Brain-derived neurotrophic factor is a regulator of synaptic transmission in the adult visual thalamus

Abstract

Brain-derived neurotrophic factor (BDNF) is an important regulator of circuit development, neuronal survival, and plasticity throughout the nervous system. In the visual system, BDNF is produced by retinal ganglion cells (RGCs) and transported along their axons to central targets. Within the dorsolateral geniculate nucleus (dLGN), a key RGC projection target for conscious vision, the BDNF receptor tropomyosin receptor kinase B (TrkB) is present on RGC axon terminals and postsynaptic thalamocortical (TC) relay neuron dendrites. Based on this, the goal of this study was to determine how BDNF modulates the conveyance of signals through the retinogeniculate (RG) pathway of adult mice. Application of BDNF to dLGN brain slices increased TC neuron spiking evoked by optogenetic stimulation of RGC axons. There was a modest contribution to this effect from a BDNF-dependent enhancement of TC neuron intrinsic excitability including increased input resistance and membrane depolarization. BDNF also increased evoked vesicle release from RGC axon terminals, as evidenced by increased amplitude of evoked excitatory postsynaptic currents (EPSCs), which was blocked by inhibition of TrkB or phospholipase C. High-frequency stimulation revealed that BDNF increased synaptic vesicle pool size, release probability, and replenishment rate. There was no effect of BDNF on EPSC amplitude or short-term plasticity of corticothalamic feedback synapses. Thus, BDNF regulates RG synapses by both presynaptic and postsynaptic mechanisms. These findings suggest that BNDF influences the flow of visual information through the retinogeniculate pathway.

NEW & NOTEWORTHY Brain-derived neurotrophic factor (BDNF) is an important regulator of neuronal development and plasticity. In the visual system, BDNF is transported along retinal ganglion cell (RGC) axons to the dorsolateral geniculate nucleus (dLGN), although it is not known how it influences mature dLGN function. Here, BDNF enhanced thalamocortical relay neuron responses to signals arising from RGC axons in the dLGN, pointing toward an important role for BDNF in processing signals en route to the visual cortex.

INTRODUCTION

The goal of this study was to test the hypothesis that brain-derived neurotrophic factor (BDNF) regulates the transmission of visual information through retinogeniculate (RG) synapses in the mature dorsolateral geniculate nucleus (dLGN), a key subcortical visual structure in the thalamus responsible for conscious vision (1–3). BDNF has established roles in the development of neuronal connectivity and cell survival, acute regulation of synaptic transmission, long-term and homeostatic synaptic plasticity, and modulation of intrinsic neuronal excitability of mature neuronal circuits (4–7). In the mature rodent dLGN, BDNF protein is localized specifically to retinal ganglion cell (RGC) axon terminals while the major BDNF receptor, tropomyosin receptor kinase B (TrkB), is found both on dLGN thalamocortical (TC) neuron postsynaptic sites and on RGC axon terminals (8). This suggests that BDNF can be released by RGC axon terminals to act both pre- and postsynaptically to influence the conveyance of visual information through the retinogeniculate pathway.

In the developing visual system, BDNF expression in the retina increases throughout the embryonic phases and is strongly expressed in RGCs at birth (9). BDNF regulates binocular segregation of RGC inputs to the rodent dLGN (10) and is important for the development of RGC axons and the formation of synapses with targets in the amphibian optic tectum (11–13). Within the mature visual system, BDNF is a key survival signal; loss of RGC axonal BDNF transport in early glaucoma appears to contribute to RGC degeneration and retinal BDNF supplementation supports RGC survival (14–18). Moreover, injection of BDNF into the eye helps prevent the loss of neurons from the dLGN and preserves their visual responses following lesions to the visual cortex (19).

In whole cell patch-clamp electrophysiology recordings in brain slices from mouse dLGN, BDNF enhanced synaptically driven spiking of TC neurons in response to RGC axon stimulation. Postsynaptically, BDNF altered TC neuron passive membrane properties leading to a subtle increase in intrinsic excitability. Presynaptically, BDNF, via TrkB receptors and phospholipase C, modulated presynaptic vesicle release to enhance the amplitude of RG excitatory postsynaptic currents (EPSCs). Together, these results indicate that BDNF regulates RG synaptic strength via pre- and postsynaptic mechanisms, thereby influencing transmission of visual information through the visual thalamus.

MATERIALS AND METHODS

Animals

Animal procedures were approved by the Institutional Animal Care and Use Committee at the University of Nebraska Medical Center. C57Bl6J mice (Jackson Labs #000664) were used for experiments involving measurements of TC neuron intrinsic excitability, mEPSCs, and optic tract and corticothalamic tract stimulation in parasagittal sections. For experiments involving optogenetic activation of retinal ganglion cell axon terminals, Chx10-Cre;Ai32 mice were used. These were produced as a cross of Chx10-Cre mice (Jackson laboratories #005105) (20) with the Ai32 reporter line (Jackson Labs #024109) (21). In the resulting Chx10-Cre;Ai32 mice, RGC axons in the dLGN express a Channelrhodopsin-2 (Chr2)-enhanced yellow fluorescent protein (EYFP) fusion protein, allowing for optical activation of excitatory RGC inputs in brain slices (22). Mice of both sexes and ages of 6–12 postnatal weeks were used for experiments. Mice were housed in a 12-h/12-h light/dark cycle with food and water provided ad libitum.

