BDNF Enhances Retinal Ganglion Cell Survival in Cats with Optic Nerve Damage

Abstract

purpose. To determine whether brain-derived neurotrophic factor (BDNF), a neuroprotectant in the small rat eye, might also serve as an effective neuroprotectant in larger vertebrate eyes.

methods. A cat optic nerve crush model was combined with standard histologic staining and analysis techniques. Twenty-nine animals were studied, with the noninjected eye serving as the control eye.

results. No treatment, or intravitreal injection of sterile water, resulted in an approximately 50% loss of ganglion cells 1 week after nerve crush. By contrast, the mean percentages of surviving ganglion cells measured in eyes receiving injections of 15, 30, 60, and 90 μg BDNF at the time of the nerve damage were 52%, 81%, 77%, and 70%, respectively. Similar values were obtained for ganglion cell density. Cell size measurements suggest a complex response among the different classes of cat ganglion cells; 30 μg BDNF treatment retained the highest number of large ganglion cells, whereas 90 μg minimized the loss of medium-sized neurons and retained normal proportions of large, medium, and small ganglion cells.

conclusions. The data show that BDNF is an effective neuroprotectant in primate-sized eyes after optic nerve injury. Although the amount required to achieve neuroprotection is much greater than that needed for the small rat eye (30 μg versus 0.5 μg), when differences in vitreal volume are considered, the effective dose is similar (0.01 μg BDNF/μl vitreal volume). High doses of BDNF induce inflammation and result in a decrease in total ganglion cell survival but appear necessary to save medium-sized neurons, which are affected most severely by nerve injury.

Primary open-angle glaucoma (POAG) is a disease of the visual system that results, in part, from pressure-induced damage to the optic nerve (ON) at the level of the lamina cribrosa (LC). ^{1 2} In addition to nerve fiber degeneration, the disease also is characterized by atrophy and a loss of ganglion cells from the retina itself. ^{3 4 5 6 7 8} Because previous work has shown that elevation of intraocular pressure (IOP) disrupts axonal transport at the LC, the ganglion cell loss associated with POAG is considered to reflect, in part, a decrease in the level of trophic material these neurons receive from their target sources. ^{9 10 11 12}  

Recent studies, using in vivo and in vitro techniques, have shown that neurons and glia within the mammalian retina contain receptors for different neurotrophic factors, and that direct application of these factors can influence the survival of injured ganglion cells. ^{13 14} In particular, studies of the rat visual system, have indicated that brain-derived neurotrophic factor (BDNF), a member of the nerve growth factor family of proteins, is highly effective in reducing the rate of axotomy-induced retinal ganglion cell death. ^{15 16 17 18 19} BDNF also has been shown to undergo both anterograde and retrograde axonal transport, and it has been implicated in reducing die-back and promoting axonal regeneration after ON injury. ^{20 21}  

Based on these data, and a longstanding interest in different treatment strategies for reducing retinal ganglion cell degeneration in the glaucomatous primate eye, we initiated a series of experiments aimed at determining whether BDNF might also exert a neuroprotective effect on injured ganglion cells in the larger cat eye, where, compared with the small rat eye, drug dose and diffusion differences might be limiting factors. 

Materials and Methods

Subjects

Twenty-nine adult cats were used in this study. All were specific-pathogen free and weighed 3 to 4 kg. The cat was selected for use based on ease of manipulating its ON, the well-defined morphologies of cat retinal ganglion cells, ^{22 23 24 25 26 27} and the similar vitreal volumes of the cat and primate eyes compared with that of the rat (∼3 to 4 ml versus ∼50 μl in the rat; Reference ²⁸ , and unpublished data). All procedures were approved by the Animal Use Committee at Michigan State University, and all adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Surgical Procedures

