Minocycline Protects Photoreceptors from Light and Oxidative Stress in Primary Bovine Retinal Cell Culture

Abstract

purpose. To determine whether minocycline, a compound known to protect the retina against light-induced damage in rodent models, and its structurally related analogues would protect photoreceptor cells in primary bovine retinal cell culture against light and oxidative stress.

methods. Minocycline and its analogues were tested in primary retinal cell culture to see whether they would inhibit light or oxidative stress–induced cell death. Primary cell cultures composed of photoreceptors, bipolar cells, and glial cells were prepared from bovine retinas. The extent of cell death induced by light or oxidative stress was assessed by using Sytox Green (Invitrogen-Molecular Probes, Eugene, OR) a nucleic acid dye uptake assay. Differential protection of photoreceptor cells from stress were examined using immunocytochemistry.

results. Minocycline and methacycline were cytoprotective against light- or oxidative stress–induced damage of bovine primary photoreceptors in culture with an EC₅₀ < 10 μM. In contrast, structurally related analogues such as demeclocycline, meclocycline, and doxycycline were phototoxic at >3 to >10 μM. Though demeclocycline was found to be phototoxic, it was cytoprotective (EC₅₀ = 5 μM) against oxidative stress in the absence of exposure to light.

conclusions. The protective action of minocycline against light-induced damage in the cell-based assays agrees with earlier reports in animal models and suggests that the in vitro assay using bovine primary retinal cell culture is a suitable model for evaluating compounds for retinal protection. Cellular protection or toxicity produced by structurally related compounds show that minor structural modifications can alter the function of minocycline and lead to potent retinal protective compounds.

Minocycline is a semisynthetic derivative of tetracycline with improved tissue adsorption into the central nervous system (CNS) and with a longer half-life. In addition to its antibiotic activity, minocycline has anti-inflammatory and antiapoptotic activities and is effective in delaying disease progression in numerous models of neurodegeneration. ¹ Visual images focused on the retina are sensed by rod and cone photoreceptors. Owing to their high content of mitochondria, photosensitizers, and pigments; high metabolic rates; and rapid responses to small changes in light, photoreceptors are particularly susceptible to excessive exposure to light and oxygen, ^{2 3 4 5 6 7 8} and they are the first retinal cell type to show signs of damage after light-induced stress. ⁹ Minocycline has been found to be protective of photoreceptors from light-induced damage in BALB/cJ mice and in rats. ^{10 11} It also delays photoreceptors degeneration in the rds mouse, which is homozygous for the null mutation of the gene peripherin 2, which is essential for the formation and maintenance of normal photoreceptor outer segments. ¹² Based on the reduction of immunolabeling of CD11b of microglial cells found in the retina, it has been suggested that minocycline may protect against light-induced loss of photoreceptors indirectly, through the inhibition of retinal microglial activation. ¹⁰ This hypothesis is consistent with the finding that conditioned medium from activated microglial cells can induce apoptosis in a transformed photoreceptor cell line, 661W. ^{13 14} Based on its anti-inflammatory activity, minocycline has been suggested for the treatment of age-related macular degeneration (AMD). ¹⁵ An earlier study, ¹² however, indicated that the delay of photoreceptor death by minocycline is independent of the reduction of microglial cells in the retina, as depletion of microglial cells using liposomal clodronate instead of minocycline showed no protection of photoreceptor apoptosis in the rds mouse model. Although there are promising results of minocycline as a neuroprotective agent, it has also been shown to have variable or even contradictory results in different models of neurodegeneration in different species. ^{16 17 18 19 20} To determine whether minocycline can protect photoreceptors in an in vitro system derived from primary retinal cell culture, we studied the effect of minocycline and its analogues on light or tertiary butyl hydroperoxide (tBOOH)–induced damage of photoreceptors in bovine primary retinal cell cultures. 

Materials and Methods

Cell Cultures

We prepared retinal cell suspensions from bovine retinas by using a procedure described previously. ²¹ Briefly, fresh bovine eyes (2 weeks old) were obtained from Lampert Co. (Duvall, WA). After a rapid rinse in ethanol, the cornea, lens, and vitreous were removed. The retina was detached from the eye cup and chopped into small fragments in HBSS++ solution (Cambrex Bio-Science Walkersville, Inc., Walkersville, MD). Retinal fragments were incubated for 15 minutes at 37°C in papain solution (20 U/mL; Worthington Bioscience, Worthington, MN). The enzyme reaction was stopped by adding 2 mL fetal bovine serum (FBS; Invitrogen, Carlsbad, CA), and tissue aggregates were eliminated by adding 200 μL of DNase I (Sigma-Aldrich, St. Louis, MO). The cell suspension was then centrifuged at 230g for 5 minutes. The cells were resuspended in defined DMEM/F12 plus insulin (25 μg/mL), transferrin (0.1 mg/mL), sodium selenite (5.18 ng/mL), and putrescine (9.66 μg/mL), and supplemented with 10% FBS. The bovine retinal suspensions were plated directly in 96-well flat-bottomed tissue culture plates at a density of 5 × 10⁵ cells/well. Plated cells were incubated in a 5% CO₂ incubator at 37°C. 

