Effect of different exocyst subunits on synaptic transmission. (A–E) Insets, sample trace of evoked AMPA-receptor-mediated synaptic responses recorded at −60 mV, from uninfected and infected cells. Scale bars, 20 pA and 40 ms. Left: average AMPAR-mediated current amplitude (i.e. the peak of the response recorded at −60 mV) from infected (inf) neurons expressing Sec8-Ct (A), Sec8-wt (B), Sec8-Ct-Δ4 (C), Exo70-N (D) or Exo70-wt (E), and control neighboring cells not expressing the recombinant protein (uninf). n represents the number of cell pairs. Middle: average NMDAR-mediated current amplitude (recorded at +40 mV at a latency when AMPA responses are fully decayed, 60 ms) from uninfected and infected cells (n also represents the number of cell pairs). Right: average GABA_A receptor-mediated current amplitude measured at 0 mV (A) or average AMPA/NMDA ratios for uninfected and infected cells (B–E).

Co-immunoprecipitation of exocyst subunits and AMPA receptors from hippocampal synaptosomes. Exocyst subunits Exo70 and Sec8 (A), AMPA receptor subunits GluR2/3 (GR2/3) and NMDA receptor subunit NR1 (B) were immunoprecipitated from Triton-solubilized synaptosomal preparation of hippocampus. IP indicates antibodies used for immunoprecipitation. Non-immune (N.I.) antibodies were used as control. Total solubilized protein (10% of the input) and the immunoprecipitated fractions were analyzed by immunoblotting using the indicated antibodies.

Role of Exo70 in local AMPAR trafficking at spines. (A) Representative confocal image of a spine and the adjacent dendritic shaft from a neuron transfected with GluR1-GFP and tCaMKII (left) or with GluR1-GFP, tCaMKII and Exo70-N-RFP (right), as indicated. GFP fluorescence signal (green) represents total GluR1 receptor distribution. Surface GluR1-GFP receptor (cyan) was assayed by immunostaining with an anti-GFP antibody coupled to Cy5 under nonpermeabilized conditions. Scale bar: 2 μm. Lower panels: representative line plot profiles of GFP and Cy5 fluorescence intensities across dendrite–spine pairs. Total (GFP) and surface (Cy5) receptor accumulation are calculated from the corresponding peaks of fluorescence intensity at spines and adjacent dendritic shafts. (B) Surface analysis (background-subtracted Cy5 values) from pairs of spines and dendrites in GluR1 plus tCaMKII transfected neurons (first panel; n=59 spine/dendrite pairs from five different cells) or in GluR1 plus tCaMKII and Exo70-N cotransfected neurons (second panel; n=40 spine/dendrite pairs from five different cells). Surface values are normalized by the mean dendrite value. Cumulative probabilities (third panel) of surface expression at spine over dendrite show significant differences between surface distributions of GluR1 plus tCaMKII versus GluR1 plus tCaMKII and Exo70-N transfected neurons. (C) Total receptor amount (GFP signal) in spines and dendrites are calculated in GluR1 plus tCaMKII transfected neurons and in neurons cotransfected with GluR1 plus tCaMKII and Exo70-N, as indicated. GFP fluorescence is normalized to the mean dendrite value. The cumulative probability (right) indicates that Exo70-N leads to the accumulation of total receptor in spines as compared to dendrites. (D) Surface ratio from the same pairs of spines and dendrites. Surface ratios were calculated by dividing the corresponding background-subtracted Cy5 and GFP values. Surface ratios are normalized by the mean dendrite value. Cumulative probabilities (right) of the spine/dendrite ratios show a significant difference between the surface ratio distribution of GluR1 plus tCaMKII versus GluR1 plus tCaMKII and Exo70-N transfected neurons.

Opposite effects of Exo70 and Sec8 dominant negatives in the association of AMPA receptors with the postsynaptic density. (A) Total proteins in synaptosomes (8% of the total fraction) and PSD-associated proteins were analyzed by immunoblotting using the indicated antibodies (AMPARs were detected with a mix of GluR1 and GluR2/3 antibodies). The CA1 area was dissected from 30 to 45 control (unin.) hippocampal slices or slices infected with Exo70-N (inf.). Therefore, an ‘n' of 1 in this experiment represents the average of 30–45 slices. Protein extracts were processed to isolate the synaptosomal and PSD fractions (see Supplementary data). (B) Quantification of multiple independent experiments as the one shown in (A). Relative PSD enrichment was calculated as the intensity of the PSD band in the infected extracts over that of the uninfected ones, after normalization by their relative band intensities at the homogenates (input). Inset shows that infection did not produce a systematic effect on protein abundance at the inputs. (C) Similar to (A), comparing uninfected and infected slices expressing Sec8-Ct (same construct as in Figure 1). Total proteins in synaptosomes (3% of the total fraction) and PSD-associated proteins were analyzed by immunoblotting using the indicated antibodies. NMDARs were detected with a combination of NR1 and NR2A/B antibodies. (D). Quantification of PSD enrichment for the different proteins in the presence or absence of Sec8-Ct was calculated as described for (B).

