatively extrapolated to other molecular species, because each one is governed by empirical molecular factors (shape, hydration, affinity with the nanopore's wall etc), which are currently unknown. The fact that the capsules are either completely closed compartments, as in the published paper [13] or open (half chamber) as in our model, makes only a minimal difference for the in vitro testing.

When the mean free path of a molecule becomes comparable to the size of the pore (for example, catalase has a radius of 5.83 nm, endostatin 5.5 nm), the molecules are just as likely to hit the walls, as they are to hit each other. In this case, the diffusion constant takes on a new limiting value, different from the classical counterpart. These systems are described by the ratio of system size to the mean free path, the Knudsen number (Kn). When Kn>1, the continuous approximations are valid, but when Kn<1, it is necessary to use Monte Carlo calculations (estimates of actual molecular trajectories) to make accurate calculations.

Recently was proposed an yet to be proven model for the transport of molecules adsorbed in slit and cylindrical nanopores at low density, considering the axial momentum gain of molecules oscillating between diffuse wall reflections [11]. In this paper, good agreement with molecular dynamics simulations was obtained over a wide range of pore sizes, including the regime of single file diffusion where fluid-fluid interactions are shown to have a negligible effect on the collective transport coefficient. It was also shown that dispersive fluid-wall interactions considerably attenuate the transport, compared to classical hard sphere theory.

Even the study of transport of a simple fluid, such as water in a cylindrical nanopore requires a combination of equilibrium and nonequilibrium molecular dynamics methods [12]. Using these approaches, it was demonstrated that a combination of viscous flow and momentum exchange at the pore wall governs the transport over a wide range of densities. For this reason, we found it appropriate to test the diffusion of a small molecule, such as ascorbate, whose transport across nanopores may still behaves in a non-classical manner.

A comprehensive therapeutic approach for the treatment of proliferative disorders in the eye, such as PDR, requires the consideration for the following needs: (a) to counter balance the pro-oxidant state usually associated with their initiation [22,23,25], (b) maintain appropriate levels of vitamins, such as vitamin C (depleted in diabetes [28] and with aging [24]), and (c) introduce in the eye angiostatic compounds to block neovascular proliferation [19]. Therefore, we studied the possibility to deliver across the Anopore^TM nanoporous filter, in significant amounts to induce biologically detectable effects, the following compounds: (a) catalase, (b) vitamin C, and (c) endostatin. We successfully used an aluminum oxide filter with pores in the nanometer size range (20 nm), as a semipermeable barrier to separate two compartments in vitro, across which we delivered for long periods (days) the above substances to HREC, influencing both their viability as well as their ability to form sprouts.

We directly exposed HREC to cytotoxic concentrations of H₂O₂, the key reactive oxygen species involved in many steps of the oxidative stress dependent pathogenesis in the eye [22,25,38,39] When on the opposite side of the nanofilter we concomitantly added catalase, the cells were protected from the damaging effect of H₂O₂. This test was also done with HREC embedded in a collagen gel as aggregates which produced sprouting upon the exposure of appropriate angiogenic factors, in a model of in vitro angiogenesis [35]. Incubation of the aggregates with H₂O₂ was cytotoxic again, and prevented the formation of sprouts. Delivery of catalase through the nanofilter was efficient in protecting the aggregates' potency to produce sprouts.

The concentration of ascorbic acid in the vitreous humor of humans is about 25 times higher than in the plasma [27]. It is considered that, besides being a strong antioxidant [21], this high ascorbate concentration might also play a role in keeping the retina avascular in normal conditions [18]. Vitreous humor is cytotoxic to bovine aortic endothelial cells, and ascorbic acid was isolated as the cell death inducing factor [28]. Furthermore, ascorbic acid inhibited the formation of tubes by endothelial cells. Ascorbate was quantified in vitreous humor of PDR and in other pathological conditions and was found to decrease up to a third comparatively to the normal eyes [28].

When dedifferentiated porcine retinal pigmented epithelial (RPE) cells were incubated in vitro with increasing concentrations of ascorbic acid (0.25-1.5 mM), this significantly inhibited cell proliferation, migration, and contraction in concentrations of 1 mM or more [40]. However, at this concentration, dedifferentiated RPE cells acquired a pigmented status within 24 h. Addition of catalase prevented the antiproliferative effect of the ascorbic acid, and the formation of pigment. As dedifferentiation of these cells is an integral part in the development of proliferative vitreoretinopathy, ascorbic acid was suggested as a supplement in the clinical management of this disease. In the same study, when mature porcine RPE cells were cultured with ascorbic acid (0.01-5 mM), this was found to have dose dependent cytotoxicity on RPE cells. Once again, catalase protected against this cytotoxicity.

