Synaptic Consolidation Normalizes AMPAR Quantal Size following MAGUK Loss

Abstract

The mechanisms controlling synapse growth and maintenance are of critical importance for learning and memory. The MAGUK family of synaptic scaffolding proteins is abundantly expressed at glutamatergic central synapses, but their importance in controlling the synaptic content of glutamate receptors is poorly understood. Here, we use a chained RNAi-mediated knockdown approach to simultaneously remove PSD-93, PSD-95, and SAP102, the MAGUKs previously shown to be responsible for synaptic localization of glutamate receptors. We find that MAGUKs are specifically responsible for creating functional synapses after initial spine formation by filling functionally silent spines with glutamate receptors. Removal of the MAGUKs causes a transient reduction in AMPA receptor quantal size followed by synaptic consolidation resulting in a normalization of quantal size at the few remaining functional synapses. Consolidation requires signaling through L-type calcium channels, CaM kinase kinase, and the GluA2 AMPA receptor subunit, akin to a homeostatic process.

Figure 1. Pan-MAGUK Knockdown Reduces Synaptic AMPAR and NMDAR-Mediated Currents

(A) CAG hybrid promoter drives EGFP with a synthetic 3′ UTR containing miRNA hairpins against PSD-93, PSD-95, and SAP102. (B) Infection of dissociated hippocampal neurons with lentivirus expressing the MAGUK miRNA construct results in reductions in the amount of PSD-95, PSD-93, and SAP102 protein without any change in the loading control actin. (C) Scatter plots showing reductions in AMPAR and NMDAR EPSCs in MAGUK miRNA-transfected neurons compared to untransfected controls (AMPAR, 20.84% ± 4.03% control, p < 0.005, n = 41; NMDAR, 33.59% ± 8.22% control, p < 0.005, n = 41). Scatter plots of EPSCs show single pairs (open circles). Bar graphs show mean ratio ± SEM. AMPAR scale bars represent 25 ms, 25 pA; NMDAR scale bars represent 100 ms, 25 pA. (D) No change in paired-pulse ratio (PPR), defined as second EPSC over first EPSC (Ctrl 1.75 ± 0.07, Expt 1.59 ± 0.06; p > 0.05, n = 17). Scale bars represent 25 pA, 50 ms. See also Figure S1.

Figure 2. All MAGUKs Play Roles in Baseline Glutamate Receptor Localization

(A) PSD-93 knockdown causes decrease of AMPAR-mediated current (49.23% ± 8.34% control, p < 0.005, n = 17) and NMDAR-mediated current (71.34% ± 11.44% control, p < 0.05, n = 17). (B) PSD-95 knockdown causes decrease of AMPAR-mediated current (45.38% ± 7.457% control, p < 0.005, n = 34) and NMDAR-mediated current (75.38% ± 11.37% control, p < 0.05, n = 31). (C) SAP102 knockdown causes decrease of AMPAR-mediated current (54.83% ± 12.45% control, p < 0.05, n = 13) and NMDAR-mediated current (63.74% ± 20.07% control, p < 0.01, n = 11). (D) Summary graphs of mean ± SEM. EPSC amplitudes, expressed as a percentage of control EPSC values. Open circles represent amplitudes for single pairs. Scale bars represent 25 pA, 50 ms.