Brain Slice Preparation

Acute brain slices were prepared for patch-clamp recording following the “protected recovery” approach (23, 24). Mice were euthanized by CO₂ asphyxiation and cervical dislocation and brains were dissected into a slush of artificial cerebrospinal fluid (aCSF; 128 mM NaCl, 2.5 mM KCl, 1.25 mM NaH₂PO₄, 24 mM NaHCO₃, 12.5 mM glucose, 2 mM CaCl₂, 2 mM MgSO₄, pH 7.4, 305–315 ×10⁻³ osmol/L, bubbled with 5% CO₂ and 95% O₂). For coronal slices, the cerebellum was removed with a razor blade and the caudal end of the brain was affixed to the stage of a vibratome (Leica VT1000S) using cyanoacrylate glue and submerged in ice-cold aCSF. Sections 250 µm thick containing the dLGN were transferred onto a nylon net in a beaker of 32°C NMDG-aCSF [107 mM N-methyl-d-glucamine (NMDG), 2.5 mM KCl, 1.25 mM NaH₂PO₄, 30 mM NaHCO₃, 20 mM HEPES, 0.5 mM CaCl₂, 10 mM MgSO₄, 2 mM thiourea, 5 mM Na-ascorbate, 3 mM Na-pyruvate, 25 mM glucose] for 12 min. After NMDG-aCSF incubation, slices were transferred to a nylon net in a beaker containing aCSF for >1 h before recording.

For preparing parasagittal slices, the brain was cut with a razor blade angled 5° from the longitudinal fissure and 20° from the vertical axis following the approach of Turner and Salt (25). Pieces were mounted cut side down on the vibratome stage and cut into 300-μm-thick sections before being incubated in the warmed NMDG-aCSF followed by >1 h in aCSF, as described for coronal sections.

Patch-Clamp Electrophysiology and Analysis

Following >1 h incubation in aCSF, a slice was transferred to a recording chamber positioned on the stage of an upright fixed-stage microscope (Olympus BX51WI), anchored with a nylon net, and constantly superfused with aCSF warmed to 30°C–33°C (26) at ∼2 mL/min. This relatively fast superfusion rate allows adequate supply of oxygenated and pH-buffered aCSF to the slice. Moreover, this rapid flow rate allows for adequate penetration of BDNF into the slices for TrkB receptor activation (27–30). Prior work has shown that superfusion rate can impact the efficacy of BDNF effects on synaptic transmission, with a perfusion rate of ∼1–4 mL/min being optimal for assessing the effects of BDNF and TrkB signaling on basal synaptic transmission (27–30). Recordings were performed using a Multiclamp 700B amplifier, a Digidata 1550B AD/DA interface, and Clampex 10 software (Molecular Devices). Individual thalamocortical (TC) relay neurons within the dLGN were targeted for whole cell patch-clamp recording based on oval soma shape and identity was confirmed by presence of a low voltage-activated Ca²⁺ current. For current-clamp recordings, the patch pipettes (1.2 mm OD, 0.9 mm ID borosilicate tubing with internal filament) had resistances of ∼5–8 MΩ and were filled with a K⁺-based solution (120 mM K-gluconate, 8 mM KCl, 2 mM EGTA, 10 mM HEPES, 5 mM ATP-Mg, 0.5 mM GTP-Na₂, and 5 mM phosphocreatine-Na₂). For most voltage-clamp recordings, the patch pipettes contained a Cs⁺-based pipette solution (120 mM Cs-methanesulfonate, 2 mM EGTA, 10 mM HEPES, 8 mM TEA-Cl, 2 mM ATP-Mg, 0.5 mM GTP-Na₂, 5 mM phosphocreatine-Na₂, and 2 mM QX-314 bromide). For current-voltage subtraction experiments to measure BDNF-mediated conductance changes, the K-gluconate pipette solution was used and the bath was supplemented with 60 µM picrotoxin, 20 µM CNQX, and 0.5 µM tetrodotoxin (TTX). Reported voltages are corrected for measured liquid junction potentials of 10 mV for the Cs-based solution and 14 mV for the K-based solution. Pipettes were pulled using a Sutter P-1000 pipette puller and positioned using a Sutter MPC-325 micromanipulator. Slices were exchanged for fresh naïve slices following BDNF application. Cells were excluded from analysis if there was a loss of seal resistance or a large increase in access resistance that could not be remedied by breaking back into the cell.

