Pharmacological Reviews Get Tables of Contents delivered automatically
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     


0031-6997/05/5702-253-277$7.00
Pharmacol Rev 57:253-277, 2005

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Palmer, C. L.
Right arrow Articles by Henley, J. M.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Palmer, C. L.
Right arrow Articles by Henley, J. M.

Article

The Molecular Pharmacology and Cell Biology of {alpha}-Amino-3-hydroxy-5-methyl-4-isoxazolepropionic Acid Receptors

Claire L. Palmer, Lucy Cotton and Jeremy M. Henley

Medical Research Council Centre for Synaptic Plasticity, Department of Anatomy, School of Medical Sciences, Bristol University, Bristol, United Kingdom

Abstract
I. Introduction
    A. Classes of Glutamate Receptors
    B. AMPA Receptor Topology
        1. The N Terminus.
        2. Hydrophobic Regions.
        3. The Intracellular C Terminus.
    C. Pharmacology
    D. Post-Transcriptional Modification
        1. Splice Variants.
        2. RNA Editing.
    E. Post-Translational Modification
        1. Glycosylation.
        2. Phosphorylation.
            a. GluR1.
            b. GluR2.
            c. GluR3.
            d. GluR4.
    F. Expression Patterns
        1. Regional Distribution in the Brain.
        2. Neuronal and Glial Expression.
        3. Developmental Regulation.
        4. Subcellular Expression Patterns.
    G. Plasma Membrane Distribution of AMPA Receptors
        1. Postsynaptic Membrane.
        2. Extrasynaptic AMPA Receptors.
        3. Presynaptic Terminal.
II. Trafficking of AMPA Receptors
    A. Assembly
    B. Visualizing AMPA Receptor Translocation
    C. Lateral Diffusion of AMPA Receptors in the Membrane
    D. AMPA Receptor Delivery to Synapses
    E. AMPA Receptor Turnover at Synapses
    F. Trafficking, Learning, and Memory
III. Interacting Proteins
    A. PDZ-Containing Proteins
        1. PDZ Architecture.
        2. AMPA Receptor Binding Protein/Glutamate Receptor-Interacting Protein.
        3. LIN-10.
        4. Protein Interacting with C Kinase.
        5. Synapse-Associated Protein-97.
        6. SemaF Cytoplasmic Domain-Associated Protein-3 and PDZ-Regulator of G-Protein Signaling-3.
        7. Syntenin.
        8. Transmembrane AMPA Receptor Regulatory Proteins.
    B. Other Interactors
        1. Neuronal Activity-Regulated Pentraxin.
        2. N-Ethylmaleimide-Sensitive Factor.
        3. 4.1.
IV. Future Directions
Abstract

{alpha}-Amino-3-hydroxy-5-methyl-4-isoxazolepropionate receptors (AMPARs) are of fundamental importance in the brain. They are responsible for the majority of fast excitatory synaptic transmission, and their overactivation is potently excitotoxic. Recent findings have implicated AMPARs in synapse formation and stabilization, and regulation of functional AMPARs is the principal mechanism underlying synaptic plasticity. Changes in AMPAR activity have been described in the pathology of numerous diseases, such as Alzheimer's disease, stroke, and epilepsy. Unsurprisingly, the developmental and activity-dependent changes in the functional synaptic expression of these receptors are under tight cellular regulation. The molecular and cellular mechanisms that control the postsynaptic insertion, arrangement, and lifetime of surface-expressed AMPARs are the subject of intense and widespread investigation. For example, there has been an explosion of information about proteins that interact with AMPAR subunits, and these interactors are beginning to provide real insight into the molecular and cellular mechanisms underlying the cell biology of AMPARs. As a result, there has been considerable progress in this field, and the aim of this review is to provide an account of the current state of knowledge.

I. Introduction

A. Classes of Glutamate Receptors

The amino acid glutamate is the major excitatory neurotransmitter in the mammalian central nervous system (CNS1), and it exerts its physiological effects by binding to a number of different types of glutamate receptors (GluRs). Glutamate receptors can be divided into two functionally distinct categories: those that mediate their effects via coupling to G-protein second messenger systems, the metabotropic glutamate receptors (mGluRs) and ionotropic ligand-gated ion channels (Simeone et al., 2004Go). Based largely upon the work of Watkins and coworkers (Watkins, 1981Go, 1991Go; Watkins et al., 1981Go), the ionotropic glutamate receptors have been separated into three distinct subgroups based upon their pharmacology (for related reviews, see Dingledine et al., 1999Go; Barnes and Slevin, 2003Go; Mayer and Armstrong, 2004Go). These are the {alpha}-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors, N-methyl-D-aspartate (NMDA) receptors, and kainate (KA) receptors. However, it is important to note that AMPARs are responsive to kainate.

The pharmacologically isolated families of receptors were subsequently found to be encoded by distinct gene families. AMPARs comprise four subunits named GluR1 to GluR4 (also called GluRA–GluRD). The GluR1 subunit was first cloned after screening an expression library (Hollmann et al., 1989Go), and subsequent cDNA homology screens revealed the remaining GluR2, GluR3, and GluR4 homologs (Boulter et al., 1990Go; Keinanen et al., 1990Go; Nakanishi et al., 1990Go; Sakimura et al., 1990Go).

