BDNF Attenuates Retinal Cell Death Caused by Chemically Induced Hypoxia in Rats

Abstract

purpose. To investigate the neuroprotective effects of brain-derived neurotrophic factor (BDNF) against potassium cyanide (KCN)–induced retinal damage.

methods. Rats were injected intravitreally with iodinated BDNF. Two days later, eyeballs were dissected into various parts, and the level of radioactivity in each part was measured. Retinal damage was induced by incubating rat eyeballs with 5 mM KCN. BDNF was injected intravitreally 2 days before KCN treatment, and subsequent morphometric analysis was carried out to evaluate the retinal cell damage. To elucidate the mechanisms of BDNF’s neuroprotective effects, the intravitreal concentrations of amino acids and the expression of calretinin were investigated.

results. Intravitreally injected BDNF was distributed evenly throughout the eyes, and the incorporation of iodinated BDNF into the retina was three times higher than in other ocular tissues. Immunohistochemical analysis demonstrated that exogenous BDNF diffused throughout the retina and was especially concentrated in the inner (INL) and outer nuclear layer. Morphometric analysis showed that the number of INL cells of the posterior area, 880 μm from the optic nerve head, was 190 ± 4 with KCN treatment and 284 ± 9 in control animals. Cell death appeared to be necrotic. When eyes injected with either phosphate-buffered saline (PBS) or BDNF were subjected to treatment with KCN, the number of INL cells was 186 ± 5 in the PBS-treated controls and 253 ± 8 in eyes treated with BDNF. Also, BDNF increased the number of calretinin-positive cells in the INL and reduced the KCN-induced elevation of intravitreal glutamate levels.

conclusions. BDNF injected intravitreally reaches the retina and attenuates the INL cell death caused by KCN-induced metabolic insult. The neuroprotective effects of BDNF are partly ascribed to the upregulation of a calcium-binding protein and the attenuation of glutamate release into the vitreous body.

Neurotrophic agents have the potential to provide new therapies for neurodegenerative diseases such as Alzheimer’s disease, Parkinson’s disease, peripheral neuropathy, and optic neuropathy. Brain-derived neurotrophic factor (BDNF) is a member of the neurotrophin (NT) family, which includes nerve growth factor (NGF), NT-3, NT-4/5, and NT-6. Neurotrophins are known to have survival and neurite outgrowth-promoting activity in the central and peripheral nervous systems. ¹  

TrkB, the high-affinity receptor for BDNF (and NT-4/5), is present in developing and adult retinas ^{2 3 4 5 6 7 8 9 10} and in the retinal pigment epithelium (RPE). ¹¹ Retinal ganglion cells can upregulate BDNF mRNA expression after optic nerve injury. ^{12 13} These findings indicate that BDNF may play an important role in the development, maturation, and maintenance of various neuronal networks. Indeed, BDNF can support the survival of chick ¹⁴ and rat ^{15 16} retinal ganglion cells in vitro. In vivo, BDNF has been shown to support cell survival and to enhance the axonal regeneration of axotomized retinal ganglion cells. ^{17 18 19 20 21 22} BDNF also protects the rat retina from ischemic injury, ²³ light damage, ²⁴ and photoreceptor degeneration. ²⁵ Possible therapeutic roles have been described for calcium-channel blockers, glutamate antagonists, antioxidants, anti-apoptotic agents, and neurotrophic factors in the treatment of ischemic injury, ischemic degeneration, and glaucoma. ^{26 27 28} These findings imply a therapeutic potential for BDNF as a neuroprotective agent in the treatment of ocular diseases and prompted us to investigate its efficacy in blocking retinal damage using animal models of retinal metabolic insult. 

Retinal cell death is seen in many ocular diseases, and a number of different stimuli (such as high intraocular pressure, blockage of axonal flow, and retinal ischemia) may cause cellular damage in the retina. ^{26 29} Ischemic damage can be mimicked by hypoglycemia and anoxia, or by chemically inducing a metabolic blockade of either glycolysis or electron transport. ^{30 31 32} Cyanide is known to block the mitochondrial respiratory chain and to produce neurotoxicity. ³³ In embryonic chick eyes, potassium cyanide (KCN) treatment results in the death of inner nuclear layer (INL) cells via excitotoxicity associated with metabolic inhibition. ^{30 31} In this study, we investigated the distribution of intravitreally injected BDNF and the neuroprotective effect of BDNF on KCN-induced retinal damage in adult rat eyes ex vivo. Our results showed that KCN-induced cell death in the INL was necrotic in nature and that BDNF attenuated INL cell death by regulating the expression of a calcium-binding protein and by limiting the KCN-induced release of intravitreal glutamate. 