Two approaches were used to activate retinogeniculate inputs to TC neurons. In the first, current stimuli generated with an AM Systems Model 2100 Isolated Pulse Stimulator (anodic, cathodic, or biphasic current; 0.2–0.5 ms, 100 µA–2 mA, adjusted to provide a saturating response) were delivered to the optic tract ∼2 mm from the dLGN in parasagittal slices using a bipolar stimulating electrode while TC neurons were voltage-clamped at −70 mV. In the second approach, a full-field 460-nm light flash (0.3–2 ms, 0.5–2 mW) was delivered through the objective lens using a Sutter TLED+ while TC neurons in slices prepared from Chx10-Cre;Ai32 mice (22) were voltage-clamped at −70 mV. These light intensities were found to be saturating and activated all available RG inputs onto recorded TC neurons. The AMPA receptor-mediated EPSC amplitudes were measured as the peak of the inward EPSC at a holding potential of −70 mV. For measurements of vesicle release probability (Pr), pool size, and replenishment rate, RG inputs were stimulated with a 10-Hz train of pulses (3-s duration) and the cumulative amplitude of the last 15 EPSCs was fit with a straight line extrapolated to the y-axis. The changes in the y-intercept provide a measure of relative pool size, whereas the ratio of the first EPSC to the y-intercept is a measure of Pr (26, 31–33). The slope was normalized to the y-intercept to measure the relative replenishment rate of the vesicle pools. For measuring corticothalamic feedback synapses, the bipolar stimulating electrode was positioned in the thalamic reticular nucleus in parasagittal slices. Miniature EPSCs (mEPSCs) were detected and analyzed from a 60-s recording in coronal sections in the absence of stimulation at a −70 mV holding potential using MiniAnalysis software (Synaptosoft, Fort Lee, NJ).

To measure intrinsic TC neuron excitability and membrane properties in current clamp, TC neurons were stimulated with depolarizing and hyperpolarizing current pulses (+40 to +560 pA and −20 to −100 pA, respectively; 500 ms). The bridge balance circuitry was activated and adjusted for each cell to compensate for voltage drops across the access resistance during current injection experiments. The number of action potentials was measured using the event detection feature of Clampfit during each 500-ms depolarizing stimulus epoch and plotted against the current stimulus amplitude. The spiking behavior of each cell was determined by measuring the maximum spike number, the half-maximal stimulus amplitude (I₅₀) from a Boltzmann sigmoid fit performed in GraphPad Prism 9, and the threshold stimulus (5% of maximum from the Boltzmann fit). dLGN TC neurons exhibit spike frequency adaptation and this was quantified by calculating a spike adaptation index (SAi) for the 360-pA current stimulus as the ratio of the average interspike intervals of the first three action potentials to the average of the intervals at the end of the current stimulus. SAi < 1 means that the spiking slows during the stimulus, whereas SAi > 1 is indicative of accelerating spiking during the stimulus. Input resistance was measured as the slope of a linear fit to the voltage changes evoked by −20 and −40 pA current injections. In a subset of recordings, an EPSC waveform recorded from a TC neuron (2,600 pA) in response to optogenetic activation of RG inputs was scaled (0.125 to 1.5-fold) and used as a current clamp stimulus.

Pharmacology

For measurements of miniature excitatory postsynaptic currents (mEPSCs) and evoked EPSCs, the aCSF was supplemented with 60 µM picrotoxin (diluted from a 60 mM stock solution prepared in DMSO; Acros No. 124-87-8). Cyclothiazide (100 µM; Santa-Cruz sc-202560A) and 200 µM γ-d-glutamylglycine (γDGG; Abcam No. Ab120307) (both diluted 500-fold from stock solutions prepared in DMSO) were added along with picrotoxin for measurements using high-frequency stimulus trains, whereas 5 µM CGP55845 (Tocris No. 1248; diluted 1,000-fold from a stock solution prepared in DMSO) was added (along with picrotoxin, cyclothiazide, and γDGG) for measurements of corticothalamic feedback synapses. BDNF (Sigma No. SRP3014) was diluted 1,000-fold in aCSF from a 20 µg/mL stock (made in aCSF) to a final concentration of 20 ng/mL and bath-applied to slices. This concentration was previously found to be saturated for acute effects on synaptic transmission in brain slice recordings (27, 28, 34). When applying to the slice, aliquots of the BDNF stock solution were kept frozen until immediately before use and diluted into the 50 mL of aCSF immediately before application. Cyclotraxin-B (Tocris No. 5062) was diluted 1,000-fold from a 500 µM stock solution prepared in dH₂O. U73122 (Tocris No. 1268) was diluted 500-fold from a 10 mM stock solution prepared in DMSO.

Statistics

Statistical analysis was performed using GraphPad Prism 9. Data are presented as individual values (gray data markers) and means ± SEM (black data markers). Every recorded cell included in the study was treated after obtaining pretreatment responses, allowing for paired statistical comparisons. Datasets fitting a normal distribution (D’Agostino-Pearson omnibus normality test in Prism 9) were assessed for statistical significance using a paired Student’s t test with significance threshold set at P < 0.05. Statistical significance of nonnormally distributed data was assessed with a Wilcoxon matched-pairs sign-rank test.

