In vitro bovine eye tissue/phosphate-buffered saline, pH 7.4, partition coefficients (Kt:b), in vitro binding to natural melanin, and in vivo delivery at 1 h after posterior subconjunctival injection in Brown Norway rats were determined for eight β-blockers. The Kt:b was in the order intact tissue, dry weight method ≥ intact tissue, wet weight method corrected for tissue water and drug in tissue water ≫ intact tissue, wet weight method > homogenized tissue. In intact tissue methods, Kt:b followed the order choroid-retinal pigment epithelium (RPE) > trabecular meshwork > retina > sclera ∼ optic nerve; propranolol > betaxolol > pindolol ∼ timolol ∼ metoprolol > sotalol ∼ atenolol ∼ nadolol. Intact tissue, wet weight log (Kt:b) correlated positively with log D for all tissues (R 2 of 0.7–0.9). Log (melanin binding capacity) correlated positively with choroid-RPE log (Kt:b) (R 2 of 0.5). With an increase in concentration, Kt:b decreased in trabecular meshwork for all β-blockers and for some lipophilic β-blockers in choroid-RPE and sclera. With an increase in drug lipophilicity, in vivo tissue distribution increased in choroid-RPE, iris-ciliary body, sclera, and cornea but exhibited a declining trend in retina, vitreous, and lens. In vitro bovine intact tissue, wet weight Kt:b correlated positively with rat in vivo tissue/vitreous humor distribution for sclera, choroid-RPE, and retina (R 2 of 0.985–0.993). In vitro tissue partition coefficients might be useful in predicting in vivo drug distribution after trans-scleral delivery. Less lipophilic solutes exhibiting limited nonproductive binding in choroid-RPE might exhibit greater trans-scleral delivery to the retina and vitreous.
Determining tissue partition coefficients of drugs will potentially enable 1) selection of a drug that is most appropriate for being retained in a target tissue for therapeutic effects, 2) selection of a drug that has the least affinity in a tissue to avoid side effects in that tissue, 3) identification of a drug with reduced nonproductive binding to barriers in transit before reaching a target tissue, and 4) prediction of in vivo tissue distribution based on in vitro tissue partition coefficients. Collectively, an understanding of drug partition coefficients in tissues might enable selection of a drug molecule with enhanced delivery to a target tissue, while minimizing its nontarget tissue accumulation. However, such tissue partition coefficients and their relationship to drug properties have not been determined for ocular tissues, especially those encountered in trans-scleral drug delivery to the retina and vitreous.
Although the intravitreal route of administration is a viable option for treating retinal disorders such as age-related macular degeneration, especially using therapeutic agents with prolonged vitreal half-life (Durairaj et al., 2009b) or those with poor solubility (Durairaj et al., 2009a), intravitreal injections can cause complications such as endophthalmitis, retinal detachment, and ocular hypertension (Ozkiriş and Erkilic, 2005). To avoid these complications trans-scleral drug delivery, wherein a drug injected adjacent to the sclera, traverses the sclera and choroid-RPE to reach retina, is a viable option in treating retinal disorders (Ayalasomayajula and Kompella, 2004; Raghava et al., 2004; Amrite et al., 2006, 2008a). This route of administration is particularly suited for slow release particulate delivery systems that can avoid circulatory clearance mechanisms (Amrite et al., 2008b). However, trans-scleral drug delivery to the retina is hindered by the choroidal barrier (Cheruvu and Kompella, 2006), with this barrier being more significant in pigmented animals compared with nonpigmented animals (Cheruvu et al., 2008). For example, celecoxib binds to melanin in pigmented choroid-retinal pigment epithelium (choroid-RPE), thereby resulting in reduced delivery of this drug to the retina in a pigmented animal compared with a nonpigmented animal (Cheruvu et al., 2008). Furthermore, this barrier hinders the delivery of drugs from slow release delivery systems more than bolus doses, possibly due to incompletely saturated drug binding to the choroid layer including RPE. Thus, to understand trans-scleral drug delivery and to optimize drug selection, a key objective of this study was to determine partition coefficients of a series of eight β-blockers with varying lipophilicity in tissues of the back of the eye. In addition to serving as model solutes, β-blockers are of potential therapeutic value as neuroprotective agents in the back of the eye. Although ophthalmic β-blocker products were originally developed for reducing intraocular pressure due to their actions in the anterior segment of the eye, growing evidence indicates that these agents might exerts neuroprotective effects in the back of the eye. For example, ischemia-induced optic nerve death has been linked to accumulation of sodium and calcium in ganglion cells of the retina (Osborne et al., 2005). It is interesting to note that of all antiglaucoma drugs, only β-blockers are capable of inhibiting sodium and calcium channels, making them potential candidates for neuroprotection in the eye (Wood et al., 2003; Osborne et al., 2005; Chen et al., 2007).