Methods

Animals

Male Wistar rats were obtained at 250 to 300 g (Charles River, Yokohama, Japan) and maintained in a cyclic light environment (12 hour light/12 hour dark) for 7 or more days before experiments. All the animal experiments were carried out according to the guidelines of the Sumitomo Pharmaceuticals Committee on Animal Research and ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Preparation of ¹²⁵I–BDNF and Determination of ¹²⁵I–BDNF Radioactivity

Human recombinant BDNF (N-terminal methionine-free) was supplied from Regeneron Pharmaceuticals (Tarrytown, NY). Iodination and isolation of ¹²⁵I–BDNF was performed in a similar manner to that of Rosenfeld et al. (1993). ³⁴ The specific activity of ¹²⁵I–BDNF was 1.07 × 10⁷ cpm/μg. Rats were anesthetized with ether and injected intravitreally with 1 μl of sterilized ¹²⁵I–BDNF (1 μg). The injection was performed with a 33-gauge needle, according to the method of Faktorovich et al. ³⁵ Two days after intravitreal injection of ¹²⁵I-labeled BDNF, rats were killed with pentobarbital and the eyeballs removed. The cornea, lens, iris, ciliary body, retina, and sclera (including the choroid) were dissected from the eyeballs under a dissecting microscope. The radioactivity in each section was measured using a scintillation gamma counter. The data are shown as the radioactivity per microgram of tissue wet weight. 

Preparation of Anti-BDNF Antibody and Preabsorption of Anti-BDNF

Anti-BDNF antiserum was produced by immunizing rabbits with Escherichia coli–derived recombinant BDNF. Anti-BDNF antiserum was loaded onto a BDNF affinity column, which had been prepared by coupling BDNF to CNBr-activated Sepharose 4B (Amersham Pharmacia Biotech, Tokyo, Japan). After extensive washing with affinity column loading buffer (Pierce, Rockford, IL), bound anti-BDNF antibody was eluted with ImmunoPure Gentle Ag/Ab Elution Buffer (Pierce). To assess the specificity of western blot analysis and immunohistochemical staining with anti-BDNF, anti-BDNF was incubated with an excess of BDNF affinity resin, overnight at 4°C, on a rotary shaker. The suspension was spun, and the supernatant fraction was used as preabsorbed anti-BDNF. 

Western Blot Analysis

Purified BDNF, NGF (Serotec, Oxford, UK), NT-3 (Peprotec, Rocky Hill, NJ), and NT-4 (Regeneron, Tarrytown, NY) were used as NT standards, and rainbow markers (Amersham Pharmacia Biotech) were used as the molecular weight markers. The electrophoresis samples were prepared as follows. The retinas were removed from the posterior halves of the eyeballs under a dissecting microscope. They were then washed twice with ice-cold phosphate-buffered saline (PBS) and immediately homogenized in lysis buffer (50 mM Tris–HCl, pH 7.5, containing 1% NP-40, 150 mM sodium chloride, 2.5 mM ethyleneglycol bis[ 2-aminoethylether] tetraacetic acid [EGTA], 0.14 U/ml aprotinin, 1 mM phenylmethylsulfonyl fluoride, 2 mM sodium vanadate, and 50 mM sodium fluoride). Homogenates were spun at 10,000g for 10 minutes, and the supernatant fractions were collected. Electrophoresis was performed on 12.5% sodium dodecyl sulfate–polyacrylamide gels (12.5 μg protein/lane) and blotted on Hybond–ECL filters (Amersham Pharmacia Biotech). The membranes were incubated for 48 hours at 4°C with the anti-BDNF antibody diluted 1:500 in 0.1% Tween-20–PBS (T–PBS) or 24 hours at 4°C with a polyclonal antibody against calretinin (Chemicon International, Temecula, CA) diluted 1:3000 in T–PBS. After washing with T–PBS, the membranes were incubated with horseradish peroxidase–linked anti-rabbit immunoglobulins (Amersham Pharmacia Biotech) diluted 1:500, for 60 minutes at room temperature. After washing with T–PBS, the bands were visualized using chemiluminescence according to the manufacturer’s protocol (ECL detection kits; Amersham Pharmacia Biotech). To assess the specificity of the anti-BDNF antibody, neurotrophic factors (0.1 μg) were loaded onto duplicate sodium dodecyl sulfate–polyacrylamide gels. One gel was subjected to western blot analysis, and the other was silver stained (Daiichi Kagaku, Tokyo, Japan) to visualize the loaded proteins. 

Immunohistochemistry

Two days after the intravitreal injection of BDNF, rats were killed and the eyeballs were enucleated in preparation for the analysis of BDNF distribution and calretinin expression. The eyes were fixed with 10% formalin–PBS, embedded in paraffin, and sectioned at 4-μm thickness for immunohistochemical analysis. Sections were incubated for 2 days at 4°C with anti-BDNF antibody diluted 1:300. The specificity of this antibody was confirmed by incubating control sections in the preabsorbed anti-BDNF diluted 1:300. Calretinin was detected with the anti-calretinin antibody diluted 1:3000 (Chemicon International, Temecule, CA), overnight at 4°C. After washing in PBS, sections were incubated in fluorescein isothiocyanate–linked anti-rabbit immunoglobulins (Cappel Research Products, Durham, NC) diluted 1:500 or horseradish peroxidase-linked anti-rabbit immunoglobulins (Amersham Pharmacia Biotech) diluted 1:500, for 60 minutes at room temperature. The distribution of intravitreally injected BDNF was analyzed under a microscope equipped for fluorescence imaging. All the calretinin-positive INL cells were counted in all the retinal sections containing the optic nerve head. 