RESULTS

To test whether BDNF can acutely modulate the transfer of visual information at the retinogeniculate (RG) synapse, dLGN TC neurons were targeted for whole cell current-clamp recording in coronal slices from Chx10-Cre;Ai32, and the ChR2-expressing RGC axons terminals were stimulated with a 3-s duration 10-Hz train of 460-nm flashes (Fig. 1). This evoked depolarizing excitatory postsynaptic potential (EPSP) and action potential firing in the TC neurons (Fig. 1A). Action potentials tended to be concentrated toward the beginning of the stimulus sequence, likely due to a combination of high synaptic vesicle release probability and consequent synaptic depression at the RG synapse and initial EPSP amplification due to the LVA Ca²⁺ spike (26, 35–37), with multiple action potentials often fired in response to the first stimulus in the train. Bath application of 20 ng/mL BDNF led to a gradual enhancement of TC neuron responses, evident as an increase in the number of action potentials fired over the course of the 10-Hz stimulus sequence (Fig. 1, B and C; P = 0.02, Wilcoxon). There was also a 1.9 ± 1.1 mV increase in EPSP amplitude (P = 0.031, Wilcoxon).

Figure 1.

BDNF enhances synaptically driven spiking of thalamocortical relay neurons in the dLGN. A: whole cell current clamp recording from the dLGN TC neuron during optogenetic stimulation of RGC inputs in a coronal brain slices from a Chx10-Cre;Ai32 mouse. The stimulus (lower signal) was a 3-s duration 10-Hz train of LED pulses (460 nm) that evoked glutamate release from retinogeniculate axon terminals leading to postsynaptic depolarization and action potential firing. Multiple action potentials were fired in response to the first stimulus in the train. Measured approximately 10 min following application of 20 ng/mL BDNF (right), the TC neuron fired a greater number of action potentials in response to the same stimulus. B: time series of the number of action potentials evoked during stimulation at 30-s intervals. C: normalized time series group data. D: group data showing that BDNF led to an increase in the number of action potentials fired in response to the 10-Hz stimulus (n = 7 cells, Wilcoxon test). BDNF, brain-derived neurotrophic factor; dLGN, dorsolateral geniculate nucleus; RGC, retinal ganglion cell; TC, thalamocortical.

Next, current-clamp experiments were performed in coronal slices from C57Bl6/J mice to test whether this enhancement of synaptically driven spiking was the result of altered spike generation by TC neurons (Fig. 2). Action potential firing was measured in response to a series of depolarizing current injections (+40 to +560 pA, 40 pA increments, Fig. 2B), whereas input resistance was measured using hyperpolarizing current injections (Fig. 2C). The effects of BDNF on TC neuron spiking were measured using several approaches. In the first, the peak number of spikes for each cell was measured and was not significantly altered following BDNF application (Fig. 2D; P = 0.10, paired t test; n = 16). The current-spike plot for each cell was also fit with a Boltzmann sigmoid and this was used to measure the threshold (stimulus evoking 5% of the maximal response), which was not significantly altered after BDNF application (P = 0.48, paired t test), although there was a shift of the half-maximal stimulus amplitude to lower values (P = 0.012, paired t test, Fig. 2E). The spike adaptation index (SAi), which quantifies the timing of spikes at the beginning and end of the stimulus, was not altered following BDNF application (P = 0.90, Wilcoxon; Fig. 2F), indicating that spike adaptation was not altered by BDNF application. BDNF application also led to a 2.5 ± 0.7 mV depolarization (P = 0.0016, paired t-test; n = 23; Fig. 2G) and an increase in TC neuron input resistance (P = 0.0064, paired t test; n = 16 Fig. 2H), from 252 + 21 MΩ to 297 + 27 MΩ. When an EPSC waveform, measured during a separate voltage-clamp recording, was scaled to several different amplitudes and used as a current stimulus, BDNF application led to a small increase in the number of spikes evoked over the stimulus sequence (Fig. 2, I–K).

Figure 2.