The affinity of drug to the specific tissue can be characterized by ether in vitro tissue/plasma or tissue/buffer partition coefficients or in vivo tissue distribution at steady state (Ballard et al., 2000). In vitro tissue partitioning methods have been reported for volatile organic compounds and drug molecules using homogenized tissue and tissue slices (Gargas et al., 1989; Murphy et al., 1995; Fridén et al., 2009). There are no reports on systematic tissue partition coefficient estimates in ocular tissues. In vitro tissue partitioning using cassette or cocktail dosing method significantly reduces the time, labor, and cost. In this article, using eight β-blockers (sotalol, atenolol, nadolol, pindolol, timolol, metoprolol, betaxalol, and propranolol) with varying lipophilicities as model solutes in cassette dosing, we investigated in vitro tissue/phosphate-buffered saline (PBS; pH 7.4) partition coefficients for bovine retina, optic nerve, choroid-RPE, sclera, and trabecular meshwork. Although sclera, choroid-RPE, and retina are tissues encountered during trans-scleral drug delivery to the retina, the optic nerve is a target for neuroprotective drugs, and trabecular meshwork is often implicated in drug-induced glaucoma during treatment of retina disorders (Fan et al., 2008). Influence of lipophilicity on drug accumulation in these tissues has yet to be established. We used several approaches to estimate tissue partition coefficients. These approaches included 1) intact tissue, wet weight method; 2) intact tissue, dry weight method; 3) intact tissue, wet weight method, with correction for tissue water content and drug in tissue water; and 4) homogenized tissue, wet weight method. Furthermore, we determined the binding affinities and capacities of β-blockers for natural melanin, because choroid-RPE is a pigment-rich tissue. Tissues and β-blockers were ranked for their partition coefficients using the above-mentioned methods. Furthermore, correlations were derived between drug lipophilicity and tissue partition coefficients as well as melanin binding. By measuring log D values of β-blockers from cocktail and comparing those with literature values for individual β-blockers, we also validated our cassette dosing approach. Finally, to determine the usefulness of in vitro partition coefficients in predicting in vivo tissue distribution, we correlated in vitro bovine tissue/PBS partition coefficients with in vivo rat tissue/vitreous distribution ratios for sclera, choroid-RPE, and retina, three key tissues involved in trans-scleral drug delivery to the back of the eye.
Materials and Methods
Atenolol (99.0% purity), sotalol hydrochloride (∼98%), nadolol (∼98%), pindolol (98%), timolol maleate (98%), metoprolol tartrate (99%), betaxolol (∼98%), labetalol hydrochloride (99%), propranolol hydrochloride (99%), and natural melanin (Sepia officinalis) were purchased from Sigma-Aldrich (St. Louis, MO). Acetonitrile, methanol, dichloromethane, and ethyl acetate, all high-performance liquid chromatography grade solvents, were purchased from Thermo Fisher Scientific (Waltham, MA). Triethyl amine (99.5%), sodium hydroxide, and formic acid (88%) were purchased from Sigma-Aldrich. Ammonium formate (99.9%) was purchased form Fluka Chemical Corp. (Ronkonkoma, NY). All other chemicals and reagents used in this study were of analytical reagent grade.
Determination of Distribution Coefficients.
The distribution coefficient (log D) of β-blockers cocktail between PBS, pH 7.4, and n-octanol was determined by a shake flask method. The buffer and n-octanol were mutually saturated by shaking at 37°C for 12 h on a shaker incubator set at 300 rpm. The cocktail of β-blockers was prepared in the above-mentioned equilibrated buffer at three different concentrations (0.5, 5.0, and 25 μg/ml). The distribution coefficient at 37°C was determined by incubating 1 ml of β-blockers solution in the equilibrated buffer with 1 ml of n-octanol at 37°C for 24 h while continuously shaking at 300 rpm to reach distribution equilibrium. At the end of 24 h, the n-octanol and buffer phases were separated by centrifugation at 13,000 rpm for 10 min, and the concentrations of β-blockers were estimated in both phases.
The distribution coefficient (DC) was calculated as DC (pH 7.4) = Cn-octanol/Cbuffer, where Cn-octanol and Cbuffer represent the concentrations of β-blockers in the organic and aqueous phases, respectively, after equilibration. The DC is therefore the quotient of two concentrations, and it has been presented as logarithm to base 10 (log D).
Freshly excised bovine eyes along with the optic nerve were obtained from a local slaughter house (Nebraska Beef, Omaha, NE) and kept on ice during transport to the laboratory. The eyes were cleaned by removing the external connective tissues and muscles and rinsed with PBS. The eyes were opened by giving a circular cut just posterior to the limbus to separate anterior and posterior segment of the eye. The trabecular meshwork was isolated from the anterior segment according to the previously reported procedure, with little modification (Busch et al., 1993). The iris-ciliary body was gently lifted from the sclera with forceps. After removal of nontrabecular tissues, the gray color band of trabecular meshwork surrounding the cornea was scraped off from the sclera using a scalpel blade.