Light-Induced Damage and Oxidative Stress

Bovine retinal cells were pretreated with different concentrations of tetracycline and its analogues between 1 and 36 μM for 16 hours. The step for pretreatment of cells with test compounds before stress was included as a general protocol for screening neuroprotective compounds to allow time for certain compounds that may function indirectly by inducing protective proteins such as those from phase 2 genes. ²² In the case of minocycline, the extent of protection from light-induced stress was similar whether minocycline was added 16 hours beforehand or immediately before light-induced stress (data not shown). The minocycline analogues were initially obtained as part of the Spectrum Collection of 2000 compounds (Microsource Discovery, Gaylordsville, CT) supplied as 10 mM stock solution in dimethyl sulfoxide (DMSO). A larger amount of each analogue in powder form (Sigma-Aldrich) was subsequently purchased for further studies. Compounds were prepared from 10 mM stock solutions in DMSO followed by serial dilutions into growth media. Cells were then exposed to continuous blue-actinic light (Super Actinic/03 Fluorescent lamps, Philips Lighting Co., Somerset, NJ) set at 500 lux and 1.1 mW/cm², with a spectral range from 390 to 460 nm and λ_max = 420 nm, or to white fluorescent light (F15T8CW; Ushio, Tokyo, Japan) set at 8000 lux, 2.7 mW/cm², with spectral range from 400 to 650 nm and λ_max = 570 nm for up to 24 hours. The intensity and the spectral ranges of the light sources were determined using a photometer (model IL 1400A; International Light, Newburyport, MA). Three blue-light tubes or four white-light tubes were mounted on the topside in each incubator kept at 37°C and 5% CO₂. A maximum of 15 96-well microtiter plates can be placed at the bottom of the each incubator in our setup. Owing to the heat generated from the light tubes, the temperature of incubators were set at 30°C and at 32°C for the blue- and white-light incubators, respectively, to attain a temperature of 37°C at the bottom of the chamber. Based on monitoring using the photometer, the lux recorded can vary by ≤10% between plates placed at the center and at a corner of the incubator. A variance of ≤10% in light intensity was found not to affect significantly the extent of light-induced damage among plates placed at various positions in the incubator. Control cells were taken from the same batch as the ones used for the light-induced damage and were kept in the incubator in the dark. For oxidative stress, cells were treated with 0.8 mM tBOOH for 20 hours. 

Sytox Green Assays

Sytox Green (Invitrogen-Molecular Probes, Eugene, OR) was used to detect stress-induced cell death and to identify compounds that can protect retinal cells from light-induced cell death or oxidative stress. Sytox Green is a DNA-binding dye that would only get into dying cells where the plasma membranes are compromised. The dye becomes fluorescent on binding to nuclear DNA. The overall fluorescence (F*) of the cells is inversely related to the intactness of the cell membranes. As the extent of cell membrane damage, and hence the F* intensity, varies among different cells and with the duration of exposure to light, F* intensity of nuclear DNA bound Sytox, rather than the percentage of cells with permeable membrane, was used to assess the viability of cells in culture and the estimation of the EC₅₀. Sytox was added directly to the cell cultures to a final concentration of 0.8 μM without changing the medium and incubated for 1 hour at 37°C. Changing the medium was found to reduce the F* readings of samples treated with most phototoxic compounds, possibly due to removal of the nonattached dead cells. F* was read with a plate reader with excitation at 485 nm and emission at 528 nm. Compounds that would lower Sytox fluorescence in the presence of light or oxidative stress are regarded as putative candidates for cytoprotection. Cell cultures were subsequently examined using immunocytochemistry for differential protection of photoreceptor cells. Sytox Green was found to be a nonfixable dye, and processing the cells for immunocytochemistry analysis led to redistribution of the Sytox dye into all cells. For observing the extent of cell death after immunostaining, an amine-reactive stain (Live/Dead Fixable Dead Cell Stain; Invitrogen-Molecular Probes) was used in lieu of Sytox. In cells with compromised membranes, this dye reacts with free amines both in the cell interior and on the cell surface, yielding intense fluorescent staining relative to viable cells, where the dye’s reactivity is restricted to cell-surface amines. Concentration-dependent studies were performed to compare the relative efficacy of compounds that show cytoprotection for photoreceptors. With coefficient of variation (CV) < 15% among the replicate samples, three replicates were used, as recommended by the NIH Chemical Genomics Center (NCGC) guidelines to get a mean CV of <10%. The percentage of inhibition for neuroprotective compounds can be calculated using the following formula:  

\[\%\mathrm{inhibition}{=}\ \frac{F{\ast}\mathrm{light}{-}F{\ast}\mathrm{compound}}{F{\ast}\mathrm{light}}{\times}100\%.\]

For phototoxic compounds, the percentage of potentiation of toxicity can be calculated using the following formula:  

\[\%\mathrm{potentiation}{=}\ \frac{F{\ast}\mathrm{compound}{-}F{\ast}\mathrm{light}}{F{\ast}\mathrm{light}}{\times}100\%.\]

where F* compound is the F* units of samples treated with test compound+light, and F*light is the F* units of the light-alone samples. The average F* values of no-light control samples were subtracted from all F* values in the equations. The EC₅₀ for neuroprotective or phototoxic compounds were determined based on the concentration of the compound required to give 50% inhibition or potentiation of Sytox F*, respectively. The EC₅₀ was calculated using a curve-fitting program for variable-slope, sigmoidal dose responses (Prism 4; GraphPad software, Inc, San-Diego, CA). For compounds such as minocycline and methacycline that showed a biphasic response, the data points at 36 μM were excluded in the calculation of EC₅₀. 