Initial anesthesia was achieved by placing the cat in a Plexiglas chamber and introducing a mixture of 4% isoflurane (IsoFlo, Abbott Labs, Abbott, IL) and pure oxygen, delivered at 3 l/min. Each cat then was intubated, and anesthesia maintained using a 2.5 to 3.5% isoflurane-oxygen mixture (0.5 l/min.). Analgesia and sedation consisted of an intramuscular injection of glycopyrrolate (0.05 mg/kg; Fort Dodge Labs, Fort Dodge, IA), and subcutaneous injections of torbugesic (0.2 mg/kg; Butler, Columbus, OH) and acepromazine (0.04 mg/kg; Butler). Hydration was maintained intravenously with sterile saline (0.9%). Heart and respiratory rates were monitored every 15 minutes. Body temperature was maintained at 37°C using a heating pad. The head was stabilized using a vacuum-activated, “beanbag-like,” restraining device (Olympic Vac; Olympic Medical, Seattle, WA). In five animals, the pupils were dilated with 1% tropicamide HCl (Mydriacyl; Alcon, Fort Worth, TX) and contact lenses containing 1 to 2 drops of 0.5% proparacaine HCl (Alcaine; Alcon) were placed on the eyes. Pre- and postsurgery fundus photographs of the retinal blood vessels were obtained using a fundus camera (TRC-50; Topcon, The Netherlands). Additional fundus photographs were obtained at the time of sacrifice. 

Using sterile procedures, the bone overlying the left frontal sinus was removed to expose the roof of the bony orbit. All openings to the frontal sinus then were sealed with bone wax. This avoided disturbing the cat’s olfactory senses, which can result in a severe loss of appetite. A fine-tipped scalpel blade was used to make an opening in the dorsal surface of the orbit. Careful blunt dissection of the overlying tissues exposed the ON without disturbing the nerve sheath or retinal artery. The ON was stabilized with a hook, and a smooth-faced bulldog clamp that exerts approximately 1024 g of force was place on the nerve for 15 seconds at a distance 2 to 3 mm behind the globe. The bone wax plugs then were checked, the frontal sinus was packed with Gelfoam (Upjohn, Kalamazoo, MI) soaked with sterile saline, and the overlying skin sutured. The contact lenses were removed, and the eyes treated with sterile ophthalmic ointment. After removal of the intubation tube, each animal was monitored until it was able to move about freely and feed. Postoperative pain medication (torbugesic, 0.2 mg/Kg) was provided as needed. 

BDNF Injections

Single or multiple intravitreal injections (15, 30, 60, or 90μ g at 1μg/μl) of sterile recombinant BDNF (Regeneron Pharmaceuticals, Tarrytown, NY) were made into the left (ON crush) eye of 19/29 cats. Of the remaining 10 cats, 3 did not undergo any surgical procedures, 5 underwent a unilateral ON crush but no treatment, 1 underwent an ON crush and an intravitreal injection of 60 μl of sterile water, and 1 received an intravitreal injection of sterile water, but no ON crush. Three animals that received 90-μg injections of BDNF also received, for 1 week, daily intraperitoneal injections (35 mg/kg) of the nitric oxide synthase blocker, N-ω-nitro-l-arginine-methylester (L-NAME). All BDNF injections were made immediately after the ON crush. In most cases the injections were made through the opening in the frontal sinus at a point approximately 5-mm posterior to the ora serrata. Three animals received a second intravitreal injection of the same dose 4 days after crush (Table 1) . For these injections, which were made just posterior to the ora, the animals were anesthetized with ketamine HCl (10 mg/kg) and the eyes treated with Alcaine (Alcon). All injections were made using a Hamilton syringe with a 30 ga. needle. Care was taken to ensure that the complete bevel of the needle was within the vitreal chamber, but that it did not hit the lens. ²⁹ Intraocular injections were made over a 1 minute period, with the needle left in place for an additional 30 seconds to allow for diffusion of the drug away from the injection site. 

Survival Period and Tissue Processing

After a 7-day survival period, 24/29 of the animals were anesthetized deeply with pentobarbital sodium (50 mg/kg), and perfused transcardially with 0.5 l of saline (0.9%) followed by 2 l of a solution containing 1.5% paraformaldehyde and 2.0% glutaraldehyde in 0.1 M sodium phosphate buffer (pH 7.4). One week was selected as the baseline survival period because initial cell measurements showed that this time period resulted in a consistent, but not overly severe, change in ganglion cell number (∼50% decrease) in the untreated eye (Table 1 , Fig. 5 ). After overnight postfixation, the retinas were dissected, wholemounted onto subbed glass slides, dehydrated, and stained with cresyl violet. Care was taken during mounting to make the relief cuts in superior temporal retina as shallow as possible. This avoided any distortion in the region from which the ganglion cell samples were obtained. 

