Dynamin-dependent endocytosis of ionotropic glutamate receptors

Abstract

Little is known about the mechanisms that regulate the number of ionotropic glutamate receptors present at excitatory synapses. Herein, we show that GluR1-containing alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors (AMPARs) are removed from the postsynaptic plasma membrane of cultured hippocampal neurons by rapid, ligand-induced endocytosis. Although endocytosis of AMPARs can be induced by high concentrations of AMPA without concomitant activation of N-methyl-D-aspartate (NMDA) receptors (NMDARs), NMDAR activation is required for detectable endocytosis induced by synaptically released glutamate. Activated AMPARs colocalize with AP2, a marker of endocytic coated pits, and endocytosis of AMPARs is blocked by biochemical inhibition of clathrin-coated pit function or overexpression of a dominant-negative mutant form of dynamin. These results establish that ionotropic receptors are regulated by dynamin-dependent endocytosis and suggest an important role of endocytic membrane trafficking in the postsynaptic modulation of neurotransmission.

Figure 1

AMPARs are internalized after exposure to AMPA. (A) Living cells were labeled for surface AMPARs before treatment. After agonist application, surface receptors were detected with secondary antibodies on nonpermeabilized cells. Cells were then permeabilized, and internalized receptors were detected by using a secondary antibody with a different fluorescent conjugate. In an untreated cell (Left), AMPARs were primarily at the cell surface (green) with minimal internalized receptor immunoreactivity (red). Exposure to 100 μM AMPA for 15 min (Right) caused a dramatic increase in the amount of internalized AMPARs (red) along with a concomitant reduction in the amount of surface AMPAR staining (green). (B) Antibody-bound AMPARs internalized from the surface during agonist treatment are visualized exclusively by acid stripping antibodies from remaining surface AMPARs. In untreated, unstripped cells (Left), surface AMPARs were visualized in numerous puncta. After acid stripping of untreated cells (Center), labeling of surface AMPARs was almost abolished. After exposure of cells to 100 μM AMPA for 15 min, prominent staining of intracellular AMPAR puncta was apparent (Right), reflecting internalization of antibody-labeled AMPARs. (C) Quantitation of the acid-stripping assay in multiple specimens. Ordinate is mean number of internalized (acid-resistant) AMPAR puncta visualized per 10 μM dendrite for untreated cells (control), AMPA-treated cells (AMPA), and cells incubated for 15 min in the presence of 100 μM AMPA + 50 μM 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX).

Figure 2

Time course of ligand-induced internalization of AMPARs. (A) Cultured neurons were surface labeled with anti-GluR1 antibody, treated with 100 μM AMPA for 0, 1, 5, and 15 min, then immediately chilled on ice, and analyzed for AMPAR internalization by using the acid-strip procedure. Minimal AMPAR internalization is detected until 5 min of AMPA exposure. (B) Internalization of AMPARs can be initiated within 1 min after ligand-induced activation. Cells were incubated in the presence of 100 μM AMPA for 1 min or 5 min and then analyzed by using the acid-strip procedure either immediately (first and third panels) or after agonist washout and chase incubation in the presence of CNQX and d-APV for a total of 15 min (second and fourth panels).

Figure 3

Internalization of AMPARs is induced by glutamate and facilitated by NMDAR activation. (A) Incubation of cells with 10 μM glutamate for 1 min followed by chase incubation in the presence of CNQX and d-APV (APV) induced readily detectable internalization of AMPARs (Middle) compared with untreated cells (Top). Inclusion of d-APV (50 μM) in the pulse incubation strongly inhibited this process (Bottom). (B) Quantitation of the number of internalized AMPAR puncta under these conditions (n = 25 for each group).

Figure 4

Internalization of AMPARs induced by synaptically released glutamate. (A) Application of KCl (30 mM for 1 min) followed by 14 min chase incubation in normal medium caused significant internalization of AMPARs (compare top two panels). Blockade of glutamate receptors with CNQX and d-APV strongly inhibited this process, confirming that KCl-induced internalization of AMPARs is mediated by endogenously released glutamate binding to receptors. AMPAR internalization induced by KCl was also strongly inhibited by the NMDAR-specific antagonist d-APV. (B) Quantitation of AMPAR internalization in the absence and presence of receptor antagonists (n = 23 for each group).

Figure 5

Evidence for role of clathrin-coated pits in AMPAR endocytosis. (A) After surface labeling AMPARs and brief application of AMPA cells were fixed and permeabilized, and antibody-labeled AMPARs and immunoreactive AP2 were detected in the same specimens by using dual-channel confocal fluorescence microscopy. Localization of AP2 (Left), GluR1 (Center), and a merged image (Right) are shown (examples of colocalized puncta are indicated by arrowheads). (B) Hypertonic medium (350 mM sucrose) blocks AMPAR internalization. Cultures were equilibrated either in normal medium (Left and Center) or in hypertonic medium containing 350 mM sucrose (Right) before antibody labeling and analysis of AMPAR internalization. AMPA (100 μM, 15 min) caused clear AMPAR internalization in cells incubated in normal medium (compared Left and Center) but had minimal effects on cells preequilibrated in hypertonic medium. (C) Quantitation of the effect of hypertonic medium on AMPAR internalization (n = 30 for each group).

Figure 6

Ligand-induced internalization of AMPARs is dynamin-dependent. HA-tagged wild-type or K44A mutant dynamin−2 were expressed in hippocampal cells via adenovirus-mediated transfection. Neurons were then examined for AMPAR internalization. (A) Micrographs showing the specific inhibition of AMPAR internalization caused by K44A mutant dynamin. The top two rows illustrate cells in which AMPAR internalization was induced by 100 μM AMPA for 15 min, conditions that induce NMDAR-independent internalization of AMPARs. The bottom two rows illustrate the same experiment conducted with the pulse–chase protocol with 10 μM glutamate applied for 1 min, conditions that reveal NMDAR-dependent internalization of AMPARs. In each set of panels, expression of HA-tagged dynamin constructs is indicated (HA), and internalized AMPARs detected in the same cells are shown (GluR1). With either protocol, substantial internalization of AMPARs was observed in cells not expressing mutant dynamin (Left, Untrans; arrow indicates neuron that has no detectable HA-tagged dynamin expression) or in neurons expressing HA-tagged wild-type dynamin−2 (Center, Wt dyn-2). In contrast, in cells expressing K44A mutant dynamin−2 (Right, K44A dyn-2, arrowhead), internalization of AMPARs was strongly inhibited. (B) Quantitation of AMPAR internalization induced by both ligand-activation protocols (n = 8 for AMPA application and n = 15 for glutamate application).