Autocrine BDNF-TrkB signalling within a single dendritic spine

Abstract

Brain-derived neurotrophic factor (BDNF) and its receptor TrkB are crucial for many forms of neuronal plasticity, including structural long-term potentiation (sLTP), which is a correlate of an animal's learning. However, it is unknown whether BDNF release and TrkB activation occur during sLTP, and if so, when and where. Here, using a fluorescence resonance energy transfer-based sensor for TrkB and two-photon fluorescence lifetime imaging microscopy, we monitor TrkB activity in single dendritic spines of CA1 pyramidal neurons in cultured murine hippocampal slices. In response to sLTP induction, we find fast (onset < 1 min) and sustained (>20 min) activation of TrkB in the stimulated spine that depends on NMDAR (N-methyl-_d-aspartate receptor) and CaMKII signalling and on postsynaptically synthesized BDNF. We confirm the presence of postsynaptic BDNF using electron microscopy to localize endogenous BDNF to dendrites and spines of hippocampal CA1 pyramidal neurons. Consistent with these findings, we also show rapid, glutamate-uncaging-evoked, time-locked BDNF release from single dendritic spines using BDNF fused to superecliptic pHluorin. We demonstrate that this postsynaptic BDNF-TrkB signalling pathway is necessary for both structural and functional LTP. Together, these findings reveal a spine-autonomous, autocrine signalling mechanism involving NMDAR-CaMKII-dependent BDNF release from stimulated dendritic spines and subsequent TrkB activation on these same spines that is crucial for structural and functional plasticity.

Extended Data Figure 1. Design and development of a FRET-based sensor for TrkB activation

a, (Upper) Western blot analysis (IB) of cell extracts from HeLa cells stimulated with either BDNF or vehicle. Extracts were immunoprecipitated with an antibody for phosphorylated tyrosine residues (pTyr) and then probed with antibodies for TrkB and GFP. (Lower) IB of BDNF and vehicle stimulated cell extracts prior to IP using antibodies for TrkB, GFP, and actin. For source data, see Supplementary Figure 1. b, FLIM images of TrkB and TrkB^Y816F activation acquired before and 2–6 min after BDNF stimulation (averaged multiple images taken over 5 min). Warmer colours indicate shorter lifetimes and higher TrkB activity. c, Time course of TrkB and TrkB^Y816F activation measured as the change in binding fraction of TrkB-mEGFP or TrkB^Y816F-mEGFP bound to mRFP-PLC-mRFP before and after BDNF or vehicle stimulation. n = 22/8 TrkB plus BDNF, 9/4 TrkB plus vehicle, and 11/4 TrkB^Y816F plus BDNF (cells/experiments). d, TrkB activation (averaged over 6–10 min) for experiments in (c). e, FLIM images of TrkB activation in a neuron in a mixed cortical dissociated culture before and after BDNF stimulation followed by K252a application at 30 min. f, Time course of TrkB activation measured as described in (c) before and after BDNF or NGF stimulation followed by K252a application. n = 8 BDNF and 4 NGF (neurons). g, TrkB activation (averaged over 10–30 min and 3–5 min following K252a application) for experiments in (f). Data are mean +/− s.e.m. *p < 0.05 as determined by a two-tailed unpaired samples t-test (g) or an analysis of variance (ANOVA) followed by Tukey’s method to correct for multiple comparisons. (d). **p < 0.05 as determined by a two-tailed paired samples t-test.

Extended Data Figure 2. Rescue of sLTP with TrkB-mEGFP following post- synaptic TrkB knockout

a-b, Time course (a) and quantification (b) of glutamate-uncaging-induced spine volume change for TrkB^fl/fl hippocampal slices transfected with mEGFP (Cre Neg), mEGFP plus Cre (Cre Pos), and mCherry, TrkB-mEGFP, and Cre (Cre Pos + TrkB-EGFP). n = 7/20 Cre Neg, 9/24 Cre Pos, and 5/11 Cre Pos + TrkB-EGFP (cells/spines). Data are means +/− s.e.m. *p < 0.05) as determined by an ANOVA followed by Tukey’s method to correct for multiple comparisons.