The sclera, choroid-RPE, retina, and optic nerve were isolated from the posterior segment. In brief, the vitreous and lens tissues were removed gently from the neural retina. The neural retina was isolated and collected from choroid-RPE by exposing the eyecup to the isotonic PBS, pH 7.4. The choroid-RPE was gently peeled and then scraped off from the sclera using scalpel blade. The optic nerve was isolated by giving a cut at the joint of sclera and optic nerve on the posterior side of the sclera and was cleaned by scraping off the muscle and peeling off the white fat layer surrounding the optic nerve.
In Vitro Tissue/PBS Partition Coefficient Determination.
All tissue partitioning studies were conducted with PBS, pH 7.4, with the following composition: 8.5 g/l NaCl, 0.95 g/l Na2HPO4, and 0.455 g/l NaH2PO4. The tissue partitioning studies were performed using intact tissue or homogenized tissue. In intact tissue, in one experiment, drug levels were measured and normalized in wet tissue. In another experiment, partition study was conducted with intact tissue, and then the tissue was dried at the end of the experiment, weighed, and the drug was extracted for partition coefficient (Kt:b) estimation based on dry weight of the tissue. Approximately 100 mg of each tissue was weighed and incubated with 0.5 ml of cocktail solution of eight β-blockers in PBS. Three concentrations of β-blockers, i.e., 0.5, 5.0, and 25.0 μg/ml, were used. In the homogenized tissue partitioning study, incubation solution was added to each tissue and homogenized by Tissue Tearor (Biospec Products, Racine, WI) to form a homogenous suspension. All tissue preparations were incubated in a shaker incubator for 6 h at 37°C with constant shaking at 300 rpm. The incubation was terminated by removing the tissue from the incubating medium by centrifugation at 5000 rpm for 5 min. The tissues were washed with 0.5 ml of PBS. For estimating dry tissue weight in an experiment separate from wet intact tissue method, the tissue was dried at the end of partitioning at 90°C for 12 h. Subsequently, the tissues were weighed and extracted for the drug. Furthermore, partition coefficient estimates were also derived from intact tissue, wet weight method after correcting the values for tissue water content and drug content in tissue water. Tissues, incubation solution, and wash solution were stored at −80°C until analysis.
Determination of Tissue Water Content.
Water content in sclera, choroid-RPE, retina, trabecular meshwork, and optic nerve was determined by a heat drying method (Floyd et al., 1975). The tissues (∼100 mg) were cut into small pieces and placed in previously weighed glass test tubes (12 × 75 mm; Thermo Fisher Scientific). The tissues were kept at 90°C in a hot-air oven for 12 h or until constant weight was achieved. The tissue water content was calculated as estimated water content = (tissue weight before drying − tissue weight after drying). The percentage of water content was then calculated as % water content = (estimated water content/tissue weight before drying) × 100.
In Vitro Melanin Binding of β-Blockers.
To study the binding of β-blockers to melanin, a suspension of 2 mg/ml natural melanin (S. officinalis) was prepared in PBS, pH 7.4. The melanin suspension was warmed up to 37°C and sonicated for 15 min to form a uniform suspension. The melanin suspension (0.75 ml) was measured in aliquots into glass test tubes (12 × 75 mm) under continuous shaking to avoid sedimentation. The cocktail solution of β-blockers in PBS (0.75 ml; 0.05–100 μg/ml) was added to the above-mentioned melanin aliquots, and the tubes were incubated in a shaker incubator at 37°C and 300 rpm for 4 h. Controls were prepared by incubating the test compounds in the absence of melanin in incubation medium (PBS) and by incubating melanin without test compounds. All incubations were done in triplicates. After incubation, the suspensions were centrifuged at 13,000 rpm for 15 min at room temperature to remove the melanin granules. The supernatant was collected, diluted 20 times with acetonitrile and analyzed by an LC-MS/MS method described latter. The kinetic parameters of binding study, i.e., maximal binding capacity (r max) and the binding affinity (k) were determined for each β-blocker according to the previously reported method (Cheruvu et al., 2008).
In Vivo Tissue Distribution Study.
All animals were treated according to the ARVO statement for the use of Animals in Ophthalmic and Vision Research. Male Brown Norway (BN; pigmented) rats (Charles River Laboratories, Inc., Wilmington, DE) weighing 200 to 250 g were used. A cocktail mixture of the eight β-blockers (250 μg/ml each) including sotalol, atenolol, nadolol, pindolol, timolol, metoprolol, betaxolol, and propranolol was prepared in PBS, pH 7.4. The rats were anesthetized with intraperitoneal injection of 50 mg/ml ketamine and 10 mg/ml xyalazine. The β-blockers solution (25 μl) was administered into the posterior subconjunctival space of right eye (ipsilateral) using a 30-gauge needle. The left eye was untreated. The animals were allowed to recover from anesthesia, and water and food were provided ad libitum. At the end of 1 h, the animals were euthanized by intraperitoneal injection of sodium pentobarbital (250 mg/kg). The eyes were enucleated and immediately frozen using a mixture of isopentane and dry ice. Tissues were stored at −80°C until analysis. Periocular tissue surrounding the site of administration was also collected and stored at −80°C until analysis. Drug levels were estimated in sclera, choroid-RPE, retina, vitreous, lens, cornea, iris-ciliary body, and periocular tissue by an LC-MS/MS method. Because optic nerve and trabecular meshwork could not be isolated reliably, drug levels in these tissues were not measured.
