Loss of MeCP2 disrupts cell autonomous and autocrine BDNF signaling in mouse glutamatergic neurons

Abstract

Mutations in the MECP2 gene cause the neurodevelopmental disorder Rett syndrome (RTT). Previous studies have shown that altered MeCP2 levels result in aberrant neurite outgrowth and glutamatergic synapse formation. However, causal molecular mechanisms are not well understood since MeCP2 is known to regulate transcription of a wide range of target genes. Here, we describe a key role for a constitutive BDNF feed forward signaling pathway in regulating synaptic response, general growth and differentiation of glutamatergic neurons. Chronic block of TrkB receptors mimics the MeCP2 deficiency in wildtype glutamatergic neurons, while re-expression of BDNF quantitatively rescues MeCP2 deficiency. We show that BDNF acts cell autonomous and autocrine, as wildtype neurons are not capable of rescuing growth deficits in neighboring MeCP2 deficient neurons in vitro and in vivo. These findings are relevant for understanding RTT pathophysiology, wherein wildtype and mutant neurons are intermixed throughout the nervous system.

Keywords: BDNF; Rett syndrome; cell autonomous; glutamatergic; mouse; neuroscience; synaptic transmission.

Figure 1.. Loss of MeCP2 alters neurite outgrowth and neuronal soma size.

(A) Co-immunostaining of glutamatergic hippocampal neurons from WT and Mecp2^Null/y mice at DIV 2, 4, 8 and 12 for Tau1 (cyan) and MAP2 (green). Scale bars represent 20 µm. (B–E) Mean axonal length and dendritic length measured at different time points for Mecp2^Null/y (B,D) and Mecp2^Tg1 neurons (C,E). (F) Somatic labeling of glutamatergic hippocampal neurons from WT and Mecp2^Null/y mice at DIV 2, 4, 8 and 12 with MAP2. Scale bar represents 20 µm. (G,H) Mean soma area measured at different time points for Mecp2^Null/y (G) and Mecp2^Tg1 neurons (H). Data shown as mean ± SEM. **p<0.01; ***p<0.001. DOI: http://dx.doi.org/10.7554/eLife.19374.003

Figure 2.. BDNF overexpression in Mecp2^Null/y autaptic neurons normalizes cell morphology and restores synaptic output.

(A) Experimental scheme of lentivirus-mediated BDNF overexpression in Mecp2^Null/yneurons, showing neuronal membrane (black), BDNF (red) and TrkB receptor (blue). (B) Representative images of neuronal morphology under the following conditions (from left to right): WT, Null, Null + BDNF, Null + BDNF + ANA12, WT + ANA12. Dendritic outgrowth (top) and glutamatergic synapses (bottom) indicated by MAP2 (green) and VGLUT1 (red) labeling. Scale bars represent 20 µm. (C–E) Bar graphs show mean dendrite length (C), glutamatergic synapse number (D) and neuronal soma area (E). (F) Representative traces of evoked EPSCs recorded from autapses under the following conditions: WT, Null, Null + BDNF, Null + BDNF + ANA12, WT + ANA12, Null + ANA12. (G) Bar graph shows mean evoked EPSC amplitude. (H) Representative traces of average postsynaptic response to 5 s application of 500 mM sucrose. Experimental groups same as (F). (J–K) Bar graphs show mean RRP size (I), vesicular release probability P_vr (J) and paired-pulse ratio with 25 ms inter-stimulus interval (K). (L) Bar graph shows EPSC amplitude, RRP size and P_vr, PPR and dendrite length, glutamatergic synapse number and soma area, normalized to WT (dashed red line). Number of neurons (n) shown in the bars for all graphs. Data shown as mean ± SEM. *p<0.05; **p<0.01; ***p<0.001; ns: not significant. DOI: http://dx.doi.org/10.7554/eLife.19374.004

Figure 2—figure supplement 1.. BDNF overexpression in Mecp2^Null/y autaptic neurons normalizes nuclei size.

Bar graph shows mean neuronal nuclei size. Number of neurons (n) shown in the bars for all graphs. Data shown as mean ± SEM. *p<0.05; **p<0.01; ***p<0.001; ns: not significant. DOI: http://dx.doi.org/10.7554/eLife.19374.005

Figure 3.. BDNF neutralization fails to rescue physiological phenotypes while exogenous application of BDNF restores physiological and morphological phenotypes, reaffirming specificity of BDNF-TrkB interaction in Mecp2^Null/y glutamatergic neurons.