As Sytox is n F*-based assay, compounds that absorb or fluoresce near the λ_max of Sytox absorbance or emission are also expected to affect the Sytox reading and can represent a source of false-positive or -negative candidates. Accordingly, an additional Sytox assay was implemented wherein Sytox was added to a microtiter well containing the test compound and 25 ng of naked DNA in growth medium, but with no cells in it. If a test compound gives a significantly different F* reading compared with no compound control, it would represent a false-positive or -negative candidate that interferes with Sytox fluorescence reading. Although minocycline and its analogues appear to be slightly yellow at high concentration (10 mM), no interference with the Sytox fluorophore reading was observed at up to 100 μM; whereas other intensely colored compounds such as FD&C Yellow, FD&C Red, cresyl violet, and Evan’s blue showed more than 50% inhibition of Sytox F* at 30 μM. 

Immunocytochemistry

Immunocytochemistry was performed directly on cells grown in 96-well plates. The cell cultures were washed with PBS, fixed with 4% paraformaldehyde solution for 10 minutes at room temperature, and then bathed with cold 95% ethanol for 10 minutes at 4°C. After they were washed with PBS, the cell cultures were incubated with primary antibodies in 0.4% Triton X-100 and 10% goat serum for 1 hour at 37°C. Different antibodies were used to identify different cell populations—namely, a monoclonal antibody against rhodopsin (1:2000 dilution; Chemicon International, Inc., Temecula, CA) for rod photoreceptors; a polyclonal antibody against recoverin (1:3000 dilution; Chemicon International, Inc.) for cone and rod photoreceptors ²¹ ; a polyclonal (1:1000 dilution; Upstate, Charlottesville, VA) or monoclonal (1:1000 dilution; Chemicon International, Inc.) antibody against protein kinase C (PKCα) for rod bipolar cells ²³ ; a monoclonal antibody against vimentin (1:1000 dilution; Chemicon International, Inc.) for Müller (glial) cells; and a monoclonal antibody against CD68 for macrophages (microglial cells; 1:1000 dilution; Chemicon International, Inc.). After the cells were washed with PBS, the binding of a rabbit-derived primary antibody was detected using a goat anti-rabbit secondary antibody (1:250 dilution) conjugated with Alexa-594 for red emission, whereas the binding of a monoclonal primary antibody was detected using a goat anti-mouse secondary antibody (1:250 dilution) conjugated with Alexa-680 for infrared emission (Invitrogen-Molecular Probes, Inc.). 4′,6′-Diamino-2-phenylindole (DAPI) used for nuclei staining was mixed with secondary antibodies and incubated for 1 hour at room temperature. 

For double immunofluorescence labeling, cells were incubated with a mixture of mouse anti-rhodopsin antibody and rabbit anti-recoverin antibody, or with rabbit anti-recoverin antibody and mouse anti-PKCα antibody. After they were washed, the cells were incubated with a mixture of corresponding Alexa-594– or -680–conjugated secondary antibodies. Negative control experiments were performed by incubation with the secondary antibody only. The cells were washed and mounted (Fluoromount-G; SouthernBiotech, Birmingham, AL). Fluorescence labeling was observed with a microscope (Olympus, Tokyo, Japan) with epifluorescence illumination. All images were recorded with the aid of a digital CCD camera (Hamamatsu Photonics KK, Hamamatsu City, Japan) and the images were processed (MetaMorph software; Universal Imaging Corp. Downingtown, PA) at 200× magnification. A minimum of three repeat experiments from different sets of culture were performed for light-induced cell death of retinal cells based on Sytox readings and immunocytochemistry observation. 

Automated Cell Counting

For cell counting, imaging was performed using an automated imaging microscope (Cell/Cell Imaging Station for Life Science Microscopy; Olympus, Tokyo, Japan) and the cell counting was performed using a proprietary image based cell-counting software developed at Acucela (Bothell, WA). The cell-counting algorithm was based on scoring cells that were positive for both nuclei staining with DAPI and for immunolabeling with recoverin. A total of 16 to 24 views (covering an area of 2.3–3.5 mm²) were taken for each well. Because of the time involved in image capturing and data processing, only selected samples were analyzed. The major cell type quantified was the recoverin-positive photoreceptors. The percentage of protection of photoreceptors for a given compound against light-induced stress can be calculated using the following formula:  

\[\%\mathrm{protection}{=}\ \frac{#\mathrm{compound}{-}#\mathrm{light}}{#\mathrm{dark}{-}#\mathrm{light}}{\times}100\%.\]

In the formula, #compound is the number of photoreceptors remaining after light-induced stress with a specific compound, #light is the number of photoreceptors remaining after light-induced stress alone, and #dark is the number of photoreceptors without exposure to light. The EC₅₀ for a given neuroprotective compound were estimated based on the concentration of the compound necessary to give 50% protection. For compounds that are phototoxic, negative values obtained can be converted to absolute values to give the percentage potentiation of light-induced damage. The EC₅₀ obtained based on automated photoreceptor cell counting was compared with the EC₅₀ obtained based on Sytox F*, to assess the correlation between these two independent assays. The extent of correlation was determined by linear regression (GraphPad Software, Inc.) using the same datasets analyzed by Sytox reading and by photoreceptor counting. 

Caspase Activation

After or without exposure to light, retinal cells pretreated with or without test compounds were monitored for caspase activation (Image-iT Live Green caspase detection kits; Invitrogen-Molecular Probes). These reagents use a carboxyfluorescein (FAM)–labeled DEVD (specific for caspases-3 and -7) or the VAD (specific for caspase-1 and -3 to -9) fluoromethyl ketone (FMK) caspase inhibitor that can covalently link to activated caspases through cysteines. 