B. AMPA Receptor Topology

The schematic topology of an AMPAR subunit is illustrated in Fig. 1. This structure has been based largely on biochemical investigations as well as homology between the AMPARs and prokaryotic amino acid receptors (reviewed in Paas, 1998Go). The molecular architecture of each AMPAR subunit (GluR1–4) is very similar; each comprises ~900 amino acids and has a molecular weight of ~105 kDa (Rogers et al., 1991Go). There is approximately 70% sequence homology between genes encoding each subunit, although genes may undergo alternative splicing in two distinct regions, resulting in subunits that have either long or short C termini, and flip or flop variants in an extracellular domain (for review, see Black and Grabowski, 2003Go).



View larger version (56K):
[in this window]
[in a new window]
  		FIG. 1. Schematic showing the topology of an AMPA receptor subunit. Each subunit consists of an extracellular N-terminal domain, four hydrophobic regions (TM1–4), and an intracellular C-terminal domain. The ligand-binding site is a conserved amino acid pocket formed from a conformational association between the N terminus and the loop linking TM3 and TM4. A flip/flop alternative splice region and R/G RNA editing site are also present within the TM3/TM4 loop. TM2 forms an intracellular re-entrant hairpin loop which contributes to the cation pore channel and is also the site for Q/R RNA editing in the GluR2 subunit. The intracellular C terminus contains phosphorylation sites and conserved sequences that have been shown to interact with a number of intracellular proteins, for example, PDZ domain-containing proteins and the ATPase NSF.

 

1. The N Terminus. Each subunit includes an extracellular N terminus, four hydrophobic domains (TM1–4), and an intracellular C terminus (Fig. 1). In eukaryotes, the N terminus contains the N-terminal domain of ~400 amino acids and a ~150 amino acid ligand-binding core. The N-terminal domain is also known as the X-domain because of its unknown function (Kuusinen et al., 1999Go). Suggestions for X-domain function include receptor assembly, allosteric modulation of the ion channel, and binding of a second ligand. The X-domains of GluR4 form dimers in solution (Kuusinen et al., 1999Go) and confer specificity for AMPARs, as opposed to KARs upon coassembly with other subunits (Leuschner and Hoch, 1999Go). However, deletion of the entire X-domain of GluR4 did not alter the function of homomers expressed in human embryonic kidney 293 cells, indicating that it is not involved in homomeric assembly of this subunit (Pasternack et al., 2002Go). The structure of this part of AMPARs is suggestive of a ligand-binding site, but no endogenous ligands have been found to bind here, although Zn2+ modulates at a similar site on NMDARs (Mayer and Armstrong, 2004Go). Intriguingly, the N terminus of GluR2 is involved in dendritic spine morphogenesis, perhaps through a receptor-ligand complex (Passafaro et al., 2003Go).

The ligand-binding core of AMPARs confers pharmacological specificity to the receptors; indeed, swapping the domains of AMPA and KARs swapped both their affinity for the ligand and desensitization properties and proved that this is the glutamate binding site (Stern-Bach et al., 1994Go, 1998Go). The structures of the ligand binding cores of GluR2 and GluR4 have been studied intensively (Jayaraman et al., 2000Go; Kubo and Ito, 2004Go; McFeeters and Oswald, 2004Go), and this is the only part of any AMPAR to be crystallized so far (Armstrong et al., 1998Go). For GluR2, the ligand-binding core has been crystallized with various pharmacological agents (Johansen et al., 2003Go). This approach gives an insight into the mode of action of some AMPAR agonists and antagonists and promises to be a valuable tool for the rational design of future drugs (reviewed in Stensbol et al., 2002Go; Mayer and Armstrong, 2004Go).

2. Hydrophobic Regions. The transmembrane orientation of the AMPAR subunits was initially elucidated by the use of specific antibodies, N-glycosylation pattern, and proteolytic sites (Molnar et al., 1994Go; Wo et al., 1995Go). Together, these studies demonstrated that the mature N terminus is expressed on the exterior surface of the neuron (Hollmann et al., 1994Go; Bennett and Dingledine, 1995Go; Seal et al., 1995Go), and subsequent work showed that the TM1, TM3, and TM4 regions are all transmembrane spanning domains, whereas TM2 forms a hairpin loop on the intracellular side of the cell membrane (see also Wo and Oswald, 1994Go) (Fig. 1). Similar to K+ channels, the re-entrant loop contributes to the cation pore channel (Kuner et al., 2003Go), although the specificity of AMPARs differ in that they gate Na+ and Ca2+ in preference to K+, perhaps because of a comparatively larger pore size (Tikhonov et al., 2002Go).

3. The Intracellular C Terminus. The intracellular C terminus of eukaryotic AMPARs has been shown to be the interaction site for a range of different proteins, many of which are involved in the receptor trafficking (reviewed in Henley, 2003Go) and synaptic plasticity (reviewed in Malenka, 2003Go; Sheng and Hyoung Lee, 2003Go). The functions of AMPAR interactors are under intense scrutiny and will be discussed in greater detail below.

C. Pharmacology

AMPARs gate Na+ and Ca2+ in response to ligand binding, with conductance and kinetic properties of the receptor depending on the subunit composition (Mat Jais et al., 1984Go; Hollmann et al., 1991Go; Jonas, 1993Go). This influx of ions causes a fast excitatory postsynaptic response, and the Ca2+ component can also activate second messenger pathways (Michaelis, 1998Go), including many protein kinases (Wang et al., 2004Go); indeed, AMPARs have been reported to have metabotropic as well as ionotropic properties (Wang and Durkin, 1995Go; Wang et al., 1997Go). For researchers investigating the function of AMPARs in in vitro preparations, it may also be interesting to note that AMPARs may be potentiated by serum factors (Nishizaki et al., 1997Go).

