BDNF Increases Survival of Retinal Dopaminergic Neurons after Prenatal Compromise

Abstract

purpose. Chronic placental insufficiency (CPI) severe enough to cause growth restriction (GR) results in alterations to the retina, including a reduction in tyrosine hydroxylase immunoreactive (TH-IR)-dopaminergic amacrine cells. Brain-derived neurotrophic factor (BDNF) plays a role in the development of the retinal dopaminergic network and may therefore be an appropriate therapy for restoring dopaminergic cells after prenatal compromise. This study was conducted (1) to establish whether BDNF and its receptor NTRK2 (Trk B) are altered in the retina after CPI and (2) to explore the potential of BDNF to enhance dopaminergic cell survival in organotypic retinal cultures from prenatally compromised animals.

methods. CPI was induced in pregnant guinea pigs at 30 days’ gestation (dg; term, ∼67 dg) via unilateral ligation of the uterine artery. Fetuses were euthanatized at 60 dg and the retinas prepared for enzyme-linked immunosorbent assay (ELISA) analysis of BDNF protein levels and for immunohistochemistry to localize BDNF and NTRK2. Organotypic cultures of retinas from GR and control fetuses at 50 to 52 dg were treated with BDNF, and dopaminergic amacrine cells counts were assessed.

results. Retinal BDNF protein levels and the intensity of BDNF-immunoreactivity (IR) in the ganglion cell layer were reduced (P < 0.05) in GR fetuses compared with control fetuses. Addition of BDNF to organotypic cultures increased (P < 0.05) the survival and neurite growth of dopaminergic neurons from both control and GR fetuses.

conclusions. Alterations to BDNF levels may underlie reductions in dopaminergic amacrine cells observed after CPI. The addition of BDNF has the potential to increase survival and neurite growth of dopaminergic amacrine cells.

A compromised intrauterine environment can adversely affect the development of the retina and visual system as very low birth weight (VLBW) and small-for-gestational-age (SGA) infants have an increased risk of visual impairment. ^{1 2} The mechanisms underlying such abnormal development should be explored further. In a guinea pig model of chronic placental insufficiency (CPI), we have previously reported that the electroretinogram (ERG), which measures retinal activity, is altered and also that retinal tyrosine hydroxylase-immunoreactive (TH-IR) dopaminergic amacrine cells are reduced in number. ³ Furthermore, the retinal content of dopamine is reduced ⁴ in these animals. Dopaminergic cells are of interest, as dopamine has been shown to modulate oscillatory potentials (OPs), a component of the ERG that originates in the inner retina. Although the role of OPs in visual processing is not well understood, OPs are sensitive indicators of retinal disturbances, including ischemia, ⁵ retinopathy of prematurity (ROP), ⁶ and diabetic retinopathy. ^{7 8}  

Treating at-risk individuals before birth is likely to be challenging; therefore, an effective approach may be to enhance the neuronal environment immediately after birth during the period of neonatal brain plasticity. Administration of growth factors may be a successful strategy, because of their efficacy in supporting surviving cells, ⁹ enhancing neurite outgrowth ^{9 10} and synaptogenesis, ^{10 11} and stimulating oligodendrocyte development and myelination. ¹² Brain-derived neurotrophic factor (BDNF) is upregulated in retinal ganglion cells (RGC) in response to axonal injury ¹³ and hypoxia/ischemia ¹⁴ and promotes RGC survival both in vitro ¹⁵ and in vivo. ¹⁶ However, if an insult is prolonged or particularly severe, BDNF expression may be suppressed by the impairment of intracellular signaling pathways and may result in increased cell death. ¹⁷ Intraocular injections of BDNF in the postnatal (P) rat (P14–P22) results in more dopaminergic fibers in the retina, ^{18 19} suggesting that BDNF may be important in determining the density of retinal dopaminergic innervation. ¹⁸ BDNF also promotes the in vitro development and differentiation of the retina in general, increasing the number of cells in the nuclear layers and thickness in the plexiform layers. ²⁰  