Fluorescence Detection of Reactive Oxygen Species

Cells with or without pretreatment with minocycline or N-acetylcysteine were loaded in the dark with 1 μM the acetyl ester of 5-chloromethyl-2′,7′-dichlorodihydrofluorescein diacetate (H₂DCFDA; Invitrogen-Molecular Probes) for 1 hour before exposure to blue light for 1 to 6 hours. After the H₂DCFDA was washed off in the media, the intracellular F* was then monitored using an F* plate reader with a 485/525-nm band-pass filter for excitation and detection. 

Results

Characterization of Primary Bovine Retinal Cells in Culture

The cells used in this study were primary cultures of mixed bovine retinal cells. Photoreceptor cells survived the cell dissociation procedure, grew neurites, and were viable in culture for about 1 month. Besides Müller cells, which form feeder layers, immunolabeling with rhodopsin (marker for rods) or recoverin (a marker for rod and cone photoreceptors) antibodies showed photoreceptors to be the major population of neuronal cells with approximately 65% rods and 35% cones. Rod bipolar cells immunolabeled with PKCα antibody were found to be the second most abundant neuronal cell type in this culture. Anti-PKCα is specific for rod bipolar cells, whereas anti-PKC-α, -β, and -γ can be use to label all bipolar cell types. ²³ All the bipolar cells were double labeled with PKC-α, -β, and -γ or with PKC-α alone (data not shown), suggesting that the bipolar cell population in bovine retinal cell culture is mostly rod bipolar cells. No microglial cells were detected in the cell culture based on immunostaining with an antibody specific for macrophages (data not shown). 

Light-Induced Cell Death in Bovine Retinal Cell Cultures

As loss of visual cells by apoptosis is a key feature of AMD, we studied light-induced photoreceptor cell death to gain insight into how to inhibit photoreceptor death and potentially the loss of vision in AMD. ^{5 6 7 8} Sytox Green nucleic acid dye uptake into retinal cells was used as the primary method of assessing cell damage and establishing a suitable light source, intensity, and exposure duration that would preferentially induce cell death of photoreceptor cells in bovine primary retinal cell culture. Blue light produced a higher differential Sytox reading with Z′-factor ²⁴ ≥ 0.5 or signal window (SW) ≥ 3 (http://www.ncgc.nih.gov/guidance/index.html/ NIH Genomic Chemicals Center, Bethesda, MD) when compared with white light (Z′ < 0.3, SW < 2) in inducing cell death in bovine primary retinal cell culture with the current settings of light intensity for > 16 hours (Fig. 1A) . Fluorescence (F*) microscopic analysis showed progressive reduction of photoreceptors with increased duration of blue- or white-light treatment. Exposure to blue light for 24 hours led to the death of most photoreceptors, but the number of bipolar cells (Figs. 1B 1D)and Müller cells (Figs. 1C 1E)remained unchanged compared with control cultures without exposure to light. These results suggest that blue- or white-light exposure leads to preferential death of photoreceptors cells in bovine primary retinal cell culture, with an undetectable effect on other cell types. Accordingly, our initial screen was based on the capacity of minocycline and structurally related tetracycline analogues of protecting photoreceptors after 20 hours of exposure to blue light. 

Effects of Minocycline and Its Analogues on Blue Actinic Light–Induced Photoreceptor Damage in Bovine Retinal Cell Cultures

As several tetracycline analogues in addition to minocycline have been found to be neuroprotective in various cell-based systems, ^{25 26} we compared the concentration-dependent effects of several tetracycline analogues in the in vitro light-induced stress model on bovine primary retinal cell culture to determine whether any insight can be gained based on the structural activity relationship (SAR) of these molecules (Table 1) . Of all the tetracycline analogues tested in our study, only minocycline and methacycline were found to be cytoprotective against blue light, with an EC₅₀ based on Sytox assays (Fig. 2A)of 5.1 ± 0.5 (n = 4) and 5.5 ± 0.2 (n = 3) μM, respectively. The EC₅₀ estimated from photoreceptor cell count analysis (Fig. 2B)was 5.5 ± 0.1 (n = 2) and 6.8 ± 0.2 (n = 2) μM for minocycline and methacycline, respectively. The extent of protection by minocycline or methacycline was found to decrease when used at concentrations more than 20 μM, as shown by the increase in Sytox F* (Fig. 2A)and decrease in the number of photoreceptor cells (Fig. 2B) , suggesting that minocycline or methacycline at more than 20 μM may have some toxicity that counteracts its cytoprotective effect. In the absence of exposure to light, all tetracycline analogues tested at up to 36 μM had a negligible effect on retinal cell viability, suggesting that any toxicity observed with minocycline or methacycline at more than 20 μM is light dependent. 