Tissue Sample Preparation and LC-MS/MS Analysis.
The β-blocker concentrations in bovine and rat ocular tissues were estimated after liquid-liquid extraction of drug from tissues as described previously (Kadam and Kompella, 2009). In brief, the ocular tissue samples were homogenized with 500 μl of 2% (wt/v) solution of NaOH in water and Tissue Tearor (Biospec Products). To this homogenate 10 μl of 25 μg/ml labetalol solution was added as an internal standard, and the solution was vortexed briefly. The drugs were extracted by adding 6 ml of the extraction solvent (dichloromethane/ethyl acetate, 1:1) followed by vortex mixing for 15 min on multitube vortexer (VWR LabShop, Batavia, IL). The tubes were then centrifuged (5000 rpm at 25°C for 10 min), and the organic layer was collected and evaporated to dryness at 40°C under a nitrogen stream (N-Evap; Organomation, Berlin, MA). The dried residue was reconstituted in 250 μl of acetonitrile/water (80:20) and centrifuged at 13,000 rpm for 10 min to remove particulate matters; 100 μl of supernatant was transferred to the injection vials; and 20 μl of aliquot was injected onto the chromatographic column.
For all tissue, tissues to buffer partitioning ratios were calculated as follows: For water content subtracted tissue, the tissue concentration was calculated as follows:
The tissue water content estimated includes intra- and extracellular water. For the purpose of the above-mentioned calculations, the concentration of the drug in tissue water was assumed to be the same as the drug concentration in the buffer at the end of incubation period.
All data in this study are expressed as the mean ± S.D. Comparisons between in vitro tissue partitioning ratios in different tissues, different β-blockers, and tissue status were performed with an analysis of variance followed by the Tukey's post hoc analysis or using Student's t test, where appropriate. Statistical significance was set at P < 0.05.
The log D of β-blockers between pre-equilibrated PBS, pH 7.4, and n-octanol were measured using the shake flask method. A mixture of β-blockers was used for this purpose. The distribution coefficient, D, was estimated as the ratio of drug concentration between n-octanol and aqueous phases. Physicochemical properties of β-blockers including log D values estimated in this study and literature values of log D measured individually for each β-blocker (Schoenwald and Huang, 1983; Pitkänen et al., 2005; Wiczling et al., 2006) are summarized in Table1. The log D values obtained from cocktail analysis have closely matched with the literature values with individual β-blockers (R 2 = 0.982).
Tissue Water Content.
The water content in bovine eye sclera, choroid-RPE, retina, optic nerve, and trabecular meshwork was estimated after tissue drying. The water content in wet tissue, expressed as percentage of water content, is summarized in Table 2. It is evident that the water content in ocular tissues is in the following order: sclera < optic nerve < choroid-RPE < trabecular meshwork ∼ retina.
In Vitro Tissue/PBS Partition Coefficients of β-Blockers.
In vitro tissue partitioning studies were performed to determine the influence of tissue type, drug lipophilicity, tissue status or treatment (intact versus homogenized, wet weight method versus dry weight method, or correction for water content and drug in tissue water in intact tissue using wet weight method), and drug concentration on partitioning of drug molecule in various ocular tissues (Tables 3 ⇓–5; Fig. 1).
Irrespective of the method of estimation, when intact tissues were used for 5 μg/ml β-blocker incubation, as shown in Fig. 1, tissue partitioning ratios of β-blockers in all five ocular tissues were generally in the order choroid-RPE > trabecular meshwork > retina > sclera ∼ optic nerve. Similar rank order was evident at other concentrations of β-blockers as well (Tables 3 and 4). When intact tissue, wet weight method was used, the tissue partitioning of highly lipophilic propranolol in choroid-RPE was 5-, 8.8-, 9.6-, and 9.8-fold higher than trabecular meshwork, optic nerve, retina and sclera, respectively. Similarly, partitioning of hydrophilic β-blocker sotalol in choroid-RPE was 1.8-, 5.0-, 6.8-, and 7.8-fold higher than trabecular meshwork, optic nerve, retina, and sclera, respectively. In intact tissue, dry weight method and intact tissue, water and drug in tissue water correction methods, whereas the trend of tissue partitioning was similar to the intact tissue wet weight method, the partitioning ratios were severalfold higher (Fig. 1, B and C). When homogenized tissues were used, sclera showed slightly more tissue partitioning for hydrophilic β-blockers compared with the homogenized retina and optic nerve (Table 5). Furthermore, optic nerve and retina partition coefficients could not be well distinguished in the homogenized tissue method.