CPI induced in the guinea pig by unilateral artery ligation mimics a form of compromise that can occur in humans, albeit at the more severe end of the spectrum, reducing nutrient and oxygen supply to the fetuses in the ligated horn and can result in fetal growth restriction. ²¹ The guinea pig is a particularly useful model for human retinal development, as it has a long gestation (term, ∼67 days’ gestation, dg) with major events in development occurring in utero. ²² CPI is induced just before midgestation (28–30 dg) during a critical period for retinal development. Neurogenesis and differentiation are under way at this stage, but the insult occurs before the onset of synaptogenesis and the formation of the photoreceptor inner and outer segments. ²² In addition to TH-IR amacrine cells, horizontal ⁴ and substance P-IR amacrine cells ²³ are reduced after CPI in the guinea pig. Axonal and dendritic growth and synaptogenesis are also compromised as demonstrated by reductions in the plexiform layers. ^{3 4} Furthermore BDNF levels are reduced and its receptor, NTKR2 (previously known as Trk B) is increased at least in the hippocampus in this model. ²⁴  

In light of these findings, the initial goal of this study was to determine whether BDNF immunoreactivity (IR) and protein levels and NTRK2-IR are altered in the retina after CPI and therefore contribute to the reduction in the number of dopaminergic cells. After it was established that BDNF levels are reduced, the second goal was to develop an in vitro system in which to test the neuroprotective effects of BDNF on cell survival after CPI. We selected 50 to 52 dg for the following reasons: Organotypic retinal cultures are viable; fetal growth restriction could be reliably identified; neurons were still undergoing programmed cell death; and retinal development at this age equates with the immediate postnatal period in humans. 

Methods

Animal Preparation

Our study conformed to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and to institutional guidelines. Unilateral ligation of the maternal uterine artery ²⁵ was performed on pregnant guinea pigs (n = 34) at 28 to 30 dg. Briefly, pregnant sows were anesthetized intramuscularly with xylazine (6 mg/kg; Ilium Division; Troy Laboratories, Smithfield, NSW, Australia) and ketamine (40 mg/kg; Troy Laboratories); a midline incision was made and the mesometrium of one uterine horn exposed. The uterine artery was then ligated near the cervical end of the arterial arcade. The ligation remained in place for the duration of the experiment. This procedure results in growth restriction in approximately 50% of fetuses in the ligated horn. Fetuses from the unligated horn served as the control. 

Part A: Effects of CPI on the Guinea Pig Retina—BDNF and NTRK2 Localization and Levels

Tissue Preparation.

Fetuses were delivered by Caesarean section at 60.8 ± 0.2 dg (control, n = 17; growth restricted [GR], n = 12) after the mothers were deeply anesthetized with pentobarbitone sodium (Nembutal, 130 mg/kg IP; Rhone Merieux, Pinkenba, QLD, Australia). The fetuses were weighed and criteria applied to establish growth restriction. ²³ The eyes were enucleated and either immediately immersion fixed in 4% paraformaldehyde in phosphate buffer (pH 7.4) for 2 hours and frozen for cryostat sectioning and immunohistochemistry, or the retinas were immediately dissected, weighed, and snap frozen for ELISA analysis of BDNF protein. 

ELISA.

Assays were performed for each retina (BDNF Emax Immuno Assay system; Promega, Madison, WI). Each retina (∼0.02 g) was homogenized individually in 100 μL of a buffer (50 mM Tris-HCl [pH 7.4]; 50 mM NaCl) containing a cocktail of protease inhibitors (10 mM EDTA, 10 mM amino-caproic acid, 0.25 mM PMSF, 5 mM ethylmaleimide, 10 mM benzamidine, and 5 U/mL aprotinin) and the supernatant collected. BDNF standards (100 μL of 4–500-pg/mL solutions) and 100 μL of unknown samples were run in duplicate according to the manufacturer’s instructions. Reaction products were measured at A_450nm using a multilabel counter (Wallac Victor ² 1420; Perkin Elmer, Boston, MA). The protein content of each retina was determined by using a BCA protein assay kit (Pierce, Rockford, IL) according to the manufacturer’s instructions. The results were expressed as concentration of BDNF (in picograms per milligram of protein). 

Immunohistochemistry.

