Mechanism by which Brain-Derived Neurotrophic Factor Increases Dopamine Release from the Rabbit Retina

Abstract

purpose. To determine whether BDNF modulates the release of dopamine from amacrine cells in the rabbit retina.

methods. Isolated retinas from rabbits killed with pentobarbital were incubated in Krebs bicarbonate medium containing pargyline, nomifensine, and bovine serum albumin. The medium was changed at 10-minute intervals, and the dopamine in the resultant samples measured by HPLC. Five samples were collected to establish the spontaneous resting release of dopamine, and then the retina was exposed to BDNF for a further two collection periods. Double-label immunohistochemistry was used to identify tyrosine hydroxylase containing neurons and to localize TrkB (BDNF) receptors.

results. Exposure of the retina to BDNF (70–150 ng/mL) caused a concentration-dependent increase in the release of dopamine. The maximum effect was produced by 150 ng/mL BDNF, which almost doubled the release. The BDNF-evoked release was abolished in low-calcium/high-magnesium medium. It was also prevented by the tyrosine kinase inhibitors k252a and genistein, the phospholipase inhibitor U73122, and the sarco/endoplasmic reticulum Ca²⁺-ATPase (SERCA) inhibitors thapsigargin and cyclopiazonic acid. Antagonists of γ-aminobutyric acid (GABA) and glutamate did not affect the BDNF-evoked release of dopamine. ELISA assay confirmed the presence of BDNF in the retina, and immunohistochemistry revealed that some tyrosine hydroxylase–containing neurons possessed TrkB receptors.

conclusions. BDNF quickly (within minutes) increases the release of dopamine from amacrine cells in the rabbit retina by an action that is likely to involve TrkB receptors. The mechanism linking TrkB receptor activation to the release of dopamine involves activation of phospholipase-Cγ, with the subsequent production of IP₃ and calcium release from the endoplasmic reticulum. The resultant capacitative entry of calcium seems to be the actual trigger for BDNF-induced release of dopamine.

The neurotrophins form a gene family, the members of which include nerve growth factor (NGF), brain-derived neurotrophic factor (BDNF), neurotrophin-3 (NT-3), and neurotrophin-4/5 (NT-4/5). They have been characterized mostly as neurotrophic factors (for reviews, see Refs. ^{1 2 3} ), but it has been increasingly realized that the neurotrophins have other important actions in the nervous system, one of these being modulation of synaptic efficacy. ^{4 5 6 7}  

Most cellular actions of the neurotrophins are mediated by their high-affinity tyrosine kinase receptors. ⁸ In the retina, BDNF is able to maintain ganglion cells in vitro, ^{9 10 11} and neurons in the ganglion cell layer express the BDNF receptor TrkB. ^{12 13 14 15 16 17 18} TrkB receptors are also expressed on cells in the inner nuclear layer. ^{13 14 17 19 20} Mostly, these cells have not been identified, but a recent study ¹⁸ has revealed that, in a wide range of species, most dopaminergic cells express the TrkB receptor. In other areas of the central nervous system, BDNF has been found to increase the release of dopamine, ^{21 22} and it has been suggested that NGF may increase dopamine release in the fish outer retina. ²³ In the present study, within minutes, BDNF increased the release of dopamine from the rabbit retina by a process involving TrkB receptors situated on the dopaminergic amacrine cells. 

Materials and Methods

Animals

Male New Zealand White rabbits (n = 54), weight range 2 to 2.5 kg, were used in the study. They were cared for and treated in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Release of Dopamine

Rabbits were killed with an intravenous injection of pentobarbital, and the eyes were enucleated and the retinas dissected. The isolated retinas were bisected, and each half was placed in a tube (Eppendorf, Fremont, CA) containing Krebs bicarbonate medium (0.3 mL). The medium contained pargyline (50 μM), nomifensine (0.5 μM), and bovine serum albumin (BSA; 0.1%) and was gassed with 5% CO₂ in oxygen. The hemiretinas were incubated in a gently shaking water bath at 25°C. At 10-minute intervals, all the medium was replaced with fresh medium, and the dopamine in the resultant samples was determined by HPLC. The effect of BDNF on the release of dopamine was examined by incubating the retina in medium containing the growth factor for two 10-minute collection periods after incubation in BDNF-free medium for at least 50 minutes. The 10-minute time point for exposure of the retina to BDNF was chosen because preliminary experiments showed a steady release of dopamine at that time. 

