Rapid dispersion of SynGAP from synaptic spines triggers AMPA receptor insertion and spine enlargement during LTP

Abstract

SynGAP is a Ras-GTPase activating protein highly enriched at excitatory synapses in the brain. Previous studies have shown that CaMKII and the RAS-ERK pathway are critical for several forms of synaptic plasticity including LTP. NMDA receptor-dependent calcium influx has been shown to regulate the RAS-ERK pathway and downstream events that result in AMPA receptor synaptic accumulation, spine enlargement, and synaptic strengthening during LTP. However, the cellular mechanisms whereby calcium influx and CaMKII control Ras activity remain elusive. Using live-imaging techniques, we have found that SynGAP is rapidly dispersed from spines upon LTP induction in hippocampal neurons, and this dispersion depends on phosphorylation of SynGAP by CaMKII. Moreover, the degree of acute dispersion predicts the maintenance of spine enlargement. Thus, the synaptic dispersion of SynGAP by CaMKII phosphorylation during LTP represents a key signaling component that transduces CaMKII activity to small G protein-mediated spine enlargement, AMPA receptor synaptic incorporation, and synaptic potentiation.

Figure 1. Dynamic Dispersion of SynGAP from Spines during LTP

(A) Dispersion of SynGAP from synapses upon LTP stimulation. GFP-tagged SynGAP was dynamically dispersed upon LTP. mCherry was used as a morphology marker to show spine enlargement during LTP. Enlarged spines (e.g., spines 1–4) dispersed SynGAP. Some “no-response spines” (e.g., spines a–b) failed to disperse SynGAP. Correlations are shown in (D). Scale bar, 5 μm. (B) Time course of averages of all spine size changes and SynGAP dispersion during LTP (n = 3 independent experiments/neurons that contain 35 spines). (C) PSD fractionation during LTP also showed dynamic SynGAP dispersion from PSDs during LTP. Error bars indicate ± SEM. (D) The relationship of “Dispersion of SynGAP from spine” and “Spine enlargement” in sustained phase (60 min) showed a strong and significant positive correlation between SynGAP dispersion and spine enlargement (n = 91 spines from seven independent experiments/neurons, R² = 0.7288, p < 0.01). Spines 1–4 and a–b in (A) are also displayed. Note that the y intercept of trend line is nearly zero, showing that there was no spine enlargement if the spine failed to disperse SynGAP. (E) Effects of pharmacological inhibition of “NMDAR-CaMKII,” “small G proteins,” and “actin polymerization” on Spine volume and SynGAP dispersion in the Sustained phase (60 min) (Spine volume: Drug F(9, 40) = 23.97, p < 0.001; SynGAP dispersion: Drug F(9, 40) = 28.00, p < 0.001). These results showed NMDAR-CaMKII pathway involved both in spine enlargement and SynGAP dispersion. Inhibition of “small G proteins, downstream kinase” and “actin polymerization” inhibited spine enlargement but not SynGAP dispersion, suggesting SynGAP dispersion is upstream cellular process of small G protein activation and actin polymerization. Error bars indicate ± SEM. (F) Effects of pharmacological inhibition of “NMDAR-CaMKII,” “small G proteins,” and “actin polymerization” on Spine volume and SynGAP dispersion in the Acute phase (10 min) (Spine volume: Drug F(9, 40) = 42.13, p < 0.001; SynGAP dispersion: Drug F(9, 40) = 17.44, p < 0.001). Note that spine size enlargement was insensitive to CaMKII inhibition, Rac1 inhibition, and low dose of Latrunculin A (20 nM) treatment only in Acute phase, whereas SynGAP dispersion was still inhibited by CaMKII inhibition also in acute phase, suggesting CaMKII activity is essential for SynGAP dispersion (n = 5 independent experiments/neurons in all conditions, which contain 61 [Ctrl], 60 [APV], 52 [W7], 48 [KN62-4 μM], 50 [KN62-20 μM], 59 [Ras^DN], 62 [Rac^DN], 60 [G1152], 51 [LatA-20nM], and 60 [LatA-100nM] spines in total, respectively). Error bars indicate ± SEM.