Oxytetracycline (Fig. 2A) , tetracycline, and chlortetracycline (Fig. 2C)did not affect the extent of light-induced damage when used at less than 20 μM, and all showed some toxicity when used at more than 30 μM, with relative phototoxicity of chlortetracycline > tetracycline > oxytetracycline. Demeclocycline, meclocycline, and doxycycline potentiated the light-induced damage in retinal cells at concentrations less than 10 μM, with the relative phototoxicity of demeclocycline > meclocycline > doxycycline (Fig. 2C) . The substantial increase in Sytox F* for demeclocycline, demeclocycline, or doxycycline (Fig. 2C)corresponded to only a slight but significant (P < 0.01) decrease in the number of photoreceptor cells (Fig. 2D) . As about two thirds of photoreceptors have already been eliminated by light alone, the potentiation of light-induced damage by these phototoxic compounds most likely is the result of the death of other retinal cell types, as well as the further decline of the remaining photoreceptors. The loss of other cell types was confirmed by analyzing bipolar cells in retinal culture (Figs. 3I 3J 3K 3L) . Good correlation based on linear regression analysis (r ² = 0.83) was found (Fig. 2E)between automated photoreceptor cell counting and Sytox F* for samples treated with neuroprotective compounds such as minocycline and methacycline when the stress affected mainly photoreceptors. The three outlying samples circled in this dataset (Fig. 2E)were cells treated with 36 μM methacycline, where the elevation in Sytox F* did not correspond to an expected large reduction in photoreceptor number, suggesting that methacycline at this concentration is toxic to other retinal cells. An even better correlation (r ² = 0.90) was found if these three data points were excluded. As expected, much less correlation was found (r ² = 0.40) for datasets with phototoxic compounds, such as demeclocycline and doxycycline, which affect other retinal cells in addition to photoreceptors. 

Protection of Photoreceptors in Retinal Cell Culture by Minocycline from Light-Induced Damage

Though the condition for light-induced cell damage was specific for photoreceptors in a mixed retinal cell culture, a primary screen of Sytox Green assays in general did not distinguish among cell types in the primary retinal cell culture that were protected by minocycline. Accordingly, immunocytochemistry was used to assess the number of photoreceptors after exposure to light with or without minocycline. The protective effect of minocycline was confirmed by immunocytochemistry (Fig. 3)using double staining with either recoverin and PKC-α antibodies (Figs. 3A 3B 3C)or with rhodopsin and recoverin antibodies (Figs. 3M 3N 3O) , or with a cell viability stain (Live/Dead Fixable Dead Cell Stain; Invitrogen-Molecular Probes) and recoverin antibody (Figs. 3Q 3R 3S 3T) . The number of both rods (shown as small round cells in green or yellow in merged pseudocolored images) and cones (shown as larger cells in red in merged pseudocolored images) decreased markedly (by observation) in cell cultures exposed to blue light. Consistent with results from the time course of light-induced damage in retinal cells (Fig. 1A) , there was no marked decrease in bipolar cells (Fig. 3B)from blue-light exposure by observation. Similar to previous reports about light-induced damage in animal models, ^{6 8} blue light imparted more damage to rods than cones (Fig. 3N)in retinal cell culture, as there were hardly any viable rods left after exposure to light, whereas some viable cones were still found. Minocycline at 9 μM (approximately twice its EC₅₀) protected both rod and cone photoreceptors from blue-light–induced cell death (Figs. 3C 3O)to a level almost comparable to control cells with no exposure to light by observation (Figs. 3A 3M) . 

Results from the Sytox Green assays and automated photoreceptor counting (Fig. 2)were consistent with the immunocytochemistry results by observation regarding the viability of photoreceptors in culture (Fig. 3) . Tetracycline (Fig. 3F) , oxytetracycline (Fig. 3E) , and chlortetracycline (Fig. 3G)at 9 μM showed no protection of photoreceptors from light, similar to light-alone samples (Fig. 3B) . Similar to minocycline (Figs. 3C 3O) , methacycline (9 μM) showed marked protection of photoreceptors (Figs. 3D 3P) . As predicted from an increase in Sytox reading (Fig. 2B) , meclocycline (9 μM; Fig. 3K ), demeclocycline (4.5 μM; Fig. 3I ), doxycycline (9 μM; Fig. 3L ), and chlortetracycline (36 μM; Fig. 3H ) caused more extensive damage to retinal cells, as evidenced by cell image analysis. 

To assess the viability of recoverin-positive cells (rod and cone photoreceptors) after exposure to light, we treated the retinal cells with a cell viability stain (Live/Dead Fixable Dead Cell Stain; Molecular Probes, Inc.) before immunolabeling with anti-recoverin. As expected, photoreceptors were reduced after exposure to light (Fig. 3R) . Most of the photoreceptors that remained were still viable. The few dead photoreceptors appeared as yellow in the merged pseudoimage (Fig. 3R) . Minocycline at 7.5 or 15 μM (Fig. 3S 3T)protected the photoreceptors and none of the recoverin-positive cells were positive for the cell viability stain, suggesting that all the remaining photoreceptors were viable. Although there was a marked increase in photoreceptor survival with minocycline after exposure to light, the number of dead cells (green) remaining did not appear to change markedly with or without light or minocycline (Figs. 3Q 3R 3S 3T) , suggesting most dead cells were washed away and that minocycline protected the photoreceptors specifically from light-induced cell death, as the remaining dead cells were recoverin negative. 

As caspase activation plays a key role in apoptosis, ²⁷ we investigated whether caspases were activated in retinal cells after exposure to blue light. Caspase activation was monitored with caspase detection kits (Image-iT Live Green; Invitrogen-Molecular Probes) specific for caspase-3 and -7 or caspase-1 and -3 to -9 with a specific fluorochrome-labeled inhibitor of caspase. ²⁸ Caspase activities were negligible with no exposure to light (Fig. 3U)and found to be elevated with exposure to blue light (Fig. 3W) . The number of caspase-activated cells (green) remaining did not appear to change markedly, with or without minocycline (Fig. 3W)despite the marked increase in photoreceptor survival with minocycline after exposure to light, suggesting that most apoptotic cells were washed away. From the merged pseudocolored image (Fig. 3W) , most caspase-activated cells appeared as green cells, which suggests that they may not be photoreceptors. However, enhancement of the image with recoverin stain only (Fig. 3Y)showed that most caspase-positive cells (Fig. 3X)also contained discernible amount of recoverin (Fig. 3Y , arrows), suggesting that recoverin level goes down as the photoreceptors are undergoing apoptosis. 