There are many drugs available that act on AMPARs, and listing them is beyond the scope of this review (but see Stensbol et al., 2002Go; Stone and Addae, 2002Go; Weiser, 2002Go; McFeeters and Oswald, 2004Go; O'Neill et al., 2004Go; Stromgaard and Mellor, 2004Go).

D. Post-Transcriptional Modification

1. Splice Variants. All four AMPAR subunits undergo alternative splicing in an extracellular region N-terminal to the fourth transmembrane domain to give "flip" and "flop" splice variants (Sommer et al., 1990Go) (see Fig. 1). This modifies the channel's kinetic and pharmacological properties with flip splice variants desensitizing four times slower than flop (Mosbacher et al., 1994Go; Koike et al., 2000Go) and confers different sensitivity to allosteric modulators cyclothiazide (Partin et al., 1994Go; Kessler et al., 2000Go), 4-[2-(phenylsulfonylamino)ethylthio]-2,6-difluoro-phenoxyacetamide (Sekiguchi et al., 1997Go, 1998Go), zinc (Shen and Yang, 1999Go), and lithium (Karkanias and Papke, 1999Go), although affinity to AMPA is unchanged (Arvola and Keinanen, 1996Go). Expression levels of the different splice variants is region and cell-type specific (Sommer et al., 1990Go; Fleck et al., 1996Go; Lambolez et al., 1996Go) and developmentally regulated, for example, in the cerebellum (Mosbacher et al., 1994Go), as well as modified by physiological insults (Zhou et al., 2001bGo), lesions (Pires et al., 2000Go), and disease (Seifert et al., 2002Go, 2004Go; Tomiyama et al., 2002Go). This means that the number of permutations of AMPARs is very large giving a potential to fine-tune the kinetic properties of the channel.

Further to the flip and flop splice variants, GluR1, -2, and -4 can also undergo alternative splicing in the C terminus to give "long" isoforms (Gallo et al., 1992Go; Kohler et al., 1994Go); however the "short" isoform of GluR1 has not been reported. The short isoform of GluR2 is the most abundant, accounting for over 90% of total GluR2 (Kohler et al., 1994Go), and the long form of GluR4 is predominant (Gallo et al., 1992Go). GluR3 has a short C terminus due to a lack of splice sites in the C terminus. The alternative splice variants are able to bind different interacting proteins because the PDZ binding motif is only present in the short form (Dev et al., 1999Go). In consequence, much of the work in the field of AMPAR interactors has focused on the short form of GluR2.

2. RNA Editing. The genomic DNA of the GluR2 subunit of AMPARs contains a glutamine (Q) residue at amino acid 607. However, the vast majority of neuronal cDNA contains an arginine (R) at this position and occurs via a process of nuclear RNA editing (Sommer et al., 1991Go; Cha et al., 1994Go; Seeburg et al., 1998Go; Seeburg and Hartner, 2003Go). The edited residue is found in the channel-forming segment of the receptor, and the amount and presence of the edited subunit alters the kinetics and divalent ion permeability of the resulting receptors (Burnashev et al., 1992Go, 1995Go; Geiger et al., 1995Go), although it is not linked to receptor desensitization (Thalhammer et al., 1999Go). GluR2(R)-containing AMPARs have a low permeability to Ca2+ and low single-channel conductance because of the size and charge of the amino acid side chain in the edited form (Burnashev et al., 1992Go, 1996Go; Swanson et al., 1997Go). Nonetheless, GluR2(R)-containing AMPARs can still participate in intracellular Ca2+ signaling (Utz and Verdoorn, 1997Go) and can be trafficked in a Ca2+-dependent way (Liu and Cull-Candy, 2000Go). Furthermore, editing at this position has been shown to regulate endoplasmic reticulum retention (Greger et al., 2002Go) and tetramerization of AMPARs (Greger et al., 2003Go). Edited and unedited forms of the receptor have different sensitivity to the pharmacological agents Joro Spider toxin and adamantine derivatives (Meucci et al., 1996Go; Magazanik et al., 1997Go; McBain, 1998Go), and editing at this position is developmentally, region- and cell-specifically regulated (Lerma et al., 1994Go; Nutt and Kamboj, 1994Go). Indeed, changes in the levels of editing occur during differentiation of neurons and glia (Meucci et al., 1996Go; Lai et al., 1997Go).

Because of the link between calcium influx and excitotoxicity, changes in the amount of edited GluR2 have been implicated in a number of diseases including schizophrenia, Huntington's disease, Alzheimer's disease (Akbarian et al., 1995Go), epilepsy (Brusa et al., 1995Go), and malignant glioma (Maas et al., 2001Go) with most research being focused on amyotropic lateral sclerosis (Takuma et al., 1999Go; Kawahara et al., 2003bGo, 2004Go), although the process is not involved in ischemia (Kamphuis et al., 1995Go; Paschen et al., 1996Go; Rump et al., 1996Go; but see Tanaka et al., 2000Go). Editing at this position is not essential for brain development (Kask et al., 1998Go), but mice have neurological deficits when editing does not take place (Feldmeyer et al., 1999Go).