(A) Experimental scheme of lentivirus-mediated BDNF overexpression and BDNF neutralization in Mecp2^Null/yneurons, showing neuronal membrane (black), BDNF (red), TrkB receptor (blue) and BDNF neutralizing antibody (cyan). (B) Representative traces of evoked EPSCs recorded from autapses under the following conditions: WT, Null, Null + BDNF, Null + BDNF + neutralization. (C) Bar graph shows mean evoked EPSC amplitude. (D) Representative traces of average current response to 5 s application of 500 mM sucrose. Experimental groups same as (B). (E–G) Bar graphs show mean RRP size (E), P_vr (F) and 25 ms ISI – PPR (G). (H) Experimental scheme depicting exogenous application of BDNF in Mecp2^Null/y neurons, showing neuronal membrane (black), BDNF (red) and TrkB receptor (blue). (I–M) Bar graphs show mean evoked EPSC amplitude (I), RRP size (J), dendrite length (K), glutamatergic synapse number (L) and neuronal soma area (M). Number of neurons (n) shown in the bars for all graphs. Data shown as mean ± SEM. *p<0.05; **p<0.01; ***p<0.001; ns: not significant. DOI: http://dx.doi.org/10.7554/eLife.19374.006

Figure 3—figure supplement 1.. NGF overexpression in Mecp2^Null/y neurons does not rescue glutamatergic synaptic output.

(A) Experimental scheme of lentivirus-mediated NGF overexpression in Mecp2^Null/y neurons, showing neuronal membrane (black), BDNF (red), TrkB receptor (blue), NGF (dark blue) and TrkA receptor (yellow). (B–E) Bar graphs show mean evoked EPSC amplitude (B), RRP size (C), P_vr (D) and 25 ms ISI – PPR (E). Number of neurons (n) shown in the bars for all graphs. Data shown as mean ± SEM. *p<0.05; **p<0.01; ns: not significant. DOI: http://dx.doi.org/10.7554/eLife.19374.007

Figure 3—figure supplement 2.. Exogenous application of BDNF does not alter release efficiency or short-term plasticity in Mecp2^Null/y hippocampal neurons.

(A,B) Bar graphs show mean P_vr (A) and 25 ms ISI – PPR (B). Number of neurons (n) shown in the bars for all graphs. Data shown as mean ± SEM. DOI: http://dx.doi.org/10.7554/eLife.19374.008

Figure 4.. BDNF overexpression in Mecp2^Null/y glutamatergic neurons restores synapse density.

(A) Representative images of WT and Mecp2^Null/y neurons labeled with MAP2, VGLUT1 and Homer1 under the following conditions (from top to bottom): WT, Null, Null + BDNF. Scale bar represents 3 µm. (B–E) Bar graphs show mean normalized VGLUT1 synapse density (B), Homer1 synapse density (C), colocalized VGLUT1-Homer1 synapse density (D), and normalized fraction of VGLUT1 puncta colocalized to Homer1 (E). Number of neurons (n) shown in the bars for all graphs. Data shown as mean ± SEM. *p<0.05; **p<0.01; ***p<0.001. DOI: http://dx.doi.org/10.7554/eLife.19374.009

Figure 4—figure supplement 1.. BDNF overexpression in Mecp2^Null/y glutamatergic neurons does not alter expression levels of synaptic markers.

(A–C) Bar graphs show mean normalized VGLUT1 fluorescence intensities in autaptic (A) and continental WT and Mecp2^Null/y neurons (B), and Homer1 fluorescence intensities (C). Number of synaptic puncta (n) shown in the bars for all graphs. Data shown as mean ± SEM. DOI: http://dx.doi.org/10.7554/eLife.19374.010

Figure 5.. TrkB activation specifically in Mecp2^Null/y neurons in an in vitro RTT model normalizes glutamatergic synapse number in a cell autonomous and autocrine manner.