BDNF causes a modest enhancement of TC neuron intrinsic excitability. A: example traces of voltage responses to 500-ms duration depolarizing and hyperpolarizing current stimulation in whole cell current clamp recordings from a dLGN TC neuron before and after application of 20 ng/mL BDNF. B: current-spike plot showing a similar spike output in response to stimulation before and after BDNF application. The arrowheads at the x-axis represent the means ± SE of I₅₀ values from individual Boltzmann fits before and after BDNF application. The closed black arrowhead is control and the open blue arrowhead is +BDNF (n = 16 TC neurons). The y-axis is absolute spike counts. C: group data showing voltage deflections in response to hyperpolarizing current injections from the same group of TC neurons (*P < 0.05, **P < 0.01, paired t test). D: the maximum number of spikes evoked in each cell (ns, P > 0.05, paired t test). E: I₅₀ of the Boltzmann fit from each TC neuron before and after BDNF application (P = 0.017, paired t test). F: spike adaptation index (SAi), which was the ratio of the steady-state spike frequency to the mean instantaneous spike frequency of the first three action potentials, was not significantly altered by BDNF application (P = 0.89, Wilcoxon). G: resting membrane potential (Vrest) was slightly depolarized following BDNF application (n = 23 cells, paired t test). H: input resistance, measured as the slope of a line fit through a plot of the hyperpolarizing voltage deflections in response to −20 and −40 pA current injections, was slightly elevated following BDNF application (n = 16, paired t test). I: an excitatory postsynaptic current (EPSC) waveform (scaled to different peak amplitudes) was used as a current stimulus and evoked action potential firing in a TC neuron before and after BDNF application. J: BDNF led to a slight elevation of action potential firing in response to the EPSC waveform stimulus. K: quantification of total number of action potentials evoked during the EPSC waveform stimulus series showed an elevation of action potential firing following BDNF application (n = 5 cells, P = 0.009, paired t test). BDNF, brain-derived neurotrophic factor; dLGN, dorsolateral geniculate nucleus; I₅₀, half-maximal stimulus amplitude; TC, thalamocortical.

The ionic basis of these effects was explored by measuring the current-voltage relationship of membrane conductances changed following BDNF application (Fig. 3). In these experiments (n = 8 cells), the current-voltage relationship was measured by subtracting the pre-BDNF responses to a series of hyperpolarizing and depolarizing voltage steps from the recorded response following BDNF treatment. The resulting I-V plot had a complex profile indicating that BDNF changed the gating of a mix of channels: one component appeared to reflect a decrease in a voltage-gated K+ current (reversal at –89 mV), whereas another appeared to reflect an increased cationic conductance (reversal at –0.49 mV). Thus, BDNF led to some subtle changes in TC neuron passive membrane properties, with modest effects on action potential firing, apparently due to alterations in multiple membrane currents.

Figure 3.

BDNF application alters multiple membrane currents. A: whole cell current responses to a series of voltage steps (150-ms duration, −104 to −4 mV) prior to (“control”) and after BDNF application. Darker traces are more hyperpolarized, whereas lighter are for more depolarized responses. Lower series: difference current obtained by subtracting the control currents from the currents recorded with BDNF. B: current-volage plot of difference currents measured in 8 cells point to a mix of currents altered by BDNF. Linear fits of each component point to a reduction in a voltage-gated K⁺ current (x-intercept approximately −90 mV) and a cation current (x-intercept ∼0 mV). BDNF, brain-derived neurotrophic factor.

Given the modest effects of BDNF on TC neuron spike generation described in Fig. 2, the next goal was to determine whether BDNF influences synaptic transmission at the RG synapse. This was accomplished by recording EPSCs in response to either optogenetic or electrical activation of RG inputs in dLGN brain slices while performing simultaneous voltage-clamp recordings of postsynaptic TC neurons (Fig. 4). In the first set of experiments, application of BDNF enhanced RG EPSCs by 27 ± 5% over baseline (n = 20; P = 0.0052, paired t test; Fig. 4, A–C). This was prevented by preincubating the slices with either the tropomyosin receptor kinase B (TrkB) allosteric inhibitor cyclotraxin-B (38) (CTX-B, 500 nM; n = 8; P = 0.47, paired t test; Fig. 4, D–F) or the phospholipase C (PLC) inhibitor U73122 (20 µM, n = 6; P = 0.90, paired t test), consistent with BDNF effects being mediated by TrkB receptor and a phospholipase C-dependent cascade in the dLGN.

Figure 4.

BDNF enhances synaptic strength by presynaptic mechanisms involving the TrkB receptor and phospholipase C. A: excitatory postsynaptic currents (EPSCs) recorded from TC neurons in response to optogenetic RGC input stimulation in a brain slice from a Chx10-Cre;Ai32 mouse were elevated in amplitude following BDNF application. B, top: time series of EPSC amplitudes for the cell shown in A. Bottom: group data (means ± SE) from six cells showing the gradual increase in EPSC amplitude following BDNF application. C: group data showing that BDNF led to an increase in EPSC amplitude in recordings with optogenetic activation of RGC inputs and electrical stimulation of the optic tract in parasagittal slices (n = 22 cells, paired t test). D: preincubation with cyclotraxin-B (500 nM) blocked the effect of BDNF. E: group data showing no effect of BDNF when the slice was preincubated with CTX-B. F: preincubation with the phospholipase C inhibitor U73122 (20 µM) prevented the BDNF-induced increase in EPSC amplitude. G: EPSCs recorded from TC neurons in response to electrical stimulation of the optic tract in a parasagittal slice. H: plot of cumulative EPSC amplitude of the responses shown in F and linear fit of the last 15 data points with a straight line extrapolated to the y-axis, which was used for measurements of Pr, replenishment rate, and relative pool size. I: group data from recordings using both electrical and optogenetic activation of RGC inputs showing that vesicle release probability (Pr; EPSC₁/y-intercept) was elevated following BDNF application (n = 9 cells, paired t test). J: the pool replenishment rate, normalized to pool size, was slightly elevated by BDNF application (n = 9 cells, paired t test). K: the y-intercept, taken as a relative measure of pool size, was slightly elevated by BDNF application (n = 9 cells, paired t test). BDNF, brain-derived neurotrophic factor; RGC, retinal ganglion cell; TC, thalamocortical, TrkB, tropomyosin receptor kinase B.