Model for the exocyst-mediated targeting and delivery of AMPA receptors at the postsynaptic membrane. (Left) Exocytic vesicles containing AMPA receptors reach dendritic spines, bud from intracellular compartments within the spine and are directed towards the PSD by interactions between Sec8 and synaptic scaffolding molecules, such as PSD95. (Upper right): Incoming vesicles dock at the PSD forming a transient complex with it. (Lower right) This complex is resolved by membrane fusion with the postsynaptic membrane. This process is mediated by Exo70. See further explanation in the main text. TGN: trans-Golgi network; RE: recycling endosome; EV: exocytic vesicle; DV: docked vesicle.

Interference of the constitutive synaptic cycling of AMPA receptors with a Sec8 C-t peptide. (A) Time course of AMPAR-mediated responses recorded from CA1 neurons during whole-cell pipette infusion of Sec8 C-t, NR2B C-t or Sec8-mut C-t peptides, as indicated. Responses are normalized to a 2-min baseline from the beginning of the recording. The statistical significance of the run-down of AMPA transmission was calculated from the last 5 min of the recordings. (B) Time course of the pipette access resistance during the recordings shown in (A). There was no systematic change in access resistance with any of the peptides.

Exo70 is not involved in the dendritic trafficking of newly synthesized AMPA receptors in hippocampal slices. Hippocampal CA1 neurons were transfected with GluR2-GFP alone or cotransfected with GluR2-GFP plus Exo70-N-RFP, Sec8-Ct-RFP or NR2B-Ct-RFP (these constructs are equivalent to their GFP counterparts, only with RFP in the place of GFP; the NR2B construct includes the last 16 amino acids of the NR2B C-terminus). (A) Representative examples of neurons expressing GluR2-GFP alone or coexpressing GluR2-GFP together with Exo70-N-RFP or with Sec8-Ct-RFP, as indicated. Scale bar: 20 μm. The efficiency of GluR2 transport along dendrites was quantified from the intensity of GFP fluorescence at the main apical dendrite 170–200 μm away from the cell soma. Fluorescence intensity is normalized to the peak value at the soma after background subtraction (see arrows on left panel and representative scheme in Supplementary Figure 4A). (B) GluR2-GFP presence at distal apical dendrites (170–200 μm away from the cell soma) in the presence of Exo70-N, Sec8-Ct or NR2B-Ct, as indicated. Data are presented as mean±s.e.m. n represents the number of cells.

Role of Exo70 and Sec8 in the constitutive AMPAR trafficking at spines. (A) Similar to Figure 5B, surface analysis from pairs of spines and dendrites in GluR2 (first panel; n=21 spine/dendrite pairs from four different cells), GluR2 plus Exo70-N (second panel; n=32 spine/dendrite pairs from five different cells), GluR2 plus Sec8-Ct (third panel; n=58 spine/dendrite pairs from five different cells) or GluR2 plus NR2B-Ct cotransfected neurons (fourth panel; n=62 spine/dendrite pairs from five different cells). (B) Similar to Figure 5C, total receptor amount from the same pairs of spines and dendrites in GluR2 or in GluR2 plus Exo70-N, Sec8-Ct or NR2B-Ct cotransfected neurons. (C) Similar to Figure 5D, surface ratio of the same pairs of spines and dendrites in GluR2 or in GluR2 plus Exo70-N, Sec8-Ct or NR2B-Ct cotransfected neurons.

Electron microscopic analysis of the role of the exocyst in AMPA receptor synaptic insertion. (A) Representative micrographs of AMPA receptor immunogold labeling in control slices (Uninf), slices infected with Exo70 dominant negative (Exo70-N) or with Sec8 dominant negative (Sec8-Ct; same construct as in Figure 1). Immunogold particles at the PSD are indicated with arrows. Note accumulation of AMPA receptors within the PSD in Exo70-N-expressing neurons. The opposite effect was observed in Sec8-Ct-infected slices. (B) Quantification of AMPA receptor presence at the PSD from electron micrographs as the ones shown in (A). Percentage of PSD labeling was calculated as the number of gold particles at the PSD versus the total number of gold particles in each synaptic terminal (only gold particles within 800 nm of the synaptic cleft were included for quantification). n represents the number of synapses analyzed in each condition. (C) Quantification of AMPA receptor local distribution with respect to the synaptic cleft. The distance to the synaptic cleft was calculated for each immunogold particle from uninfected slices (black line), slices expressing Exo70-N (red line) or Sec8-Ct (blue line), and plotted as cumulative distributions. In order to assess local synaptic distribution of AMPA receptors, only immunogold particles within 60 nm of the synaptic cleft were included in this quantification (the average thickness of the PSD from our micrographs was 18 nm-thick line on the x-axis). n represents the number of gold particles analyzed in each condition. (D) Quantification of AMPA receptor lateral distribution within the PSD and on extrasynaptic membranes from uninfected slices (white columns) and Exo70-N-expressing slices (red columns). The lateral distance to the closest edge of the PSD was calculated for each immunogold particle and divided by the total length of the PSD (the average length of the PSD from our micrographs was 290 nm). Distances within the PSD were computed as negative values (relative lateral distances within the PSD cannot exceed −0.5, since the position of the immunogold particle is always calculated with respect to the closest PSD edge).