Figure 2. Degree of SynGAP Dispersion Foretells the Long-Term Spine Enlargement and Maintenance Showing SynGAP as “Negative Synaptic Marker” for Sustained Phase

(A) Three typical spine responses during LTP. (A1) “Stable” synapses (57.6% ± 2.9%) dispersed SynGAP after LTP and spines size enlargement was well retained for sustained phase. (A2) “Transient” synapses (14.4% ± 2.0%) failed to disperse SynGAP, but spine enlargement in acute phase was normal. All transient spines which failed to disperse SynGAP shrank back to the basal level in the sustained phase. (A3) Some portions of spines (28.0% ± 3.6%) were “No response” (No Res.) type, where both spine size change and SynGAP dispersion did not occur. (Stable: Spines with volume increased over 15% both at 10 and 60 min; Transient: Spines with volume increased over 15% at 10 min, but back to less than 15% changes at 60 min; No Res.: Spines with volume increased less than <15% at 10 and 60 min). Scale bar, 5 μm. (B) Relationships between SynGAP dispersion in acute phase and spine volume changes. There is strong and significant positive correlation between SynGAP dispersion in “acute” phase and spine enlargement in “sustained” phase (R² = 0.73, p < 0.001). There was less positive correlations between SynGAP dispersion and spine enlargement in “acute” phase (R² = 0.35, p < 0.001), suggesting that SynGAP dispersion in “acute” phase predicts spine enlargement and maintenance for long term rather than in acute phase. Ninety percent prediction bands of trend lines are also displayed. Examples of track changes (arrows) of unique “Stable” spines (S 1–3) and “Transient” spines (T 1–3) are presented (n = 103 spines from nine independent experiments/neurons).

Figure 3. CaMKII Inhibitor KN62 Maintained SynGAP in Spines and Changed “Stable” to “Transient” Synapses

(A) CaMKII inhibitor blocked SynGAP dispersion from spines. Spines were still enlarged in the acute phase but returned to the basal level in the sustained phase. Scale bar, 5 μm. (B) Time course of spine enlargement and SynGAP dispersion with or without CaMKII inhibitor. Note that the CaMKII inhibitor blocks SynGAP dispersion both in acute and sustained phase, and spine size returned to the basal level in the sustained phase (Ctrl: n = 3 independent experiments/neurons that contain 35 spines, KN62: n = 3 independent experiments/neurons that contain 39 spines). Error bars indicate ± SEM. (C) Population of “Stable,” “Transient,” and “No response” synapses with or without CaMKII inhibitor. KN-62 dramatically reduced “Stable” synapses and changed them into “Transient” synapses (Ctrl: n = 103 spines from nine independent experiments/neurons, KN62: n = 50 spines from five independent experiments/neurons). Error bars indicate ± SEM.