Dopamine Assay

Samples (100 μL) were acidified with 1 M perchloric acid (10 μL) and the dopamine content determined by HPLC using an electrochemical detector (Intro; Antec, Leyden, The Netherlands) with the electrode set at 0.7 V oxidation. The system comprised a column (15 cm × 4.6 mm; Spherisorb S5 ODS2; Waters, Milford, MA) with a 20-μL injection loop. The mobile phase (0.07 M KH₂PO₄, 0.0035% EDTA, 0.023% octyl sodium sulfate and 10% methanol [pH 2.75]) was degassed with helium and pumped at a flow rate of 1.0 mL/min and a temperature of 35°C. 

Extraction of BDNF

Seven individual retinas were processed to extract detectable levels of BDNF-LI, by ELISA. Several methods were used to maximize the recovery of BDNF from the tissue. ²⁴ A cocktail of inhibitors was added to the homogenization buffer to reduce the enzymatic breakdown of BDNF (137 mM NaCl, 20 mM Tris HCl, 1% NP40, 10% glycerol, 0.1% BSA, 1 mM α-toluenesulfonyl fluoride, 10 μg/mL aprotinin, and 0.5 mM sodium vanadate). Samples were centrifuged and assayed by ELISA. Protein content was determined by the Bradford assay. 

ELISA of BDNF

Ninety-six-well plates were used (MaxiSorp; Nunc, Roskilde, Denmark). BDNF standards (100 μL of 4–500-pg/mL solutions) and 100 μL of unknown samples were run in duplicate and triplicate, respectively, according to a protocol slightly modified from the manufacturer’s instructions (Emax ImmunoAssay kit; Promega, Southampton, UK). ²⁴ The assay showed less than 3% cross reactivity with human recombinant NGF. Human recombinant BDNF standard was used to assess the recovery of BDNF. Recovery was 60% for 10 pg/mL BDNF. 

Immunohistochemistry

The retina was removed, fixed in 4% paraformaldehyde in 0.1 M phosphate buffer (PB; pH 7.4) for 24 hours, and cryoprotected in 30% sucrose in PB. Twenty-micrometer coronal sections were cut and collected directly onto slides (Superfrost; Fisher Scientific, Pittsburgh, PA). Sections were treated for 10 minutes in 0.3% hydrogen peroxide in PB containing 0.9% NaCl (PBS) to block potential endogenous peroxidase. After several washes in PB, sections were incubated overnight at room temperature in PBS containing 0.3% Triton X-100 (PBST) with a mix of chicken anti-TrkB antibody (1:20,000, a gift from Louis F. Reichardt) and mouse anti-tyrosine hydroxylase (TH; 1:500; catalog no. 5280; Chemicon, Temecula, CA). The secondary antibodies were anti-mouse-tetramethylrhodamine isothiocyanate (TRITC; 1:300; Jackson Laboratories, Bar Harbor, ME) and biotinylated anti-chicken (1:400; Vector Laboratories, Peterborough, UK) in PBST. The anti-mouse was left to incubate on the slides for 3 hours in total. At the end of the second hour of incubation, the anti-chicken was added and incubated for 1 hour. After three washes, sections were incubated in streptavidin-horseradish peroxidase (HRP) for 45 minutes (1:200, tyramine peroxidase revelation kit; NEN Life Sciences, Cologne, Germany). After three washes, the sections were incubated in FITC-linked tyramine (1:400, NEN Life Sciences) for 10 minutes. After several washes, slides were coverslipped in antifade medium (Vectashield; Vector). 

Fluorescent images were captured with a digital camera (Hamamatsu, Hamamatsu City, Japan) and the cognate software provided, using the blue (for FITC-labeled TrkB) or the green (for TRITC-labeled TH) filter of the microscope. 

Control experiments were performed on sister sections. The omission of either the anti-TrkB or the anti-TH antibody induced a specific absence of labeling for each of them, proving specific labeling and non–cross talk with the other fluorescent filter. 