Intravitreal Injection of BDNF and KCN Treatment

BDNF was diluted in PBS to a final concentration of 0.1 to 10 mg/ml and intravitreally injected into the eyes of rats anesthetized with ether, 2 days before ex vivo KCN treatment. Intact eyes were removed immediately after death, and each eye was incubated in 3 ml bicarbonate-buffered Krebs–Ringer solution (K–R), which was kept at 37°C and equilibrated with 5% CO₂/95% O₂. After preincubation in K–R for 10 minutes, KCN was added to the medium to a final concentration of 5 mM, and the eyes were incubated for 5, 30, and 60 minutes. Ca²⁺ chelaters and glutamate receptor antagonists were added to the incubation medium 10 minutes before KCN treatment. 

TdT-Mediated dUTP Nick-End Labeling Method

After 30 minutes of incubation with KCN, eyes were fixed in 10% formalin–PBS. TdT-mediated dUTP nick-end labeling (TUNEL) analysis was performed on 4-μm–thick paraffin sections using the TUNEL kit (Apop Taq; Oncor, Gaithersburg, MD). Eyes prepared from rats exposed to fluorescent light at 190 foot-candles for 48 hours were used as a positive control for apoptosis. 

Retinal Cell Counts

After incubation in K–R with 5 mM KCN, eyeballs were immediately fixed in a solution of 2.5% formalin–1% glutaraldehyde in PBS. Samples were embedded in Thechnovit 7100 (Heraeus Kulzer, Zweigniederlassung, Germany) and sectioned at 2-μm thickness. The sections containing the optic nerve head were stained with 0.1% cresyl violet for 1 minute (Nissl staining) and observed under the microscope. In our study, the KCN-induced cell loss in the INL and the attenuation of INL cell death by BDNF treatment were similar in all regions of the superior and inferior retina from the center to the periphery. Therefore, for the quantitative analysis, the numbers of Nissl stain–positive cells were counted in four microscopic fields, each 220μ m in length, of the posterior portions of the retina lying 880 and 1320 μm away from the optic nerve head in the superior and inferior regions. The total number of positive cells in the four regions was used for the evaluation of INL degeneration and rescue. 

Amino Acid Analysis

After incubating the eyeball in K–R, the vitreous humor was collected by cutting the eye. The humor was acidified with a one-tenth volume of 4 N perchloric acid. Before performing amino acid analysis, the medium was neutralized with 2.5 M potassium carbonate. It was then spun at 10,000g for 10 minutes at 4°C, and the supernatant fraction was analyzed by a micro–high performance liquid chromatography system for automated analysis of amino acids using precolumn o-phthalaldehyde derivatization and fluorescence detection (CMA 1200 refrigerated Microsampler, CMA/280 Fluorescence Detector; BAS, Tokyo, Japan). 

Results

Distribution of Intravitreally Injected BDNF

Two days after intravitreal injection of ¹²⁵I–BDNF, the levels of radioactivity in the cornea, iris, ciliary body, lens, retina, and sclera containing the choroid were determined. As shown in Figure 1 , the incorporation of ¹²⁵I–BDNF into the retina was three times higher than into the other regions. The measured radioactivity levels were as follows (cpm/mg wet-weight tissue): cornea, 10.7 ± 3.7; iris, 41.9 ± 41.9; ciliary body, 81.7 ± 53.5; lens, 65.1 ± 25.1; retina, 287.4 ± 97.9; and sclera, 95.7 ± 14.1. The radioactivity in the retina was competed by a 100-fold excess of cold BDNF. The amount of BDNF incorporated into the retina was calculated to be approximately 0.3% of the total injected. Thus, we found that intravitreally injected BDNF reached the retina and remained there even 2 days after administration. 

Next, the subretinal distribution of BDNF 2 days after intravitreal injection was determined using immunohistochemistry. First, we characterized our anti-BDNF antibody (see the Methods section). The anti-BDNF antibody used in this study did not have a strong affinity for NGF, NT-3, or NT-4 (Fig. 2 A; lanes 6–8), and was highly specific for BDNF, as shown in Figure 2A (lane 5). Endogenous BDNF was not detected by western blot analysis (Fig. 2A ; lane 1), but exogenous BDNF was detected in the retinas of eyeballs injected with BDNF (Fig. 2A ; lane 2). No bands were observed after incubation with pre-absorbed anti-BDNF (Fig. 2A ; lanes 3, 4). The immunohistochemical analysis showed a fluorescent signal in the outer segment and outer plexiform layer in the PBS-injected retina (Fig. 2B) . These signals were also detected in retinas incubated with pre-absorbed anti-BDNF (Fig. 2D) , showing they were caused by nonspecific binding. In contrast, positive fluorescence signals for BDNF were detected throughout the retina, especially in the INL and outer nuclear layer (Fig. 2C) . This result shows that the intravitreally injected BDNF diffused well throughout the retina. 