Lipophilicity (log D) correlated positively with the tissue partitioning in all tissues (Fig. 2). When estimates were made using intact tissues, tissue partitioning coefficients were generally in the following order for β-blockers: propranolol > betaxolol > metoprolol ∼ timolol ∼ pindolol > nadolol ∼ atenolol ∼ sotalol (Fig. 3). The lipophilic β-blockers showed higher tissue partitioning in all five tissues compared with the hydrophilic β-blockers. Using wet weight, intact tissue method, the tissue partitioning of propranolol was approximately 50.1-fold more in choroid-RPE compared with sotalol.
As can be seen in Fig. 3, the tissue partitioning ratios for lipophilic β-blockers were generally higher in intact tissues compared with homogenized tissues. Among intact tissue methods, the partition coefficients were in the order: intact tissue, dry weight method ≥ intact tissue, wet weight method corrected for tissue water and drug in tissue water > intact tissue, wet weight method. Tissue partitioning ratios of water content subtracted tissues were slightly lower than dry tissue for hydrophilic drugs but approached the values of dry weight method for lipophilic drugs (Fig. 3). For propranolol, the partition coefficient was 2.6-fold higher for sclera, 3.6-fold higher for choroid-RPE, 2.7-fold higher for retina, and 1.5-fold higher for optic nerve in intact tissue, wet weight method compared with homogenized tissue, wet weight method (Fig. 3). In trabecular meshwork, there was no significant difference in partition coefficients between the intact, wet weight method and homogenization method (Fig. 3E).
The starting drug concentration has significant influence on tissue partition coefficients in trabecular meshwork. The partition coefficients decreased with increasing solute concentration (Fig. 4E). In intact sclera and choroid-RPE (wet weight), for hydrophilic β-blockers and propranolol there is no clear influence of drug concentration on tissue partition coefficients, whereas for other lipophilic β-blockers there was a decrease in partition coefficients with increase in concentration (Fig. 4, A and B). In choroid-RPE, these differences were more prominent in the homogenized tissue compared with the intact tissue. For retina and optic nerve, there was no significant effect of drug concentration on the observed tissue partition coefficients (Fig. 4, C and D).
Melanin Binding of β-Blockers.
Affinity of β-blockers to natural melanin (S. officinalis) was determined. The cocktail solution of β-blockers in PBS was incubated with natural melanin at various ligand concentrations to determine the binding parameters. The maximal number of moles of solute bound per milligram of melanin (r max) and k values are summarized in Table 6. Propranolol shows the highest binding to natural melanin followed by pindolol at all ligand concentrations. Propranolol and pindolol exhibit both high binding affinity and high binding capacity, whereas atenolol, nadolol, timolol, and betaxolol exhibit high affinity but low binding capacity. Metoprolol and sotalol have exhibited both low binding affinity and capacity to natural melanin. The binding capacity to natural melanin exhibited a positive correlation with the tissue partitioning in intact choroid-RPE (R 2 = 0.493) as well as in homogenized choroid-RPE (R 2 = 0.536) (Fig. 5, A and C). With an increase in melanin binding affinity, there was an increase in tissue partitioning in both intact (R 2 = 0.208) and homogenized choroid-RPE (R 2 = 0.545) (Fig. 5, B and D). Log D correlated positively with melanin binding capacity (R 2 = 0.305) (Fig. 5E). With an increase in log D, melanin binding affinity tended to increase, although the correlation coefficient was low (R 2 = 0.105; Fig. 5F).
In Vivo Tissue Distribution of β-Blockers in Rat Model.
The concentrations of β-blockers in various eye tissues in BN rats at 1 h after subconjunctival injection are shown in Fig. 6. As evident from Fig. 6, tissue distribution was the highest in melanin containing tissues, i.e., choroid-RPE and iris-ciliary body, for all β-blockers. The general rank order for distribution of β-blockers in various tissues was iris-ciliary body > choroid-RPE > sclera > cornea > retina > vitreous > lens. These differences and rankings were most prominent for lipophilic solutes, with the relative differences between tissues diminishing for hydrophilic solutes. With an increase in lipophilicity of the β-blockers, in vivo drug distribution to sclera, choroid-RPE, cornea, and iris-ciliary body increased positively. Of these tissues, the dependence on the lipophilicity was most prominent in pigment-rich choroid-RPE and iris-ciliary body (Fig. 7).
Of the tissues analyzed in vivo in the rat model, we have comparative bovine in vitro tissue/PBS partition coefficients for sclera, choroid-RPE, and retina. Therefore, we investigated whether the bovine in vitro partition coefficients correlate with the in vivo tissue/vitreous drug distribution ratios for sclera, choroid-RPE, and retina in the rat model. It is interesting to note that in vitro tissue partitioning coefficients of β-blockers in bovine sclera, choroid-RPE, and retina to PBS, pH 7.4, correlated positively with in vivo tissue/vitreous distribution ratios for sclera, choroid-RPE, and retina in Brown Norway rats. The correlation coefficients for these relationships were high, with R 2 values being ≥0.985 (Fig. 8). Thus, it appears that in vitro tissue partition coefficients might be useful in predicting in vivo tissue distribution ratios.