Ganglion Cell Measurements and Identification

Retinal ganglion cells from normal, untreated, and BDNF-treated eyes were compared using a computer-based imaging system. The region selected for quantitative analysis occupied 1.72 mm², and was located 3 mm above and 1.5 mm temporal to the area centralis (Fig. 1) . This region was chosen because of the relatively constant size and density of ganglion cells in this location of the cat retina. ²² A stage digitizer (AccuStage, Shoreview, MN) was used to properly orient ³⁰ each retina on the microscope, and to standardize the starting point and stage movements used for cell sampling. From the starting point, 42 digital images (41,000μ m²/image) were obtained systematically using a high resolution video camera (chilled CCD, model C5985; Hamamatsu) and 40× objective. The retinal images were collected as three dorsal-ventral passes composed of 14 images each. Double counting was avoided by separating each sample column horizontally by 500 μm, and vertically by using the previous image as a reference. Cell size, density, and number were determined directly from the digital images using image analysis software (Image Pro Plus, Media Cybernetics). Neurons were classified as ganglion cells based on the criteria of Stone. ^{24 25} In brief, they had to display a distinct nucleus and nucleolus, and have a clear ring of cytoplasm completely surrounding the nucleus. Although this conservative approach most likely underestimated the number of small ganglion cells, it also reduced the number of nonganglion cells included in the measurements. ^{26 27} Based on these criteria, our estimates of ganglion cell density, as well as the proportions of small, medium, and large ganglion cells in the normal sample region were comparable to those reported by Stone. ²⁴  

Statistical Analysis

All data are presented as means ± 1SD. Differences in mean cell number, size, and density were compared using one-way analysis of variance (ANOVA), combined with the Bonferroni adjustment for multiple comparisons (SPSS, Chicago, IL). Paired comparisons of cell size distributions were made using the Kolmogorov–Smirnov test (SPSS) for two independent samples. In all cases, P = 0.05 was used as the level of significance. 

Results

Effects of ON Crush on Retinal Vasculature

In all cases, ON damage was achieved by placing a small bulldog clamp on the ON 2 to 3 mm behind the left eye for 15 seconds. Fundus photographs obtained pre- and postsurgery, as well as after the 1 week survival period for five randomly-selected animals, indicated that this brief period of pressure had no adverse effect on the vasculature of the inner retina. 

Cellular Changes in Untreated Retinae

While brief ON compression did not compromise the retinal vasculature, it did result in a significant loss of ganglion cells (∼50%) from the retina within the 1-week test period. Neurons undergoing atrophic changes were identified by their irregular shape, pale-staining cytoplasm, clumped chromatin, and displaced nuclei (Figs. 2A 2B ). Although no degenerating neurons were observed in normal retinas, systematic examination (×250) of the sample region in animals receiving an ON crush but no treatment with BDNF revealed a high density of atrophic profiles (∼26.9 profiles/mm²) in these retinas. 

The data presented in Figure 3 compare the cellular morphology of the normal cat retina with that from cats that received a unilateral ON crush, a 1-week survival period, and no BDNF treatment. For the 24 normal retinas examined, we found the sample regions to contain an average of 539 ± 48 ganglion cells, with a mean population soma size of 411 ± 47μ m², and a mean cell density of 313 ± 28 neurons/mm² (Table 1) . Twenty six percent of the ganglion cells in the sample region had small (65–300μ m²) somata, 68% had medium-sized cell bodies (301–800 μm²), and 6% contained large (800–2000 μm²) somata. 