Extended Data Figure 3. Characterization of prolonged TrkB activation and spine volume change during single spine sLTP

a, Prolonged time course of spine volume change following 2p-glutamate uncaging in rat hippocampal slices transfected with the TrkB sensor (TrkB) or mEGFP (GFP). n = 50/54 for TrkB sensor (9/10 for experiments longer than 20 min) and 8/8 for mEGFP (cells/spines). b, Prolonged time course of TrkB activation in stimulated spines (Stim spine), the base of the spine neck (Spine base), adjacent spines (Adj spine), and the dendritic shaft adjacent to the stimulated spine (Dendrite). n = 50 cells with 54 Stim spine, Spine base, and Dendrite plus 59 Adj spine. c-d, Time course (c) and quantification (d) of the transient (averaged over 1–2 min) and sustained (averaged over 20–40 min) phases of glutamate uncaging-induced spine volume change in rat hippocampal slices in the absence and presence of anisomycin (25 μM). n = 12/14 Ctrl and 5/5 anisomycin (cells/spines). Data are mean +/− s.e.m.

Extended Data Figure 4. Comparison of temporal dynamics of BDNF release, TrkB activation, and spine volume change during single spine sLTP

a, Time course of normalized changes in TrkB activity and spine volume change (percent of maximal activity and volume change). b, Magnified view of normalized changes of BDNF-release, TrkB activation, and spine volume during and 1 minute after the uncaging epoch.

Extended Data Figure 5. Determination of the specificity of glutamate uncaging evoked TrkB activation

a, Time course of TrkB activation following glutamate uncaging before (Ctrl) and at least 30 min after K252a application to the perfusion bath (K252). n = 41/45 Ctrl and 4/9 K252a (cells/spines). b, Peak (averaged over 1–2 min) and sustained (averaged over 10–20 min) TrkB activation for experiments in (a). c, Time course of spine volume change for experiments in (a). d, Transient and sustained spine volume change for experiments in (a). e–h, Similar experiments with (a-d) but in TrkB^F616A hippocampal slices transfected with the TrkB^F616A sensor before (Ctrl) and at least 30 min after 1NMPP1 application (1NMPP1; 2μ;M). n = 4/5 Ctrl and 3/6 1NMPP1 (cells/spines). i–l, Similar experiments with (a-d) but with the TrkB (Ctrl) and TrkB^Y816F (Y816F) sensors. n = 9/10 Ctrl and 7/11 Y816F (cells/spines). Data are mean +/− s.e.m. *p<0.05 as determined by two-tailed unpaired samples t-test.

Extended Data Figure 6. Effect of temperature on the spatiotemporal dynamics of TrkB activation

Time course of TrkB activation at room temperature (RT; 24–26°C) and 30–32°C in the stimulated (Stim spine) and dendrite. n = 20 spines and dendrites for RT and 25 for 30–32°C. Data are mean +/− s.e.m.

Extended Data Figure 7. Effects of sensor expression levels on changes reported by the sensor

a-d, Effect of TrkB-mEGFP concentration as measured in individual neurons on corresponding change in binding fraction of the stimulated spine (a), change in spine volume (b), binding fraction prior to uncaging (basal binding fraction) (c), and change in binding fraction of the dendrite (d). n = 25/28 (cells/spines). Data are mean values and were fit to a linear regression model with corresponding coefficients of determination (R²) provided for each.

Extended Data Figure 8. Basal spine size and CaMKII activation in the presence and absence of post- synaptic BDNF

a-b, Quantification (a) and representative 2p images (f) of basal spine size/morphology in Bdnf^fl/fl slices transfected with EGFP (Cre Neg) or EGFP plus Cre (Cre Pos). n = 14/50 Cre Neg and 29/117 Cre Pos (cells/spines). Scale bar equates to 1μm. c-d, Time course (c) and quantification (averaged over 0–45 s) (d) of CaMKII activation in Bdnf^fl/fl slices transfected with the CaMKII sensor (Cre Neg) or CaMKII plus Cre (Cre Pos). n = 7/13 Cre Neg and 7/15 for Cre Pos (cells/spines). e-f, Time course and quantification of the transient phase of spine volume change for experiments in (c). Data are mean +/− s.e.m. *p < 0.05 as determined by a two-tailed unpaired samples t-test.