This is the first study to report 1) the influence of lipophilicity of drug molecules on tissue partitioning in various eye tissues including sclera, choroid-RPE, retina, optic nerve, and trabecular meshwork; and 2) a correlation between in vitro tissue partition coefficients and in vivo tissue distribution after trans-scleral drug delivery. Specifically, partition coefficients of a series of β-blockers were measured at three different concentrations using intact tissue or homogenized tissue and in vivo tissue distribution was measured in BN rats after posterior subconjunctival administration.
In all tissues assessed, the rank order for tissue partitioning was as follows: intact, dry weight method ≥ intact, water content and drug in water subtracted wet weight method > intact, wet weight method > homogenized, wet weight method (Fig. 3; Tables 3 ⇑–5). Higher partition coefficients in intact tissues compared with the homogenized tissues can be explained on the following basis. With homogenization, it can be anticipated that intracellular drug accumulation is absent or not accounted for, and therefore the lower partition coefficients can be explained. It should be noted that basic drugs tend to accumulate in acidic compartments within the cell. Indeed, Rodgers et al. (2005) observed greater distribution of β-blockers in lung tissue compared with adipose and skin tissues. These differences were explained on the basis of greater quantity of lysosomes in lung tissue compared with the other tissues. Because all β-blockers used in this study are basic drugs with a pK a value of 9.3 to 10.3, they are all charged positively at physiological pH as well as lysosomal pH, with the ionization being greater at acidic lysosomal pH. Thus, a significant quantity of β-blockers can be present in the cellular fluid compartments, accounting for greater partition coefficients in intact, wet weight method compared with homogenized tissue. We observed that the partition coefficients are much higher with the dry weight method. This can be anticipated because lower tissue weight devoid of water content is used in the denominator for estimating tissue partition coefficients. As an additional measure, we determined tissue partition coefficients using the wet weight approach after subtracting the measured tissue water contents and the drug amount in tissue water. The tissue water content for our tissue isolation methods was in the order retina ∼ trabecular meshwork > choroid-RPE > optic nerve > sclera (Table 2). This approach essentially allowed us to measure drug levels in the solid tissue, without including drug levels in tissue water. Partition coefficients obtained using these approaches were slightly less in most cases and equal in a few cases to what was observed with the dry weight method (Fig. 3). The differences were significant for hydrophilic β-blockers and not lipophilic β-blockers. This is consistent with greater relative contribution of drug in extracellular or intracellular water to the partition coefficients of hydrophilic β-blockers. We investigated partition coefficients using wet weight as well as dry weight methods, presuming that one of these methods might be more reproducible than the other. However, we observed similar coefficients of variation for the two methods.
Irrespective of the method of assessment, lipophilic β-blockers have much higher tissue partitioning in all five ocular tissues compared with hydrophilic β-blockers (Fig. 1) and the measured log D of β-blockers correlated positively with log (tissue partition coefficients) (R 2 = 0.7–0.9; Fig. 2A). Higher tissue partition coefficients for lipophilic molecules might be a result of greater drug binding to lipid bilayer, enhanced drug entry into cells, and drug binding to proteins or other macromolecules.
Of the five ocular tissue studied, intact tissue partitioning coefficients followed the order; choroid-RPE > trabecular meshwork > retina > sclera ∼ optic nerve (Fig. 1, A–C). Physiologically, a major difference between choroid-RPE and other four tissues is that choroid-RPE is a melanin rich tissue (Cheruvu et al., 2008). The β-blockers are known to bind to melanin, potentially explaining their preferential accumulation in the iris and ciliary body after topical application as eye drop (Abrahamsson et al., 1988; Maurice, 2002). Earlier reports indicated that basic drugs and lipophilic molecules preferentially bind to melanin (Zane et al., 1990; Leblanc et al., 1998). Pitkänen et al. (2005) have reported that choroid-RPE permeation lag time increases with an increase in lipophilicity. In vivo distribution studies have shown that, tissue/plasma partition coefficients of drug molecules in several other tissues of the body increase with an increase in lipophilicity (Poulin and Theil, 2000). All β-blockers used in this study are ionized and positively charged almost completely (98.8–99.9%) at the pH assessed, so we assumed that lipophilicity discriminates the binding of these drugs to melanin.