One week after ON crush, the mean number of ganglion cells within the sample region decreased to 263 ± 39 neurons, a reduction of 51.2% (Table 1 , Figs. 3 5 ). Mean soma size decreased 23%, to 316 ± 57μm², and cell density now was only about one-half normal (151 ± 28 neurons/mm²). The percentage of neurons with large cell bodies decreased from 6% to 3.2%, whereas those with medium-sized somata decreased from 68% to 40.6% of the total population. While the proportion of small ganglion cells within the sample region more than doubled (56.2% versus 26%), this increase was due primarily to the loss of large (decrease from 98 to 25 neurons) and medium-sized neurons (from 1099 to 316 neurons), and not a significant increase in the number of small neurons (from 420 to 438 neurons) due to cell shrinkage. These cellular changes are reflected in the cell size histograms of Figure 3 , which show a clear decrease in the number of large and medium-sized ganglion cells, and a shift toward smaller soma sizes. The untreated retinas also showed a clear increase in the number of glial cells (compare normal with untreated, Fig. 3 ). Despite these changes, however, most surviving ganglion cells retained their round-oval shape, and well-defined nucleus and nucleolus. 

Cellular Changes in Sham and BDNF-Treated Retinae

Single Injections.

The photomicrographs and cell size histograms in Figure 4 compare the cellular morphology of the 24 normal retinas with that from the 12 animals (three per treatment condition) that received intravitreal injections of BDNF at the time of nerve crush. In all cases, the BDNF-treated animals were examined after a 1-week survival period. 

Injecting either sterile water (60 μl) or 15 μg of BDNF into the vitreal chamber at the time of the nerve crush had little or no beneficial effect. The retinas from these cats did not differ qualitatively or quantitatively from those of cats that received an ON crush, but no BDNF treatment (Table 1 , Figs. 3 5A ). Both the untreated and the 15 μg BDNF-treated retinas contained many microglia and a high density of atrophic profiles (∼28 profiles/mm²). Both also showed comparable ganglion cell loss (51% and 48%, respectively). The density of ganglion cells in the sample regions of the untreated and 15μ g–treated retinas were 151 neurons/mm² and 163 neurons/mm², approximately one half the cell density measured in the normal retinas (313 neurons/mm²). Cell size measurements (Fig. 5B ) indicated an approximately twofold greater decrease in mean soma size in the untreated eyes (23%) than in those receiving 15 μg of BDNF (11%), but neither reduction was significantly different from normal. The sample region in the BDNF-treated eyes contained a lower percentage of small ganglion cells (46.8% versus 56.2%), and higher percentages of medium (48.2% versus 40.6%) and large (5.2% versus 3.2%) ganglion cells compared with the untreated eyes. Although this difference is indicated by a reduced amount of skew in the cell size histogram of the 15 μg BDNF versus untreated animals (compare Figs. 3 and 4 ), the cell size distributions for both groups of animals were statistically different from normal. 

Intravitreal treatment with 30 μg of BDNF at the time of nerve crush resulted in a significant improvement in the number and appearance of surviving ganglion cells (Figs. 4 5) . Neurons in these retinas had well-defined membranes, uniformly distributed Nissl substance, and a clear nucleus and nucleolus. The mean percent difference in ganglion cell number between the normal and treated eyes for these animals was 19%, indicating a survival level of 81%. This represents a significant saving of ganglion cells compared with the untreated and 15μ g BDNF-treated eyes (49% and 52%, respectively). Although ganglion cell density in the sample region of these animals was only 81% of normal, it was 60% higher than that measured in the untreated and 15μ g BDNF-treated retinas. The density of atrophic profiles in the 30μ g BDNF-treated animals was only 4.6 profiles/mm², significantly lower than the approximately 28 profiles/mm² measured in the untreated and 15 μg BDNF–treated retinas. Ganglion cells in the 30μ g BDNF–treated eyes were approximately 5% smaller than normal (389μ m² versus 411 μm²), but as a population had a mean soma size that was 13% to 23% greater than that of the untreated (316 μm²) and 15μ g BDNF–treated (367 μm²) eyes. These differences were not statistically significant. The percentages of ganglion cells within the sample region with small, medium, and large somata in the 30 μg BDNF–treated eyes were 39.9%, 53.9%, and 6.2%, respectively. This represented an increase in the proportions of cells with large and medium-sized somata, which resulted in a broadening of the cell size distribution compared with normal. The two distributions, however, were not statistically similar. 