Extended Data Figure 9. Design and validation of BDNF-SEP

a, Schematic of BDNF-SEP and BDNF-RFP. HA, hemagglutinin tag; Pro, amino acids 19–128 of human BDNF; BDNF, amino acids 129 – 247 of human BDNF corresponding to the mature chain; FLAG, FLAG tag; SEP, supereclipitic pHluorin; RFP, red fluorescent protein. b, Mechanistic model linking changes in SEP fluorescence with BDNF release. c, Change in BDNF-SEP fluorescence following glutamate uncaging under control (Ctrl), acidic (pH 6.5), and basic (pH 8.0) conditions. d, Confocal images of a CA1 pyramidal neuron transfected with mEGFP and BDNF- mRFP. Arrowheads indicate dendritic spines. e, Prolonged time course of BDNF-SEP fluorescence change (left) and spine volume change (right) in response to glutamate uncaging. n = 11/20 (cells/spines). f–g, Time course (f) and quantification (g) of spine volume change for experiments in Fig. 4c, d. Ctrl – control conditions; TeTx – neurons transfected with tetanus toxin, an inhibitor of exocytosis; POMC – neurons transfected with the POMC peptide, an inhibitor of activity- dependent BDNF release; AP5 – NMDAR inhibitor; AP5+NBQX – NMDAR and AMPAR inhibitors; and CN21 – CaMKII inhibitor. n = 31/218 Ctrl, 6/82 TeTx, 2/29 POMC, 3/50 AP5, 2/46 AP5+NBQX, 4/40 NBQX, and 7/88 CN21 (cells/spines). h, Data from Fig. 4d presented as medians +/− interquartile range. Data are mean +/− s.e.m. unless otherwise indicated. *p < 0.05) as determined by an ANOVA followed by Dunnet’s method to correct for multiple comparisons. **p < 0.05 as determined by a Kruskal-Wallis test followed by Dunn’s test.

Extended Data Figure 10. CA1-LTP requires exogenous BDNF

a, Time course of average EPSC amplitude changes recorded in CA1 pyramidal cells evoked by Schaffer collateral stimulation before and after LTP induction in the absence (Ctrl) or presence of human-IgG (H-IgG) or TrkB-Ig. Representative traces are above the graphs. n = 22 Ctrl, 9 H-IgG, and 12 TrkB-Ig (animals). b, Quantification of EPSC amplitude changes averaged over 10–20 min following LTP induction. c–d, Time course (c) and quantification (d) of the transient and sustained glutamate-uncaging-induced spine volume change in rat hippocampal slices in the absence (Ctrl) or presence of human-IgG (H-IgG) or TrkB-Ig. n = 8/8 Ctrl, 6/8 TrkBIg, and 4/6 HIgG (cells/spines). e, Model of spine autonomous, autocrine, BDNF release and post-synaptic TrkB activation. Data are mean +/− s.e.m. *p < 0.05 as determined by an ANOVA followed by Tukey’s method to correct for multiple comparisons.

Figure 1. sLTP induces rapid, persistent, and largely spine specific TrkB activation

a, Sensor design. b, 2pFLIM images of TrkB activation averaged across indicated time points. Arrowhead represents point of uncaging. Warmer colours indicate shorter lifetimes and higher TrkB activity. Image size is 6.8 × 4.4 μm. c, Time course of volume change for the stimulated spine. n = 50 cells/54 spines. d, e, Time course (d) and quantification (e) of peak (1.25–2 min) and sustained (10–20 min) activation for experiments in c measured as the change in sensor binding fraction in stimulated spines, adjacent spines and dendrites. Right panel in d shows magnified time course. n = 50/54 for stimulated spines and dendritic shafts, and 50/59 for adjacent spines (cells/spines). f, Spatial profile of TrkB activation—change in binding fraction of the dendrite plotted as a function of the distance from the stimulated spine. n = 48/52 (cells/stimulated spines). Data are mean ± s.e.m. *P < 0.05, analysis of variance (ANOVA) with Dunnet’s test.

Figure 2. TrkB activation during sLTP depends on NMDAR-CaMKII signalling and post-synaptic BDNF

a, b, Time course (a) and quantification (b) of peak and sustained TrkB activation in stimulated spines in the presence of pharmacological inhibitors. Ctrl, control; AP5 denotes an NMDAR inhibitor; CN21 denotes a CaMKII inhibitor. n = 19/19 control, 6/10 AP5, and 7/16 CN21 (cells/spines). c, d, Time course (c) and quantification (d) of transient (1–2 min) and sustained (10–20 min) spine volume change for experiments in a and b. e–h, Similar experiments to a–d but with different pharmacological conditions. HIgG, human IgG; TrkB-Ig, an extracellular scavenger of BDNF. n = 16/18 control, 6/11 HIgG, and 8/14 TrkB-Ig (cells/spines). i–l, Similar experiments to a–d but in Bdnf^fl/fl hippocampal slices transfected with the TrkB sensor with or without Cre-recombinase (Cre⁻ or Cre⁺, respectively). n = 7/15 Cre⁻ and 9/17 Cre⁺ (cells/spines). Data are mean ± s.e.m. *P < 0.05, ANOVA with Dunnet’s test (b, d), Tukey’s test (f, h) or a two-tailed t-test (j, l).