Protection of Selected Tetracycline Analogues from Oxidative Stress–Induced Damage in Retinal Cells

Light-induced damage may involve the generation of reactive oxygen species (ROS) that trigger retinal cell death. ^{5 7 8} Minocycline, but not tetracycline, has been found to be an effective antioxidant with radical scavenging potency. ^{29 30} We compared the concentration-dependent effects of tetracycline analogues in an oxidative stress model using 0.8 mM tBOOH in bovine primary retinal cell cultures (Figs. 4 and 5) . Similar to light-induced stress, minocycline, and methacycline showed partial protection of oxidative stress between 9 and 20 μM. Though demeclocycline was the most phototoxic compound (Fig. 2C) , it also shows protection of oxidative stress between 9 to 20 μM (Fig. 4B) . Doxycycline is the only compound that shows potentiation of oxidative damage, whereas tetracycline, oxytetracycline, chlortetracycline, and meclocycline had a minimal effect on oxidative stress induced by tBOOH (Fig. 4) . Unlike the light-induced stress model, which is specific for photoreceptors, immunocytochemistry analysis indicated that treatment with tBOOH led to reduction of both photoreceptors and rod bipolar cells (Fig. 5) . Minocycline (Fig. 5C)protected both photoreceptors and rod bipolar cells from oxidative stress. Demeclocycline (Fig. 5F)and methacycline (Fig. 3G)showed a similar extent of protection as minocycline at between 9 and 18 μM against oxidative stress induced by tBOOH. Tetracycline (Fig. 5E) , oxytetracycline (Fig. 5H) , and chlortetracycline (Fig. 3I)showed no protection from oxidative stress induced by tBOOH. Doxycycline showed further reduction of photoreceptors and bipolar cells at 36 μM (Fig. 5D) , compared with tBOOH-alone samples (Fig. 5B) . The finding that tBOOH induced retinal cell death was not specific to photoreceptors was also consistent with the observation that the extent of photoreceptor reduction was less than twofold (Z′ factor < 0.3, SW < 2) with the addition of 0.8 mM tBOOH based on automated counting of photoreceptors (data not shown), whereas there was a more than threefold increase in Sytox signals with tBOOH. To compare the relative potency of the minocycline analogues in protecting photoreceptors against tBOOH treatment by automated cell counting, the concentration of tBOOH was increased to 1 mM, a level found empirically to eliminate most photoreceptors in culture. Minocycline, methacycline, and demeclocycline were again found to protect photoreceptors (Fig. 6) . Meclocycline was also found to protect photoreceptors against tBOOH treatment under this condition. No protection was found for any compounds when tBOOH concentration was elevated to ≥2 mM (data not shown). 

To determine whether minocycline would protect retinal cells from light or oxidative stress by inhibiting the generation of ROS, we examine whether minocycline would intervene in this process by observing the light-dependent conversion of a nonfluorescent probe, ³¹ the acetyl ester of H₂DCFDA, to a fluorescent product as a result of intracellular oxidation. Accordingly, retinal cells were preloaded with H₂DCFDA with or without minocycline, exposed to blue light for various intervals. The intracellular F* was then compared after washing off H₂DCFDA in the medium. After the retinal cells were exposed to blue light for at least 1 hour, an increase of ROS was recorded (Fig. 7) . The production of ROS was markedly decreased when minocycline at 10 μM, but not at 1 μM, or a standard antioxidant N-acetylcysteine ³² at 1 mM was added during exposure to light, suggesting that the antioxidant activity of minocycline is partly responsible for its protection against light-induced stress. The concentration of minocycline required to inhibit light-induced ROS generation is consistent with those required to protect photoreceptors from light or oxidative stress. 

Discussion

It has been demonstrated that different sources of light can mediate photoreceptor cell death in in vivo animal models. ^{6 8 10} Life-long exposure to blue light, as a component of sunlight or bright artificial light sources, may contribute to pathogenic processes in AMD. ^{2 33} In this study, we established a primary retinal cell culture from bovine retinas as a model for the evaluation of light-induced damage to photoreceptors. ²¹ We were able consistently to show light-induced damage in primary bovine retinal photoreceptors grown on 96-well culture plates exposed to blue or white light. Immunocytochemistry confirmed that blue- or white-light exposure led to preferential death of photoreceptors cells versus other cell types such as neuronal bipolar cells and Müller cells. This is consistent with preferential cell death of photoreceptors found in an in vivo model of light-induced damage. ^{6 8} The importance of glial cells for long-term photoreceptor survival and neurite outgrowth have been demonstrated in other in vitro retinal cell culture systems. ^{34 35}  

Minocycline was found to protect photoreceptors in primary retinal cell culture from light-induced damage, with an EC₅₀ of approximately 5 μM based on Sytox readings, automated photoreceptor cell counting (Fig. 2) , and immunocytochemistry of photoreceptor cells and rod bipolar cells in retinal culture (Figs. 3C 3F) . Almost complete protection of photoreceptors from light-induced damage was observed when the culture was pretreated with 9 μM minocycline. The extent of protection was diminished when more than 20 μM minocycline was used, possibly due to increased phototoxicity of tetracycline-like compounds when used at high concentrations. ^{36 37 38}  