Increasing the dose of BDNF to either 60 μg or 90 μg also resulted in a significant improvement in the appearance and number of surviving ganglion cells when compared with either no treatment or treatment with 15 μg of BDNF. Similar to the 30 μg BDNF–treated animals, the sample regions of these eyes contained low densities of atrophic profiles (6.6 and 1.3 profiles/mm², respectively). However, unlike the 15 μg and 30 μg BDNF–treated animals in which increased levels of BDNF produced increased numbers of surviving ganglion cells, in these eyes ganglion cell number decreased with the application of higher doses of neuroprotectant (30 μg: 81%; 60 μg: 77%; 90 μg: 70%). The mean number of ganglion cells in the sample region of the 60 μg BDNF–treated eyes (414 cells) was significantly greater than that measured in the untreated (263 cells) and 15 μg BDNF–treated eyes (281 cells), but the mean number measured in the 90 μg BDNF–treated eyes (378 cells) was not (Fig. 5A , Table 1 ). Overall, the ganglion cells in the 60 μg–treated eyes were approximately 16% smaller than normal (345μ m² versus 411 μm²), whereas those in the 90 μg–treated eyes were slightly larger than normal (420 μm²). The sample region in the 60μ g–treated eyes showed a slightly lower than normal proportion of ganglion cells (4.2% versus 6%) with large somata and a higher than normal proportion (53% versus 26%) with small somata. The cell size distributions indicated a continued reduction in the number of ganglion cells with medium-sized somata. By contrast, the proportions of ganglion cells with small, medium, and large somata in the eyes treated with 90 μg of BDNF were almost identical with those measured in the normal eyes (30.5% versus 26%; 64.2% versus 68%; 5.4% versus 6%). Mainly, this was due to the increased survival of medium-sized ganglion cells. Nevertheless, the cell size distributions for both the 60 μg and 90 μg remained statistically different from normal. 

One noticeable difference between the 30 μg, 60 μg, and 90 μg BDNF–treated retinas was a clear increase in the number of inflammatory cells present with increased levels of the drug (Figs. 4 and 6) . In most cases the inflammatory cells were distributed near blood vessels, or scattered randomly across the retina. However, in some areas these cells appeared to be clustered over specific neuronal profiles (Fig. 6B) . 

Multiple BDNF Injections.

Three cats with unilateral ON crush received dual injections of BDNF. In all animals, the first injection was made at the time of the ON crush and the second injection 4 days post-crush. Two cats received double injections of 30 μg of BDNF, while the third cat received multiple 60 μg BDNF injections. In all cases, the percentage of surviving ganglion cells was comparable to that of a single injection given at the time of the nerve crush (30 μg: 81% versus 84%; 60μ g: 77% versus 76%). Similarly, no differences were seen with respect to single versus multiple injections and mean soma size (30μ g: 389 μm² versus 408μ m²; 60 μg: 345 μm² versus 345 μm²) or ganglion cell density (neurons/mm²) within the sample area (30 μg: 254 versus 264; 60 μg: 244 versus 237). 

Combined BDNF and L-NAME Injections

Three cats that received high doses of BDNF (90 μg) also were treated during the 1-week survival period with daily injections of L-NAME, a nitric oxide synthase specific inhibitor (35 mg/kg per day). Although treatment with L-NAME eliminated the inflammatory response induced by the high levels of BDNF, it did not result in a significant change in the size, number, or density of ganglion cells measured when compared with animals that received 90 μg injections of BDNF alone (size: 315.3 ± 32.6 μm² versus 420 ± 64 μm²; number: 383 ± 62.9 cells versus 378 ± 31cells; density: 223 ± 36.4 neurons/mm² versus 219 ± 18 neurons/mm²). 

