Synapse Elimination Triggered by BMP4 Exocytosis and Presynaptic BMP Receptor Activation  Takahito Higashi, Shinji Tanaka, Tadatsune Iida, Shigeo Okabe  Cell Reports  Volume 22, Issue 4, Pages 919-929 (January 2018) DOI: 10.1016/j.celrep.2017.12.101 Copyright © 2018 The Authors Terms and Conditions

Cell Reports 2018 22, 919-929DOI: (10.1016/j.celrep.2017.12.101) Copyright © 2018 The Authors Terms and Conditions

Figure 1 Regulated Synthesis of BMP4 and Presence of BMP4-Dependent Signaling Pathways in Hippocampal Neurons (A) Microarray analysis of TGF-β-related genes with or without manipulations of neuronal activity. Neuronal cultures were treated with the sodium channel blocker TTX for 2 days before RNA extraction. (B) qPCR of mRNA prepared from hippocampal neurons treated with either TTX or bicuculline for 2 days. (n = 3 samples for each condition. Unpaired t test, ∗∗p < 0.01, ∗∗∗p < 0.001.) (C) Immunoblot and quantification of proprotein of BMP4 (proBMP4) in cultured hippocampal neurons after treatment with TTX or bicuculline for 2 days. (n = 6 samples for each condition. Unpaired t test, ∗p < 0.05; NS, not significant, p > 0.05.) (D) Immmunoblot of BMPRI proteins in cultured hippocampal neurons. Neuronal cell lysates taken at the indicated culture stages (DIV 3, 7, 14, and 21) were analyzed. Anti-tubulin antibody was used as a control. (E) Phosphorylation and nuclear translocation of Smad1/5/8 in cultured neurons stimulated with BMP4 at DIV 14. Neurons were treated with BMP4 or vehicle for 0 or 180 min. Localization of activated Smad protein was visualized by anti-pSmad1/5/8 antibody. Cell nuclei were stained with Hoechst 33342. Scale bar, 50 μm. (F) Quantification of pSmad1/5/8 immunofluorescence in individual nuclei 180 min after BMP4 application. (Control [BSA], n = 331 cells/3 cultures; BMP, n = 286 cells/3 cultures. Comparison of the mean by unpaired t test, p < 0.01.) Cell Reports 2018 22, 919-929DOI: (10.1016/j.celrep.2017.12.101) Copyright © 2018 The Authors Terms and Conditions

Figure 2 Axonal Sorting, Transport, and Release of BMP4-Containing Vesicles (A) Live imaging of BMP4-EGFP and NPY-mCherry in the axon of cultured hippocampal neurons. Arrowheads indicate the positions of mobile BMP4-EGFP clusters. Scale bar, 10 μm. (B) Kymograph of axonal BMP4-containing vesicles and NPY-containing vesicles in dissociated neurons. Scale bar, 5 μm. (C) Quantitative analysis of colocalization between BMP4-EGFP and NPY-mCherry signals in stationary, anterogradely transporting, and retrogradely transporting fractions. (n = 3 cells.) (D) Time-lapse images of fluorescently labeled BMP4 (BMP4-EGFP or BMP4-mCherry; green), together with synaptophysin-tagRFP (magenta) or PSD-95-EGFP (magenta), in neurons at DIV 14. Stationary BMP4-EGFP clusters tend to be near pre- or postsynaptic structures (arrows). Scale bar, 2 μm. (E) Ratio of fluorescent clusters associated with other markers, relative to total number of clusters. (BMP4-Synph, BMP4 clusters associated with the synaptophysin signal per total BMP4 clusters; Synph-BMP4, synaptophysin clusters associated with the BMP4 signal per total synaptophysin clusters; BMP4-PSD-95, BMP4 clusters associated with the PSD-95 signal per total BMP4 clusters. Number of clusters counted: BMP4-EGFP, n = 195 clusters, 12 cells/5 cultures; BMP4-mCherry, 138 clusters, 11 cells/4 cultures; synaptophysin-tagRFP, n = 128 clusters, 12 cells/5 cultures.) (F) Release of BMP4-SEP by field stimulation. Images of a fusion event detected by increase in SEP fluorescence (upper left panel), together with the temporal profiles of fluorescence intensity (upper right panel), and the cumulative numbers of DCV fusion events after field stimulation (lower panel). Data from 5 cells before and after stimulation. Scale bar, 1 μm. (G) High-magnification images illustrating the spatial relationship of surface BMP4 puncta, presynaptic synaptophysin-EGFP, and postsynaptic PSD-95 immunolabeling. Panels show surface BMP4 clusters with both (1) pre- and postsynaptic markers, (2) with a postsynaptic marker only, (3) with a presynaptic marker only, and (4) without synaptic markers. Scale bar, 5 μm. (H) Live imaging of membrane recycling by FM1-43 and subsequent immunocytochemistry of surface and total BMP4-HA in the axon. The distance between FM1-43 labeling and nearby surface BMP4-HA clusters was measured. (n = 3 cells.) Scale bar, 1 μm. Cell Reports 2018 22, 919-929DOI: (10.1016/j.celrep.2017.12.101) Copyright © 2018 The Authors Terms and Conditions

