Dynamic Interaction of Stargazin-like TARPs with Cycling AMPA Receptors at Synapses

See allHide authors and affiliations

Science  05 Mar 2004:
Vol. 303, Issue 5663, pp. 1508-1511
DOI: 10.1126/science.1090262


Activity-dependent plasticity in the brain arises in part from changes in the number of synaptic AMPA receptors. Synaptic trafficking of AMPA receptors is controlled by stargazin and homologous transmembrane AMPA receptor regulatory proteins (TARPs). We found that TARPs were stable at the plasma membrane, whereas AMPA receptors were internalized in a glutamate-regulated manner. Interaction with AMPA receptors involved both extra- and intracellular determinants of TARPs. Upon binding to glutamate, AMPA receptors detached from TARPs. This did not require ion flux or intracellular second messengers. This allosteric mechanism for AMPA receptor dissociation from TARPs may participate in glutamate-mediated internalization of receptors in synaptic plasticity.

AMPA-type glutamate receptors mediate most fast excitatory synaptic transmission in the brain. Interestingly, AMPA receptors recycle rapidly at the plasma membrane (17), and dynamic changes in the number of synaptic AMPA receptors regulate synaptic strength (812). The first transmembrane protein found to interact with AMPA receptors is the tetraspanning protein stargazin (13, 14), which is mutated in stargazer mice (15). Stargazin is required for synaptic trafficking of functional AMPA receptors in cerebellar granule cells (13). Furthermore, an extended family of stargazin-related TARPs regulate AMPA receptors in forebrain neurons (16). In addition to interacting with AMPA receptors, TARPs bind to PDZ domains from the postsynaptic density protein PSD-95, and this mediates the synaptic targeting of AMPA receptors (17).

To study TARP function in forebrain neurons, we generated a specific antiserum to γ-3 (Fig. 1A) (18), which is enriched in cerebral cortical neurons together with stargazin and γ-8 (16). Immunofluorescent labeling of primary cerebrocortical cultures revealed that γ-3 occurs selectively at punctate sites along the dendrites. These puncta reflect excitatory synapses, given that they overlap with AMPA receptor subunit GluR2 and PSD-95 but not the γ-aminobutyric acid (GABA)–synthesizing enzyme GAD-65 (Fig. 1B). γ-3 also colocalizes at synapses with erbB4, a marker for excitatory synapses in GABAergic neurons.

Fig. 1.

Synaptic TARPs are stable at the cell surface during AMPA-induced receptor internalization. (A) Antibody to γ-3 recognizes specifically a single 32-kD band in adult rat cerebral cortex. (B) Immunofluorescence for γ-3 occurs at punctate sites (white arrows) along the dendrites that colocalize with the glutamatergic markers GluR2, PSD-95, and erbB4 but not with the GABAergic marker GAD65. Preabsorbing the γ-3 antibody with antigen blocks labeling (small boxes in second row). (C) Surface biotinylated cortical cultures were treated for 1 min with 100 μM AMPA plus 50 μM d-APV and were then incubated for the indicated times. After cleavage of extracellular biotin, internalized receptors or TARPs were quantitated relative to calibration standards. GluR1 but not NR1, stargazin, or γ-3 underwent AMPA-stimulated endocytosis. (D) Time course of protein endocytosis in cultures treated with 100 μM AMPA plus 50 μM d-APV (*, P < 0.03; ** P < 0.005). (E) AMPA-induced dissociation of the cell surface γ-3/AMPA receptor complex in cortical neuronal cultures. Cultures were treated for 1 min with 100 μM AMPA plus 50 μM d-APV or control and were then incubated for another 14 min. Surface proteins were cross-linked and the cell surface γ-3/AMPA receptor complex was detected by coimmunoprecipitation followed by Western blotting. (F) Quantitation of the γ-3/AMPA receptor complex on the cell surface after AMPA treatment or control (*, P < 0.05). (G and H) High extracellular K+ depolarization dissociates the cell surface AMPA receptor from γ-3 in a CNQX-sensitive fashion. Cortical cultures were treated for 15 min with 50 mM KCl, 50 μM D-APV, and 100 nM picrotoxin minus or plus 100 μM CNQX, and levels of the surface γ-3/AMPA receptor complex were determined as described in (E) and (F) (n = 4; *, P < 0.05). Error bars in (D), (F), and (H) show mean + SEM.

We next asked whether TARPs undergo activity-dependent endocytosis along with AMPA receptors. Neuronal activity was blocked by tetrodotoxin (2 μM), and protein internalization was monitored with a biotinylation assay (1). Endocytosis of AMPA receptors but not N-methyl-D-aspartate (NMDA) receptors was accelerated by treating cultures with AMPA (Fig. 1, C and D) (1, 3, 4). In contrast to AMPA receptors, stargazin and γ-3 turned over very slowly at the plasma membrane and their endocytosis was not influenced by AMPA (Fig. 1, C and D).