Agent Composition and Sources

The Krebs bicarbonate had the following composition (mM): NaCl 118, KCl 4.84, CaCl₂ 2.4, NaHCO₃ 25, MgSO₄ 1.8, KH₂PO₄ 1.2, and glucose 9.5. It was oxygenated with 95% O₂-5% CO₂. The low-calcium/high-magnesium medium contained 20 mM MgCl₂ and 0.24 mM CaCl₂. 

BDNF was obtained from Promega, k252a from Calbiochem (Nottingham, UK), and genistein from Alexis Corp. (Nottingham, UK). All other drugs were obtained from Sigma-Aldrich Co., Ltd. (Poole, UK) or Merck (Poole, UK). 

Results

Effect of BDNF on Release of Dopamine

A steady spontaneous release of dopamine from the isolated rabbit retina of 3.26 ± 0.09 fmol/min/mg wet weight (n = 12; 1.14 ± 0.03 pmol/10 min per hemiretina) was maintained for at least 120 minutes. Exposure of the retina to BDNF (70–150 ng/mL) caused a concentration-dependent increase in release of dopamine. However, at concentrations above 150 ng/mL, BDNF caused a progressively smaller increase in release of dopamine, and 250 ng/mL BDNF did not significantly increase the release. A typical experiment showing the effect of 150 ng/mL BDNF is illustrated in Figure 1A , and the results are summarized in Figure 1B . The release of dopamine caused by exposure of the retina to 150 ng/mL BDNF was at its maximum during the first 10 minutes and was maintained for a further 10 minutes (Fig 1A) . A preliminary time-course experiment revealed that at longer exposure times (i.e., longer than 20 minutes) the evoked release of dopamine was not statistically significant (data not shown). 

Effect of Low-Calcium/High-Magnesium Medium on BDNF-Evoked Dopamine Release

The effect of low-Ca/high-Mg medium on the BDNF-evoked release of dopamine is illustrated in Figure 2 . Exposure of retinas to 150 ng/mL BDNF in normal medium increased the release of dopamine to 179.5% ± 12.8% of the spontaneous release (P < 0.001, n = 12). In contrast, when the retina was exposed to BDNF in low-Ca/high-Mg medium, the release of dopamine was unchanged, remaining at 98% ± 3.6% (n = 4) of the spontaneous resting release. 

Effect of Tyrosine Kinase Inhibitors on BDNF-Evoked Release of Dopamine

Exposure of the retina to the tyrosine kinase inhibitors k252a (5 μM) and genistein (100 μM) had no effect on the release of dopamine. However, when these concentrations were doubled, both drugs caused an increase in the release, and as with BDNF itself, this release was calcium dependent. Typical experiments with k252a are illustrated in Figure 3A and the results with k252a and genistein are summarized in Figure 3B . 

When the retinas were exposed to k252a (5 μM) and genistein (100 μM), the concentrations of the inhibitors that had no effect on release of dopamine, the effect of BDNF on the release was completely blocked. A typical experiment illustrating this effect of genistein on the BDNF-evoked release of dopamine is shown in Figures 4A and 4B , and the results with k252a and genistein are summarized in Figure 4C . 

Effect of Inhibition of Phospholipase-Cγ on BDNF-Evoked Release of Dopamine

One of the transduction mechanisms linking BDNF receptor binding to its cellular effects involves the activation of phospholipase-Cγ (PLC-γ) and synthesis of IP₃. To investigate whether this pathway may be involved in the BDNF-evoked release of dopamine, we exposed the retina to the PLC-γ inhibitor U73122 (5 μM) for 30 minutes before and during the application of 150 ng/mL BDNF. U73122 alone had no effect on retinal release of dopamine (data not shown), but it completely blocked the effect of BDNF on dopamine release (Fig. 5) . 

Effect of SERCA Inhibitors on BDNF-Evoked Release of Dopamine

To examine the requirements for intracellular Ca²⁺ stores in the BDNF-evoked release of dopamine, we exposed the retinas to the sarcoendoplasmic reticulum Ca²⁺-ATPase (SERCA) inhibitors thapsigargin (5 μM) and cyclopiazonic acid (CPA; 5 μM). Neither drug alone affected retinal release of dopamine (data not shown) but both inhibitors prevented the BDNF-evoked release of dopamine (Fig. 5) . 