Although melanin binding capacity (R 2 = 0.3) as well as affinity (R 2 = 0.1) increased with increasing drug lipophilicity (Fig. 5, E and F), the correlations were poor. Correlations were worse with pK a and percentage of ionization of the β-blockers, because these parameters were distributed more narrowly (Table 1). One possible reason for the lack of dependence on lipophilicity is the role of the composition of the medium in melanin binding. Drug-melanin interactions can be influenced by ionic composition of the incubation medium. For example, chloroquine, a lipophilic, positively charged drug binds to melanin through ionic interactions as well as hydrophobic interactions (Stepień and Wilczok, 1982). The degree of binding of chloroquine can be influenced by the concentration of Na+ in the medium. Similar to chloroquine, β-blockers used in this study are positively charged. Thus, we can anticipate that ionic composition influences β-blocker binding to melanin. However, this possibility remains to be assessed. It is noteworthy that although log D correlated very well with choroid-RPE partition coefficient with intact tissue methods (R 2 > 0.9; Fig. 2), this relationship became poor once choroid-RPE was homogenized, thereby releasing melanin (R 2 = 0.4). Thus, although the mechanisms of β-blocker interactions with melanin are unclear, the natural melanin binding studies indicating poor correlation with log D might be somewhat reflective of melanin binding in the actual tissue.
The concentration of β-blockers can influence the tissue partition coefficients. In the range assessed, the concentration of β-blockers have a significant influence on tissue partitioning (intact, wet weight method) in trabecular meshwork, choroid-RPE, and sclera, with partition coefficients decreasing for all or some of the β-blockers with an increase in drug concentrations (Table 3). As shown in Fig. 4E, partition coefficients in trabecular meshwork are significantly lower at 5 and 25 μg/ml compared with 0.5 μg/ml concentration of β-blockers. Similar results were also observed with choroid-RPE and sclera, with the differences being significant for pindolol, timolol, metoprolol, and betaxolol (Fig. 4, A and B). In homogenized choroid-RPE, these differences were more prominent (Table 5), possibly due to enhanced exposure of melanin to the β-blockers. In the range assessed, drug concentration had no statistically significant effect on tissue partition coefficients in optic nerve and retina in both intact (Fig. 4, C and D) and homogenized tissues. Thus, potentially due to saturable binding process in the range of concentrations assessed, a decrease in partition coefficients was observed in trabecular meshwork, choroid-RPE, and sclera.
In this study, we applied a cassette dosing approach for assessing trans-scleral drug delivery in a rat model. It is well known that ocular pharmacokinetic studies are animal intensive requiring sacrifice of approximately 36 animals (four animals per time point and nine time points) to obtain time course data for each compound. Although microdialysis experiments are well established for vitreous humor drug levels (Majumdar et al., 2006; Hosoya et al., 2009), currently there are no reliable methods available to quantify drug levels in other ocular tissues in a continuous manner. Furthermore, trans-scleral drug delivery is not well understood in terms of influence of drug properties on target tissue delivery. To overcome these limitations, we are taking a two pronged approach in this study. 1) We are beginning to establish the usefulness of tissue/buffer partition coefficients as predictors of in vivo tissue/vitreous ratios of drug. 2) To begin establishing drug properties suitable for enhanced trans-scleral delivery, we performed a cassette dosing study with eight β-blockers and used a 1-h time point for drug analysis. The peak levels of retinal drug absorption for small molecules are typically observed at approximately 1 h in trans-scleral drug delivery studies (Cheruvu et al., 2008).
Similar to any other high throughput method, cassette dosing is fraught with its limitations. These include potential interactions between drugs for a common binding site (e.g., in melanin rich tissues in the present study), metabolic enzyme, or a transporter (Smith et al., 2007). Although lipophilicity could be correlated with drug partitioning and delivery in this study, we cannot rule out the dependence of β-blocker delivery or tissue uptake on other factors. For example, β-blockers might be transported via transporters such as P-glycoprotein, organic cation transporters, and ATP-binding cassette G2/breast cancer resistance protein transporter that are known to be present in the retinal pigment epithelium (Mannermaa et al., 2009). In this study also, we cannot rule out such possibilities for drug interactions. Because we collected in vitro partition coefficient data as well as in vivo tissue distribution data using a cocktail of β-blockers, similar interactions might be present in both in vitro and in vivo scenarios, thereby enhancing the potential value of our in vitro-in vivo correlations in this regard. Tools such as the ones described in this study might be of value in the discovery phase of drug development. That is, identification of suitable candidate drugs for further development from among a series of molecules might be enabled by a cassette dosing approach. Although we assessed the influence of drug lipophilicity on in vitro tissue partitioning and in vivo tissue distribution after trans-scleral delivery, it should be borne in mind that lipophilicity is only one of several factors that need to be kept in mind as lead compounds are identified in drug discovery. Other parameters that need to be considered include drug efficacy and toxicity profiles.