Vertical cryostat (Leica, Wetzlar, Germany) sections (15 μm) of the entire eye including the retina were collected at the level of the optic nerve head and processed for IR by using the avidin-biotin peroxidase complex. Antibodies were used at the following dilutions: rabbit anti-BDNF (1:1000; donated by Qiao Yan, Amgen, Inc., Thousand Oaks, CA ²⁶ ) and rabbit anti-NTRK2 (1:1000: H-181, recognizing the extracellular domain; Santa Cruz Biotechnology, Santa Cruz, CA). The specificity of the BDNF antigen for guinea pig tissue has been characterized by Western blot analysis. ²⁷ The specificity of the NTRK2 has been characterized in rat ²⁸ and human tissue. ²⁹ The sections were incubated overnight in the appropriate dilution of primary antibody, incubated in biotinylated secondary antibody (1:200; anti-mouse IgG; Vector Laboratories, Burlingame, CA), incubated in the avidin-biotin complex (1:200; Vector Laboratories), and reaction product visualized with 3,3′-diaminobenzidine (DAB; Sigma-Aldrich, St. Louis, MO) in 0.01% hydrogen peroxide. For each antibody, all control and GR retinas were reacted simultaneously to minimize procedural variation. Control experiments were performed by omitting the primary antibodies; in these experiments, staining failed to occur. 

Analysis.

BDNF and NTRK2-IR cells in the ganglion cell layer (GCL) were counted in vertical sections at the level of the optic nerve head with an image-analysis system (Image Pro Plus ver. 4.1, magnification ×6.75; Media Cybernetics, Frederick, MD). Five sections were randomly selected per animal, and five regions of peripheral retina analyzed per section (25 measurements/animal). Peripheral retina was selected from the inferior temporal quadrant, and a strip was collected halfway between the optic disc, extending to the ora serrata. The total number of immunoreactive cells was counted and the results expressed as cells per millimeter of GCL. Qualitative assessment was also made of the intensity and distribution of BDNF and NTKR2 between groups. 

Part B: Establishment of Optimal Tissue Culture Conditions for Dopaminergic Cell Survival and for BDNF Treatment Levels

Retinas were cultured in static organotypic cultures on a semipermanent filter at the interface between the gas and culture medium phase (culture medium supplied from below the filter). Nonsurgical pregnant sows were anesthetized with pentobarbitone, and the fetuses (n = 8 at each of 40, 50, 55, and 60 dg) were removed via caesarean section. The eyes were enucleated, and the retinas were dissected from the pigment epithelium (RPE) and immediately transferred to a dissecting buffer (Ca²⁺/Mg²⁺ Hanks’ balanced salt solution [Invitrogen-Gibco, Grand Island, NY], distilled H₂O, 0.3 M HEPES [pH 7.4; Sigma Aldrich], and penicillin-streptomycin). The retinas were cultured in six-well plates (Nalgene; Nalge Nunc Int., Roskilde, Denmark) on 0.45-μm transparent membrane inserts (30 mm diameter; Millipore, Bedford, MA) in DMEM culture medim (Invitrogen-Gibco), 10% bovine serum albumin (Sigma-Aldrich), 1% penicillin-streptomycin, and 200 mM/L glutamine (Sigma-Aldrich). The retinas were mounted ganglion cell side uppermost in the center of the membrane inserts (1 retina per insert), cultured at 37°C for 2 days in 10% CO₂, and transferred to 5% CO₂ for 1 to 8 days, with the medium changed daily. The cultures were terminated by fixing retinas with 4% paraformaldehyde in phosphate buffer for 1 hour. 

Pilot studies were performed to establish the appropriate length of time to culture retinas (3, 5, 7, or 10 days) for optimal neuronal survival at 40, 50, 51 52, 55, and 60 dg; 50 to 52 dg for 5 days was selected as the ideal preparation to retain retinal integrity. In a second series of experiments, the cultures were treated with increasing concentrations of BDNF (1, 3, 10, 30, and 100 ng/mL rhBDNF; Amgen, Inc., Thousand Oaks, CA), to determine the optimal concentration to enhance TH-IR dopaminergic cell survival (n = 4 retinas at 50–52 dg for each dose). 

Retina Preparation.

After fixation, the retinas were prepared either for TH-IR (wholemounts or vertical strip collected and embedded in paraffin; 8 μm) or fixed for a further 2 hours in 1% glutaraldehyde in 4% paraformaldehyde in phosphate buffer, with the vertical strips (∼3 mm, at the level of the optic nerve head) embedded in Epon-Araldite (Pro Sci Tech, Thuringowa, QLD, Australia) for structural analysis. 

TH-IR.