These results indicate that ascorbic acid is toxic to ocular cells, probably (although not exclusively [41]) through an H₂O₂ mediated mechanism [39,42]. In our hands, ascorbate was inhibitory on HREC starting at concentrations above 0.7 mM when directly exposing adherent HREC, above 0.25 mM when the cells were in aggregates, and above 2 mM in the Alamar blue cytotoxicity assay. These results, although in a close range of values which confirms the published data, may reflect both differences in the assays, as well as in the status of the cells during the experiments.

Available published data suggest that formation of ascorbyl radical is necessary for vitamin C induced cytotoxicity in various culture systems [43,44]. Because the spontaneous conversion of ascorbic acid into ascorbyl may take place on either side of the filter, we did not address the separate rate of diffusion of intact vitamin C.

When we delivered vitamin C to HREC from a stock solution across the Anopore^TM filter, it showed a clear effect above 1 mM in the upper compartment, although signs of the inhibition were present from the lowest concentration tested in this case (0.025 mM). A cooperative cytotoxic effect of aluminum released from the filter could not be completely ruled out, although this compound is in oxidated form in the commercially available tissue culture system used for our assays. Even in non-oxidated state, aluminum had only a small cytotoxicity on brain microvascular endothelial cells [45,46]. Moreover, ascorbate was found to be instrumental in aluminum resistance in some cell lines [45]. In any case, if the nature of the material from which the nanaofilter is made would ever be an issue, it can be easily replaced with filters with nanopores in the same diameter ranges made out of silicon [9].

In retinal microvessels, the endothelial cells are exposed with their abluminal side towards the content of the vitreous cavity, from where a drug would be administered by an implanted sustained release delivery system of the type devised in this paper. Therefore, we found it appropriate to test if it will be still efficient after the exposure of HREC to a compound coming from its abluminal side. To this end, we seeded the cells directly on the inner surface of the Anopore^TM insert, and after overnight adhesion, we immersed the inserts in solutions of endostatin, a very potent anti-angiogenic agent [19], at various concentrations. The filter system is one of the few modalities to study the potentially asymmetric effects of drugs on polarized cells, such as epithelial and endothelial cells.

We directly assessed diffusion of endostatin by ELISA, and we found that it was released across the filter with a linear rate for up to 7 days. The lack of leveling off of the diffusion might be explained by the small percentage of the total protein passed (about 6.5%), which suggests that the concentration gradient did not change significantly during this interval. However, that fraction was enough, supposedly by its cumulative action, to produce a clear inhibition of both filter attached and sprouting HREC. With this rate, we calculated that the content of the reservoir would be delivered in about 4 months. While it may depend on capsule's geometry, in another study based on a filter with pore size of 18 nm, for example, the percentage of diffusion for IgG molecules was less than 0.4% of the content at 24 h [7]. Dextran of 20 kDa (5.56 nm) diffused trough a 55 nm pore filter at a rate of 0.1% per day [13]. These data support the notion of "sustained release systems" applied to nanofilter based containers.

So far, the administration of endostatin for inhibition of choroidal neovascularization or retinal angiogenesis was induced by intravenous [47] or intraocular [48] injections of adenoviral vectors expressing secretable endostatin. In order to prove the feasibility of our alternative approach, which consists in extended delivery of the drug from a reservoir, we cultivated HREC on the bottom of the chamber with compartments separated by Anopore^TM membranes. Endostatin was added on the other side of the filter and within 7 days, most of the cells died, indicating that the diffusion across the nanofilter was enough to produce cytotoxic concentrations of drug in the second compartment.