(A) Left panel: Neuronal pair depicting lentivirus-mediated labeling of nucleus (NLS) and synapses of WT (NLS: red, Synaptophysin: green) and Mecp2^Null/y neurons (NLS: green, Synaptophysin: red), co-stained for MAP2 to identify synapses localized on dendrites. Right panel: Experimental scheme of lentivirus-mediated BDNF overexpression in WT or Mecp2^Null/y neurons, showing neuronal membrane (black), BDNF (red) and TrkB receptor (blue). (B) Representative images of co-cultured WT and Mecp2^Null/y neurons labeled with MAP2 under the following conditions (from left to right): WT/Null; WT+BDNF/Null and WT/Null+BDNF. Bottom panel shows WT (green) and Null (red) Synaptophysin+ synapses localized on a single dendrite. Scale bars on top and middle panel represent 20 µm. Scale bar on bottom panel represents 3 µm. (C,D) Bar graphs show mean glutamatergic synapse density (C) and nucleus size (D) for all co-cultured groups. Number of neurons (n) shown in the bars for all graphs. Data shown as mean ± SEM. **p<0.01; ***p<0.001; ****p<0.0001; ns: not significant. DOI: http://dx.doi.org/10.7554/eLife.19374.011

Figure 6.. Cell autonomous BDNF-TrkB signaling regulates glutamatergic synapse number by functioning as a presynaptic rate-limiting factor.

(A) Experimental scheme of lentivirus-mediated BDNF overexpression and exogenous application of TrkB agonist in WT or Mecp2^Null/y neurons, showing neuronal membrane (black), BDNF (red), TrkB receptor (blue) and TrkB agonist (green). (B) Bar graph shows glutamatergic synapse density for all co-cultured groups upon application of TrkB agonist, 7,8-DHF. (C) Example two-neuron scheme illustrating pre- and postsynaptic neurons (WT or Mecp2^Null/y) forming synapses onto itself or onto the partner neuron, in a mixed WT/Null co-culture system. (D) Bar graphs show glutamatergic synapse density measured from a postsynaptic WT (clear bars) and Null (filled bars) neuron for all experimental groups. Number of neurons (n) shown in the bars for all graphs. Data shown as mean ± SEM. ***p<0.001; ns: not significant. DOI: http://dx.doi.org/10.7554/eLife.19374.012

Figure 6—figure supplement 1.. WT and Mecp2^Null/y hippocampal neurons reveal transient and persistent fusion events upon stimulation, show equivalent activity-dependent BDNF secretion and membrane-resident BDNF-SpH fraction.

(A) Representative images of a WT hippocampal neuron expressing BDNF-superecliptic pHluorin perfused with standard extracellular solution, 60 mM KCl, 50 mM NH₄Cl and washout with extracellular solution (left to right). Scale bar represents 10 µm. (B) BDNF fluorescence (ΔF/F₀) over time reflecting fusion events upon application of KCl, MES and NH₄Cl solutions. Black line indicates duration of KCl, MES and NH₄Cl application. (C) Average of BDNF fluorescence (ΔF/F₀) after NH₄Cl application (WT, n = 1521 events, Null, n = 948 events). (D) Bar graph shows activity-dependent BDNF release of WT (1464 events) and Mecp2^Null/y neurons (895 events). (E,F) Representative traces of BDNF fluorescence (ΔF/F₀) indicating transient and persistent fusion events upon KCl-induced membrane depolarization in WT (E) and Mecp2^Null/y neurons (F). (G) Bar graph shows BDNF-SpH fraction in WT and Mecp2^Null/y neurons indicating surface accumulation of BDNF upon acid wash (MES, pH 5.5; WT, 1480 events; Mecp2^Null/y, 900 events). Data shown as mean ± SEM. DOI: http://dx.doi.org/10.7554/eLife.19374.013

Figure 7.. MeCP2-deficient hippocampal CA1 neurons have smaller somata and nuclei in Mecp2^+/- heterozygous mice in vivo.

(A) Representative images of hippocampal CA1 from Mecp2^+/- mice labeled for DAPI (blue), MAP2 (green) and MeCP2 (red) at two- (top) and eight-weeks of age (bottom). Scale bar represents 50 µm. (B) Representative images of hippocampal CA1 from Mecp2^+/- mice labeled for DAPI (blue) and MeCP2 (green) at two- (top) and eight-weeks of age (bottom). Scale bar represents 50 µm. (C–F) Scatter plots show mean soma area and nucleus size of MeCP2+ and MeCP2- neurons at two- (C,E) and eight-weeks of age (D,F). Number of neurons (n) shown in all graphs. Data shown as mean ± SEM. ***p<0.001; ****p<0.0001. DOI: http://dx.doi.org/10.7554/eLife.19374.014