Retinas were processed for TH-IR using the avidin-biotin peroxidase complex, as described previously, ²³ using mouse anti-TH (1:1000, 72 hours; Chemicon International, Temecula, CA). All control and GR retinas were reacted simultaneously to minimize procedural variation. Control experiments were performed by omitting the primary antibodies; in these experiments, staining did not occur. 

Analysis.

Qualitative analysis was performed in semithin (1 μm) vertical sections stained with methylene blue to assess cell survival and morphologic preservation of the retina after culture. Cells undergoing apoptosis were identified by nuclear fragments, which appear as discrete basophilic masses of clumped chromatin (apoptotic bodies) and a clear cytoplasm. ³⁰ Apoptotic figures in the GCL were assessed with an image-analysis system (magnification, ×2500; 25 measurements/retina for two retinas at each survival time i.e., 3 to 10 days; 50 measurements in total) and the results expressed as cells/mm of GCL. Counts of total type I (TI) and TII TH-IR amacrine cells were performed (CASTGRID system, ver. 1.10; Olympus, Birkeroed, Denmark), with the software set for random sampling of 100 fields per retina (field area 0.04 mm²). ⁴ Retinal areas were determined from projected images of wholemounts by using a computerized digitizing pad (Sigma Scan Pro, ver. 4.0; Media Cybernetics). Total counts for each cell type were calculated from the mean density and retinal areal measurements. Somal areas were assessed in 50 randomly sampled neurons for each retina using an image-analysis system and a mean area calculated. Vertical TH-IR sections were assessed qualitatively, to determine whether BDNF treatment caused alterations to the normal stratification of TH-IR cells. In the intact guinea pig retina, the TH-IR TI population have large cell bodies with stellate dendrites stratifying in sublaminar 1; TII cells are smaller in size and are located in the INL and GCL, but do not have processes identifiable with TH-IR. ³¹  

Part C: BDNF Application to Retinal Explants from GR Animals (50–52 dg)

Animal Preparation and Organotypic Retinal Cultures.

From the initial experiments (part B) 50 to 52 dg was established as the optimum age for retinal explant survival (see the Results section). Pregnant sows, that had undergone uterine artery ligation at 28 to 30 dg, were euthanatized with an overdose of pentobarbitone at 50 to 52 dg and control (n = 10) and GR (n = 8) fetuses collected. The eyes were enucleated and the retinas cultured as just described. Left and right retinas from each animal were randomly allocated to receive either control or supplemented media. The optimal dilution to enhance TH-IR amacrine cell survival was determined to be 30 ng/mL (see the Results section) and was consequently the working dilution used in this series of experiments. The cultures were terminated on day 5 by fixing for 1 hour in 4% paraformaldehyde in phosphate buffer. An additional experiment (n = 7, control; n = 5, GR) was performed to establish whether TH-IR dopaminergic amacrine cells were reduced in the intact retina at 50 to 52 dg, as has been described at 60 dg. ⁴ Cell counts confirmed that the total number and somal areas of TI and TII TH-IR amacrine cells were lower in GR retinas than in control retinas at 50 to 52 dg (Table 1 ; P < 0.05). 

Retinal Preparation.

The retinas were prepared as wholemounts, and the retinal areas measured in wetmount preparations, with a computerized digitizing pad, before they were processed for TH-IR. Counts of total TI and TII TH-IR amacrine cells were performed by using the system described earlier (CASTGRID; Olympus). 

Analysis of Dendritic Parameters.

To assess the effects of BDNF on process outgrowth, dendritic arbors were quantified by using Sholl’s concentric ring method of analysis. ³² Tracings of 25 cells per retina were made with a projector scope (magnification, ×1000; Olympus). Cells were randomly sampled from the entire retina. Traced neurons were covered with a transparent overlay of concentric rings set apart at 10-μm intervals. The number of dendrites intersecting each consecutive ring was counted to give the total number of dendritic intersections and branching points for each cell. The total dendritic length for each cell was estimated by multiplying the total number of dendritic intersections by the distance between the concentric rings (10 μm). 

Statistical Analysis.

All quantitative and qualitative assessment was performed on coded slides; t-tests were used for comparison between control and GR animals. To detect any differences in control and GR groups with and without BDNF treatment, the mean values of each dependent variable (total number of cells, somal area, total TI dendrite length, or total number of branching points) were compared by two-way ANOVA. Treatment was the independent variable. Significant treatment effects were subjected to post hoc analysis with the Bonferroni test. All data are expressed as the mean of means ± SEM, and significance was accepted if P < 0.05. 