Further, to the Q/R site of GluR2, GluR2, -3, and -4 may be edited at another site, the R/G site, in a region that immediately precedes the flip/flop splice module in the N terminus of the molecule (Lomeli et al., 1994Go; Fig. 1). This modification changes the desensitization and resensitization of the resulting AMPAR (Lomeli et al., 1994Go; Krampfl et al., 2002Go) and may be involved in epilepsy (Vollmar et al., 2004Go) and, in contrast to the Q/R site, ischemia (Yamaguchi et al., 1999Go).

The RNA-dependent adenosine deaminase 2 carries out RNA editing of AMPARs (Higuchi et al., 2000Go; Ohman et al., 2000Go), and like the edited and unedited forms of AMPAR subunits, expression levels of the enzyme are developmentally and regionally regulated. Indeed, it has been shown that RNA-dependent adenosine deaminase 2 abundance regulates levels of edited GluR2 (Kawahara et al., 2003aGo).

E. Post-Translational Modification

1. Glycosylation. All AMPARs have sites for asparagine (N)-linked glycosylation (Hullebroeck and Hampson, 1992Go; Breese and Leonard, 1993Go; Keinanen et al., 1994Go) in the extracellular domains of the protein, with two conserved sites in the S1 domain that forms part of the ligand-binding domain (Arvola and Keinanen, 1996Go; reviewed in Standley and Baudry, 2000Go; Pasternack et al., 2003Go). The functional consequence of the addition of these oligosaccharides is not clear. Inhibition of glycosylation with tunicamycin was found to prevent functional expression of recombinant AMPARs (Musshoff et al., 1992Go; Kawamoto et al., 1994Go), but later the drug itself was reported to inhibit AMPARs regardless of glycosylation state (Maruo et al., 2003Go). Studies of recombinantly expressed S1-S2 domain fusion proteins show that this form of post-translational modification does not affect the ligand-binding site (Arvola and Keinanen, 1996Go; Pasternack et al., 2003Go), however, the desensitizing lectin concanavalin A potentiates AMPAR currents by binding to these carbohydrates, with GluR2 remaining unaffected (Everts et al., 1997Go). This suggests that glycosylation of different AMPAR subunits may have different functional effects. It appears that only surface and synaptically expressed AMPARs possess the mature glycosylated form (Hall et al., 1997Go; Standley et al., 1998Go), and the exact nature of the oligosaccharides involved has been identified (Clark et al., 1998Go). It is likely that this protein modification is involved in the maturation and transport of the receptor or could protect AMPARs from proteolytic degradation. Indeed, the aberrant glycosylation of GluR3 can lead to proteolysis and release of an autoimmunogen leading to Rasmussen's encephalitis (Gahring et al., 2001Go).

2. Phosphorylation. Phosphorylation of ligand-gated ion channels can regulate the properties of the channel, its intermolecular interactions, and trafficking of the protein (reviewed in Swope et al., 1999Go). The regulation of AMPAR phosphorylation adds a further complex level of receptor modulation beyond subunit composition, splice variants, and other post-translational modifications (Carvalho et al., 2000Go; Gomes et al., 2003Go). Phosphorylation of receptor subunits can occur basally or in response to particular types of synaptic activity, and each AMPAR subunit has its own calcium and kinase profile (Wyneken et al., 1997Go). In particular, the role of GluR1 phosphorylation in synaptic plasticity has been studied in detail (reviewed in Roche et al., 1994Go; see Yakel et al., 1995Go). However, some phosphorylation sites are thought to be shared by all AMPARs: a residue between 620 and 638 of GluR1 and the equivalent sites in all other AMPARs is phosphorylated in vitro by CaMKII leading to enhanced responses in synaptic plasticity (Yakel et al., 1995Go), and although most research has centered on the role of CaMKII, it is possible that CaMKIV may play a similar role (Kasahara et al., 2000Go). There also appears to be a general role for developmental regulation of AMPAR properties by phosphorylation (Shaw and Lanius, 1992Go; Li et al., 2003Go). Furthermore, a GluR2 Ser 696 phosphospecific antibody helped demonstrate phosphorylation of this and analogous sites in all other AMPARs in response to agonist by PKC (Nakazawa et al., 1995aGo,bGo). This phosphorylation is thought to underlie long-term desensitization.



View larger version (18K):
[in this window]
[in a new window]
  		FIG. 2. Schematic illustrating that phosphorylation of the GluR1 AMPA receptor subunit may act as a bidirectional switch in synaptic plasticity.

 
In addition to modulation of AMPAR phosphorylation in response to glutamatergic synaptic activity, PKA phosphorylation of AMPARs, primarily GluR1, is enhanced by D1 dopamine receptor activation (Price et al., 1999Go; Snyder et al., 2000Go; Chao et al., 2002Go) and may play a role in Parkinson's disease (Chase et al., 2000Go; Oh et al., 2003Go). Indeed, dopamine receptors may influence synaptic plasticity of AMPARs (Wolf et al., 2003Go) by increasing PKA-mediated AMPAR insertion (Mangiavacchi and Wolf, 2004Go). Serotonin hydroxytryptamine 1A receptors can, on the other hand, inhibit CaMKII phosphorylation of AMPARs (Cai et al., 2002bGo).


a. GluR1. GluR1 is phosphorylated at multiple sites on the C terminus. PKC and CaMKII phosphorylate Ser 831 (Roche et al., 1996Go; Mammen et al., 1997Go), whereas PKA phosphorylates Ser 845 (Mammen et al., 1997Go). Differential phosphorylation of both sites occurs according to activity (Blackstone et al., 1994Go) leading to changes in synaptic efficacy (Fig. 2) and phosphorylation by PKA and PKC may play a role in nociception (Fang et al., 2003aGo,bGo; Nagy et al., 2004Go). Furthermore, phosphorylation at both PKA and CaMKII sites controls synaptic incorporation of these receptors (Esteban et al., 2003Go) and may be enhanced by the constitutive activity of phospholipase A2 (Menard et al., 2005Go).