For neurodegenerative indications, minocycline has been applied either as an anti-inflammatory agent or an antiapoptotic agent. ¹ It has been suggested to protect against light-induced loss of photoreceptors indirectly through the inhibition of retinal microglial activation. ¹⁰ Because there are no detectable microglial cells present in our primary retinal cell culture, it is more likely that minocycline acts directly as an antioxidant and/or as an antiapoptotic agent, ^{1 11 12} to protect photoreceptor cells from stress in retinal culture. This observation is supported by the partial protection by minocycline and methacycline against tBOOH-induced oxidative stress and by the partial inhibition of light-dependent generation of ROS by minocycline. A possible antiapoptotic mechanism of minocycline may involve inhibition of caspase activation, ^{11 25 39} upregulation of antiapoptotic proteins such as Bcl-2 ⁴⁰ and the X-linked inhibitor of apoptosis protein (XIAP) ³⁹ in mitochondria, and inhibition of the release of apoptotic proteins such as cytochrome c, ⁴¹ apoptosis inducing factor (AIF), and Smac/Diablo ^{39 41} from mitochondria. We have been able to show that blue light induces caspase activation in photoreceptor and that minocycline may inhibit this process. We also tested a caspase-independent apoptosis marker, AIF, in our retinal cell culture system. Most cells immunolabeled with anti-AIF are not rhodopsin positive (data not shown). It is not clear whether AIF played a role in photoreceptor apoptosis in our culture system due to its low expression in rod photoreceptors. The observation that demeclocycline and possibly meclocycline protected retinal cells from tBOOH but potentiated light-induced cell death suggests that the light-protective activity of certain tetracycline analogues is not fully dependent on their antioxidant property. Further experiments are needed to define precisely the molecular targets of minocycline against light- or oxidative stress–induced retinal photoreceptor cell damage. 

We have compared the concentration-dependent effects of several tetracycline analogues in the in vitro retinal light-induced stress model to determine whether any insight can be gained based on the structure–activity relation (SAR) of these molecules. Minocycline, oxytetracycline, tetracycline, chlortetracycline, methacycline, doxycycline, meclocycline, and demeclocycline were tested in bovine retinal primary cell cultures. Besides minocycline, only methacycline showed protection from blue light–induced damage of photoreceptors. Minocycline and methacycline were found to be phototoxic at more than 20 μM, suggesting that these two compounds have a narrow therapeutic index for neuroprotection of photoreceptors from light-induced stress. Analogues with slight structural modifications such as demeclocycline, doxycycline, and meclocycline were found to be highly toxic to retinal cells in the presence of light. The phototoxicity of some tetracycline analogues have been noted in fibroblasts on UV irradiation, ⁴² showing toxicity of demeclocycline ∼ doxycycline ≫ tetracycline and no toxicity of minocycline. Blue light–induced damage of photoreceptors showed a similar trend, but with phototoxicity of demeclocycline > doxycycline and the demonstration of the protective effect of minocycline. This trend also correlates with the relative capacity of demeclocycline, tetracycline, and minocycline in generating ROS on excitation with long-wave UV. ⁴³ Fluoroquinolone derived antibiotics have been reported to have phototoxic effects on both human and animal subjects due to the generation of ROS, including hydrogen peroxide and the hydroxyl radical. ⁴⁴ The most phototoxic ones have the fluoro-group at the C-8 position, whereas most phototoxic compounds except for doxycycline in the tetracycline family have the chloro-group at the R1 position, suggesting that the phototoxicity of doxycycline, similar to those phototoxic fluoroquinolones without a C-8 halogen atom, may work through a different mechanism. ⁴⁵ No toxicity to photoreceptors has been observed for any of the tetracycline analogues without exposure to light. Our data on protection from oxidative stress seem to follow the same trend from a previous study that showed that minocycline, but not tetracycline, protected rodent derived retinal cells or neurons from glutamate-induced cell death. ²⁵ Demeclocycline has been found to protect against oxidative stress, and demeclocycline and chlortetracycline have been found to protect against glutamate-induced neuronal death in vitro and cerebral ischemia in vivo to the same extent as minocycline. ²⁶ A compound of similar efficacy but with minimal phototoxicity is preferable for clinical application, to minimize any damage to photoreceptors. 

The demonstration of minocycline’s protective action against light-induced damage or oxidative stress in the cell-based assays and in animal models ^{10 11 12} suggests that the in vitro assay based on protection of photoreceptors from light-induced damage or oxidative stress in bovine primary retinal cell culture is a suitable model for evaluating compounds that are neuroprotective of photoreceptors. The finding of toxic compounds within the tetracycline family also suggests the utility of the retinal cell–based assay for identifying compounds that are toxic or phototoxic to photoreceptors or other retinal cell types during the early drug development process. As minocycline may not work in all models of neurodegenerative diseases, ^{16 17 18 19 20} this cell-based system may be used to screen for neuroprotective candidates with more potency and a wider therapeutic window than minocycline or methacycline and especially those without the side effects often associated with broad-spectrum antibiotics. 

Because the activities of minocycline and its analogues that we studied varied by minor structural modifications, our data suggest that the structural activity development of minocycline-related molecules ^{46 47} could lead to a potent molecule for AMD therapy. 

 Submitted for publication May 10, 2006; revised July 17, 2006; accepted October 11, 2006.