Characterization of KCN-Induced Retinal INL Cell Damage

First, we analyzed the retinal morphology after KCN treatment. In KCN-treated adult rat eyes, a decrease in the total number of Nissl stain–positive retinal cells in the INL was observed in uninjected eyes, whereas there was no significant change in cell number in the outer nuclear layer (Fig. 3) , which agrees with the results seen with chick embryonic retinas. ^{30 31} KCN treatment caused more cell loss in the ganglion cell side of the INL, where the amacrine cells are located. Swelling of retinal cells in the INL was also observed (Fig. 3C , asterisks). KCN treatment did not affect the thickness of the INL; rather, it resulted in a low density of INL retinal cells. In addition, retinal edema was found in the nerve fiber layer (Fig. 3B , arrow) and the inner plexiform layer (Fig. 3B , arrowhead). The extent of KCN-induced INL cell damage observed was similar in all regions of the superior and inferior retina, from the center to the periphery. The total number of Nissl stain–positive cells in four regions of the posterior retina was counted for the quantitative analysis, as described in the Methods section. The mean (± SEM) number of cells in the INL of PBS-treated eyes was 256 ± 20, 186 ± 5, and 165 ± 11 at 5, 30, and 60 minutes of KCN incubation, respectively (Fig. 7) . The type of cell death in the retina was also analyzed. As shown in Figure 4 A, light-induced apoptosis made outer nuclear layer cells TUNEL positive (arrows). In contrast, TUNEL-positive cells were not observed after 30 (Fig. 4C) or 60 (data not shown) minutes of KCN treatment. This observation and the cell swelling caused by the KCN treatment (Figs. 3B 3C) suggest that the KCN-induced cell death of the INL cells was necrotic. 

Second, the role of extracellular Ca²⁺ was studied. To investigate the effect of extracellular Ca²⁺ deprivation, EGTA or 1,2-bis (2-amino-4-fluorophenoxy) ethane-N,N,N8242’27,N′-tetraacetic acid tetrapotassium salt (BAPTA) was added to the incubation medium. EGTA (1 mM) and BAPTA (5 mM) blocked 39% and 82%, respectively, of the KCN-induced cell death. Pretreatment with an intracellular Ca²⁺ chelater, thapsigargin (250 μM), also blocked the cell death (Table 1) . This result indicates that an influx of extracellular or intracellular Ca²⁺ is involved in KCN toxicity. 

Finally, the role of glutamate receptors in KCN toxicity was evaluated by using the glutamate antagonist, MK-801 (10 μM) or 1,2,3,4-tetrahydro-6-nitro-2,3-dioxo-benzo(f) quinoxaline-7-sulfonamide, (NBQX) (150 μM). NBQX, an AMPA/kainate receptor selective antagonist, reduced the amount of cell death by 29.8% (150 μM), although MK-801, an NMDA receptor selective antagonist, had no effect on cell survival in our system (Table 1) . This result indicates that the activation of at least the AMPA/kainate-type glutamate receptor is also involved in KCN toxicity. 

Protective Effect of BDNF on KCN-Induced Cell Death

BDNF (0.1, 1, and 10 μg) was injected intravitreally 2 days before KCN treatment, to examine the efficacy of BDNF in preventing KCN-induced cell death. After a 30-minute incubation with KCN, intravitreal BDNF protected INL cells in a dose-dependent manner, whereas the injection of PBS had no effect (Fig. 5) . Although only four regions of the posterior retina were analyzed for the quantitative evaluation, a similar protective effect of BDNF was observed in the other parts of the retinal sections as well. Compared with the PBS treatment (186 ± 5/880 μm), 1 and 10 μg of BDNF produced a statistically significant reduction in KCN-induced INL cell death, by 28.6% (214 ± 6/880 μm) and 68.4% (253 ± 8/880μ m), respectively (Fig. 6) . Interestingly, a lag time after BDNF treatment appeared to be required for neuroprotection to occur, because 10 μg of BDNF showed no protection when it was injected 1 hour instead of 2 days before KCN treatment (Fig. 6) . 

Next, we studied whether BDNF protected INL cells from more severe injury, by incubating the eyes longer with KCN. Sixty minutes of KCN treatment caused only slightly more cell death than the 30-minute treatment, but no protective effect of BDNF was seen in this case (Fig. 7) . Thus, BDNF appeared to enhance the resistance of INL cells to injury rather than to block KCN toxicity directly. 