Figure 4. Phosphorylation of SynGAP Regulates Its Synaptic Localization

(A) Localization of SynGAP WT, phospho-deficient (2SA; S1108/1138A), and phospho-mimetic (2SD; S1108/1138D) before and after LTP stimulus. Note that 2SA failed to be dispersed upon LTP and cells expressing S2A showed a failure of spine enlargement. 2SD was not concentrated even in basal state, and spines were already enlarged. (B) Quantification of (A) (n = 5 independent experiments/neurons in each condition that contains 48 [WT], 59 [2SA], and 50 [2SD] spines in total, respectively) showing relative SynGAP enrichment and spine size change upon LTP for each SynGAP construct transfected. Two-way ANOVA followed by Tukey’s post hoc test was performed (for SynGAP enrichment [left panel], Phospho-mutation F(2, 24) = 20.82, p < 0.001; chemLTP F(1, 24) = 7.98, p < 0.001; Interaction F(2, 24) = 4.49, p < 0.001; for Spine size area [right panel], Phospho-mutation F(2, 24) = 45.23, p < 0.001; chemLTP F(1, 24) = 14.07, p < 0.001; Interaction F(2, 24) = 17.18, p < 0.001). Error bars indicate ± SEM. (C) Rapid phosphorylation at Ser1108 and 1138 upon LTP. (D) PSD fractionations from neurons with basal state or after LTP. Note that SynGAP was dispersed from PSD fraction and moved to triton-soluble synaptosomes (Syn/Tx). Phosphorylated SynGAP after LTP was mainly located in cytosolic fraction (S2) and Syn/Tx. (PNS, postnuclear supernatant; P2, membrane fraction; S2, cytosolic fraction; Syn, total synaptosomal fraction; Syn/Tx, triton soluble synaptosomal fraction; PSD, postsynaptic density fraction.) (E) HEK cells cotransfected with myc-tagged SynGAP and constitutive active CaMKII (T286D) were lysed and blotted with indicated antibodies. Only WT SynGAP was phosphorylated by active CaMKII. (F) Coimmunoprecipitation of PSD-95 and SynGAP from transfected HEK293 cells with or without active (T286D) or inactive (K42M) CaMKII constructs. Myc-PSD95 coprecipitates SynGAP (Lane4). Interaction was disrupted by active CaMKII T286D (Lane6) but not inactive CaMKII (Lane5). (G) Coimmunoprecipitation of PSD-95 and various SynGAP constructs expressed in HEK293 cells. Interaction was diminished in phospho-mimetic SynGAP 2SD (Lane 6) compared to WT (Lane 4) or phospho-deficient (Lane 5) constructs. (H) Rapid dissociation of SynGAP from PSD-95 upon LTP stimulus in neurons. During LTP, levels of phosphorylation at S1108 and 1138 were increased, and SynGAP was concurrently released from PSD-95 (Lanes 2–5). Inhibition of CaMKII by KN62 blocked this dissociation (Lane 6).

Figure 5. Phosphorylation of SynGAP Regulates Ras Activity during LTP

(A) Imaging of Raichu-Ras, a FRET-based sensor for cellular Ras activity with or without SynGAP knockdown as well as rescued by WT or phospho-mutants. Error bars indicate ± SEM. (A1) Control: Upon LTP stimulus, synaptic Ras activity was increased. (A2) Knockdown of SynGAP: (shRNA-SG#5) increased the basal Ras activity, thus occluding the Ras activity change upon LTP. (A3) Knockdown of SynGAP rescued with WT: shRNA-resistant SynGAP WT rescued this occlusion. (A4) Knockdown of SynGAP rescued with phospho-deficient SynGAP 2SA: shRNA-resistant SynGAP 2SA failed to rescue Ras activation upon LTP, likely because this mutant could not be dispersed, since it cannot be phosphorylated. (A5) Knockdown of SynGAP rescued with phospho-mimetic SynGAP 2SD: shRNA-resistant SynGAP 2SD failed to rescue, likely because 2SD could not localize to spines (N = 7 independent experiments/neurons respectively that contain 72 [A1], 81 [A2], 75 [A3], 76 [A4], and 95 [A5] spines in total, respectively). Two-way ANOVA followed by Tukey’s post hoc test was performed (shRNA⁺Rescue F(4, 90) = 76.64, p < 0.001; Time F(2, 90) = 55.79, p < 0.001; Interaction F(8, 90) = 10.07, p < 0.001). Scale bar, 2 μm. (B) Amount of GTP bound (active) Ras was quantified by pull-down assay using Raf-RBD (Ras effector domain) beads with or without electroporation of SynGAP knockdown constructs and shRNA-resistant SynGAP rescues (N = 6 independent experiments). Two-way ANOVA followed by Tukey’s post hoc test was performed (shRNA⁺Rescue F(4, 75) = 63.53, p < 0.001; Time F(2, 75) = 46.56, p < 0.001; Interaction F(8, 75) = 7.526, p < 0.001). Error bars indicate ± SEM.

