Activity-dependent regulation of synaptic AMPA receptor composition and abundance by beta3 integrins

Abstract

At synapses, cell adhesion molecules (CAMs) provide the molecular framework for coordinating signaling events across the synaptic cleft. Among synaptic CAMs, the integrins, receptors for extracellular matrix proteins and counterreceptors on adjacent cells, are implicated in synapse maturation and plasticity and memory formation. However, little is known about the molecular mechanisms of integrin action at central synapses. Here, we report that postsynaptic beta3 integrins control synaptic strength by regulating AMPA receptors (AMPARs) in a subunit-specific manner. Pharmacological perturbation targeting beta3 integrins promotes endocytosis of GluR2-containing AMPARs via Rap1 signaling, and expression of beta3 integrins produces robust changes in the abundance and composition of synaptic AMPARs without affecting dendritic spine structure. Importantly, homeostatic synaptic scaling induced by activity deprivation elevates surface expression of beta3 integrins, and in turn, beta3 integrins are required for synaptic scaling. Our findings demonstrate a key role for integrins in the feedback regulation of excitatory synaptic strength.

Fig 1. RGD Peptides Reduce mEPSC Amplitude

(A) Overlay of mEPSC traces recorded from a hippocampal pyramidal neuron of rat primary cultures over a 5 min control period (−), the last 5 of 10 min bath application of echistatin (100 nM), and the last 5 of 10 min wash. Grey traces are aligned original mEPSCs and black traces represent the population average. (B) Cumulative distribution plot of mEPSC amplitudes from the experiment shown in A (p<0.01, control vs. echistatin). (C) Average mEPSC amplitude during control and echistatin application shown pairwise for each cell (p<0.05 for each pair). (D) The reduction in mEPSC amplitude is maximal after 5 min bath application of echistatin (n=13). Each data point represents the mean amplitude over 100 s intervals normalized to baseline (**p<0.01 relative to baseline). (E) Summary of the peptide effects on mEPSC amplitude. Mean amplitudes from the last 5 of 10 min application are plotted relative to 5 min baseline for untreated control cells (n=4) and echistatin (n=6); for the control GRGESP peptide (RGE, 200 μM; n=5) and GRGDSP (RGD, 200 μM; n=5) (*p<0.04).

Fig 2. Echistatin-Mediated Reduction in mEPSC Amplitude Requires Elevation of Intracellular Ca²⁺

(A) mEPSC population averages before (−) and after echistatin application (100 nM), and (B) cumulative distributions of mEPSC amplitudes from representative neurons where echistatin was applied at least 5 min after treatments with latrunculin A (LatA, top) or ifenprodil (bottom), or applied in the presence of BAPTA in the intracellular solution (BAPTA_post, middle). (C) Summary of the echistatin effects on mEPSC amplitude in the presence of blockers or Ca²⁺ chelators: LatA (20 μM; *p=0.04 relative to baseline, n=4), D-APV (50 μM; n=6), ifenprodil (3 μM; n=6), BAPTA-AM (100 μM, n=6), BAPTA_post (37.5–50 mM, n=6). Echistatin does not reduce mEPSC amplitude in the presence of Ca²⁺ chelators or NMDAR blockers. (D) Traces of evoked autaptic AMPAR and NMDAR synaptic currents from a representative experiment before (baseline) and after echistatin application (100 nM). The unclamped Na⁺ spikes have been blanked for clarity. (E) Time courses of the echistatin effect on peak AMPAR current (left, n=5, p=0.03) and on NMDAR charge transfer (right, n=6, p=0.53; open triangles: responses used to evaluate statistical significance). Echistatin selectively reduces the AMPAR current.

Fig 3. Integrins Regulate GluR2 Endocytosis

(A) Neurons were loaded with D15 (left) or a scrambled S15 peptide (right) for at least 10 min prior to the baseline period. mEPSC population averages are from representative neurons during 5 min baseline (−) and the last 5 of 10 min application of echistatin (100 nM). (B) Time course of mEPSC amplitudes from neurons loaded with GTPγS (0.5 mM; n=6), or D15 or S15 peptides (0.25–2 mM; D15, n=5; D15+echistatin, n=6; S15+echistatin, n=6). Each data point represents the average mEPSC amplitude over 100 s interval normalized to the first 100 s (*p=0.03, **p<0.01). Echistatin does not reduce mEPSC amplitudes under conditions that block endocytosis. (C) Internalization assay for endogenous GluR1 (left) and GluR2 (right) subunits. Surface (top) vs. internalized fractions (bottom) from representative 10 min mock-(untreated) and echistatin-treated neurons (300 nM) are shown. Scale bars are 30 and 10 μm. (D) Summary of GluR1 (n=21 images each from 4 independent cultures) and GluR2 (n=36 images each from 5 independent cultures) internalization assay. Internalized fraction indicates: (mean intensity_surface)/(mean intensity_surface + mean intensity_internalized). ***p=0.00002. Echistatin specifically increases GluR2 endocytosis.