Mechanism of BDNF Action

Western blot analysis showed that retinas from BDNF (1 or 10μ g)-injected eyes expressed 1.5 times more calretinin than did PBS-injected eyes (Fig. 8) . In an immunohistochemical study, calretinin was detected in retinal ganglion cells and in the INL cells and was especially strong in amacrine cells and in the inner plexiform layer (Figs. 9 A, 9B). Intravitreal injection of BDNF increased the number of calretinin-positive cells in the INL. In contrast, we observed very little increase in the calretinin signal in the ganglion cell layer after BDNF treatment, probably because the normal level of calretinin expression in ganglion cells, as seen in the PBS-treated control retina, was high, which obscured any increase. The numbers of calretinin-positive cells in the INL were 33 ± 5, 64 ± 7, and 69 ± 11 (mean ± SEM/1000 μm) in PBS-treated and 1-μg– and 10-μg–BDNF–treated eyes, respectively (Fig. 9C) . 

Next, the effect of BDNF on the elevation of amino acids in the vitreous body was examined. The amount of glutamate in the vitreous body increased to 4 times its normal level after a 30-minute incubation with KCN, compared with untreated controls (Table 2) . GABA and aspartate were also elevated by KCN treatment, to 3 and 2 times above the levels seen with PBS treatment, respectively (Table 2) . BDNF attenuated the elevation of glutamate levels in a dose-dependent manner, and 10 μg BDNF reduced the elevation of glutamate levels by 52.3% (Table 2) . In contrast, BDNF had no effect on the increase in GABA and aspartate levels. 

Discussion

Distribution of Exogenously Administered BDNF in Ocular Tissues

Intravitreal injection of BDNF has been shown to have a protective effect on retinal cells. ^{17 18 19 20 21 22 23 24 25} The distribution of injected BDNF, however, has not been clarified. In this study, we examined the tissue distribution of intravitreally administered ¹²⁵I-labeled BDNF. Our results showed that more radioactivity was detected in the retinal tissue than in the cornea, iris, ciliary body, lens, or sclera 2 days after the intravitreal injection of radiolabeled BDNF. Furthermore, our immunohistochemical study showed that exogenous BDNF was extensively distributed throughout the retina. Because TrkB, the high-affinity receptor for BDNF, is expressed in the retina, ^{2 3 4 5 6 7 8 9 10 11} we assume that BDNF’s biological effect occurs in the cells that express TrkB on their surfaces. 

KCN-Induced Retinal Damage by Necrosis

Previous reports showed that the INL cells from chick embryo retinas were selectively damaged by KCN. In the embryonic retina, the INL cells showed edema and degeneration after approximately 30 minutes of treatment with KCN. ³⁰ In contrast, it was reported that cyanide induced apoptotic cell death in PC12 cells. ³⁶ In our adult rat retinal system, we observed swelling and acute cell loss in the retina after KCN treatment (Figs. 3B 3C) . This damage was partially dependent on extracellular Ca²⁺, and it was sensitive to AMPA/kainate glutamate receptor blockage (Table 1) , in agreement with a previous report. ³¹ Furthermore, chromatin condensation was not observed by ethidium bromide staining (data not shown), and there were no positive cells in the TUNEL analysis (Fig. 4C) . These lines of evidence suggest that the cell death in the KCN-treated adult rat retinas in our system was necrosis rather than apoptosis. The differences between apoptotic and necrotic cell death in retinas treated with KCN can be attributed to differences in the severity of the damage, as reported for glutamate toxicity. ³⁷  

Neuroprotective Effects of BDNF on the INL

BDNF attenuated the cell loss in the INLs of retinas treated with KCN for 30 minutes. The effect of BDNF was transient, and the attenuation of the cell loss was not observed with a longer KCN treatment (60 minutes, Fig. 7 ). Thus, BDNF showed its protective effects on retinas that were lightly damaged but did not prevent cell death in cases of severe injury. BDNF exerts its trophic effects on neurons by several mechanisms. Among these actions, it prevents an overload of free Ca²⁺ by elevating a Ca²⁺-binding protein, it reduces the elevation of glutamate levels, and it enhances cellular resistance to antioxidative stress by elevating glutathione peroxidase activity, as discussed below. It seems that the cell death signal cannot be blocked by BDNF in conditions in which severe cell damage occurs. 

The protective effect of BDNF against KCN-induced INL cell damage in this study is consistent with a previous study that demonstrated that BDNF attenuated ischemic damage to INL cells induced by raising the pressure in the anterior chamber of the eye. ²³ Although ganglion and INL cell death in the ischemic injury model was caused by apoptosis or necrosis, our result at least indicates that BDNF attenuates the nonapoptotic INL cell death induced by KCN treatment. 

Neuronal apoptosis has been suggested to be widely involved in the various types of injury against which BDNF shows protective effects, as described previously. ^{17 38} In contrast, Koh et al. ³⁹ demonstrated that BDNF enhanced acute necrotic neuronal cell damage induced by oxygen-glucose deprivation in cortical cell cultures, although BDNF also attenuated the apoptotic death induced by serum deprivation in the same cells. Although their result is not consistent with ours, the differences in our experimental systems and in the severity of injury may account for this discrepancy. 