Figure 6. Phosphorylation of SynGAP Regulates Spine Enlargement and GluA1 Trafficking during LTP

(A and B) (A) Spine enlargement and (B) GluA1 trafficking upon LTP with or without SynGAP ([A1] and [B1]) Control: (shRNA-Ctrl), ([A2] and [B2]) Knockdown of SynGAP: (shRNA-SG#5), and ([A3] and [B3]) Knockdown of SynGAP rescued with WT: shRNA-resistant SynGAP WT rescue. ([A4] and [B4]) Knockdown of SynGAP rescued with phospho-deficient SynGAP 2SA; ([A5] and [B5]) Knockdown of SynGAP rescued with phospho-mimetic SynGAP 2SD. N = 7 independent experiments/neurons that contain 70, 66, 68, 72, and 78 spines in total, respectively. Two-way ANOVA followed by Tukey’s post hoc test was performed (shRNA⁺Rescue F(4, 90) = 68.76 [A]/73.30 [B], p < 0.001 [A, B]; Time F(2, 90) = 22.54 [A]/22.75 [B], p < 0.001 [A, B]; Interaction F(8, 90) = 8.88 [A]/10.90 [B], p < 0.001 [A, B]). Scale bar, 2 μm. Error bars indicate ± SEM.

Figure 7. Phosphorylation of SynGAP Regulates Synaptic Strength during LTP

(A) Uncaging EPSC (uEPSC) changes upon LTP. Caged-glutamate was uncaged at each spine before or after chemLTP induction, and the changes were compared (n = 24, 20, 26, 20, and 20 spines from three independent experiments/neurons, respectively; shRNA⁺Rescue F(4, 105) = 14.49, p < 0.001). (A1–A5) Constructs were the same as Figures 5 and 6. Scale bar, 5 μm. (B) Location of SynGAP nonsense mutation found in human intellectual disability patients that is used in this study (S738X, L813RfsX22 [frame shift at L813 position that leads to premature stop codon after 22 amino acids], and Q893RfsX184 [frame shift at Q893 position that leads to premature stop codon after 184 amino acids]). Note that all mutants lack the phosphorylation sites by CaMKII. (C) Synaptic targeting of SynGAP mutants found in the patients of human intellectual disability patients. Scale bar, 2 μm. (D) Synaptic dispersions of SynGAP and spine enlargements were abolished in the mutants of human intellectual ability. (D1) Representative images of SynGAP localization and spine shapes during LTP. (D2) Quantification of SynGAP dispersion and spine enlargement during LTP. Two-way ANOVA followed by Tukey’s post hoc test was performed (n = 4 independent experiments/neurons that contain 39, 41, 45, and 42 spines, respectively. Mutation F(3, 24) = 22.92 [enrichment]/5.671 [size], p < 0.001; chemLTP F(1, 24) = 15.54[enrichment]/7.05[size], p < 0.001; Interaction F (3, 24) = 11.06[enrichment]/5.50 [size], p < 0.001). Scale bar, 1 μm.

Figure 8. CaMKII Phosphorylation of SynGAP Transmit Signals to Small G Protein

(A) Top panel: Artificially highly concentrated SynGAP reduced spine size and SEP-GluA1 at the basal state and blocked spine enlargement and insertion of SEP-GluA1 into spines during chemLTP. Dynamics of AMRAR trafficking (SEP-GluA1 channel) and spine enlargement (mCherry channel) during chemLTP were observed when endogenous SynGAP was replaced with WT^Res#5 GFP-SynGAP or with WT^Res#5:GL-NR2B^C GFP-SynGAP. Yellow arrows indicate spines with newly inserted SEP-GluA1 upon chemLTP, while green arrowheads indicate spines without newly inserted SEP-GluA1 upon chemLTP. Note that the population of the spines with green arrowheads was increased with WT^Res#5:GL-NR2B^C compared to WT ^Res#5 SynGAP replacement. Scale bar, 2 μm. Bottom panel: Quantification of the relative ratio of SEP-GluA1 per spine before/after chemLTP with WT ^Res#5 or WT^Res#5:GL-NR2B^C SynGAP replacement. Error bars indicate ± SEM. (B) Schematic diagram of relationships between CaMKII activity and small G protein activation that leads to cellular changes upon LTP. (C) Schematic model of the cellular events that link CaMKII activity, SynGAP dispersion, and small G protein activation.