Discussion

The primary goal of this study was to determine whether BDNF, a well-known neuroprotectant in the small rat eye, might also serve as an effective neuroprotectant in primate-sized eyes, where dose and diffusion differences may be limiting factors. The importance of these data derives from their potential use in the treatment of retinal diseases, and in particular glaucoma, where a reduction in target-derived neurotrophin levels has been implicated as playing a role in retinal ganglion cell death. ^{11 12}  

In agreement with previous studies in the rat, ^{15 16 17 18 19} our data show that intravitreal application of BDNF also can enhance retinal ganglion cell survival in cats after ON injury. This result was not unexpected, because ganglion cells in the cat retina (Chen H, unpublished data, 2000) like those in the rat and other species, ^{31 32 33} express TrkB, the high-affinity BDNF receptor. However, in contrast with the rat, in which relatively small amounts of BDNF (∼0.5–5 μg), and sometimes vehicle solution alone, have been shown to promote ganglion cell survival, we found that in the larger cat eye, vehicle solution alone had no beneficial effect, and that approximately 30 μg of BDNF was needed to achieve a significant level of neuroprotection. Although this amount of drug may appear excessive, when it is taken into consideration that the vitreal chamber of the cat eye is approximately 60-fold larger than that of the rat (3 ml versus 50 μl), the effective dose for these different eyes is approximately the same (∼0.01 μg BDNF/μl of vitreal volume). 

Increasing the amount of BDNF injected above 30 μg resulted in a decrease, rather than increase, in ganglion cell survival (Fig. 5A) . Eyes receiving 30 μg of BDNF showed the highest level of survival (81%), whereas those receiving 60 and 90 μg showed progressively fewer surviving cells (77% and 70%, respectively). A similar dose-related limitation in BDNF effectiveness has been reported in the rat retina, ³⁴ as well as in other areas of the central nervous system (CNS). ^{35 36 37} Although many factors may be involved, recent work has focused on two specific mechanisms. The first concerns BDNF-induced nitric oxide (NO) neurotoxicity, ^{36 38 39 40} and the second involves BDNF-induced downregulation of the TrkB receptor. ^{35 41 42 43} Nitric oxide is a relatively ubiquitous molecule that modulates a number of different physiological processes. Typically, NO is localized in a tissue by immunocytochemical recognition of its synthesizing enzyme, nitric oxide synthase (NOS). Of the different isoforms, neuronal NOS (nNOS) and inducible NOS (iNOS) have been studied most completely in the retina. Species differences aside, there is good evidence that nNOS is found in all five of the major cell types of the vertebrate retina (ganglion, amacrine, bipolar, horizontal and photoreceptor), and iNOS is associated primarily with Müller cells and microglia. ^{38 39 44 45} Recent studies in the rat have shown that both nNOS and iNOS activity are elevated after ON section and/or intravitreal injection of BDNF. ^{38 39} That increased NOS activity affects retinal ganglion cell survival is indicated by the neuroprotective action of concurrent administration of NOS inhibitors. ^{34 38 39}  

In addition to BDNF-induced iNOS activation in microglia, it also has been hypothesized that BDNF induces iNOS activity in immune-competent cells. ³⁸ Both these mechanisms are relevant to the present study, where ON crush produced an increase in the number of microglia (Figs. 3 4) and high doses of BDNF (90 μg), with ON crush or alone, generated a strong inflammatory response within the retina (Figs. 4 6) . We unexpectedly found that treatment with the NOS-inhibitor L-NAME blocked the BDNF-induced inflammatory response, but did not enhance ganglion cell survival, a result in direct contrast with that obtained in the rat. Although it is possible that differences in dose (50 mg/kg per day versus 35 mg/kg per day) ³⁸ and route of administration (intravitreal versus intraperitoneal) ³⁹ are the cause of this variation, that the retinas of the L-NAME treated cats appeared normal suggests that iNOS derived from BDNF-activated immune cells was not a limiting factor. It does not rule out, however, incomplete blockade of other NO sources. 