Effect of Picrotoxin on BDNF-Evoked Release of Dopamine

BDNF has been reported to affect the release of GABA in cortical and hippocampal neurons, ^{25 26 27} and dopaminergic amacrine cells possess GABA_A receptors. ²⁸ However, exposure of the retina to picrotoxin (20 μM) had no effect on the increase in dopamine caused by 150 ng/mL BDNF (Fig. 6) , suggesting that an indirect action involving GABA is unlikely. 

Effect of Glutamate Antagonists on BDNF-Evoked Release of Dopamine

It is unlikely that the increase in release of dopamine from the retina is secondary to an increase in release of glutamate, because exposure of the retina to cis-2,3-piperidine dicarboxylic acid (PDA; 2 mM) or (−)2-amino-5-phosphonopentanoic acid (AP5; 100 μM) had no effect on the BDNF-evoked release of dopamine (Fig. 6) . 

BDNF Content

ELISA assays revealed that the concentration of BDNF in the rabbit retina was 5.95 ± 0.45 pg/mg protein (48.7 ± 3.94 pg/retina). Similar amounts of BDNF have been reported in the hamster retina. ²⁹  

Double Labeling of TrkB and TH

Immunolocalization of TH in the rabbit retina showed a punctiform labeling in the inner plexiform layer (IPL) and a cytoplasmic labeling of amacrine cells at the border between the IPL and inner nuclear layer (INL; Figs. 7A 7B ). We observed TrkB immunolabeling in the cytoplasm and processes of ganglion cells (Fig. 7C) and in amacrine cells in the IPL (Fig. 7D) . Some amacrine cells were the only cells to stain for both TrkB and TH (Fig. 7F) . 

Discussion

Several recent studies have shown that BDNF modulates neurotransmission in the brain and spinal cord. ^{4 5 6 30 31 32} In the present study, BDNF quickly increased the release of dopamine from amacrine cells in the rabbit retina. Because dopaminergic amacrine cells possess glutamate, glycine, and GABA_A receptors, ²⁸ the possibility of an indirect mechanism of action exists. Neurotrophins have been shown to reduce GABA_A-receptor–mediated inhibition, ^{25 26 27} but it is unlikely that GABA was involved in the BDNF-evoked release of dopamine from the retina, because it was unaffected by the GABA_A antagonist picrotoxin. BDNF has been shown to enhance the release of glutamate in cultured hippocampal neurons, ^{4 33} cultured cerebellar neurons, ³⁴ and cultured cortical neurons ³⁵ and in the visual cortex, ^{36 37} but because neither PDA nor AP5 affected the BDNF-evoked release of dopamine it is unlikely that the increase in dopamine release is consequential to an increase in glutamate release. 

The mechanism by which BDNF increases the release of dopamine appears to involve TrkB receptors because the evoked release was completely blocked by the tyrosine kinase inhibitors k252A and genistein. At concentrations above those needed to block the effects of BDNF, both tyrosine kinase inhibitors increased dopamine release, probably because these drugs can act as partial agonists. ^{38 22} Tyrosine kinase inhibitors have been reported to inhibit dopamine uptake, ³⁹ but such an action cannot be important in our experiments, because they were performed in the presence of nomifensine, a dopamine uptake inhibitor. 

BDNF binding to TrkB receptors causes receptor dimerization and kinase activation. Subsequent phosphorylation of the receptors creates docking sites for adaptor proteins that couple the receptors to intracellular signaling cascades, including the Ras/extracellular signal regulated kinase (ERK) protein kinase pathway, the phosphatidylinositol-3-OH kinase (PI3K)/Akt kinase pathway, and phospholipase C-γ1 (PLC-γ1; for reviews, see Refs. ⁴⁰ , ⁴¹ ). 