We consider that the nanofilter approach is a promising new way of delivering the drugs to the eye because: (a) its potential for controlled diffusion (dependent on the size of the molecule) [13]; (b) its semi-permeable character (small molecular weight drugs would diffuse outwards, but larger organic macromolecules would be prevented to enter in the capsule) [49]; (c) In opposition with the release from polymeric media, the filter walled capsules will have a larger active volume, since their interior is completely available to load with any drug formulation; (d) these implantable devices could be retrieved at any time when they are no more needed, or for replenishing or changing their content; (e) good biocompatibility of the materials from which the filters (either alumina or silicon) and the capsules (titanium, for example) can be made; (f) the materials of these capsules make them compatible, as parts of larger platforms, with the emerging industry of implantable electronic devices with potential ophthalmic applications, such as biosensors for probing the local milieu [50] and/or with electronic retinal prosthesis [51,52]. These electronic devices are still in the early stages of development, but have an immense potential and are likely to be made of inorganic materials of various compositions and surface topography. For this purpose, the study of the interaction of subretinal cells with these materials is of considerable importance. The principle of sustained release would be further strengthened by partially immobilizing the drugs in gels or in crystalline state inside of the drug capsule. The rate of the diffusion could be controlled by the pore density and size, by the active surface of the filter compared to the size of the capsule, and/or by controlled application of electrical fields across the filter, which might slow the exit of charged or polar molecules.

Our data also confirm that Anopore^TM filter is a very good substrate for cell growth, and is not toxic for HREC and for human RPE cells (data not shown). The interaction of the porous aluminum oxide with other cell types and with animal tissue in vivo remains to be further investigated.

In conclusion, we proved that drugs of ophthalmic interest could diffuse across a nanoporous filter and provide anti-angiogenic and anti-oxidant effects in vitro. We took advantage of the commercial availability of an aluminum oxide filter with uniform pores of 20 nm (Anopore^TM). We confirmed that the diffusion properties of this material could support the construction of containers destined for the sustained release of drugs in the ocular cavity. Collectively, the data shown here represent the proof of concept regarding the possibility to manufacture and use implantable, biocompatible capsules based on nanoporous filters, able to provide controlled delivery of anti-angiogenic and anti-oxidant molecules.

Acknowledgements

This work was done with the encouragement and the financial support from Patti Blow Research Fund in Ophthalmology. The authors are also very grateful to Mrs. Laura Sladoje for the exceptional logistic and administrative skills, and Dr. N. L. Parinandi for useful suggestions.

References

1. Wolfensberger TJ. Pharmacology of the retinal pigment epithelium. In: Marmor MF, Woflensberger TJ, editors. The retinal pigment epithelium: function and disease. New York: Oxford University Press; 1998. p. 604-20.

2. Sanborn GE, Anand R, Torti RE, Nightingale SD, Cal SX, Yates B, Ashton P, Smith T. Sustained-release ganciclovir therapy for treatment of cytomegalovirus retinitis. Use of an intravitreal device. Arch Ophthalmol 1992; 110:188-95.

3. Tao W, Wen R, Goddard MB, Sherman SD, O'Rourke PJ, Stabila PF, Bell WJ, Dean BJ, Kauper KA, Budz VA, Tsiaras WG, Acland GM, Pearce-Kelling S, Laties AM, Aguirre GD. Encapsulated cell-based delivery of CNTF reduces photoreceptor degeneration in animal models of retinitis pigmentosa. Invest Ophthalmol Vis Sci 2002; 43:3292-8.

4. Kompella UB, Bandi N, Ayalasomayajula SP. Subconjunctival nano- and microparticles sustain retinal delivery of budesonide, a corticosteroid capable of inhibiting VEGF expression. Invest Ophthalmol Vis Sci 2003; 44:1192-201.

5. Bourges JL, Gautier SE, Delie F, Bejjani RA, Jeanny JC, Gurny R, BenEzra D, Behar-Cohen FF. Ocular drug delivery targeting the retina and retinal pigment epithelium using polylactide nanoparticles. Invest Ophthalmol Vis Sci 2003; 44:3562-9.

6. Qaddoumi MG, Gukasyan HJ, Davda J, Labhasetwar V, Kim KJ, Lee VH. Clathrin and caveolin-1 expression in primary pigmented rabbit conjunctival epithelial cells: role in PLGA nanoparticle endocytosis. Mol Vis 2003; 9:559-68 <http://www.molvis.org/molvis/v9/a68/>.

7. Desai TA, Hansford DJ, Kulinsky L, Nashat AH, Rasi G, Tu J, Wang Y, Zhang M, Ferrari M. Nanopore technology for biomedical applications. Biomedical Microdevices 1999; 2(1):11-40.

8. Desai TA, Hansford DJ, Ferrari M. Micromachined interfaces: new approaches in cell immunoisolation and biomolecular separation. Biomol Eng 2000; 17:23-36.