This combination of ligand-stimulated endocytosis of AMPA receptors and membrane stability of stargazin and γ-3 suggests that these molecules may dissociate for receptor internalization. We thus treated neurons with AMPA to induce endocytosis, cross-linked surface proteins, and immunoprecipitated γ-3. Treatment with AMPA decreased the association of surface AMPA receptors with TARPs (Fig. 1, E and F). Depolarizing cultured neurons with elevated extracellular K+ also reduced surface levels of the AMPA receptor/TARP complex in a 6-cyano-7-nitroquinoxaline-2, 3-dione (CNQX)–sensitive fashion, indicating that the release of endogenous glutamate at excitatory synapses disrupts the association of AMPA receptors with TARPs (Fig. 1, G and H). Depolarization decreases surface AMPA receptor/TARP complex levels by ∼10%, which is consistent with previous studies showing that the majority of AMPA receptors are non-synaptic (19).

Mechanisms that control AMPA receptor endocytosis are uncertain (1, 3, 4). Some studies suggest that internalization of activated AMPA receptors may be triggered by an agonist-induced conformation change of the receptor (1, 4). To evaluate whether glutamate binding to AMPA receptors induces dissociation from TARPs, we immunoprecipitated γ-3 from detergent-solubilized cerebrocortical extracts, treated the immunoprecipitates with agonists, and quantitated AMPA receptor association. Treating immunoprecipitates with either AMPA or glutamate stimulated the dissociation of GluR1 and GluR2 from γ-3, but NMDA, GABA, and glutamine had no effect (Fig. 2A). This AMPA effect was dose dependent, with a median effective concentration of 10 μM (Fig. 2B). CNQX blocked the AMPA effect, but D-aminophosphovalerate acid (DAPV) did not (Fig. 2C); the median inhibitory concentration for CNQX was 1 μM (Fig. 2D). γ-3 binding to PDZ domains from PSD-95 was not influenced by AMPA (Fig. 2, A and C).

Fig. 2.

Agonist binding dissociates AMPA receptors from TARPs. [(A) to (D) and (F)] Solubilized membranes from cerebral cortex were immunoprecipitated with an antibody to γ-3. Beads were then washed with glutamate receptor agonists or other agents, and bound proteins were detected by Western blotting. (A) AMPA or glutamate causes dissociation of AMPA receptor subunits GluR1 and GluR2 from γ-3, whereas the related compounds—NMDA, glutamine, and GABA— do not cause dissociation. AMPA receptor agonists do not affect the interaction between γ-3 and PSD-95. (B) Dose dependence for AMPA-induced receptor dissociation from γ-3 (n = 4). (C) CNQX but not d-APV (D-AP5) prevents the dissociation of AMPA receptors from γ-3 induced by AMPA. (D) Dose-dependent inhibition by CNQX of AMPA-induced receptor dissociation from γ-3 (open square; n = 4). (E) Solubilized membranes from cerebral cortex were immunoprecipitated (IP) with an antibody to GluR2/3 or nonimmune rabbit immunoglobulin G (rb IgG). Beads were then washed with AMPA and/or CNQX, and bound proteins were detected by Western blotting. STG, stargazin. (F) Cyclothiazide (CTZ; 200 μM) has no effect on agonist-induced AMPA receptor dissociation from TARP.

We performed reciprocal experiments by immunoprecipitating GluR2/3. Again, AMPA stimulated dissociation of TARPs (γ-8, γ-3, and stargazin) from GluR2/3, and CNQX blocked this effect (Fig. 2E). AMPA did not influence GluR1 association with GluR2/3. Cyclothiazide, which blocks AMPA receptor desensitization, did not prevent agonist-induced stargazin dissociation from AMPA receptors, suggesting that allosteric changes that are associated with receptor activation rather than desensitization induce the dissociation (Fig. 2F).

We next determined the site(s) on stargazin that regulate AMPA receptor trafficking. We used a chimeric analysis with γ-5, the protein most similar in sequence to TARPs that does not regulate AMPA receptors (16). After mRNA injection into Xenopus laevis oocytes, stargazin or γ-3 vastly increased the functional surface expression of coexpressed GluR1 or GluR2 (20), whereas γ-5 had no effect (Fig. 3A) (21). We systematically swapped each domain in stargazin with the corresponding segment of γ-5. These experiments identified critical roles for both the first extracellular loop and the intracellular tail of stargazin in AMPA receptor trafficking (Fig. 3A). Furthermore, swapping in just these two domains efficiently conferred TARP activity into γ-5 (Fig. 3B).

Fig. 3.

GluR1 trafficking requires extracellular and intracellular domains of stargazin. (A) Oocytes were injected with 100 pg of complementary RNA (cRNA) encoding GluR1 and each stargazin/γ-5 construct (left). Stargazin (green) enhanced glutamate (5 μM)–evoked GluR1-mediated currents, but γ-5 did not. Chimeric constructs are composed of the γ-2 backbone and the indicated region(s) of γ-5. The chimera containing the cytoplasmic region of γ-5 (Cyto) is inactive, and the chimera containing the first extracellular domain of γ-5 (Ex1) shows a marked decrease in glutamate-evoked currents as compared with those that occurred in the presence of stargazin (n = 5; *, P < 0.05; ***, P < 0.001; black asterisk shows Student's t test value against γ-5; blue asterisk shows difference compared with stargazin). Norm., normalized; TM, transmembrane domain. (B) Swapping the first extracellular loop and cytoplasmic tail from stargazin into γ-5 efficiently conferred TARP activity for glutamate (40 μM)–evoked GluR1-mediated currents (n = 14; ***, P < 0.001).