 Disclosure: D.W. Leung, Acucela, Inc. (E); L.A. Lindlief, Acucela, Inc. (E); A. Laabich, Acucela, Inc. (E); G.P. Vissvesvaran, Acucela, Inc. (E); M. Kamat, Acucela, Inc. (E); K.L. Lieu, Acucela, Inc. (E); A. Fawzi, Acucela, Inc. (E); R. Kubota, Acucela, Inc. (E)

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

 

Figure 1.

 

Time course of light-induced cell death. Fourteen-day-old primary bovine retinal cells were exposed to blue actinic light or white fluorescent light for different times. Cell death was monitored by evaluation of Sytox Green fluorescence (shown as arbitrary units on the y-axis ofA). Data are the mean ± SD (n = 3). Examples of pseudocolored images of cells kept in the dark (B, C) or exposed to blue light for 24 hours (D, E) are shown. Cells were immunolabeled, and rod and cone photoreceptors (red) and rod bipolar cells (green) are shown (B, D). Müller cells are red in (C) and (E). Scale bar, 10 μm.

Table 1.

 

Structural Activity of Tetracycline and Its Analogues

 

 These compounds contain a common structure made up of four fused six-membered rings’ with variations found at one or more of the R-groups. EC₅₀ values based on nucleic acid stain (Sytox; Invitrogen) assays under the Protection column refer to concentrations needed to protect retinal cells from light damage, whereas EC₅₀ values under the Toxic column refer to concentrations needed to potentiate light-induced damage in retinal cells.

Figure 2.

 

Concentration-dependent effect of minocycline and its analogues on light-induced damage in retinal cells. Primary bovine retinal cell cultures were pretreated with minocycline and its analogues at different concentrations before exposure to blue light for 20 hours. The change in Sytox Green F* (A, C) or change in the number of photoreceptor cells (B, D) was used to assess the extent of protection or the potentiation of light-induced damage. Data are the mean ± SD (n = 3). The control was cells treated under similar conditions without test compound. All curves for test compounds without light are similar. Only data for minocycline or demeclocycline without light are shown (A, C) for illustration. The correlation between photoreceptor cell number and Sytox F* for samples treated with neuroprotective compounds such as minocycline and methacycline (E) or with phototoxic compounds such as demeclocycline and doxycycline (F) plus light was analyzed by linear regression. The three outlying samples circled (E) were cells treated with 36 μM methacycline.

Figure 3.

 

Effect of minocycline and its analogues on light-induced damage of photoreceptors in retinal cell culture. Examples of merged pseudocolored images of primary bovine retinal cells pretreated with tetracycline analogues (C–L, O, P, S, T, X, Y) at various concentrations before exposure to blue actinic light for 20 hours, along with the no-light control cells (A, M, Q, U) and the light-alone samples (B, N, R, W) are shown. (A–L) Cells were double immunolabeled either with recoverin (red) + PKCα (green); or with rhodopsin (yellowish green) + recoverin (red) (M–P); or either stained with a fixable dye for dead cells (Q–T) (yellow for photoreceptors, green for other retinal cells) or with a fluorochrome-labeled inhibitor of caspase-3 or -7 (green) and immunolabeled with recoverin (red) (U–X). (Y) The same field as (X) with only recoverin labeling shown and enhanced. Scale bar, 10 μm.

Figure 4.

 

Concentration-dependent effect of tetracycline and its analogues on tBOOH-induced damage in retinal cells. Primary bovine retinal cell cultures were pretreated with tetracycline and its analogues at different concentrations before treatment with 0.8 mM tBOOH for 20 hours. Change of Sytox Green F* was used to assess the extent of protection or potentiation of oxidative stress. Data are the mean ± SD (n = 3). The controls refer to cells treated in similar conditions without test compound.

Figure 5.

 

Effect of minocycline and its analogues on oxidative stress of photoreceptors in retinal cell cultures. Primary bovine retinal cell cultures were pretreated with minocycline and its analogues at various concentrations before treatment with 0.8 mM tBOOH for 24 hours. Examples of merged pseudocolored images of control cells without tBOOH treatment, cells treated with tBOOH alone, and cells treated with tBOOH + 9 μM minocycline, tetracycline, demeclocycline, methacycline, oxytetracycline, or chlortetracycline or 36 μM doxycycline are shown. Cells were double immunolabeled with recoverin (red) and PKC-α (green). Scale bar, 10 μm.

Figure 6.

 

Certain minocycline analogues partially protected photoreceptors against oxidative stress in bovine retinal cell cultures. Primary bovine retinal cell cultures were pretreated with minocycline and its analogues at 15 μM before treatment with 1.0 mM tBOOH for 24 hours. Change in Sytox Green F* and corresponding change in the number of photoreceptor cells were used to assess the extent of protection. Data are means ± SD (n = 3).

Figure 7.

 

Effect of minocycline and N-acetylcysteine on light-dependent generation of ROS. Primary bovine retinal cell cultures were preloaded with the ROS probe, H₂DCFDA, with or without minocycline or N-acetylcysteine before exposure to blue light for various intervals. After medium was changed to remove H₂DCFDA, the generation of ROS was monitored by evaluation of conversion of nonfluorescent H₂DCFDA to fluorescein (shown as arbitrary F* units on the y-axis). Data are the mean ± SD (n = 3).

The authors thank Kyoko Murata, Corinne C. Manmoto, and Tim E. McGinn for assistance in preparing cells, immunocytochemistry, and automated cell counting analysis.