9. Moldovan NI, Ferrari M. Prospects for microtechnology and nanotechnology in bioengineering of replacement microvessels. Arch Pathol Lab Med 2002; 126:320-4.

10. Conlisk AT, McFerran J, Zheng Z, Hansford D. Mass transfer and flow in electrically charged micro- and nanochannels. Anal Chem 2002; 74:2139-50.

11. Jepps OG, Bhatia SK, Searles DJ. Wall mediated transport in confined spaces: exact theory for low density. Phys Rev Lett 2003; 91:126102.

12. Bhatia SK, Nicholson D. Hydrodynamic origin of diffusion in nanopores. Phys Rev Lett 2003; 90:016105.

13. Gong D, Yadavalli V, Paulose M, Pishko M, Grimes CA. Controlled molecular release using nanoporous alumina capsules. Biomedical Microdevices 2003; 5:75-80.

14. Cadman J. Ganciclovir implants: one year later. GMHC Treat Issues 1997; 11:3-6.

15. Verdejo C, Marco P, Renau-Piqueras J, Pinazo-Duran MD. Lipid peroxidation in proliferative vitreoretinopathies. Eye 1999; 13:183-8.

16. Delamere NA. Ascorbic acid and the eye. Subcell Biochem 1996; 25:313-29.

17. Jacob RA, Sotoudeh G. Vitamin C function and status in chronic disease. Nutr Clin Care 2002; 5:66-74.

18. Hanashima C, Namiki H. Reduced viability of vascular endothelial cells by high concentration of ascorbic acid in vitreous humor. Cell Biol Int 1999; 23:287-98.

19. O'Reilly MS, Boehm T, Shing Y, Fukai N, Vasios G, Lane WS, Flynn E, Birkhead JR, Olsen BR, Folkman J. Endostatin: an endogenous inhibitor of angiogenesis and tumor growth. Cell 1997; 88:277-85.

20. Kerbel R, Folkman J. Clinical translation of angiogenesis inhibitors. Nat Rev Cancer 2002; 2:727-39.

21. Richer SP, Rose RC. Water soluble antioxidants in mammalian aqueous humor: interaction with UV B and hydrogen peroxide. Vision Res 1998; 38:2881-8.

22. Ellis EA, Guberski DL, Somogyi-Mann M, Grant MB. Increased H2O2, vascular endothelial growth factor and receptors in the retina of the BBZ/Wor diabetic rat. Free Radic Biol Med 2000; 28:91-101.

23. Rose RC, Richer SP, Bode AM. Ocular oxidants and antioxidant protection. Proc Soc Exp Biol Med 1998; 217:397-407.

24. van der Pols JC. A possible role for vitamin C in age-related cataract. Proc Nutr Soc 1999; 58:295-301.

25. Beatty S, Koh H, Phil M, Henson D, Boulton M. The role of oxidative stress in the pathogenesis of age-related macular degeneration. Surv Ophthalmol 2000; 45:115-34.

26. Peponis V, Papathanasiou M, Kapranou A, Magkou C, Tyligada A, Melidonis A, Drosos T, Sitaras NM. Protective role of oral antioxidant supplementation in ocular surface of diabetic patients. Br J Ophthalmol 2002; 86:1369-73.

27. Wunderlich K, Knorr M, Dartsch PC, Steuhl HP, Thiel HJ. [Ascorbic acid. Cytotoxic effect on cultivated bovine lens epithelium cells]. Ophthalmologe 1992; 89:313-8.

28. Takano S, Ishiwata S, Nakazawa M, Mizugaki M, Tamai M. Determination of ascorbic acid in human vitreous humor by high-performance liquid chromatography with UV detection. Curr Eye Res 1997; 16:589-94.

29. Doly M, Droy-Lefaix MT, Braquet P. Oxidative stress in diabetic retina. EXS 1992; 62:299-307.

30. Yorek MA. The role of oxidative stress in diabetic vascular and neural disease. Free Radic Res 2003; 37:471-80.

31. Nicosia RF, Belser P, Bonanno E, Diven J. Regulation of angiogenesis in vitro by collagen metabolism. In Vitro Cell Dev Biol 1991; 27A:961-6.

32. Pe'er J, Shweiki D, Itin A, Hemo I, Gnessin H, Keshet E. Hypoxia-induced expression of vascular endothelial growth factor by retinal cells is a common factor in neovascularizing ocular diseases. Lab Invest 1995; 72:638-45.