Fig 4. Echistatin-Mediated Activation of Rap1 Is Required for mEPSC Amplitude Decrease

(A) Pulldown with GST-RalGDS-RBD from mock- (Untreat) and echistatin-treated (Echi, 300 nM, 10 min at 37°C) culture lysates (500 μg each) reveals activation of endogenous Rap1 by echistatin (top right). Comparable amounts of Rap1 were detected from the respective lysates (20 μg per lane, bottom right). Top left, lysates were treated with GDP (1 mM; negative control) or GTPγS (0.1 mM; positive control). Exposure times are different for the three panels shown. Similar results were obtained in four independent experiments. (B) mEPSC population averages from representative neurons expressing Rap1DN (top) or Rap2DN (bottom), before (−) and after echistatin application (100 nM). Basal mEPSC amplitude is not significantly different between Rap1DN and Rap2DN transfected neurons (30.6 ± 3.1 pA vs. 27.5 ± 2.2 pA, respectively; n=6–8, p=0.44). (C) Summary of experiments as in B shown pairwise for individual cells and as bar graph for the average cell population (**p=0.002). Echistatin does not reduce mEPSC amplitudes under conditions that prevent Rap1 activation.

Fig 5. Postsynaptic Expression of β3 Integrins Alters mEPSC Amplitude

(A) Sample mEPSCs and (B) histograms of mEPSC amplitudes from representative neurons overexpressing WTβ3 and mRFP (WTβ3, top), mRFP (control, middle), or CTβ3 and mRFP (CTβ3, bottom). (C) Summary of group data from experiments as in A. WTβ3 increases and CTβ3 decreases mEPSC amplitudes. (D) Images of dendrites from pyramidal neurons overexpressing WTβ3 and GFP (WTβ3, top), GFP (control, middle), or CTβ3 and GFP (CTβ3, bottom). Synapsin (red) was used as a presynaptic marker. Scale bar is 4 μm. (E) Summary of spine measurements. None of the differences were statistically significant (n=19 cells for control and WTβ3; n=20 for CTβ3 and CTβ1).

Fig 6. Postsynaptic β3 Integrins Affect AMPAR Subunit Composition

(A) mEPSC population averages from representative neurons overexpressing WTβ3 and mRFP (WTβ3, left), mRFP (Control, middle), or CTβ3 and mRFP (CTβ3, right), during 5 min baseline (−) and the last 5 of 10 min application of the GluR2-lacking AMPAR blocker PhTx (10 μM). (B) Representative images of GluR1 (red) and GluR2 (green) surface labeling for neurons transfected with WTβ3 and GFP (WTβ3, left), GFP (Control, middle), or CTβ3 and GFP (CTβ3, right). GluR1 and GluR2 were stained with Cy3- and Cy5-conjugated secondary antibodies, respectively. GFP fluorescence is not shown. Scale bars are 30 and 5 μm. (C) Agonist-evoked AMPAR I-V relationships from untransfected neurons (middle), and neurons overexpressing WTβ3 (left) or CTβ3 (right). Recordings were performed in outside-out somatic patches in response to a voltage-ramp from +60 to −90 mV delivered over 3 s. Currents were elicited by bath application of kainate (100 μM) and blocked by SYM 2206 (Sym; 25–50 μM). (D) Summary of the effects of PhTx on mEPSC amplitudes. The reduction of mEPSC amplitudes is shown relative to baseline (n=6 for each condition, **p=0.01 relative to baseline). Only mEPSCs recorded from CTβ3-expressing neurons are reduced by PhTx. (E) Summary of the GluR1 and GluR2 surface levels for experiments as in (B). Fluorescence intensity was quantified on the full dendritic arbor (n=27, 23, and 26 images for WTβ3, control, and CTβ3, respectively; *p=0.02 relative to control). (F) Summary of I-V rectification indices (−I₋₆₀/I₊40) from recordings as in (C) (WTβ3, n=5; control, n=8; CTβ3, n=6; *p = 0.04 relative to control). WTβ3 expression decreases and CTβ3 expression increases the rectification of agonist-evoked AMPAR currents. (G) Time course of mEPSC amplitudes from untransfected neurons (Control), and neurons overexpressing WTβ3 or CTβ3, using intracellular solutions supplemented with pep2m or pep4c peptides (150 μM; Control+pep2m, n=4; CTβ3+pep2m, n=6; WTβ3+pep2m, n=6; WTβ3+pep4c, n=4). Each data point represents the average mEPSC amplitude over 100 s interval normalized to the first 100 s (*p=0.02 and **p<0.007). Only mEPSCs recorded from WTβ3 neurons are decreased by pep2m.