BDNF Enhanced Calretinin Expression and Reduced Glutamate Levels

When BDNF was administered intravitreally 1 hour before KCN treatment instead of 2 days before, no protective effect of BDNF was observed (Fig. 6) . This result implies that the induction of some substance, other changes in retinal tissue, or both are required for BDNF to show its protective effect. We observed the involvement of Ca²⁺ influx and glutamate in the KCN-induced INL cell damage. Therefore, to understand the mechanism of the protective effect of BDNF, we investigated the influence of BDNF on the regulation of intracellular Ca²⁺ and on the levels of intravitreal glutamate. 

First, we examined the possibility that BDNF blocked Ca²⁺ influx. In the central nervous system, BDNF is known to enhance the expression of the Ca²⁺-binding protein calbindin. ^{40 41 42 43} This protein chelates free Ca²⁺ within cells. Previous studies of hippocampal neurons have shown that NGF, basic fibroblast growth factor, and BDNF prevent the increase in intracellular Ca²⁺ and the subsequent cell damage induced by hypoglycemia. ^{44 45 46} These findings prompted us to investigate whether BDNF upregulates a Ca²⁺ -binding protein in our system. In the retina, there are several kinds of Ca²⁺-binding proteins. ^{47 48 49 50 51 52} Among them, calretinin is known to be expressed in the amacrine cells of the rat INL, ⁵⁰ where TrkB receptors are highly expressed. ^{5 10} Therefore, we examined the change in the expression pattern of calretinin after the intravitreal injection of BDNF in naive adult rats. Two days after the injection of BDNF, the number of calretinin-positive INL cells increased (Figs. 9B 9C) . Because ganglion cells also express TrkB and INL cells, ^{2 3 4 5 6 7 8 9 10} ganglion cells should have responded to exogenous BDNF. However, we did not observe this increase in our study. This result could be ascribed to the fact that the preexisting expression level of calretinin in ganglion cells is already quite high, thus obscuring any increase in the signal (Fig. 9A) . Our finding suggests that the upregulation of the Ca²⁺-chelating system in the INL is one mechanism by which BDNF acts to protect the INL against KCN-induced insult. 

In our study, there was no obvious increase in calretinin expression in the 10-μg BDNF–injected group compared with the 1-μg BDNF–treated group. However, there was clearly more cell survival from KCN injury in the retinas treated with 10 μg BDNF than in those treated with 1 μg BDNF (Fig. 6) . This observation implied that an additional mechanism of protection by BDNF treatment exists. Therefore, we next focused on the levels of glutamate in the vitreous body (Table 2) , because a glutamate receptor antagonist partially reduced the INL cell damage in our system. BDNF treatment attenuated the release of glutamate from retinal tissue into the vitreous body after KCN treatment, and the rate of suppression by 10 μg BDNF was higher than that by 1 μg BDNF. This result coincides with the dose-dependency of the protective action of BDNF against KCN-induced cell damage, suggesting that the efficacy of BDNF could be ascribed to the reduction of the elevation of glutamate levels in the vitreous body rather than to the upregulation of a Ca²⁺-binding protein in the cytoplasm. Because INL cells use glutamate as a neurotransmitter, a possible interpretation is that the attenuation of the rise in intravitreous glutamate levels is a secondary effect of BDNF’s preventing cell death. This is unlikely, however, because BDNF did not prevent the concomitant increase in GABA concentration that was caused by the damage to the INL by KCN treatment. Thus, our results suggest that BDNF shows a specific effect on glutamate. It is well known that the glutamate uptake system is ubiquitous in INL cells and that INL cells express the TrkB receptor. ⁵³ Therefore, BDNF might have an effect on the glutamate uptake system in the retina, as has been reported for GABA uptake in the cerebral cortex. ^{41 43}  

It has been reported that nitric oxide and reactive oxygen species are involved in cyanide-induced neurotoxicity. ^{36 54} BDNF is known to enhance the activity of the antioxidant system by elevating glutathione peroxidase activity, ^{46 55 56} and this enhancement may provide protection from toxicity. 

BDNF as an Optic Neuroprotective Agent

Retinal damage can have various causes. Ischemic insults cause ocular damage and lead to glaucoma, retinal ischemic disease, retinal degeneration, and optic neuropathy. Many ocular diseases are treated preliminarily by removing the source of the insult. However, application of a new approach, neuroprotection therapy, which can enhance the resistance to disease, is now anticipated. ^{26 27 28 29 57} Our present results show that BDNF has a protective effect when KCN-induced energy depletion in the retina is used to mimic ischemia. Our results and previous reports clearly demonstrate that exogenously administered BDNF shows a therapeutic potential for many retinal injuries. Thus, as a neuroprotectant, BDNF can be expected to provide new strategies for treating retinal ischemic diseases. 