Figure 3 Axonal Sorting of BMPRIa and Its Role in Surface Anchoring of BMP4 (A) Relative expression levels of TGF-β family receptors in dissociated hippocampal cultures at DIV 12 determined by microarray analysis. (B) Low-magnification views of axons expressing BMPRIa-HA and EGFP at DIV 15. Scale bar, 20 μm. (C) High-magnification images of axonal regions expressing BMPRIa-HA, together with a volume marker (EGFP). Scale bar, 5 μm. (D) Representative images of hippocampal neurons transfected with shRNA for BMPRIa. Distribution of cell surface and total BMP4 clusters in the axons was analyzed. The effect of coexpressing a shRNA-resistant form of BMPRIa was also evaluated. Scale bar, 5 μm. (E) Quantification of surface BMP4 signals normalized by intracellular BMP4 in hippocampal neurons. (Control of short hairpin RNA [sh-control], n = 13 neurons/5 cultures; short hairpin RNA of BMPRIa [sh-BMPRIa], n = 13 neurons/5 cultures; sh-BMPRIa plus BMPRIa∗-EGFP, n = 12 neurons/3 cultures. ANOVA with Turkey’s post hoc test, ∗p < 0.05; NS, not significant, p > 0.05.) Cell Reports 2018 22, 919-929DOI: (10.1016/j.celrep.2017.12.101) Copyright © 2018 The Authors Terms and Conditions

Figure 4 BMP4 Deletion in Neurons Increases Synaptic Vesicle Clustering (A) Western blot analysis of cell lysates taken from dissociated hippocampal cultures at DIV 14. Neurons with the Bmp4flox/flox genotype were infected with Cre-expressing adenovirus at DIV 10 to excise the floxed Bmp4 exon. Anti-tubulin immunoblot is shown as a control for protein loading. (B) Representative images of dendrites in BMP4 knockout (KO) and control neuron cultures. Dendritic morphology was visualized by EGFP expression. Scale bar, 10 μm. (C) Dendritic segments of BMP4 KO and control neurons at DIV 14. Morphology of dendrites and spines was detected by EGFP fluorescence. Neurons with the Bmp4flox/flox genotype were transfected with either the EGFP expression vector or a plasmid with floxed stop codon sequences between beta-actin promoter and EGFP (beta-actin-loxSTOP-EGFP) at DIV 8 or 9. After transfection, culture preparations transfected with beta-actin-loxSTOP-EGFP were infected with Cre-expressing adenoviruses at DIV 10 or 11 to produce Bmp4−/− and EGFP-positive neurons. The positions of excitatory presynaptic structures were identified by anti-VGluT1 staining. Scale bar, 5 μm. (D) Density of VGluT1-positive puncta in neurons with different genotypes. (Bmp4flox/flox, n = 27 cells/10 cultures; Bmp4−/−, n = 28 cells/10 cultures. Unpaired t test, ∗∗∗p < 0.001.) (E) Density of dendritic protrusions in neurons with different genotypes. (Bmp4flox/flox, n = 40 cells/15 cultures; Bmp4−/−, n = 38 cells/14 cultures. Unpaired t test, ∗∗p < 0.01.) (F) Length of dendritic protrusions in neurons with different genotypes. (Bmp4flox/flox, n = 9 cells, 204 protrusions/3 cultures; Bmp4−/−, n = 15 cells, 404 protrusions/3 cultures. Unpaired t test, ∗∗∗p < 0.001.) (G) Distribution of GAD65-positive inhibitory presynaptic structure along dendrites of Bmp4−/− and control neurons at DIV 14. Bmp4flox/flox neurons were transfected with either EGFP expression plasmid or beta-actin-loxSTOP-EGFP plasmid at DIV 8 or 9. Additional infection of neurons expressing beta-actin-loxSTOP-EGFP with Cre-expressing adenoviruses at DIV 10 or 11 induced Bmp4 gene knockout and EGFP expression. Scale bar, 5 μm. (H) Density of GAD65-positive puncta in neurons with different genotypes. (Bmp4flox/flox, n = 14 cells/5 cultures; Bmp4−/−, n = 11 cells/4 cultures; Unpaired t test; NS, not significant, p > 0.05.) Cell Reports 2018 22, 919-929DOI: (10.1016/j.celrep.2017.12.101) Copyright © 2018 The Authors Terms and Conditions