To test for presynaptic mechanisms underlying the BDNF/TrkB-dependent enhancement of synaptic transmission, slices were incubated in 100 µM cyclothiazide and 200 µM γDGG to prevent AMPA receptor desensitization and saturation, respectively, and RG inputs were stimulated with 3-s duration 10-Hz train of optogenetic or electrical pulses (Fig. 4G). Extrapolation from the cumulative amplitude of the last 15 pulses to the y-intercept (Fig. 4H) provides relative measurements of vesicle pool size while the ratio of the first EPSC to the y-intercept provides Pr. The slope, normalized to the y-intercept, is a measure of the vesicle pool replenishment rate relative to the vesicle pool size. In this analysis, each of these three parameters was enhanced following BDNF application (Fig. 4, I–K), indicating that BDNF/TrkB-dependent processes can regulate the presynaptic vesicle recruitment and release at RG synapses.

Although Pr, pool size, and replenishment rate all involve a presynaptic locus of BDNF’s effects, synaptic strength can be influenced by mechanisms converging on postsynaptic glutamate receptor properties including the number, composition, and modulatory state of AMPA receptors. To test this possibility, single-vesicle miniature excitatory postsynaptic currents (mEPSCs) were recorded from TC neurons in the absence of stimulation (Fig. 5A). BDNF application led to a reduction in mEPSC frequency (P = 0.0059, paired t test; Fig. 5B) and a reduction in mEPSC amplitude (P = 0.0098, Wilcoxon; Fig. 5C). When the TrkB inhibitor CTX-B (500 nM) was applied before BDNF, there was no significant effect of BDNF on mEPSC amplitude (n = 6 cells, P = 0.094, Wilcoxon) or mEPSC frequency (P = 0.22, paired t test). Consistent with a PLC-dependent signaling cascade, there was no significant effect of BDNF on mEPSC amplitude (n = 6 cells, P = 0.094, Wilcoxon) or frequency (P = 0.91, paired t test) when slices were pretreated with U73122 (20 µM). The reduction in mEPSC amplitude would be consistent with effects on AMPA receptors at postsynaptic sites. Surprisingly, however, when Ca²⁺ in the extracellular solution was replaced with 3 mM Sr²⁺ to desynchronize exocytosis and evoked quantal release was measured following optogenetic stimulation of the optic tract, there was no detectable difference in mEPSC amplitude following BDNF application (Fig. 5F, n = 7 cells, P = 0.86, paired t test).

Figure 5.

BDNF alters mEPSCs recorded in TC neurons. A: trace of mEPSCs recorded from a TC neuron in the absence of stimulation before and after BDNF application. Middle: average detected mEPSC waveforms. Right: amplitude/noise histograms for the displayed recordings showing the amplitude distributions of the baseline recording noise and the mEPSC amplitudes fit with a Gaussian distribution. B: the mEPSC frequency was reduced following BDNF application (n = 11, paired t test). C: mEPSC amplitude was decreased following BDNF application (Wilcoxon). D: the effects of BDNF on mEPSC amplitude and frequency were blocked when the slice was pretreated (10 min) with the TrkB antagonist cyclotraxin-B (CTX-B, 20 µM) (n = 6 cells; frequency: paired t test; amplitude: Wilcoxon). E: the effects of BDNF on mEPSC amplitude and frequency were blocked when the slice was pretreated (10 min) with the phospholipase-C inhibitor U73122 (20 µM) (n = 6 cells; frequency: paired t test; amplitude: Wilcoxon). F: retinogeniculate synaptic vesicle release evoked by optogenetic stimulation in Chx10-Cre;Ai32 mice was desynchronized by replacing extracellular Ca²⁺ with Sr²⁺. Four example records from a single TC neuron are displayed before and after BDNF application. Middle: average mEPSC waveforms of all events detected 200 ms poststimulus from the example cell. Right: group data showing BDNF did not significantly alter the amplitude of mEPSCs recorded in this way (n = 6 cells, paired t test). BDNF, brain-derived neurotrophic factor; mEPSCs, miniature excitatory postsynaptic currents; TC, thalamocortical, TrkB, tropomyosin receptor kinase B.