33. Eisenstein R, Grant-Bertacchini D. Growth inhibitory activities in avascular tissues are recognized by anti-transforming growth factor beta antibodies. Curr Eye Res 1991; 10:157-62.

34. Funatsu H, Yamashita H. Pathophysiology of diabetic retinopathy. Drug News Perspect 2002; 15:633-639.

35. Vernon RB, Sage EH. A novel, quantitative model for study of endothelial cell migration and sprout formation within three-dimensional collagen matrices. Microvasc Res 1999; 57:118-33.

36. Mosavi-Mohavedi AA, Wilkinson AE, Jones MN. Characterization of Aspergillus niger catalase. Int J Biol Macromol 1987; 9:327-32.

37. Watson CJ, Rowland M, Warhurst G. Functional modeling of tight junctions in intestinal cell monolayers using polyethylene glycol oligomers. Am J Physiol Cell Physiol 2001; 281:C388-97.

38. Zhang H, Agardh E, Agardh CD. Hydrogen peroxide production in ischaemic retina: influence of hyperglycaemia and postischaemic oxygen tension. Diabetes Res 1991; 16:29-35.

39. Iwasaka K, Koyama N, Nogaki A, Maruyama S, Tamura A, Takano H, Takahama M, Kochi M, Satoh K, Sakagami H. Role of hydrogen peroxide in cytotoxicity induction by ascorbates and other redox compounds. Anticancer Res 1998; 18:4333-7.

40. Bohmer JA, Sellhaus B, Schrage NF. Effects of ascorbic acid on retinal pigment epithelial cells. Curr Eye Res 2001; 23:206-14.

41. Nemoto S, Otsuka M, Arakawa N. Effect of high concentration of ascorbate on catalase activity in cultured cells and tissues of guinea pigs. J Nutr Sci Vitaminol (Tokyo) 1997; 43:297-309.

42. Woo KI, Lee J. The effects of ascorbic acid on free radical injury in cultured retinal pigment epithelial cells. Korean J Ophthalmol 1995; 9:19-25.

43. Makino Y, Sakagami H, Takeda M. Induction of cell death by ascorbic acid derivatives in human renal carcinoma and glioblastoma cell lines. Anticancer Res 1999; 19:3125-32.

44. Satoh K, Sakagami H. Effect of metal ions on radical intensity and cytotoxic activity of ascorbate. Anticancer Res 1997; 17:1125-9.

45. Vorbrodt AW, Dobrogowska DH, Lossinsky AS. Ultracytochemical studies of the effects of aluminum on the blood-brain barrier of mice. J Histochem Cytochem 1994; 42:203-12.

46. Vorbrodt AW, Trowbridge RS, Dobrogowska DH. Cytochemical study of the effect of aluminium on cultured brain microvascular endothelial cells. Histochem J 1994; 26:119-26.

47. Mori K, Ando A, Gehlbach P, Nesbitt D, Takahashi K, Goldsteen D, Penn M, Chen CT, Mori K, Melia M, Phipps S, Moffat D, Brazzell K, Liau G, Dixon KH, Campochiaro PA. Inhibition of choroidal neovascularization by intravenous injection of adenoviral vectors expressing secretable endostatin. Am J Pathol 2001; 159:313-20.

48. Auricchio A, Behling KC, Maguire AM, O'Connor EM, Bennett J, Wilson JM, Tolentino MJ. Inhibition of retinal neovascularization by intraocular viral-mediated delivery of anti-angiogenic agents. Mol Ther 2002; 6:490-4.

49. Desai TA, Hansford DJ, Leoni L, Essenpreis M, Ferrari M. Nanoporous anti-fouling silicon membranes for biosensor applications. Biosens Bioelectron 2000; 15:453-62.

50. Frost MC, Meyerhoff ME. Implantable chemical sensors for real-time clinical monitoring: progress and challenges. Curr Opin Chem Biol 2002; 6:633-41.

51. Chow AY, Pardue MT, Perlman JI, Ball SL, Chow VY, Hetling JR, Peyman GA, Liang C, Stubbs EB Jr, Peachey NS. Subretinal implantation of semiconductor-based photodiodes: durability of novel implant designs. J Rehabil Res Dev 2002; 39:313-21.

52. Humayun MS. Intraocular retinal prosthesis. Trans Am Ophthalmol Soc 2001; 99:271-300.