 Reprint requests: Chikao Nakayama, Sumitomo Pharmaceuticals Research Center, 3-1-98 Kasugadenaka, Konohanaku, Osaka 554-0022, Japan.

 Submitted for publication July 29, 1998; revised January 28, 1999; accepted March 2, 1999.

 Proprietary interest category: EN.

 

Figure 1.

 

Distribution of intravitreally injected ¹²⁵I-labeled BDNF in cornea, iris, ciliary body, lens, retina, and sclera. These data (in counts per minute) indicate the mean ± SEM of four independent experiments. Each bar indicates the counts of radioactivity of tissues from ¹²⁵I–BDNF–treated rat eyes (closed bars) or from eyes treated with ¹²⁵I–BDNF plus a 100-fold excess of cold BDNF (hatched bars). The counts from retina were higher than from any other tissues and were displaced by the addition of cold BDNF.

Figure 2.

 

Distribution of intravitreally injected BDNF in the retina. (A) Western blots with anti-BDNF. Retinal protein from the PBS-injected eyes (lane 1) or 10 μg–BDNF–injected eyes (lanes 2, 3) was loaded onto a sodium dodecyl sulfate–polyacrylamide gel. BDNF (0.1 μg) was loaded onto lanes 4 and 5. NGF, NT-3, and NT-4 were loaded onto lanes 6, 7, and 8, respectively. Anti-BDNF (lanes 1, 2, 5, 6, 7, and 8) and preabsorbed anti-BDNF (lanes 3, 4) were used as primary antibodies. Lanes 9, 10, 11, and 12 show a silver-stained gel with bands corresponding to BDNF, NGF, NT-3, and NT-4, respectively. (B, C, D) Immunohistochemical analysis of the retina. (B) Eye injected with PBS. (C, D) Eyes injected with 10 μg BDNF. Fixed eyes were treated with anti-BDNF (B, C) or preabsorbed anti-BDNF (D). Scale bars, (B, C, D) 100 μm.

Figure 3.

 

Photomicrographs of KCN-treated posterior retina. (A) Eye incubated in K–R for 30 minutes. (B) Eye incubated in 5 mM KCN–K–R for 30 minutes. KCN treatment selectively reduced the number of INL cells compared with the K–R–incubated eye. The arrow and arrowhead indicate edema in the nerve fiber layer and inner plexiform layer, respectively. Scale bar, (A, B) 50 μm. (C) High magnification of the region shown within the square in (B). The asterisk indicates the swollen INL cells. Scale bar, (C) 25 μm.

Figure 4.

 

TUNEL analysis of KCN-induced INL cell death. (A) Retina from constant light–exposed rat. Forty-eight constant hours of light exposure resulted in positive outer nuclear cells (arrows). (B) Retina from light-exposed rat without TdT enzyme. None of the retinal cells was positive. (C) KCN-treated retinas with 30 minutes of incubation. With 5-mM KCN treatment, the INL cells were not positive for TUNEL. These data indicate that KCN treatment induced necrotic cell death. Scale bars, 100 μm.

Table 1.

 

Effect of Ca²⁺ Deprivation and Glu Antagonist on KCN–Induced Cell Damage

KCN	CaCl2	Reagents	Cells/880 μm	n
−	+	−	284 ± 9	(11)
+	+	−	190 ± 4	(12)
+	+	MK-801	187 ± 15	(4)
+	+	NBQX	218 ± 8**	(4)
+	−	EGTA	227 ± 7**	(10)
+	−	BAPTA	267 ± 14**	(6)
+	−	Thapsigargin	222 ± 11**	(6)

 Nissl stain–positive cells were counted in four areas, which were 220 μm in length: superior and inferior retina, 880 and 1320μ m from the optic nerve head. Values (mean ± SEM) are the sums of these four areas. The experiments were performed with MK-801 (10μ M), NBQX (150 μM), EGTA (1 mM), BAPTA (5 mM), and thapsigargin (250 μM). The number of eyes (n) is shown in parentheses.

 Statistical analysis was done with Student’s t-test; results from treated groups were compared with the control (+KCN,+ CaCl₂, −Reagents) group (

**   P < 0.01).

Figure 5.

 

Photomicrographs of the KCN-treated posterior retina from BDNF-injected eye. (A) PBS-injected eye after incubation in K–R for 30 minutes. (B) PBS-injected eye after incubation in 5 mM KCN–K–R for 30 minutes. Eyes treated with 1 μg BDNF (C) or 10 μg BDNF (D) and KCN incubation for 30 minutes. BDNF treatment selectively protected the INL cells against KCN-induced injury. BDNF injection was performed 2 days before KCN treatment. Scale bars, 50 μm.

Figure 6.