Fig 7. Postsynaptic β3 Integrins Mediate the Echistatin-Induced Reduction in mEPSC Amplitude

(A) Left, representative images of GluR1 (red) and GluR2 (green) surface labeling for hippocampal pyramidal neurons from β3^−/− and β3^+/− mouse primary cultures. Scale bar is 30 μm. Right, summary of the surface levels of GluR1 and GluR2 (n=29, 24, and 16 images for β3^−/−, β3^+/−, and β3^+/+, respectively). (B) Left, sample traces of mEPSC recordings from hippocampal pyramidal neurons of β3^−/− (top) and β3^+/− (bottom) mouse primary cultures. Right, summary of group data from experiments as on the left but also including recordings from β3^+/+ mouse cultures (average amplitude: −30.4.6±3.8 pA for β3^−/− [n=21], −28.7±1.8 for β3^+/− [n=16], and −28.4±2.3 pA for β3^+/+ [n=13]). None of the differences were statistically significant. (C) Left, mEPSC population averages of a hippocampal pyramidal neuron from β3^−/−mouse primary cultures before (left trace) and after (right trace) echistatin application (100 nM). Middle, cumulative distribution of mEPSC amplitudes for the experiment shown on the left. Right, summary for individual cells shown pairwise and for the average cell population shown as bar graph. (D) As in C but for hippocampal pyramidal neurons from β3^+/− mouse primary cultures. **p=0.009. (E) As in C but for β3^−/− hippocampal pyramidal neurons transfected with WT β3. **p=0.01.

Fig 8. Postsynaptic β3 Integrins are Required for Synaptic Scaling

(A) Surface labeling of endogenous β3 (top) and β1 (bottom) integrins for untreated (middle) and bicuculline- (20 μM, 48 h; left) or TTX-treated (1 μM, 48 h; right) hippocampal cultures. Scale bars are 30 and 10 μm. (B) Summary of the effects of 48 h bicuculline or TTX treatments on the surface expression of GluR2 and β1 and β3 integrins (bicuculline, β3, n=28 images; TTX, β3, n=65; bicuculline, β1, n=30; TTX, β1, n=30; bicuculline, GluR2, n=16; TTX, GluR2, n=18; *p=0.03, **p=0.006, ***p<0.0002 relative to untreated controls). Chronic changes in neuronal network activity alter surface expression of β3 integrins and GluR2 but not β1 integrins. (C) β3 integrin surface expression in response to TTX treatments of different durations (6 h, n=25; 12 h, n=24; 24 h, n=26; 48 h, n=30; *p=0.05, **p=0.004, ***p<0.00001 relative to untreated controls). (D) Western blot of protein extracts (20 μg per lane) from untreated (−) and TTX-treated (1 μM, 48 h) sister cortical cultures reveals no changes in the total amount of β3 integrins. The same membrane was re-probed with an anti-actin antibody. Bar graphs indicate quantification of three independent experiments. (E) Surface labeling of β3 integrins for untreated (top) and TNFα-treated (0.1 μg/μl, 20 min; bottom) hippocampal cultures. Surface β3 levels are higher in TNFα-treated cultures. Scale bar is 10 μm. (F) Summary of the increase in β3 and GluR2 surface expression induced by TNFα (n=27 each for β3 and n=39 each for GluR2; **p=0.002, ***p<0.0004 relative to untreated controls). (G) Summary of mEPSC amplitudes from hippocampal pyramidal neurons of β3^−/− and β3^+/− mouse primary cultures incubated with (red) or without (black) TTX (1 μM, 48 h; β3^−/−, n=21; β3^−/−+TTX, n=21; β3^+/−, n=16; β3^+/−+TTX, n=18; p=0.008 between β3^+/− and β3^+/− +TTX). Insets, mEPSC population averages from representative experiments. Scale bars are 4 ms and 10 pA. (H) As in G but for CA1 pyramidal neurons of mouse hippocampal slice cultures (β3^−/−, n=18; β3^−/−+TTX, n=19; β3^+/−, n=18; β3^+/−+TTX, n=18; p=0.000003 between β3^+/− and β3^+/−+TTX). Scale bars for insets are 8 ms and 5 pA.