Figure 3. Dependence of SAP102 Phenotype on Method of RNAi Delivery

(A) SAP102 knockdown by viral transduction does not decrease AMPAR currents (88.95% ± 8.87% control, p > 0.05, n = 16), in contrast to biolistic knockdown (54.83% ± 12.45% control) (biolistics versus virus p < 0.05, n = 13 biolistics; n = 16 virus). SAP102 knockdown by viral transduction slightly decreases NMDAR currents (83.03% ± 11.04% control, p < 0.05, n = 15). Biolistic knockdown (63.74% ± 20.07% control) results in a greater decrease of NMDAR currents (biolistics versus virus p < 0.01, n = 11 biolistics; n = 15 virus). (B) Biolistic knockdown of SAP102 in wild-type mice causes a decrease of AMPAR-mediated current (52.03% ± 7.06% control, p < 0.01, n = 13). Biolistic knockdown of SAP102 in wild-type mice causes a decrease of NMDAR-mediated current (65.58% ± 9.49% control, p < 0.05, n = 13). (C) Biolistic knockdown of SAP102 in SAP102 knockout mice results in no change in AMPAR-mediated currents (102.70% ± 20.06% control, p > 0.05, n = 15) or NMDAR-mediated currents (90.16% ± 15.22% control, p > 0.05, n = 14). (D) Knockdown of SAP102 in wild-type mice causes a statistically significant decrease in AMPAR-mediated current compared to knockdown in SAP102 knockout mice (wild-type versus KO p < 0.05, n = 13 WT, n = 15 KO). Knockdown of SAP102 in wild-type mice causes a statistically significant decrease in NMDAR current compared to controls, but not compared to knockdown in SAP102 knockout mice, although there is a trend toward significance (wild-type versus KO p > 0.05, n = 13 WT, n = 14 KO). Open circles represent amplitudes for single pairs. Scale bars represent 25 pA, 50 ms.

Figure 4. MAGUK Knockdown Causes Loss of Functional Glutamatergic Synapses

(A) Representative sample traces of asynchronous EPSCs (aEPSCs) recorded in the presence of Sr²⁺ in neurons expressing MAGUK miRNA or control neurons. 50 ms following stimulation (gray box) was excluded from analysis. Scale bars represent 50 ms, 15 pA. (B) Representative average aEPSC traces showing no change in average amplitude. Black trace is control; green trace is experimental. Scale bars represent 20 ms, 4 pA. (C) aEPSC frequency in neurons expressing MAGUK miRNA. Plot shows single pairs (open circles) and mean ± SEM (filled circles). aEPSC frequency is significantly reduced (p < 0.05) in neurons expressing MAGUK miRNA. (D) aEPSC amplitude in neurons expressing MAGUK miRNA. Plot as in (C). There is no change in amplitude between control and MAGUK miRNA neurons (p = 0.15, n = 13). (E and F) Cumulative distribution plots of aEPSC frequency and amplitude. Control shown in black, experimental in green. Cumulative distribution functions show no irregularities. (G and H) Coefficient of variation analysis of simultaneously recorded pairs of control/miRNA neurons. CV⁻² graphed against ratio of mean amplitude within each pair. Results along the horizontal y = 1 line are consistent with change in quantal size (q), results along gray dashed identity (45°) line are consistent with change in quantal content (N × P_r). Analysis of AMPAR and NMDAR responses suggests decrease is due to reduction in quantal content. Small solid and dashed lines indicate linear regression line and 95% confidence intervals, respectively. (I) Summary of coefficient of variation analysis. Both AMPAR and NMDAR average fall on the identity line (p < 0.05 versus horizontal line), while average response after D-APV falls on horizontal y = 1 line (p > 0.05). See also Figure S2.