The effects on mEPSCs are somewhat enigmatic, as processes that enhance presynaptic neurotransmitter release generally lead to an enhancement of nonevoked vesicle fusion events, which would be evident as an increase in mEPSC frequency rather than the decrease observed here. However, TC neurons receive ∼90% of their synaptic inputs from cortical feedback synapses (corticothalamic, or C-T synapses) with only ∼10% being RG synapses, although the C-T synapses have a much lower vesicle release probability. To test whether the reduction in mEPSC frequency might represent BDNF effects on C-T synaptic function, EPSCs were recorded from C-T synapses in parasagittal slices by positioning a stimulating electrode in the thalamic reticular nucleus, a region through which C-T fibers pass on their way to the dLGN (Fig. 6). C-T axons were stimulated with pairs of stimuli (50-ms interval) before and after BDNF application to test for effects on EPSC amplitude and vesicle release probability (Fig. 6A). There was no statistically significant change in EPSC amplitude accompanying BDNF application (P = 0.3, paired t test; Fig. 6B), nor was there a change in paired-pulse ratio (P = 0.86, paired t test; Fig. 6C).

Figure 6.

Corticothalamic feedback synapses were not affected by BDNF application. A: EPSCs recorded from a TC neuron in a parasagittal slice in response to paired pulse stimulation (50-ms interval) delivered with an electrode positioned in the thalamic reticular nucleus to activate corticothalamic fibers. B: group data showing EPSC amplitude was not significantly altered by BDNF application (n = 5 cells; P = 0.30, paired t test). C: the paired pulse ratio (EPSC2/EPSC1) was not altered by BDNF application, suggesting that Pr was not affected (P = 0.86, paired t test). BDNF, brain-derived neurotrophic factor; EPSCs, excitatory postsynaptic currents; TC, thalamocortical.

DISCUSSION

This study sought to determine whether BDNF can modulate synaptic and neuronal function in the dLGN, a key subcortical structure for conscious vision. Prior studies have shown that the BDNF receptor TrkB is present at both presynaptic RGC axon terminals and on postsynaptic TC neuron dendrites adjacent to RGC terminals (8). Consistent with this, acutely applied BDNF enhanced spiking activity triggered by retinogeniculate synaptic inputs. BDNF application also led to a slight TC neuron depolarization and increased input resistance, which had modest effects on TC neuron spiking. BDNF also altered retinogeniculate synaptic function by apparently presynaptic mechanisms involving increases in synaptic vesicle release probability, vesicle pool size, and pool replenishment rate.

Numerous studies have documented presynaptic effects of BDNF on excitatory and inhibitory synaptic transmission. For instance, BDNF enhances glutamate release from isolated CA1 synaptic terminals, cultured hippocampal neurons, and from Schaeffer collateral terminals in hippocampal slices (29, 30, 39–42). Slice experiments showed that this originates from increases in both Pr and vesicle pool size, based on FM1-43 staining, electrophysiology, and electron microscopy (29, 30). BDNF also causes presynaptic enhancement of GABA release in hippocampus by enhancing Pr and replenishment rate (43–45), although the evidence for this is mixed (41). Effects of BDNF on synaptic transmission differ in other parts of the brain. For instance, at the calyx of Held synapse, BDNF activation of TrkB receptors ultimately decreases Pr by triggering postsynaptic release of endocannabinoids and activation of presynaptic CB1 receptors to suppress PKA activity and reduce presynaptic Ca²⁺ influx (46). Other studies have documented similar links between BDNF and endocannabinoids in cortex and hippocampus (47–50). CB1 receptors are expressed in the dLGN (51) and involved in modulating visual signal transmission to the cortex (52–54), making it possible that the endocannabinoid system might be involved in BDNF effects in the dLGN. Exploration of this possibility will be a fruitful area for future study.

BDNF binding to TrkB leads to activation of phospholipase Cγ1 (PLCγ1) (55). PLCγ1, in turn, recruits the IP₃/Ca²⁺/Calmodulin (CaM)/CaM kinase and the diacylglycerol (DAG)/PKC pathway, each of which is tied to synaptic plasticity via regulation of Pr, pool size, and replenishment. For instance, Ca²⁺ store activation has been linked to BDNF-dependent enhancement of mEPSC frequency and CaM is involved in calcium-dependent acceleration of vesicle pool replenishment (39, 56–58). BDNF/TrkB can also activate the Ras/MAP kinase (MAPK) pathway. MAPK phosphorylation of synapsin1 enhances Pr and pool size in the hippocampus (59) and phosphorylation of RIM1a, a protein involved in vesicle priming by ERK2 contributes to enhanced glutamate release (42). In the current study, the blockade of BDNF effects by U73122 points to a PLC-dependent signaling pathway, but the specific endpoints responsible for the upregulation of Pr, pool size, and replenishment remain to be determined.