BDNF exerts its influence on ganglion and other cells in the retina via TrkB receptors. ^{46 47} There are two types of TrkB receptor: full length and truncated. The basic difference between the two is that the truncated form shows some amino acid residue variations and lacks the cytoplasmic tyrosine kinase domain. Neurotrophin binding to the extracellular domain of the full-length receptor induces phosphorylation of tyrosine residues within the cytoplasmic domain. Downstream of the phosphorylated internal domain are several intracellular signaling pathways, and activation of these pathways has been shown to regulate gene expression related to cell death and survival. ^{48 49 50 51 52} Recent studies have demonstrated that continuous application of BDNF to the brain or cultured neurons results in a decrease in TrkB receptor protein and/or mRNA, ^{35 41 42 43} and we have found this also to be true in the rat retina. ⁵³ A single injection of BDNF (5 μg) produces approximately a 96% decrease in retinal TrkB protein over the first 24 hours. Recovery is slow, achieving only approximately 31% of normal at 14 days after injection. Studies using chimeric structures of the TrkA and TrkB receptors indicate that a short sequence in the juxtamembrane region of the cytoplasmic domain is responsible for neurotrophin-induced downregulation of the TrkB receptor. ⁴² Based on these data, it is reasonable to hypothesize that BDNF-induced downregulation of the TrkB receptor may also have played a role in limiting drug effectiveness in the present study. In addition, it may also have been responsible for our failure, and that of others, to achieve enhanced ganglion cell survival through the administration of multiple injections of BDNF. Given the rapid receptor downregulation and long recovery time, ⁵³ it is not surprising that our administration of a second BDNF injection just 4 days after ON crush did not increase ganglion cell survival over that seen with the initial treatment. DiPolo et al., ¹⁹ did not find a similar decrease in TrkB effectiveness (until ∼10 days after axotomy) with prolonged delivery of BDNF; however, this result may reflect a positive side to their approach. By transfecting retinal Müller cells to produce and release BDNF, their method of drug delivery was much less invasive than that used here and in the other rat studies. In addition, the relatively slow delivery of BDNF in their retinas may have allowed adjacent truncated TrkB receptors to better buffer the concentration of drug within the eye, ^{54 55} thereby preventing rapid activation of various inhibitory mechanisms. 

The cell size measurements (Figs. 3 4) indicate that although both large and medium-sized ganglion cells are affected, ON crush has a much more severe effect on the medium-sized cells. One week after ON crush and no BDNF treatment, both populations of neurons showed approximately a 75% reduction in ganglion cell number within the sample area. However, because medium-sized ganglion cells comprise a much larger proportion of all ganglion cells in the cat retina, ^{22 23 24 25 26 27} the number of medium-sized ganglion cells lost from these retinas was approximately 10 times greater than that of large ganglion cells (783 medium versus 72 large). Shrinkage of medium-sized cells did not appear to be a significant factor; there was only a 4% increase in the number of small ganglion cells in these animals. A rapid and severe loss of medium-sized ganglion cells is consistent with other studies of ON damage in the cat, ^{56 57 58} including elevation of IOP. ⁵⁹ A similar selective loss of medium-sized cells also has been reported in the avian retina with experimental glaucoma. ⁶⁰ Although these results appear to be in contrast with human glaucoma, in which it generally is thought that large ganglion cells are most susceptible, anatomic ⁷ and physiological ^{61 62} evidence indicates that small and medium-sized ganglion cells also can be affected severely in the glaucomatous human retina. 

The primary effect of the BDNF treatments was to restore balance to the proportions of small-, medium-, and large-sized ganglion cells within the sample region. Treatment with 30 μg BDNF saved the largest number of ganglion cells, including the highest number of large ganglion cells. The increase in medium, and particularly small ganglion cells, may reflect the increased ability of the BDNF at this dose to block the rapid loss of medium-sized cells, but not maintain their normal size. Increasing the amount of BDNF injected to 90 μg caused a reduction in the number of large and small ganglion cells, but produced the largest saving of medium-sized neurons. This resulted in a normal balance in the cell proportions. Although more cell-specific studies are needed, the data suggest differential sensitivities of large and medium-sized ganglion cells to intravitreal application of BDNF. Large ganglion cells appear to respond well to low doses, but their survival is limited by high doses. Medium-sized cells appear to respond less well to low doses, but do not show a decline with high levels of drug application. This differential effect may reflect differences in the number and type (full length versus truncated) of TrkB receptors present on large versus medium ganglion cells, or it might reflect differences in retinal circuitry. The close association of large ganglion cells with amacrine cells, which also contain TrkB receptors and are capable of producing NO, may be a disadvantage to these neurons in the presence of high levels of BDNF. Ongoing studies are designed to isolate the potential differential influences of BDNF on the various classes of cat ganglion cells. 