To analyze further TARP's association with AMPA receptors, we immunoprecipitated stargazin from crude brain membranes, separated the immunoprecipitates by SDS–polyacrylamide gel electrophoresis, and identified associated proteins by silver staining. Only a single protein band of 105 kD, which reflects AMPA receptor subunits, coimmunoprecipitates with stargazin (fig. S1). This indicates that stargazin uniquely and directly binds to AMPA receptors. To study this binding in detail, we developed an assay in which C-terminal green fluorescent protein (GFP)–tagged stargazin, γ-5, and chimeras were each expressed in human embryonic kidney (HEK) 293 cells and, after their extraction from the HEK 293 cell membranes, were incubated with solubilized membranes from rat brain P2 extracts. Stargazin-GFP associated specifically with AMPA receptors from brain extracts, whereas γ-5–GFP did not (fig. S2). The first extracellular domain of stargazin was essential for binding because replacing this region with that from γ-5 disrupted association with AMPA receptors (Fig. 4C). The inclusion of a hemagglutinin (HA) tag near the end of the first extracellular loop (Fig. 4, A, B, and D; HA3) but not in the middle (HA1 and HA2) largely disrupted stargazin binding and function, indicating that AMPA receptor interaction is likely in this region. Deletion analysis of the C-terminal domain of stargazin identified a critical role for residues 212 to 268 in the tail of stargazin (Fig. 4A) for both binding (Fig. 4E) and trafficking (Fig. 4B) AMPA receptors. These experiments indicate that glutamate binding likely causes allosteric changes in both the extra- and intracellular domains of the AMPA receptor to induce TARP dissociation.

Fig. 4.

Extracellular and intracellular domains of stargazin interact with AMPA receptors. (A) Diagram shows the domain structure of the various stargazin constructs. HA, HA epitope tag (Tyr-Pro-Tyr-Asp-Val-Pro-Asp-Tyr-Ala). (B) Oocytes were injected with 100 pg of cRNA encoding GluR1 and each construct as shown on (A). STG-HA3, which contains the HA epitope close to TM2, shows substantial decrease in glutamate (5 μM)–evoked GluR1-mediated currents. Stargazin (1 to 212) failed to enhance glutamate-evoked currents whereas STG (1 to 268) was active. (C) γ-5–GFP and Ex1-GFP, as described in Fig. 3A, show no detectable binding to AMPA receptors, whereas stargazin-GFP binds to AMPA receptors. (D) Inclusion of an HA tag near the end of the stargazin first extracellular loop (STG-HA3) but not in the middle (STG-HA1,2) largely disrupts stargazin binding to AMPA receptors. (E) STG (1 to 268) binds to AMPA receptors, but γ-5 and STG (1 to 212) do not.

Taken together, our studies show that the AMPA receptor/TARP complex is not always stable at the synapse, because agonist binding can induce receptor dissociation. The regulated interaction with TARPs at the synapse may participate in activity-dependent turnover of AMPA receptors. This rapid cycling of AMPA receptors contrasts with the stability of NMDA receptors (15, 8), and the glutamate-mediated dissociation of AMPA receptors from TARPs may contribute to this differential regulation.

Pharmacological studies suggest that the activation of NMDA and AMPA receptors elicits internalization of AMPA receptors by distinct mechanisms. NMDA receptor activation induces AMPA receptor internalization through calcium-dependent pathways (1, 3, 10) that involve phosphorylation of GluR2 (22), dephosphorylation of GluR1 (1, 23, 24), or depalmitoylation of PSD-95 (25). In addition, phosphorylation of the PDZ-binding site in stargazin by protein kinase A may induce AMPA receptor internalization (26, 27).

In contrast to the roles for intracellular second messengers downstream of NMDA receptors, some studies have shown that glutamate binding to AMPA receptors may stimulate receptor internalization (1, 4) and subsequent degradation. Although the role for calcium in this process is controversial (1, 3, 4), we find that glutamate binding directly induces AMPA receptor dissociation from stargazin, which may explain receptor degradation (1). These effects do not appear to require traditional ion channel activation or second messengers but rather result from agonist-mediated allosteric changes in AMPA receptor conformation. Consistent with this, structural studies indicate that agonist but not antagonist binding causes a large conformational change in AMPA receptors (28). Other work suggests that use-dependent down-regulation of NMDA receptors is also independent of ion flux (29). These emerging modes for agonist-mediated internalization of glutamate receptors provide mechanisms for highly localized plasticity of the synaptic membrane.

Supporting Online Material

Materials and Methods

Figs. S1 and S2


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article