Results

Part A: Effects of CPI on the Guinea Pig Retina—BDNF and NTRK2 Localization and Levels

At 60 dg, body (P < 0.001), brain (P < 0.005), and liver (P < 0.001) weights and crown rump length (P < 0.001) were reduced in GR fetuses compared with the control (Table 2) . In GR fetuses, brain-sparing was evidenced by an increase in brain-to-body weight ratios (P < 0.001). There was no significant difference in eye weight (P > 0.05) between the GR and the control fetuses (Table 2) . 

BDNF.

ELISA assays demonstrated a reduction in the total amount (P < 0.05) and concentration (P < 0.05) of BDNF protein in the retina in GR compared with control fetuses (Table 2) . At 60 dg, both large and small cells in the GCL were strongly immunoreactive for BDNF. Pale staining was also observed in cells in the inner aspect of the inner nuclear layer (INL; Fig. 1A ). There was no difference in the number of BDNF-IR cells in the GCL (Table 2 ; P > 0.05); however, qualitative observation revealed a reduction in BDNF-IR in these cells in GR (Fig. 1B)compared with control fetuses. 

NTRK2.

At 60 dg, both large and small cells in the GCL were strongly immunoreactive for NTRK2 (Fig. 1C) . Cells throughout the INL stained but were less immunoreactive, as were photoreceptor inner and outer segments. There was no difference in the number (Table 2 ; P > 0.05) or intensity of NTRK2-IR cells in the GCL between GR and control fetuses (Figs. 1C 1D) . 

Part B: Establishment of Optimal Tissue Culture Conditions for Dopaminergic Cell Survival and BDNF Treatment Levels

Retinal Morphology.

Assessment of retinas in vertical sections revealed good morphologic preservation in organotypic cultures (5-day survival) prepared from fetuses at 40 and 50 to 52 dg (Figs. 2A 2B) . All neuronal and plexiform layers were intact and appeared comparable to, although thinner than, layers observed in vivo at the same gestational ages ³¹ ; photoreceptor outer segments were less developed. Substantial disruption to the GCL and INL was observed at 55 dg (Fig. 2C) , and severe disruption and degeneration was evident in the INL and ONL with apoptotic cells at 60 dg (Fig. 2D) . Retinal wholemount preparations remained intact at 40, 50, and 55 dg, although degeneration was observed around the optic nerve head. At 60 dg, degeneration was more extensive, and intact wholemounts were difficult to obtain. TH-IR cells were evident at 40, 50, and 55 dg (Figs. 2E 2F 2G) ; however, staining was highly variable at 60 dg. Figure 2Hshows strongly stained cells. We selected 50 to 52 dg with 5-day survival as the optimum time point for further experiments, as apoptotic cells were rarely observed. After 7 days some apoptosis was observed in the INL and ONL, and by 10 days apoptosis was observed throughout the GCL, INL, and ONL. The number of apoptotic cells (cells per millimeter of GCL) in the peripheral retina at 50 to 52 dg after 3 to 10 days of organotypic culture is shown in Figure 3A . 

BDNF Application.

The addition of BDNF to organotypic cultures increased the number of surviving TI TH-IR amacrine cells at 50 to 52 dg at dilutions above 1 ng/mL (Fig. 3B) . Maximum cell survival was observed with 30 ng/mL of BDNF and was selected as the optimal concentration for later experiments. 

Part C: BDNF Application to Retinal Explants from GR and Control Animals at 50 to 52 dg

All comparisons in Table 3are between retinas in organotypic cultures. 

TH-IR Amacrine Cell Counts.

The total number of TI and TII cells in organotypic culture was reduced (P < 0.05) in untreated GR (Fig. 4B)versus control (Fig. 4A)retinas. BDNF administration increased the number of TI and TII TH-IR amacrine cells in both control (P < 0.05; Fig. 4C ) and GR retinas (P < 0.05; Fig. 4D ). 

TH-IR Amacrine Cell Somal Areas.

Somal areas of TH-IR TI and TII cells were reduced in untreated GR (P < 0.05) compared with control retinas. Addition of BDNF increased TI and TII somal areas in GR (P < 0.05) retinas and increased TII somal areas in control (P < 0.05) retinas. 