In summary, the data presented here indicate that BDNF is a suitable neuroprotectant for use in primate-sized eyes. The best intravitreal dose for short-term treatment, which may be sufficient when combined with a reduction in IOP, appears to be approximately 0.01 μg BDNF/μl vitreal volume. Improving the long-term neuroprotective effectiveness of BDNF will require a better understanding of the differential effect the drug has on different classes of ganglion cells, as well as the relation between these neurons and other BDNF-sensitive elements of the retina. 

 Supported by National Eye Institute Grant EY11159 (AJW) and a grant from the Michigan State University Foundation.

 Submitted for publication August 22, 2000; revised December 11, 2000; accepted January 8, 2001.

 Commercial relationships policy: N.

 

Table 1.

 

Summary of Cell Number, Size, and Density Measurements

	Normal	No Treatment					BDNF Treatment
				1 wk	2 wk	4 wk	15 μg	30 μg	60 μg	90 μg	30 μg × 2*	60 μg × 2*
Cell number	539 ± 48	263 ± 39 (49)	123 ± 0 (23)	50 ± 0 (9)	281 ± 41 (52)	438 ± 42 (81)	414 ± 92 (77)	378 ± 31 (70)	454 ± 5 (84)	407 ± 0 (76)
	24, †	3	1	1	3	3	3	3	2	1
Cell size (μm2)	411 ± 47	316 ± 57 (77)	476 ± 0 (116)	443 ± 0 (108)	367 ± 51 (89)	389 ± 130 (95)	345 ± 14 (84)	420 ± 64 (102)	408 ± 9 (100)	345 ± 0 (84)
	24	3	1	1	3	3	3	3	2	1
Cell density (N/mm2)	313 ± 28	151 ± 25 (48)	72 ± 0 (23)	29 ± 0 (9)	163 ± 24 (52)	254 ± 25 (81)	244 ± 58 (78)	219 ± 18 (70)	264 ± 3 (84)	237 ± 0 (76)
	24	3	1	1	3	3	3	3	2	1

 Data are means ± SD, with percentage of normal shown in parentheses and the number of subjects on the next line.

*  BDNF injections on day 0 and 4.

†  Number of eyes examined: includes fellow eyes from crush only and BDNF-treated animals, but not sham or L-NAME treated.

Figure 1.

 

Schematic drawing showing the approximate location of the retinal sample area (stippled) for the left retina. The area centralis (AC) and optic disc (OD) were used as references to properly orient the retina.

Figure 2.

 

Low (A) and high (B) power photomicrographs showing examples of normal (stars) and degenerating (arrowheads) ganglion cells after ON crush. Examples of glial cells are indicated by arrows. Atrophic neurons are recognized by their irregular shape and clumped chromatin. Scale bar, 30 μm (A); 15 μm (B).

Figure 3.

 

Photomicrographs and cell size histograms comparing the cellular morphology of the normal retina with that from cats that received an ON crush but no BDNF treatment. ON injury resulted in a significant loss of medium- and large-sized ganglion cells, and an increased level of gliosis. The data are based on 3 animals under each condition. Scale bar, 50 μm.

Figure 4.

 

Photomicrographs and cell size histograms comparing the cellular morphology of the normal retina with that of cats that received an ON crush and different levels of BDNF treatment. The highest percentage of surviving ganglion cells were measured after 30 μg injections. Higher doses resulted in inflammation and reduced ganglion cell survival. The data are based on 3 animals under each condition. Scale bar, 50 μm.

Figure 5.

 

Summary histograms comparing the changes in cell number and size after different BDNF treatment paradigms. Eyes receiving 30 μg and 60 μg injections of BDNF showed significantly greater ganglion cell survival than the untreated or 15 μg treated eyes. Values are the number of eyes examined under each condition. Error bars are ± SD.

Figure 6.

 

Photomicrographs showing the inflammatory response produced by injection of 90 μg of BDNF. Most inflammatory cells were either randomly distributed or associated with blood vessels (A, arrows). Others were clustered over unidentified retinal profiles (B, arrowheads). Scale bar, 50 μm.

The authors thank Judith McMillan for technical assistance with the surgical procedures, histology, and cell measurements.