The molecules used in current cassette study are β-blockers. Rodgers et al. (2005) indicated that there are no significant differences in pharmacokinetic behavior of β-blockers when administered alone or as a cassette. There is one published report on cassette dosing approach for ocular pharmacokinetic studies. In this recent study, intravitreal mode of drug administration was used for 18 drugs, administered in four cassettes containing four or five molecules per cassette (Proksch and Ward, 2008). One of these cassettes was further validated using a single drug dosing strategy. Their results indicated similar rank order for drug distribution between single drug dosing and cassette dosing strategies. In addition, preliminary reports exist indicating similar tissue distribution trends for drugs administered as a single drug or in a cassette after subtenon mode of administration for trans-scleral drug delivery in rabbits (Gale et al., 2005). Thus, cassette-dosing approach is probably of value in drug discovery for the identification of lead compounds with suitable tissue distribution.
A key finding made in this study using cassette dosing and trans-scleral delivery is that tissue partition coefficients as well as drug lipophilicity can be predictive of in vivo drug distribution during trans-scleral delivery. A drug administered periocularly for trans-scleral delivery to the retina can potentially access back of the eye tissues as well as those in the front of the eye. The β-blockers assessed in this study reached all eye tissues within 1 h after periocular administration (Fig. 6). It is interesting to note that the in vitro bovine tissue/PBS partition coefficients correlated very well (R 2 = 0.99) with in vivo tissue/vitreous drug distribution ratios for sclera, choroid-RPE, and retina (Fig. 8). Thus, interspecies correlations as well as in vitro-in vivo correlations are probably feasible for trans-scleral drug delivery. In addition, the in vivo tissue drug levels increased with an increase in log D for iris-ciliary body, choroid-RPE, sclera, and cornea (Fig. 7A). In addition, drug levels in these tissues exhibited a clear positive trend with an increase in lipophilicity (Fig. 6). Furthermore, possibly due to drug retention in the choroid-RPE for more lipophilic solutes, retinal and vitreal tissue levels declined with an increase in log D (Fig. 7B). The drug levels in retina and vitreous were the lowest for propranolol, the most lipophilic drug assessed (Fig. 6). This is consistent with our earlier observations with trans-scleral delivery of celecoxib, wherein retinal celecoxib delivery was lower in pigmented rats compared with albino rats (Cheruvu et al., 2008).
In this study, we have for the first time demonstrated quantitatively the partitioning of drug molecules in trabecular meshwork and optic nerve. Lipophilicity has a greater influence on drug accumulation in trabecular meshwork compared with optic nerve (Fig. 2). However, in each of these tissues, an increase in lipophilicity results in an increase in drug accumulation. Based on these findings, it is likely that the accumulation of therapeutically beneficial neuroprotective agents in optic nerve can be increased by increasing drug lipophilicity. Conversely, the trabecular meshwork accumulation of corticosteroids responsible for intraocular pressure elevation (Smithen et al., 2004; Jones and Rhee, 2006; Kubota et al., 2006; Fan et al., 2008) can potentially be reduced by choosing less lipophilic corticosteroids.
Although the primary purpose of this study is to begin understanding the relationship between drug lipophilicity and tissue partitioning, melanin binding, and trans-scleral drug delivery, the compounds used in this study may have therapeutic potential in the back of the eye (Wood et al., 2003; Osborne et al., 2005; Chen et al., 2007).
In summary, drug partitioning is greater in intact tissues compared with homogenized tissue, possibly due to significant intracellular accumulation of β-blockers. Solute lipophilicity has significant influence on tissue partitioning. Lipophilic molecules have more affinity to isolated tissues and therefore higher tissue partition coefficients compared with hydrophilic molecules. The highest tissue partitioning occurs in choroid-RPE for all eight β-blockers, compared with sclera, retina, trabecular meshwork, and optic nerve. The β-blockers have significant binding affinity to melanin, which is a possible explanation for the highest partitioning of these drugs in choroid-RPE. In vivo tissue distribution of β-blockers in Brown Norway rats shows more accumulation of drug in choroid-RPE and iris-ciliary body and the tissue/vitreous distribution ratios for sclera, choroid-RPE, and retina correlate well with the in vitro bovine tissue partition coefficients. Thus, along with the solute lipophilicity, binding of drugs to macromolecules such as proteins and melanin present in tissue might determine tissue partitioning in various ocular tissues. The tissue partition coefficients and melanin binding properties estimated in this study are expected to be of value in choosing drug molecules with appropriate properties to target or avoid a given eye tissue. In particular, selection of less lipophilic solutes might be beneficial in avoiding trabecular meshwork and any associated side effects. Similarly, selection of a less lipophilic solute might reduce nonproductive binding to choroid-RPE, thereby allowing greater trans-scleral delivery of the solute to the retina and vitreous.
- Received September 11, 2009.
- Accepted November 18, 2009.
This work was supported by National Institutes of Health National Eye Institute [Grant EY017533]; and by an unrestricted research gift from Pfizer Global Research.
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
- retinal pigment epithelium
- phosphate-buffered saline
- distribution coefficient
- liquid chromatography/tandem mass spectrometry
- Brown Norway.
- Copyright © 2010 by The American Society for Pharmacology and Experimental Therapeutics