The phospholipase-C inhibitor, U73122 completely prevented BDNF from stimulating the release of dopamine from the retina indicating an involvement of the PLC-γ1 pathway. TrkB phosphorylates PLC-γ1, which then hydrolyzes phosphatidylinositol to produce diacylglycerol and IP₃. IP₃ induces Ca²⁺ release from stores in the endoplasmic reticulum and increases the concentration of the cytoplasmic Ca²⁺. In our experiments, CPA and thapsigargin abolished the BDNF-evoked release of dopamine. These drugs reduced the endoplasmic Ca²⁺ stores by inhibiting the SERCA. These results support the idea that BDNF increases the release of dopamine by a mechanism involving IP₃-mediated release of Ca²⁺. We have not measured the putative changes in cytosolic Ca²⁺, but BDNF has been shown to increase the level of Ca²⁺ in hippocampal cell bodies ³³ and in presynaptic nerve terminals. ³⁰ Although our study suggests that a release of Ca²⁺ from internal stores is crucial for the BDNF-evoked release of dopamine, we found that the release is also dependent on extracellular Ca²⁺. In frog motoneurons, autophosphorylation of TrkB was not affected by Ca²⁺-free conditions, and an increase in release of acetylcholine caused by BDNF did not involve voltage-gated Ca²⁺ channels. ⁴²  

In many non-neuronal cells, the release of Ca²⁺ from limited-capacity stores triggers a more sustained increase in cytoplasmic Ca²⁺ by capacitative entry of Ca²⁺ from the extracellular solution. ⁴³ There is evidence that a capacitative influx of Ca²⁺ also occurs in neurons, and it is possible that the dependence on extracellular Ca²⁺ in our experiments occurred because the entry of Ca²⁺ is triggered by Ca²⁺ released from intracellular stores. ^{44 45 46}  

The mechanism by which we think BDNF causes release of dopamine from amacrine cells is summarized in Figure 8 . This mechanism in the retina is very similar to, if not identical with those involved in the release of glutamate from cultured hippocampal and cortical neurons ³³ and acetylcholine from cultured Xenopus spinal motor neurons. ⁴² However, BDNF also increases transmitter release by other mechanisms. In cultured cerebellar neurons BDNF increases intracellular Ca²⁺ through activation of PLC-γ and IP₃, but this seems to be linked to an influx of Na⁺, which reverses the glutamate transporter. ³⁴ In rat cerebral cortical synaptosomes, BDNF has been reported to increase release of glutamate by a mechanism involving phosphorylation of MAP kinase-dependent synapsin-1. ⁴⁷ However, in cultured cerebral cortical neurons BDNF apparently increases release of glutamate by the PLC-γ-IP₃-Ca²⁺ pathway. ⁴⁸  

The actions of BDNF on neurons are believed to involve mainly the high-affinity TrkB receptors. ⁸ These receptors have been found on most dopaminergic amacrine cells in some species, ¹⁸ but in the rabbit retina we found a sparse distribution of TrkB immunoreactivity on neurons located in the INL and a small number of dopaminergic neurons that possessed TrkB immunoreactivity. We do not know whether the small number of cells staining for both TrkB and TH reflects the true incidence of such cells or whether the (chicken) TrkB antibody failed to stain some rabbit dopaminergic neurons. Clearly, some dopaminergic amacrines have TrkB receptors, which suggests that BDNF may increase the release of dopamine by a direct action, but our immunohistochemical results do not rule out the possibility of an indirect action through other neurons possessing TrkB receptors. 

Our study does not provide information as to whether a physiological release of BDNF normally affects dopaminergic amacrine cells, but it seems likely that a physiological release of BDNF (from unknown cells) is important during development, because in mice homozygous for a null mutation of the Bdnf gene, dopaminergic cells were atrophic. ¹⁹ Although the rabbit retina contains BDNF, its cellular location is unknown. There have been relatively few studies of the retinal cell types containing BDNF. Rat red–green sensitive cones ⁴⁹ and ganglion cells contain BDNF and TrkB receptors. ^{50 51} Displaced amacrine cells were also labeled by antibodies to both BDNF and TrkB. ⁵¹ Some of the BDNF may be synthesized in the ganglion cells, but in the chick retina most of the BDNF apparently accumulates in ganglion cells by anterograde transport after its release from more distal retinal cells. ⁵² In the rat retina, in contrast, optic nerve crush strikingly increases expression in putative ganglion cells after 24 to 48 hours, an effect not associated with a change in expression of the TrkB receptor. ⁵³ Exogenous BDNF has several beneficial effects (e.g., reduction of gliosis and preservation of synaptic structure) in a feline model of retinal detachment. ⁵⁴ Taken together, the findings in these studies suggest that BDNF may have a neuroprotective role after retinal damage. However in our study, it is unlikely that retinal damage affected the BDNF-evoked release of dopamine because the experiments were completed within 2 hours, whereas the changes after optic nerve crush and retinal detachment were significant only at times well beyond 6 hours. 