AMPARs present in the postsynaptic density are phosphorylated by CamKII, and there is evidence that the PKA site is occluded in this location (Figurov et al., 1993Go; Vinade and Dosemeci, 2000Go). CaMKII phosphorylation plays a role in synaptic unsilencing (Liao et al., 2001Go) and enhancement (Figurov et al., 1993Go; Nishizaki and Matsumura, 2002Go), ischemia (Takagi et al., 2003Go; Fu et al., 2004Go), inflammation (Guan et al., 2004Go), and LTP (Hayashi et al., 1997Go; of depressed synapses: Strack et al., 1997Go; Lee et al., 2000Go) perhaps by increasing the AMPAR-mediated current (Derkach, 2003Go; Vinade and Dosemeci, 2000Go) and/or by modulating an interaction between GluR1 and a PDZ-containing protein (Hayashi et al., 2000Go). Interestingly, amyloid {beta} protein prevents the activation of CaMKII and AMPAR phosphorylation during LTP (Zhao et al., 2004Go). Dephosphorylation of this site by protein phosphatase 1 leads to depotentiation (Lee et al., 2000Go; Vinade and Dosemeci, 2000Go; Huang et al., 2001Go).

PKA phosphorylation leads to potentiation of homomeric peak current (Roche et al., 1996Go; Vinade and Dosemeci, 2000Go) by increasing the peak open probability (Banke et al., 2000Go) and leads to LTP in naive synapses (Lee et al., 2000Go). Dephosphorylation occurs via a calcineurin-mediated pathway (Snyder et al., 2003Go) and is a feature of NMDA-dependent LTD expression at naive synapses (Kameyama et al., 1998Go; Ehlers, 2000Go; Lee et al., 2000Go; Launey et al., 2004Go). The interacting proteins SAP-97 and A-kinase-anchoring protein 79 are thought to direct basal PKA phosphorylation of GluR1 and calcium-dependent dephosphorylation at this site (Colledge et al., 2000Go; Lisman and Zhabotinsky, 2001Go; Tavalin et al., 2002Go).

It is clear that the effects of GluR1 (de)phosphorylation at the major CaMKII and PKA sites on synaptic plasticity depends on the history of the synapse (Lee et al., 2000Go), and it is becoming increasingly evident that these kinases and phosphatases play important roles in development, synaptic plasticity, learning, and memory (Vianna et al., 2000Go; Genoux et al., 2002Go; D'Alcantara et al., 2003Go; Lee et al., 2003aGo; Lu et al., 2003Go), although a role for GluR1 dephosphorylation in LTD maintenance rather than induction has been postulated (Brown et al., 2005Go). See Fig. 2 for a model of how the phosphorylation state of GluR1 relates to the plastic state of the synapse. GluR1 phosphorylation by the Src family tyrosine kinase Fyn also occurs in vitro and protects the receptors from calpain digestion which could modulate trafficking events or channel properties (Rong et al., 2001Go).


b. GluR2. GluR2 is phosphorylated by PKC at Ser 880 in the PDZ binding site. This differentially regulates binding of AMPAR binding protein (ABP)/glutamate receptor-interacting protein 1 (GRIP1) and protein interacting with C kinase (PICK)1, with a decrease in ABP/GRIP binding but not PICK1 in response to phosphorylation (Matsuda et al., 1999Go; Seidenman et al., 2003Go). Furthermore, ABP binding to GluR2 prevents this phosphorylation (Fu et al., 2003Go). Phosphorylated GluR2 has been shown to recruit PICK1 to synapses apparently causing the release of the GluR2-PICK1 complex from synapses facilitating internalization (Chung et al., 2000Go; Seidenman et al., 2003Go) and LTD (Matsuda et al., 2000Go; Xia et al., 2000Go; Kim et al., 2001Go; Chung et al., 2003Go). Interestingly, cerebellar LTD has also been shown to involve a mitogen-activated protein kinase step that converts transiently phosphorylated AMPARs to stably phosphorylated ones (Kuroda et al., 2001Go), and LTD-inducing protocols lead to tyrosine phosphorylation of GluR2 and, consequently, endocytosis of the receptors (Ahmadian et al., 2004Go). Indeed, phosphorylation of Tyr 876 near the C terminus of the receptor by a src family tyrosine kinase has a similar effect on GRIP1 and PICK1 binding to PKC phosphorylation of Ser 880, although this phosphorylation is involved in AMPA- and NMDA-induced internalization of these receptors (Hayashi and Huganir, 2004Go). Furthermore, potential PKC phosphorylation sites have also been identified in the C terminus of GluR2, including Ser 863 (Hirai et al., 2000Go; McDonald et al., 2001Go).


c. GluR3. There has been no systematic study into the regulation of GluR3 subunit phosphorylation.


d. GluR4. Recombinant homomeric GluR4 is the most rapidly desensitizating of the AMPARs, a property that may be regulated by (de)phosphorylation. This subunit is phosphorylated by PKC in response to activation of the kinase/receptor (Carvalho et al., 2002Go). PKC directly interacts with the membrane-proximal C terminus of the receptor, phosphorylating at Ser 482, and leading to increased surface expression of recombinant receptors (Correia et al., 2003Go; Esteban et al., 2003Go), and Thr 830 is also a potential PKC site (Carvalho et al., 1999Go). Ser 842 can be phosphorylated by PKA which modulates surface expression of the receptor (Gomes et al., 2004Go).