TI TH-IR Dendrite Length.

The estimated total dendrite length for TI, TH-IR amacrine cells was reduced (P < 0.05) in untreated GR compared with control retina. BDNF administration increased the estimated total dendrite length in control (P < 0.05) and GR (P < 0.05) retinas. 

TI TH-IR Branching Points.

The estimated mean total number of branching points for TI TH-IR amacrine cells was reduced (P < 0.05) in untreated GR compared with control retina. BDNF administration increased the estimated mean total number of branching points per cell in both control (P < 0.05) and GR (P < 0.05) retinas. 

TH-IR Dendrite Structure.

Although additional BDNF increased both the number and length of dendrite branches in both control and GR retinas, there was no alteration to the dendritic stratification with TI TH-IR dendrites localized to sublamina 1 of the inner plexiform layer (IPL), as in the intact retinas in all cases. 

Discussion

The significant findings of this study are first that retinal BDNF levels were reduced as a result of CPI and second that exogenous BDNF enhanced the survival and growth of dopaminergic amacrine cells in retinal explants from guinea pigs chronically compromised in utero. We know from a previous study that retinal dopaminergic amacrine cells are reduced after such an intrauterine compromise. ⁴ The present study is novel in that it demonstrates that neurons can respond to exogenous BDNF, even after a chronic insult. Earlier studies reported improved survival after acute insults. ^{33 34}  

Level of BDNF in the Retina after CPI

BDNF-IR was localized to ganglion cells and populations of cells in the INL; the distribution was similar to that reported for rat ³⁵ retinas. It is possible that the cells in the INL were displaced ganglion cells ³⁵ as has been suggested in other species. In GR fetuses compared to the control, there was no difference in the number of BDNF-IR cells in the GCL; however, the intensity of BDNF-IR in these cells and the total level of BDNF protein in the retina was reduced. BDNF levels are also reduced in the hippocampus after CPI in the guinea pig ²⁴ and sheep ³⁶ and in the porcine hippocampus after intravenous administration of the corticosteroid dexamethasone. ³⁷  

In the present study, no differences were observed in the number or intensity of cells immunoreactive for NTRK2. Thus, there was no compensatory increase in receptor levels in the face of decreased BDNF. The distribution of NTRK2-IR in the guinea pig is similar to that reported in other species, with the majority of ganglion cells and some INL cells, possibly amacrine cells, being immunoreactive. ^{35 38}  

Mechanisms Underlying Reduction of BDNF

The mechanisms underlying the altered structure and neurotrophin levels in the retina in this model are likely to be multifactorial. The regimen of CPI results in GR fetuses that are chronically hypoxemic and hypoglycemic ²¹ and have an altered endocrine status. ³⁹ Hypoxemia leads to increased extracellular glutamate levels and triggers a cascade of events that involves calcium (Ca²⁺) influx into cells via NMDA or non-NMDA glutamate receptors and increased transcription and translation of the BDNF and NTRK2 genes. ¹⁷ BDNF may exert its protective effects via mechanisms involving prevention of Ca²⁺overload by elevation of Ca²⁺-binding proteins, reducing the elevation of glutamate levels and enhancing cellular resistance to antioxidant stress by elevating glutathione peroxidase activity. ⁴⁰ However, it is likely that in situations of severe damage, BDNF is unable to prevent cell death signals. ⁴⁰ Increased levels of glutamate also lead to increased generation of nitric oxide (NO). In vitro, it has been shown that NO can induce a rapid downregulation of BDNF, possibly via the cGMP-activated protein kinase G pathway, ⁴¹ which would in turn reduce trophic support for dopaminergic amacrine cells. 

Effect of Exogenous Application of BDNF on Growth of Dopaminergic Amacrine Cells

In initial studies to determine optimal explant culture conditions, we established that normally grown retinas maintained good cytoarchitectural arrangement after 5 days of culture at 40 and 50 to 52 dg, substantial disruption was observed at 55 dg, and severe degeneration was apparent at 60 dg. At all ages, neuronal and plexiform layers were thinner than in vivo as previously reported. ⁴² In vitro treatment with BDNF increased the number of TI and TII TH-IR amacrine cells surviving at 40 to 55 dg, in a dose-dependent manner. Maximum survival was obtained at a dose of 30 ng/mL. High levels of BDNF induce an inflammatory reaction and decreased ganglion cell survival in vitro ⁴³ ; a similar situation may occur in dopaminergic amacrine cells at concentrations above 30 ng/mL. Cell survival is not increased at 60 dg as this is after the period of programmed cell death. ³¹ As the guinea pig is functionally mature at 55 to 56 dg, ⁴⁴ we cultured retinas just before this time point at 50 to 52 dg, to determine whether BDNF can rescue and stabilize cells. 

