S-SCAM is essential for synapse formation

Abstract

Synapse formation is critical for the wiring of neural circuits in the developing brain. The synaptic scaffolding protein S-SCAM/MAGI-2 has important roles in the assembly of signaling complexes at post-synaptic densities. However, the role of S-SCAM in establishing the entire synapse is not known. Here, we report significant effects of RNAi-induced S-SCAM knockdown on the number of synapses in early stages of network development in vitro. In vivo knockdown during the first three postnatal weeks reduced the number of dendritic spines in the rat brain neocortex. Knockdown of S-SCAM in cultured hippocampal neurons severely reduced the clustering of both pre- and post-synaptic components. This included synaptic vesicle proteins, pre- and post-synaptic scaffolding proteins, and cell adhesion molecules, suggesting that entire synapses fail to form. Correspondingly, functional and morphological characteristics of developing neurons were affected by reducing S-SCAM protein levels; neurons displayed severely impaired synaptic transmission and reduced dendritic arborization. A next-generation sequencing approach showed normal expression of housekeeping genes but changes in expression levels in 39 synaptic signaling molecules in cultured neurons. These results indicate that S-SCAM mediates the recruitment of all key classes of synaptic molecules during synapse assembly and is critical for the development of neural circuits in the developing brain.

Keywords: neuronal morphology; scaffolding protein; spine; synapse formation; synaptic transmission.

Figure 1

ShRNA mediated knockdown of S-SCAM in cultured hippocampal neurons. (A) Schematic diagram of the experimental timeline. (B) Design of lentiviral shRNA vectors for knockdown of all three S-SCAM isoforms. U6, U6 RNA polymerase III promotor; Ub, ubiquitin promotor. (C–F) Specificity and efficacy of S-SCAM-RNAi. (C) Western blot confirming knockdown of endogenous S-SCAM at protein level in low-density cultures of hippocampal neurons. (D) Semi-quantitative analysis of band intensities. (E) Representative images of S-SCAM knockdown neurons immunostained for S-SCAM and GFP. Bar, 4 μm. Quantification of the data is provided in (F). ^***p ≤ 0.001, unpaired t-test.

Figure 2

Loss of function of S-SCAM causes a severe reduction of synapse formation. (A) Schematic diagram of the experimental timeline. (B–K) Representative images of hippocampal neurons that were transfected with S-SCAM knockdown or control vectors and analyzed by immunofluorescence with antibodies to the presynaptic proteins Bassoon, VAMP2, Synaptophysin, VGlut1, VGAT, and the post-synaptic proteins PSD95, Neuroligin-1/2/3/4, pan-Shank, and Gephyrin. Bar, 10 μm, bar in zoom, 3 μm. (L, M) Summary graphs of the effect of S-SCAM knockdown on the density of presynaptic (L) and post-synaptic (M) markers. A 100–155 puncta (10–15 cells) were quantified. N = 3 independent culture experiments. **p < 0.01, two-tailed unpaired t-test.

Figure 3

Reversal of synapse loss by overexpression of S-SCAMα, -β, or -γ in S-SCAM knockdown neurons. (A) Design of expression vectors for overexpression of S-SCAMα, -β, or -γ in S-SCAM knockdown neurons. CMV, cytomegalovirus promotor; IRES, internal ribosome entry sequence. (B) Verification of S-SCAM rescue vectors. HEK293 cells were transfected with S-SCAM RNAi or control plasmids and cotransfected with either S-SCAMα or S-SCAMα, -β, or -γ rescue vectors (S-SCAMαr, S-SCAMβr, and S-SCAMγr) containing silent mutations. Anti-tubulin blot is shown as a loading control. (C) Schematic diagram of the experimental timeline. (D–G) Rescue experiments with S-SCAMαr, -βr, and -γr. Representative images of S-SCAM RNAi or control neurons cotransfected with S-SCAMαr, S-SCAMβr, or S-SCAMγr stained for Bassoon (D) or PSD95 (F). Bar, 4 μm. Quantification of the rescue experiments for Bassoon (E) and PSD95 (G) dendritic puncta density. A 85–125 puncta (10–12 cells) were quantified. N = 3 independent culture experiments. ***p < 0.001, ANOVA Tukey's test.

Figure 4

Loss of function of S-SCAM affects neuronal morphology and synaptic maintenance. (A, D) Schematic diagram of the experimental timeline. (B, C) Sholl analysis revealed a loss of neuronal processes upon S-SCAM knockdown. One-way ANOVA, **P < 0.01. (E, F) Neurons in older cultures (DIV14) were transfected with S-SCAM knockdown or control constructs and immunostained for Bassoon. Quantification shows a reduction in Bassoon puncta density, control RNAi, n = 96; S-SCAM RNAi, n = 112. N = 3 independent culture experiments; **p < 0.01, two-tailed unpaired t-test.