 

Dose-dependency of BDNF efficacy against the 30 minutes of KCN damage. The numbers of Nissl stain–positive cells were counted in the four microscopic fields as described in the Methods section, and the total number of positive cells in the four regions was indicated as the“ Number of INL cells” on the y axis. Bars represent the mean ± SEM values of 6 to 17 experiments. −KCN and +KCN represent the incubation of isolated eyeballs without and with KCN, respectively. PBS (hatched bars), or 0.1 or 10 μg of BDNF (closed bars) was injected intravitreally 2 days before the experiments. The gray bar indicates the eye in which 10 μg BDNF was injected 1 hour before the KCN incubation. Open bars indicate uninjected eyes. After the preincubation in K–R buffer for 10 minutes, eyes were incubated in 5 mM KCN–K–R for 30 minutes. n = 6 to 17. Statistical analysis was done with Dunnett’s test; results from the treated eyes were compared with those of the PBS-injected KCN treatment group (*P < 0.05, **P < 0.01).

Figure 7.

 

Time course of KCN-induced injury and the efficacy of BDNF. Numbers (mean ± SEM) of Nissl stain–positive INL cells were determined in a similar manner as for Figure 6 . PBS (hatched bars) or 10 μg of BDNF (closed bars) was injected intravitreally 2 days before the enucleation. After a preincubation in K–R buffer for 10 minutes, intact eyes were incubated in 5 mM KCN–K–R for various times. The number of eyes was n = 5 to 17. BDNF transiently showed a protective effect against KCN-induced damage. Statistical analysis was done with Dunnett’s test; the results from treated eyes were compared with those of the PBS-injected KCN treatment group (**P < 0.01).

Figure 8.

 

Western blot analysis of calretinin expression in the retina. (A) Western blots with anti-calretinin polyclonal antibody. Retinal proteins from the eye injected with PBS (lanes 1–4) or 1 μg (lanes 5–8) or 10 μg (lanes 9–12) of BDNF were fractionated by electrophoresis. PBS or BDNF administration was performed 2 days before the enucleation. All these samples expressed a 29-kDa band identified as calretinin (arrow). BDNF treatment enhances the expression of calretinin. (B) Quantification of the amount of calretinin expression in the experiment shown in (A) and another independent experiment. The amount of calretinin expression was estimated by scanning the density of the band using an NIH image analyzer (1.59/fat). Closed bars indicate the PBS- and BDNF-injected eyes. Calretinin expression was shown as the percentage of the PBS-treated group. The number of eyes was n = 8. Statistical analysis was done with Dunnett’s test; results from treated eyes were compared with those of the PBS-treated group (**P < 0.01).

Figure 9.

 

Immunohistochemical analysis of calretinin expression in the retina. (A) Photograph of the expression pattern of calretinin in the PBS-injected retina. The arrow and arrowhead indicate the retinal ganglion and INL, respectively. All the retinal ganglion cells and some of the INL cells stained positively with anti-calretinin polyclonal antibody. Scale bar, 100 μm. (B) Photograph of the expression pattern of calretinin in the 10 μg BDNF–injected retina. BDNF was injected 2 days before the preparation of paraffin sections. The number of calretinin-positive INL cells (arrowhead) was larger than that seen in PBS-treated eyes. Scale bar, 100 μm. (C) Quantification of the number of calretinin-positive INL cells. All calretinin-positive INL cells were counted throughout the sagittally sectioned retina, and the total length of the retina was measured. Each dot indicates the number of positive cells in 1-μg BDNF and 10-μg BDNF treatment groups per 1000-μm length of retina. Statistical analysis was done with Dunnett’s test; results from treated eyes were compared with those of the PBS-treated group (*P < 0.05, **P < 0.01).

Table 2.

 

Effect of BDNF Administration on the Levels of Selected Amino Acids in the Vitreous Body

Treatment	KCN	Glu (pmol/μl)	Asp (pmol/μl)	GABA (pmol/μl)
None	−	103 ± 15 (14)	41 ± 3 (14)	75 ± 13 (14)
None	+	399 ± 80 (14)	78 ± 10 (14)	206 ± 19 (14)
PBS	+	389 ± 49 (12)	106 ± 10 (12)	225 ± 24 (12)
BDNF (1 μg)	+	337 ± 49 (12)	103 ± 14 (12)	256 ± 26 (12)
BDNF (10 μg)	+	245 ± 20 (11)*	83 ± 6 (12)	211 ± 18 (12)

 Concentration (mean ± SEM) of amino acids in the vitreous body. The number of eyes (n) is shown in parentheses. The injection of BDNF or PBS was performed 2 days before the 30-min KCN incubation. Vitreous humor was collected by cutting the eyeballs and analyzed as described in the Materials and Methods section.

 Statistical analysis was done with Dunnett’s tests; results from the BDNF treatment group were compared with the PBS-injected KCN treatment group (

*   P < 0.05).

The authors thank Akiyoshi Kishino, Hirohsi Ogo, Osamu Konishi, and Shigeyuki Honda for their valuable technical advice concerning the immunohistochemical and TUNEL methods; Akira Itoh for providing valuable comments on the manuscript; and Mutsuko Sakai for technical assistance.