BDNF has been shown to have effects on both AMPA and NMDA receptors at excitatory synapses, either up- or downregulating receptors depending on brain region and cell type (6, 60–67). The current study also found that BDNF also had postsynaptic effects on dLGN TC neurons; mEPSC measurements showed a slight reduction in amplitude. However, when vesicle release was evoked from RG terminals in the presence of Sr²⁺ to desynchronize release, we did not detect any significant change in event amplitude, suggesting that the altered mEPSC amplitude might be due to postsynaptic effects at nonretinal inputs to TC neurons such as corticothalamic feedback synapses, which comprise ∼90% of the excitatory inputs to TC neurons by number. In addition, although BDNF application also reduced mESPC frequency, the origin of this effect is unclear, especially since RG inputs showed an increase in Pr (typically associated with an increase in mEPSC frequency) and evoked release from cortical feedback synapses, did not show evidence of altered Pr. Some dLGN TC neurons also receive excitatory input from nonretinal and noncortical sources (i.e., superior colliculus) (68), so it remains possible that BDNF modulation of these underlies the effects on mEPSC frequency.

There appeared to be fairly modest effects of BDNF on TC neuron intrinsic neuronal excitability. This indicates that the BDNF enhancement of synaptically driven spiking involved contributions from both modulation of the RG synapse as well as effects on TC neuron spike generation. Reports of BDNF effects on neuronal excitability show varied results, possibly depending on neuronal type, duration of BDNF treatment, and which ion channels are modulated downstream of BDNF (4, 69–72). In primary cultures of cortical pyramidal neurons and interneurons, BDNF blocked a compensatory excitability increase triggered by activity blockade (4). In various hippocampal neurons, BDNF can either increase or decrease excitability (69, 71) and these effects have been tied to enhancement of I_h (70), decreased voltage-gated K⁺ currents (72), or enhanced M-type K⁺ currents (71). Here, BDNF led to a small depolarization and increase in input resistance, both of which might be attributable to modulation of TC neuron membrane currents such as I_h or K_ir, etc. (26, 73–75) in addition to the reduced mEPSC frequency (Fig. 5). A current-voltage analysis of the change in membrane conductances evoked by BDNF in the presence of synaptic blockers suggests that it decreases a voltage-gated K⁺ current and might enhance a nonselective cation current of unknown origin. Resting membrane potential in TC neurons plays an important role in regulating tonic versus burst firing properties, in part due to regulation of the extent of inactivation of low-voltage-activated Ca²⁺ currents (26, 36, 37, 74, 76, 77). However, the depolarization did not lead to any detectable effects on spike frequency adaptation, which quantifies some of the burst-like evoked firing properties of TC neurons.

Previous studies have explored the influence of superfusion speed in studies of BDNF application to brain slices or cultured neurons, documenting how different flow rates can regulate different neuronal and synaptic functions on varying timescales (27, 28). This has led to a multifaceted understanding of BDNF’s roles in neuronal signaling, with BDNF appearing to function as both a rapidly acting neurotransmitter-like signal and a slower neuromodulator (78, 79). As BDNF appears to be stored and released from RGC axon terminals and is capable of acting on both RGC terminals and postsynaptic TC neurons, its function on multiple timescales might allow for dynamic modulation of the visual signal at RG synapses.

Together, the findings of this study indicate that BDNF can regulate the flow of visual information through the retinogeniculate pathway by acting at both pre- and postsynaptic sites in the dLGN. This is consistent with prior work showing BDNF is present in RGC axon terminals while TrkB receptors are found pre- and postsynaptically (8). BDNF is released in an activity-dependent manner (80, 81) and this would be in support of a model wherein visually driven RGC activity triggers BDNF secretion to maintain synaptic homeostasis in the dLGN.

Moreover, BDNF appears to play an important role in neuronal survival in the visual system. For instance, supplementation of BDNF to the retina following ablation of the visual cortex helps preserve TC neurons in the dLGN by downregulating proapoptotic signaling pathways (19, 82). This indicates that BDNF is an anterogradely transported survival signal in the visual system. Neurodegenerative diseases of the optic nerve such as glaucoma impair BDNF transport along the optic nerve (16, 17), which is likely to deprive dLGN of BDNF and thereby contribute to neuronal and synaptic dysfunction and eventual degeneration (31, 83).

In conclusion, the findings of this study suggest that BDNF can regulate retinogeniculate synaptic function in adulthood in a manner involving both the pre- and postsynaptic sites, consistent with a model where BDNF released by RGCs acts on receptors present on both RGC axon terminals and postsynaptic TC neurons. This provides insights into the role(s) played by BDNF in the retinal projection and maintenance of information transmission through the visual pathway. It is likely that this is the result of activity-dependent BDNF secretion from RGC axon terminals, which serves as a signal to maintain and strengthen RG synaptic function in adulthood. Future work will need to explore this possibility as well as the consequences for neuronal survival and synaptic function of impaired BDNF delivery to central targets in diseases such as glaucoma.

GRANTS

This work was supported by National Institutes of Health/National Eye Institute R01 Grant EY030507.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

M.J.V.H. conceived and designed research; performed experiments; analyzed data; interpreted results of experiments; prepared figures; drafted manuscript; edited and revised manuscript; and approved final version of manuscript.