Figure 5. L-Type Calcium Channels Are Required for Synaptic Consolidation

(A) MAGUK miRNA transfection in the presence of 20 μM nifedipine causes significant reductions in both AMPAR and NMDAR EPSCs compared to neighboring untransfected neurons (AMPAR 20.17% ± 3.24%; NMDAR 21.35% ± 4.17%; n = 10 and p < 0.05 for both). (B and C) Summary data showing no additional change in AMPAR or NMDAR EPSCs in MAGUK miNRA-expressing neurons with nifedipine compared to MAGUK miRNA alone (p > 0.05 for both). Dashed line and shaded area show mean ± SEM of normalized synaptic responses for MAGUK miRNA without nifedipine. (D) No change in paired-pulse ratio (PPR), defined as second EPSC over first EPSC (Ctrl 1.63 ± 0.11, Expt 1.72 ± 0.17; p > 0.05, n = 9). (E and F) Coefficient of variation analysis of simultaneously recorded pairs of control/miRNA neurons. CV ⁻² graphed against ratio of mean amplitude within each pair. Results along the horizontal line are consistent with change in quantal size (q), results along identity (45°) line are consistent with change in quantal content (N × P_r). Decrease in AMPAR EPSC is due to reduction in quantal size. Decrease in NMDAR EPSC is due to reduction in quantal content. Small solid and dashed lines indicate linear regression line and 95% confidence intervals, respectively. (G) Summary of coefficient of variation analysis. AMPAR average falls near the horizontal line (p < 0.05 compared to horizontal line) and NMDAR average falls on the identity line. Data from MAGUK miRNA without nifedipine incubation (Figure 4G) are re-plotted to aid comparison. (H) Coefficient of variation analysis of simultaneously recorded pairs of control/PSD-95 miRNA neurons (black circle) shows the reduction in AMPAR EPSC is due to reductions in quantal content. Reduction in AMPAR EPSC following PSD-95 knockdown and nifedipine incubation (green circle) is due to reductions in quantal size. (I and J) Coefficient of variation analysis of simultaneously recorded pairs of control/miRNA neurons either 24 hr (I) or 96 hr (J) after nifedipine washout. Twenty-four hours after washout, the decrease in AMPAR EPSC is due to reductions in quantal content and quantal size. Ninety-six hours after nifedipine washout, the decrease in AMPAR EPSC is due to pure reduction in quantal content. (K) Summary graph showing slope of CV dataset regression line versus hours post-nifedipine washout. As a control, nifedipine was not washed out of some slices (dashed line and black square). See also Figure S3.

Figure 6. CaM Kinase Kinase Is Required for Synaptic Consolidation

(A) Coefficient of variation analysis of simultaneously recorded pairs of control/miRNA neurons in slices incubated with 3 μM STO-609. The decrease in AMPAR EPSCs is due to reduction in quantal size and quantal content. Small solid and dashed lines indicate linear regression line and 95% confidence intervals, respectively. (B) Summary of coefficient of variation analysis. Data from MAGUK miRNA alone (Figure 4G), and plus nifedipine incubation (Figure 5E), are re-plotted to aid comparison. (C) Coefficient of variation analysis of simultaneously recorded pairs of control/miRNA + CaMKK L233F neurons in slices incubated in STO-609. The decrease in AMPAR EPSC is due to reduction in quantal content. (D) Summary of coefficient of variation analysis. Data from MAGUK miRNA alone (Figure 4G), and plus nifedipine incubation (Figure 5E), are re-plotted to aid comparison. (E) MAGUK miRNA transfected in the presence of 3 μM STO-609 causes reductions in both AMPAR and NMDAR EPSC (AMPAR 31.93% ± 6.66%, n = 12; NMDAR 46.98.1% ± 13.14%, n = 12; p < 0.05 for both). (F) MAGUK miRNA co-transfected with STO-609 insensitive CaMKK L233F in the presence of STO-609 causes reductions in both AMPAR and NMDAR EPSC (AMPAR 22.31% ± 4.32%, n = 9; NMDAR 22.86% ± 3.14%, n = 6; p < 0.05 for both). (G) Summary data showing no additional change in AMPAR EPSC reduction due to incubation in STO-609 and co-expression of CaMKK L233F (p > 0.05 for all).

Figure 7. GluA2 AMPAR Subunit Is Required for Synaptic Consolidation

(A and B) Coefficient of variation analysis of simultaneously recorded pairs of control/miRNA neurons in GluA2 ^−/− slices. The decrease in AMPAR EPSC is due to reduction in quantal size and quantal content (p < 0.01 compared to horizontal line). The decrease in NMDAR EPSC is due to reduction in quantal content. (C) Summary of coefficient of variation analysis. Data from MAGUK miRNA (Figure 4G) are re-plotted to aid comparison. (D and E) Sr²⁺-evoked aEPSC amplitude and frequency are reduced in neurons from GluA2 ^−/− slices shot with MAGUK miRNA compared to neighboring GluA2 ^−/− neurons (p < 0.05). (F) Model of synaptic consolidation. MAGUK loss initially causes reductions in number of AMPARs present at individual synapses. Over time, compensatory processes normalize the number of AMPARs present at the few remaining synapses, at the cost of complete loss of other synapses. The glutamate receptors lost from synapses are re-distributed on the plasma membrane at extrasynaptic sites and have been omitted for clarity. The total number of surface-localized receptors remains unchanged. See also Figure S4.