Figure 5 BMP4 Overexpression and Knockout in Single Cells Affect Synapse Development (A) Low-magnification images of transfected cells. Morphology of transfected cells was visualized by immunostaining with β-galactosidase (β-gal). Scale bar, 20 μm. (B) Accumulation of synaptophysin-EGFP clusters along axons of transfected neurons (identified by anti-β-gal staining) associated with MAP2-positive dendrites of non-transfected neurons. Scale bar, 10 μm. (C) Quantification of synaptophysin-EGFP cluster density along axons associated with MAP2-positive dendrites of non-transfected neurons. (Bmp4flox/flox, n = 11 cells/4 cultures; Bmp4−/−, n = 14 cells/6 cultures; BMP4 rescue, n = 6 cells/4 cultures. ANOVA with Turkey’s post hoc test, ∗∗∗p < 0.001; NS, not significant, p > 0.05.) (D) Low-magnification images of neurons transfected with BMP4 or a control plasmid. Morphology of transfected neurons was visualized by anti-β-gal staining. Scale bar, 20 μm. (E) Accumulation of synaptophysin-EGFP along axons of transfected neurons associated with MAP2-positive dendrites of non-transfected neurons. Scale bar, 10 μm. (F) Quantification of synaptophysin-EGFP cluster density along axons associated with MAP2-positive dendrites of non-transfected neurons. (Control, n = 18 cells/6 cultures; BMP4, n = 16 cells/5 cultures. Unpaired t test, ∗∗∗p < 0.001.) Cell Reports 2018 22, 919-929DOI: (10.1016/j.celrep.2017.12.101) Copyright © 2018 The Authors Terms and Conditions

Figure 6 Stability of Synaptophysin Clusters Is Affected by Nearby Surface BMP4 Clusters (A) Live cell imaging of axons expressing BMP4-SEP and synaptophysin-tagRFP at DIV 14. Scale bar, 10 μm. (B) Time-lapse images of the areas marked by rectangles in (A) at intervals of 60 min for 4 hr. Arrowheads mark the position of a synaptophysin-tagRFP cluster associated with BMP4-SEP in the first frame. This synaptophysin-tagRFP cluster disappeared during imaging. Scale bar, 5 μm. (C) Position survival rates of synaptophysin clusters with or without nearby BMP4 clusters. (BMP4+ Synph, synaptophysin clusters with BMP4, n = 92 clusters; BMP4− Synph, synaptophysin clusters without BMP4, n = 445 clusters from imaging of 3 cells. Unpaired t test, ∗p < 0.05.) Cell Reports 2018 22, 919-929DOI: (10.1016/j.celrep.2017.12.101) Copyright © 2018 The Authors Terms and Conditions

Figure 7 In Vivo Imaging of Presynaptic Components and Axons in Neurons Genetically Manipulated to Eliminate BMP4 (A) In vivo two-photon images of axons extending from cortical pyramidal neurons expressing synaptophysin-EGFP and DsRed2 at postnatal week 3. Cortical neuroblasts were electroporated in utero with an expression plasmid of Cre recombinase, together with an expression plasmid for synaptophysin-EGFP and a plasmid that has a floxed stop codon sequence between beta-actin promoter and DsRed2. Expression of DsRed2 indicates successful expression of Cre recombinase. In the control group, in utero electroporation was performed with a mixture of two expression plasmids for synaptophysin-EGFP and DsRed2 without a Cre expression plasmid. Scale bar, 10 μm. (B) Densities of synaptophysin-EGFP clusters in axons of control or Cre-expressing neurons at postnatal week 3. (n = 150 axonal segments/5 animals for both control and Cre. Unpaired t test, ∗p < 0.05.) (C) In vivo time-lapse imaging of synaptophysin-EGFP clusters with an interval of 2 days at postnatal week 3. White arrows indicate gain or loss of fluorescent puncta. Scale bar, 10 μm. (D) Turnover rate of synaptophysin-EGFP puncta over 2 days in axons of control and Cre-expressing neurons at postnatal week 3. (Control, n = 12 axon segments, 4 independent in vivo experiments; Cre, n = 13 axon segments, 4 independent in vivo experiments. Unpaired t test, ∗p < 0.05.) (E) Proposed model of synapse destabilization by BMP4 clusters on the axonal surface. Cell Reports 2018 22, 919-929DOI: (10.1016/j.celrep.2017.12.101) Copyright © 2018 The Authors Terms and Conditions