Figure 3. Endogenous BDNF localizes to axons, dendrites, and dendritic spines

a, Immunoperoxidase labelling of HA in hippocampal areas CA1 (left) and CA3 (right) from Bdnf-HA and wild-type (WT) mice, visualized by light microscopy. dSR, distal stratum radiatum; PCL, principal cell layer; pSR, proximal stratum radiatum; SLM, stratum lacunosum-moleculare; SO, stratum oriens. b–d, Immunoperoxidase labelling of HA in CA1 pyramidal neuron axon terminals (b), dendrites (c), and dendritic spines (d) of Bdnf-HA mice, visualized by electron microscopy. e, f, Quantification of observed immunoperoxidase labelling of HA in various cellular types (e) and subcellular compartments (f) in proximal and distal stratum radiatum in hippocampal slices from wild-type and Bdnf-HA mice. n = 3 animals each.

Figure 4. Glutamate uncaging induces rapid release of postsynaptic BDNF

a, Two-photon images of glutamate-uncaging-evoked changes in BDNF–SEP fluorescence in dendritic spines of CA1 hippocampal neurons. Each row represents the uncaging-triggered average of the BDNF–SEP signal in response to individual uncaging pulses for the designated time window. Image size is 3.9 × 5.5 μm. b, Averaged time course of BDNF–SEP fluorescence change in spines and adjacent dendritic shafts in response to glutamate uncaging (timing of glutamate pulses indicated by black bars (top)). Inset shows the change in mCherry (mCh) fluorescence (red) in response to glutamate uncaging, indicative of spine volume change (sLTP). n = 26/187 (cells/spines). c, Uncaging-triggered average of the increase in BDNF–SEP fluorescence with glutamate uncaging. TeTx, neurons transfected with tetanus toxin, an inhibitor of exocytosis; POMC, neurons transfected with the POMC peptide, an inhibitor of activity-dependent BDNF release. n = 31/218 control, 6/82 TeTx, 2/29 POMC, 3/50 AP5, 2/46 AP5 + NBQX, 4/40 NBQX, and 7/88 CN21 (cells/spines). d, Peak of the uncaging-triggered averaged increase of BDNF–SEP fluorescence in c. e, Time course of glutamate-uncaging-induced spine volume change for Bdnf^fl/fl hippocampal slices transfected with eGFP (Cre⁻), eGFP plus Cre (Cre⁺), or eGFP, Cre and BDNF–SEP. n = 9/13 Cre⁻, 6/11 Cre⁺ and 8/13 Cre⁺ plus BDNF–SEP (cells/spines). f, Transient (1–2 min) and sustained (10–40 min) spine volume change for experiments in e. g, h, Similar experiments to c and d but in Bdnf^fl/fl hippocampal slices in the absence or presence of Cre. n = 10/105 Cre⁻ and 15/132 Cre⁺ (cells/spines). Data are mean ± s.e.m. See Extended Data Fig. 9h for data in d represented as median ± interquartile interval. *P < 0.05, Kruskal–Wallis test with Dunn’s test (d) or an ANOVA with Tukey’s test (f).

Figure 5. Functional and structural LTP depends on post-synaptic BDNF-TrkB signalling

a, b, Time course (a) and quantification (b; 30–45 min) of excitatory postsynaptic current (EPSC) change recorded in CA1 pyramidal cells of hippocampal slices from Trkb^F616A and wild-type mice, before and after LTP induction in the presence of vehicle or 1NMPP1. Representative traces of Trkb^F616A slices with vehicle or 1NMPP1 are shown above the graphs. n = 11 Trkb^F616A vehicle, 10 Trkb^F616A 1NMPP1, 11 wild-type vehicle, and 13 wild-type 1NMPP1 (cells). c, d, Time course (c) and quantification (d) of transient and sustained glutamate-uncaging-induced spine volume change for Trkb^F616A hippocampal slices in the absence or presence of vehicle or 1NMPP1. n = 20/20 control, 10/13 vehicle, and 16/20 1NMPP1 (cells/spines). e–h, Similar experiments to a and b but from Bdnf^fl/fl mice infected with or without Cre. Representative traces are shown above the graphs. n = 20 Cre⁻ and 19 Cre⁺ (cells). g, h, Time course (g) and quantification (h) of transient and sustained glutamate-uncaging-induced spine volume change for Bdnf^fl/fl hippocampal slices transfected with eGFP or eGFP plus Cre. For Cre⁺ plus BDNF, Cre-positive cells were treated with BDNF for 10 min before glutamate uncaging. n = 13/14 Cre⁻, 22/32 Cre⁺, and 6/7 Cre⁺ plus BDNF (cells/spines). Data are mean ± s.e.m. *P < 0.05, two-tailed t-test (b, f) or ANOVA with Tukey’s test (d, h).