Figure 5

S-SCAM knockdown reduces spontaneous synaptic activity in conventional hippocampal cultures via a post-synaptic effect. (A) Schematic diagram of the experimental timeline. (B) Example traces of mEPSC events in DIV9–10 neurons infected with control RNAi (black) or S-SCAM RNAi (gray). (C) S-SCAM RNAi reduces mEPSC frequency, whereas the amplitude of mEPSC events between S-SCAM knockdown and control neurons is unchanged. n = 13 for S-SCAM RNAi and control neurons. N = 4 independent culture experiments. ***p < 0.001, Mann–Whitney U test.

Figure 6

S-SCAM knockdown reduces spontaneous and evoked excitatory synaptic activity in autaptic hippocampal micro-island cultures. (A) Schematic diagram of the experimental timeline. (B) Example traces of evoked EPSC responses. (C) Mean EPSC amplitude measured in neurons after lentivirus expression of the control RNAi and SSCAM knockdown constructs. (D) Example traces of current responses induced by the application of 0.5 M sucrose solution. (E) Mean RRP sizes as estimated by the charge integral measured inward currents during the application of 0.5 M sucrose solution. (F) Calculated mean vesicular release probabilities calculated by dividing the charge transfer during a single EPSC by the charge transfer measured during RRP release. (G) Evoked EPSC depression during 10 Hz stimulation train. Data were normalized to the first response in the train. (H) Mean mEPSC frequencies. (I) Mean mEPSC amplitudes. (J) Example traces of current responses induced by the application of 100 μM glutamate solution. (K) Mean peak amplitudes of responses to exogenous 100 μM glutamate. (L) Example traces of current responses induced by the application of 10 μM GABA solution. (M) Mean peak amplitudes of responses to exogenous 10 μM GABA. ***p < 0.001, two-tailed unpaired t-test; n, neurons recorded.

Figure 7

Expression of housekeeping genes is unchanged, whereas expression of signaling proteins is altered upon S-SCAM knockdown. (A) Schematic diagram of the experimental timeline. (B) Selected results of the deep sequencing analysis of neuronal cultures treated with S-SCAM shRNA or control shRNA. The graph shows the log₂-fold change in the shRNA-treated cultures compared to control cultures of housekeeping genes (black), synaptic marker proteins tested by immunofluorescence in our study (green), and known S-SCAM interaction partners (red). *p < 0.05; **p < 0.01; ***p < 0.001, false discovery rate. N = 3 independent experiments. (C) RT-PCR of selected RNAs from Figure 7B. RNA was isolated from n = 3 cultures transduced with SCAM-RNAi or EGFP. Ct values were normalized to α-tubulin (δct). RNAi values were normalized to EGFP (δδct), and the fold change (FC) was calculated. Values are plotted as log₂FC.

Figure 8

Gene ontology and functional clustering of differentially expressed genes by the knockdown of S-SCAM. (A) The GO terms for biological processes with the lowest adjusted p-value (teal bars) by using the g:Profiler tool, and the number of differentially expressed genes in each category (black bars). (B) STRING network analysis of differentially expressed genes functionally connected to S-SCAM (MAGI-2, in bold). Blue nodes represent protein products present in the nucleus, red nodes indicate synaptic proteins, and green nodes indicate proteins present in cell junctions but not proven to be synaptic. Nodes that have more than one color are proteins that localize to both subcellular compartments, whereas gray nodes indicate proteins that localize to none of these three compartments. Edges represented by full lines are functional associations within a cluster, whereas edges represented by dotted lines are connections between clusters.

Figure 9

S-SCAM is involved in spine formation in vitro and in vivo. (A, E) Schematic diagram of the experimental timeline. (B) Images of dendrites of hippocampal neurons transfected with S-SCAM knockdown or control constructs and immunostained for Bassoon. Bar, 5 μm. (C) Quantification of the effect of S-SCAM knockdown on dendritic spine density, control RNAi, n = 124; S-SCAM RNAi, n = 134. N = 3 independent culture experiments; **p < 0.01, unpaired t-test. (D) Quantification of the effect of S-SCAM knockdown on the percentage of spines with Bassoon staining; control RNAi, n = 155; S-SCAM RNAi, n = 136. N = 3 independent culture experiments. *p < 0.05, unpaired t-test. (F–I) Lentiviral particles expressing shRNAs against S-SCAM or control shRNAs were injected into the cortex of P3 rat brains (E). Brains were analyzed at P19 by immunohistochemistry with an antibody against GFP, and confocal images of dendrites of cortical neurons were analyzed (F). Bar, 10 μm. Quantification of the effect of S-SCAM knockdown on dendritic spine density, number of mature mushroom spines (G), average spine neck length (H), and spine head diameter (I); control RNAi, n = 13 neurites S-SCAM RNAi, n = 16 neurites Mean ± SEM (N = 2 control RNAi-injected animals; N = 4 S-SCAM RNAi-injected animals); **p < 0.01, ^****p ≤ 0.001, unpaired t-test.