It is not known what type of synaptic activity may cause release of BDNF in the retina. It is possible that ganglion cell depolarization may stimulate local release of BDNF, which would result in a feedback enhancement of the release of dopamine. There is no evidence at present that light stimulation increases the release of BDNF in the retina. However, in the fish retina, NGF has been shown to act in a light adaptive signaling mechanism that involves an increase in release of dopamine. ²³  

The neurotrophins were discovered as survival-promoting agents, but the present study adds to the rapidly increasing evidence that they are involved in regulating synaptic function and plasticity. ^{3 55}  

 Submitted for publication June 10, 2002; revised August 29, 2002; accepted September 30, 2002.

 Disclosure: M. Neal, None; J. Cunningham, None; I. Lever, None; S. Pezet, None; M. Malcangio, 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.

 

Figure 1.

 

Effect of BDNF on release of dopamine from the rabbit retina. (A) Typical experiment. Each histobar is a 10-minute collection period. Hatched histobars: exposure of the retina to BDNF (150 ng/mL). (B) Concentration-response curve. Each result is the mean ± SEM of 4 to 15 experiments. Significant increases in release were determined with Students t-test, *P < 0.01, **P < 0.001.

Figure 2.

 

Effect of low-Ca/high-Mg medium on BDNF-evoked dopamine release. (A) Control experiment showing that exposure of the retina to 150 ng/mL BDNF increased dopamine release. (B) Experiment in low-Ca/high-Mg showing absence of BDNF-evoked dopamine release. Hatched histobars: exposure to BDNF.

Figure 3.

 

Effect of tyrosine kinase inhibitors on dopamine release. (A) Typical experiments showing that 10 μM k252a (cross-hatched histobars) increased dopamine release, an effect that was Ca dependent. (B) Summary of results showing that k252a (10 μM) and genistein (200 μM) increased dopamine release (P < 0.005), but lower concentrations of k252a (5 μM) and genistein (100 μM) had no effect on dopamine release. Each result is the mean ± SEM of four to seven experiments.

Figure 4.

 

Effect of tyrosine kinase inhibitors on BDNF-evoked dopamine release. (A) Typical control experiment showing an increase in dopamine release by BDNF (150 ng/mL; hatched histobars). (B) Experiment showing that k252a (5 μM; horizontal bar) inhibited the effect of BDNF on dopamine release. (C) Summary of results showing that k252a (5 μM) and genistein (100 μM) completely block the effect of BDNF on the release of dopamine. Each result is the mean of four to seven experiments. *BDNF alone significantly increased the release of dopamine P < 0.001.

Figure 5.

 

Summary of results showing that the PLC-γ inhibitor U73122 and the SERCA inhibitors thapsigargin and CPA blocked the effect of BDNF on the release of dopamine. *Control BDNF (150 ng/mL) significantly increased release (P < 0.01). Each result is the mean ± SEM of four to six experiments.

Figure 6.

 

Summary of results showing that picrotoxin, PDA, and AP5 did not affect the BDNF-evoked release of dopamine. * BDNF significantly increased dopamine release (P < 0.001); **no significant difference from control effect on release. Each result is the mean ± SEM of five to seven experiments.

Figure 7.

 

Immunolocalization of TH (A, B), TrkB receptors (C, D), and double labeling (E, F) in the rabbit retina. TH was restricted to amacrine cells in the IPL (A, B, arrow). TrkB immunoreactivity occurred in ganglion cells (C) and in some amacrine cells (D, arrow). Double labeling revealed that some TH containing amacrine cells also possessed TrkB receptors (F, arrow), although many apparently did not (E). Magnification bar, 50 μm.

Figure 8.

 

BDNF evoked the release of dopamine from amacrine cells. A mechanism that is consistent with the current results involves BDNF’s binding to TrkB receptors and activating PLC-γ. The formation of IP₃ then leads to the release of Ca²⁺ from internal stores and a subsequent influx of Ca²⁺ from the extracellular space. The resultant increase in intracellular Ca²⁺ levels triggers the release of dopamine.