F. Expression Patterns

1. Regional Distribution in the Brain. In situ hybridization studies (Keinanen et al., 1990Go; Pellegrini-Giampietro et al., 1991Go), receptor autoradiography using [3H]AMPA and [3H]glutamate (Monaghan et al., 1984Go), and immunocytochemistry using antibodies directed against individual AMPAR subunits (Petralia and Wenthold, 1992Go; Martin et al., 1993Go) demonstrated the widespread and varied distribution of AMPARs in the brain (reviewed in Hollmann and Heinemann, 1994Go). The hippocampus, outer layers of the cortex, olfactory regions, lateral septum, basal ganglia. and amygdala of the CNS are enriched in GluR1, GluR2, and GluR3 (Keinanen et al., 1990Go; Beneyto and Meador-Woodruff, 2004Go). In contrast, GluR4 mRNA and immunolabeling is low to moderate throughout the rat CNS, except in the reticular thalamic nuclei and the cerebellum where levels are high (Petralia and Wenthold, 1992Go; Martin et al., 1993Go; Spreafico et al., 1994Go).

2. Neuronal and Glial Expression. Interestingly, AMPARs have also been found on glial cells (Gallo and Russell, 1995Go; Garcia-Barcina and Matute, 1998Go; Janssens and Lesage, 2001Go), where they appear to be involved in excitotoxicity (Yoshioka et al., 1996Go; Park et al., 2003Go) and ischemia pathology (Gottlieb and Matute, 1997Go; Meng et al., 1997Go). Activation of these receptors on some glia can lead to the release of ATP or nitric oxide, which may act as autocrine or paracrine messengers (Queiroz et al., 1999Go; Comoletti et al., 2001Go) and can affect glial morphology (Ishiuchi et al., 2001Go). It has recently been demonstrated that mouse hippocampal astrocytes may be categorized into AMPAR-expressing cells or glutamate transporter cells, which adds yet another layer of complexity (Wallraff et al., 2004Go). Glia-glia coupling (Muller et al., 1996Go) and neuron-glia signaling is an emerging area of interest in this field, with work centering on the Bergmann glia of the cerebellum (Mennerick et al., 1996Go; Clark and Barbour, 1997Go; Iino et al., 2001Go; Dziedzic et al., 2003Go; Millan et al., 2004Go). Indeed, it appears to be possible to induce a form of LTP in cerebellar glia by stimulating neighboring neurons (Linden, 1997Go), and a form of epilepsy appears to be coupled to a change in AMPAR splice variant expression in hippocampal glia (Seifert et al., 2002Go). It has been proposed that neuron-glia signaling in the hypothalamus may also control sexual development (Dziedzic et al., 2003Go).

3. Developmental Regulation. AMPAR mRNA can be detected at very early stages of development. In embryonic rat brain, GluR2 mRNA is near ubiquitous, with GluR1, GluR3, and GluR4 more differentially expressed (Monyer et al., 1991Go). GluR1 protein was found in rat brain as early as E15.5 and GluR4 at E11 in mouse brain (Durand and Zukin, 1993Go; Martin et al., 1998Go). Later in development, studies on postnatal tissue have suggested that expression levels of GluR1–4 increase gradually, concurrent with synapse development, and appear to peak in the third postnatal week (Insel et al., 1990Go; Pellegrini-Giampietro et al., 1991Go; Durand and Zukin, 1993Go; Standley et al., 1995Go; Arai et al., 1997Go; Martin et al., 1998Go). However, immunoreactivity to GluR4 in rat brain was not apparent until P14, after which its levels of immunoreactivity increased gradually until adult-hood (Hall and Bahr, 1994Go). Developmental changes in the expression levels of AMPARs in hippocampal organotypic slice cultures (Fabian-Fine et al., 2000Go) and living cultured hippocampal neurons have also been reported (Pickard et al., 2000Go; Molnar et al., 2002Go), concluding that maturation of synapses may be retarded in vitro. AMPAR expression is also developmentally regulated in the spinal cord (Jakowec et al., 1995Go; Kalb and Fox, 1997Go), the visual system (Silveira dos Santos Bredariol and Hamassaki-Britto, 2001Go; Batista et al., 2002Go; Hack et al., 2002Go), and the auditory system (Sugden et al., 2002Go). During neonatal development, AMPAR incorporation into the plasma membrane occurs prior to synaptogenesis when GluR1-containing AMPARs cluster at potential postsynaptic sites (Martin et al., 1998Go). As well as developmental regulation of receptor subunit expression, splicing of AMPARs changes during development (Monyer et al., 1991Go; Tonnes et al., 1999Go), as can the kinetics of channel opening (Koike-Tani et al., 2005Go; Wall et al., 2002Go). The Ca2+-permeability of some glia also appears to be developmentally regulated (Backus and Berger, 1995Go).