Administration of BDNF to retinal explants increased the number of TH-IR amacrine cells in both control and GR retinas at 50 to 52 dg. This result suggests that BDNF may act either to increase the survival of TH-IR cells preventing the natural loss that occurs in vitro ⁴⁵ or to induce expression of TH-IR in precursor neurons. In addition, in GR BDNF may also act to prevent CPI-associated cell loss. Although this study was conducted after the peak period of cell death, ³¹ apoptotic cells were still observed in control retina at 50 to 52 dg. It is possible that dopaminergic amacrine cell precursors reside in the retina at 50 to 52 dg and that the addition of BDNF induces their differentiation; however it is more likely that BDNF supports the survival of existing cells. BDNF exerts its effects on a wide range of neuronal subtypes in the retina, including ganglion cells. ^{15 16} It should be noted that, although BDNF levels were assayed in the intact retina at 60 dg, they were not assayed directly in cultured retinas after exogenous application of BDNF. 

Somal areas of TI and TII TH-IR amacrine cells were reduced in the intact retina and in organotypic retina cultured from GR animals at 50 to 52 dg compared with the control. As somal size correlates with the extent of dendritic ⁴⁶ and axonal arborization, ⁴⁷ decreased size may reflect decreases in afferent and/or efferent connectivity of neurons and may have functional implications. BDNF treatment also affected neurite growth, increasing the total dendrite length and number of branching points of TI cells in both control and GR retinas. BDNF, however, did not appear to alter the normal dendritic stratification of TH-IR cells, with dendrites remaining localized within sublaminar 1 of the IPL in all cases. Our studies have established strong evidence to suggest that BDNF has the potential to influence dopaminergic cell survival and structure and future studies will involve intraocular injections of BDNF in GR animals in the postnatal period and assessment of both retinal structure and function which is vital to a beneficial outcome. These studies are important, as remodeling by retinal neurons in retinal diseases can have negative consequences in some situations (reviewed in Ref. ⁴⁸ ). 

Although dopaminergic amacrine cells have been the focus of the present study, it is likely that BDNF is also acting on other retinal neurons including RGCs and photoreceptors. In vivo administration of BDNF rescues photoreceptor cell bodies from light damage in the Royal College of Surgeons (RCS) rat model of inherited retinal dystrophy. ⁴⁹  

Conclusion

BDNF is reduced in the retina after CPI. This reduction may underlie the decrease in the number of dopaminergic amacrine cells associated with CPI. In retinal explants we have demonstrated that exogenous BDNF increases the number of dopaminergic cells and growth of neurites. This study has established a rationale for future studies to examine whether in vivo administration of BDNF in GR fetuses can support dopaminergic cell survival and increase neurite growth, factors that may improve visual function after birth. 

 Supported by the Australian Research Council.

 Submitted for publication April 30, 2007; revised October 3 and November 14, 2007; accepted January 29, 2008.

 Disclosure: M.M. Loeliger, None; T. Briscoe, None; S.M. Rees, None

 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.

 

Table 1.

 

TH-IR in Intact 50- to 52-dg Retinas

Parameter	Control	GR
Total number TH-IR, TI amacrine cells	4758 ± 279	3700 ± 127*
Somal area TH-IR, TI amacrine cells (μm2)	95 ± 2	79 ± 3*
Total number TH-IR, TII amacrine cells	9982 ± 810	7524 ± 374*
Somal area TH-IR, TII amacrine cells (μm2)	49 ± 3	36 ± 2, **

*   P < 0.05.

**   P < 0.01.

Table 2.