4. Subcellular Expression Patterns. Subcellular fractionation experiments have indicated an enrichment of AMPARs in both synaptic membrane and postsynaptic density preparations (Rogers et al., 1991Go; Blackstone et al., 1992Go; Archibald and Henley, 1997Go). For a diagram of the morphology of an excitatory synapse at the electron-micrograph level, please see Fig. 3. Microscopy confirmed these findings in many brain areas (Petralia and Wenthold, 1992Go; Craig et al., 1993Go; Baude et al., 1994Go; Spreafico et al., 1994Go; Petralia et al., 1997Go). However, some AMPARs have also been detected extrasynaptically and within the cytoplasm of individual neurons (Baude et al., 1994Go, 1995Go). Consistent with this observation, a large number of intracellular AMPARs have been identified by biochemical studies (Barnes and Henley, 1993Go; Hall and Soderling, 1997Go; Hall et al., 1997Go; Lee et al., 2001Go). Overall, these studies suggest 60 to 70% of the total AMPAR population is intracellular. As discussed in more detail below, it has been speculated that this intracellular pool of AMPARs may play a role in synaptic plasticity and development via the "silent synapse" hypothesis.



View larger version (52K):
[in this window]
[in a new window]
  		FIG. 3. Electron micrograph and corresponding diagram depicting a single excitatory synapse. The electron dense PSD directly opposes the neurotransmitter release sites located on the presynaptic bouton and the spine neck and spine apparatus are clearly defined (reproduced with permission from Fischer et al., 2000Go).

 
G. Plasma Membrane Distribution of AMPA Receptors

1. Postsynaptic Membrane. It is well established that AMPARs reside in the postsynaptic compartment and that their appropriate targeting and clustering in this region is critical for the formation and maintenance of excitatory synapses. The mechanisms of receptor insertion and removal at the postsynaptic membrane and of incorporation and maintenance in functional clusters within the postsynaptic density have been the topics of intensive research (Nusser, 2000Go; Savtchenko et al., 2000Go; Sheng, 2001Go; Franks et al., 2003Go). The interactions between AMPARs and many accessory proteins are of critical importance in these trafficking steps, and insertion and removal of synaptic AMPARs plays an important role in synaptic plasticity (Lu et al., 2001aGo). These interactors will be discussed later in the review.

GluR2/3 and GluR4 subunits colocalize throughout the postsynaptic density in the rat organ of Corti, with higher concentrations of receptors located around the periphery of the PSD. These subunits were not detected at extrasynaptic membranes, but some GluR4 subunits appeared to be presynaptic (Matsubara et al., 1996Go). This suggests that AMPARs are inserted into the postsynaptic membrane in a very precise manner and that receptor density increases upon moving away from the center of the synapse. In the rat hippocampus, it was estimated that there were 3 to 140 individual AMPAR present at synapses on CA3 pyramidal spines (Nusser et al., 1998Go). However, more recent investigations into the relationship between spine morphology and AMPAR distribution suggest that different types of spines contain differing amounts of AMPARs (Matsuzaki et al., 2001Go). In addition to the differential expression of AMPAR subunits within and between brain regions (McBain and Dingledine, 1993Go; Hack et al., 2001Go), AMPARs of different composition may be targeted to different synapses within a single cell (Rubio and Wenthold, 1997Go) or the distribution of AMPARs inside a neuron may be heterogeneous (Andrasfalvy and Magee, 2001Go).

It has been shown that the abundance of postsynaptic AMPARs correlates with both the size of the synapse and the dimensions of the dendritic spine head (Matsuzaki et al., 2004Go). These findings suggest that silent Schaffer collateral-commissural (SCC) synapses (see below) are smaller than the majority of SCC synapses at which AMPA and NMDARs are colocalized and that synapse size may determine important properties of SCC synapses (Takumi et al., 1999Go). Furthermore, NMDAR activation has been reported to cause the formation of new spines as well as synaptic delivery of AMPARs and perhaps is involved in the mechanism of LTP (Shi et al., 1999Go). Recently, it has been reported that the N-terminal domain of GluR2 increases spine size and density in hippocampal neurons suggesting that, in addition to being involved in rapid neurotransmission, GluR2 is important for spine growth and/or stability (Passafaro et al., 2003Go).

2. Extrasynaptic AMPA Receptors. Electrophysiological experiments indicate that AMPARs are widely distributed throughout the cell surface plasma membrane. However, antibody surface labeling of neurons indicates that surface-expressed AMPARs do not have a homogenous distribution. Typically, distinct immunopositive areas are observed that are thought to correspond to synaptic puncta. This discrepancy may be explained if nonsynaptic AMPARs are present at an insufficient density to be detected efficiently by immunocytochemistry.

Recent studies have focused on the mechanisms of receptor recruitment to the plasma membrane, incorporation of receptors into the synapse, and their clustering, both during synaptogenesis and synaptic plasticity (for example, see Cottrell et al., 2000Go; Andrasfalvy and Magee, 2004Go). This will be discussed in more detail in the trafficking sections below.