 

BDNF and NTRK-2 in Guinea Pig Retina at 60 dg

Parameter	Control	GR
Body wt (g)	110.7 ± 2.7	62.2 ± 2.4, ***
Brain wt (g)	2.55 ± 0.04	2.31 ± 0.06, **
Brain/body wt	0.023 ± 0.002	0.037 ± 0.004, ***
Liver wt (g)	6.5 ± 0.4	3.0 ± 0.2, ***
Eye wt (g)	0.24 ± 0.01	0.22 ± 0.01
Crown rump length (cm)	15.3 ± 0.4	12.1 ± 0.4, ***
Total amount BDNF (pg/retina)	6.7 ± 0.3	5.9 ± 0.1*
BDNF concentration (pg/mg protein)	21.7 ± 1.6	16.1 ± 1.5*
BDNF-IR cells in GCL (cells/mm)	33.9 ± 1.4	38.5 ± 3.1
NTRK2-IR cells in GCL (cells/mm)	28.4 ± 1.7	32.2 ± 2.1

*   P < 0.05.

**   P < 0.005.

***   P < 0.001.

Figure 1.

 

(A) Distribution of BDNF-IR in control guinea pig retina at 60 dg. Cells in the GCL (arrow) are strongly immunoreactive, and cells at the vitreal border of the INL (*) are palely immunoreactive. Qualitative observation revealed a reduction in BDNF-IR in cells in the GCL in GR (B) compared with the control (A) fetuses. (C) Distribution of NTRK2-IR in the guinea pig retina at 60 dg. Cells in the GCL (arrow) were strongly immunoreactive, whereas cells throughout the INL and photoreceptor (PR) inner and outer segments were more palely immunoreactive. Qualitative observation did not reveal any difference in NTRK2-IR between GR (D) and control (C) fetuses. IPL, inner plexiform layer; OPL, outer plexiform layer, ONL, outer nuclear layer. Scale bar, 60 μm.

Figure 2.

 

Methylene blue–stained vertical sections (1 μm) of the retina after 5 days in organotypic culture at 40 (A), 50 to 52 (B), 55 (C), and 60 dg (D). Cytoarchitectural arrangement was well maintained at 40 and 50 to 52 dg, substantial disruption to the INL and ONL was observed at 55 dg; severe disruption and degeneration were evident at 60 dg with apoptotic cells (arrows) observed in the GCL, INL, and ONL. TH-IR in retinal wholemounts after organotypic culture at (E) 40 dg, (F) 50 to 52 dg, (G) 55 dg, and (H) 60 dg. Both TI (arrow) and TII (arrowhead) TH-IR amacrine cells were observed at all ages. Scale bar: (A–D) 30 μm; (E–H) 20 μm.

Figure 3.

 

(A) The number of apoptotic cells in the GCL increased after 5 days in organotypic culture at 50 to 52 dg. (B) BDNF administration increased TH-IR, dopaminergic cell survival in organotypic culture in a dose-dependent manner with optimal survival at 30 ng/mL at 50 to 52 dg.

Table 3.

 

TH-IR in Organotypic Retinal Cultures at 50 to 52 dg

Parameter	Control	Control+BDNF	GR	GR+BDNF
TI, TH-IR amacrine cells
Total number	3,513 ± 181	6,177 ± 1,201*	1,714 ± 264, †	3,345 ± 181, ‡
Somal area (μm2)	96 ± 3	113 ± 7	71 ± 5, †	94 ± 6, ‡
Total dendrite length (μm)	532 ± 56	873 ± 113*	206 ± 48, †	573 ± 80, ‡
Total dendritic branching points (number/cell)	3.5 ± 0.5	4.5 ± 0.4*	1.4 ± 0.2, †	3.5 ± 0.5, ‡
TII, TH-IR amacrine cells
Total number	8,172 ± 946	12,323 ± 804*	3,815 ± 116, †	5,819 ± 403, ‡
Somal area (μm2)	46 ± 2	51 ± 2	36 ± 2, †	44 ± 2, ‡

*   P < 0.05 control vs. control+BDNF.

†   P < 0.05 control vs. GR.

‡   P < 0.05 GR vs. GR+BDNF.

Figure 4.

 

BDNF administration in vitro increased the survival of TI (arrow) and TII (arrowhead), TH-IR amacrine cells in control (C) and GR (D) retinas compared with untreated control (A) and GR (B) retinas. Scale bar, 50 μm.

The authors thank Sandra Dieni and Alexandra Rehn for assistance with some of the animal husbandry and surgery and Anthony Hughes for his interest and input into these studies.