3. Presynaptic Terminal. The investigation into presynaptic AMPARs has not received much attention to date. Nonetheless, some years ago it was reported that AMPA increased glutamate release from rat hippocampal synaptosomes. This effect was shown to be related specifically to AMPARs and suggests that AMPARs are present on presynaptic terminals and that they may play a role in the regulation of neurotransmitter release (Barnes et al., 1994Go). It has also been shown that the movement of axonal filopodia is strongly inhibited by glutamate and requires the presence of axonal AMPA/kainate glutamate receptors (Chang and De Camilli, 2001Go). Furthermore, functional GluR1 and GluR2 are expressed in axonal growth cones of hippocampal neurons (Martin et al., 1998Go; Hoshino et al., 2003Go), and a pool of presynaptic AMPAR subunits have also been isolated biochemically (Pinheiro et al., 2003Go; Schenk et al., 2003Go).

II. Trafficking of AMPA Receptors

A. Assembly

Each mature AMPAR is assembled from four individual subunits (Wu et al., 1996Go; Mano and Teichberg, 1998Go; Rosenmund et al., 1998Go; Safferling et al., 2001Go), although early work had indicated that the receptors may be pentamers (Ferrer-Montiel and Montal, 1996Go). Assembly is thought to occur through dimer pairing (Armstrong and Gouaux, 2000Go; Ayalon and Stern-Bach, 2001Go; Mansour et al., 2001Go; Robert et al., 2001Go). The early stages of assembly may be mediated by the proximal extracellular N-terminal domain of the subunits (Leuschner and Hoch, 1999Go; Ayalon and Stern-Bach, 2001Go; Horning and Mayer, 2004Go), but involvement of other regions, such as the membrane region and S2 portion, could also be necessary for formation of the mature tetrameric receptor (for differing viewpoints, see Wells et al., 2001Go; Pasternack et al., 2002Go).

Functional assembly of AMPAR subunits expressed in mammalian cells or Xenopus oocytes is selective, with no association with kainate or NMDAR subunits (Brose et al., 1994Go; Leuschner and Hoch, 1999Go), although coassembly with the orphan GluR{delta}2 with GluR1 and GluR2 has been reported (Kohda et al., 2003Go). The stoichiometry of the receptor complexes in these systems seems to be largely controlled by the expression levels of individual subunits, and it remains to be determined what processes govern assembly in neurons, especially since the type of complex can vary dramatically between synapses on a single neuron. Evidence that GluR2 is the preferred binding partner of GluR1 during heteromeric receptor assembly was uncovered by Mansour et al. (2001Go) in an elegant study involving physiological tagging of recombinant subunits and modeling of the results. They suggest that mature receptors are composed of a dimer of heteromers (i.e., that a GluR1/2 dimerizes with another GluR1/2), not a pair of homomers, and that they are arranged with identical subunits on opposite sides of the pore and not side by side, the stoichiometry and spatial arrangements of subunits affecting the phenotype. However, assembly of homomeric receptors is a stochastic process. Furthermore, recent studies in GluR2 knockout mice reported AMPAR complexes comprising abnormal heteromers of GluR1 and GluR3 as well as increased numbers of GluR1 and GluR3 homomers (Sans et al., 2003Go) with less efficient synaptic expression. This confirms that GluR2 is the preferred subunit partner in the assembly process and that it is important for synaptic expression of AMPARs.

There is considerable evidence that the subunit composition of functional AMPARs at specific synapses can change rapidly in response to synaptic activation (Shi et al., 2001Go; Liu and Cull-Candy, 2002Go; Lee et al., 2004Go), possibly due to targeted delivery of specific AMPAR complexes or subunit rearrangement of existing receptors in the spine. For example, it was observed that AMPARs containing only long C-terminal tails (i.e., GluR1) require plasticity-inducing synaptic activity for delivery to synapses, and those containing only short tails (i.e., GluR2/3) are constitutively expressed there (Ehrlich and Malinow, 2004Go). An attractive but as yet untested hypothesis is that binding of interacting proteins could favor the assembly of certain subunit combinations and prevent or disfavor others.

B. Visualizing AMPA Receptor Translocation

The isolation of green fluorescent protein (GFP) and its variants has enabled the experimenter to label cell structures and proteins for use in microscopy. For example, the morphology of neurons expressing GFP can be visualized using confocal microscopy, down to the shape and size of individual spines (Fig. 4). GFP-tagged AMPAR subunits may be expressed in neurons using a variety of techniques, and this approach has provided new insight into trafficking (Sheridan et al., 2002Go; Ashby et al., 2004Go), e.g., by viral transfer (Okada et al., 2001Go). For example, tetanic stimulation of hippocampal slice cultures causes the rapid NMDAR-dependent delivery of GFP-GluR1 into dendritic spines (Shi et al., 1999Go). GFP-labeled subunits have also been visualized in combination with electrophysiological tagging, where the channel rectification properties of recombinant AMPARs comprising specific subunits are altered by point mutations, e.g., GluR2 (R586Q)-GFP. Furthermore, expression of GFP-tagged AMPAR subunits in knockout mice has been used to "rescue" the wild-type phenotype, giving an insight into the cell biology of GluR1-containing heteromers (Mack et al., 2001Go). Using these approaches, it has been proposed that there are differential targeting mechanisms for AMPARs comprising either GluR1/GluR2 or GluR2/GluR3 subunit combinations (for example, see Hayashi et al., 2000Go). More specifically, GluR1/GluR2 receptors are added to synapses during plasticity via interactions between GluR1 and group I PDZ domain proteins, and CaMKII and LTP drive the synaptic expression of GluR1-containing AMPARs. In contrast, GluR2/GluR3 receptors replace existing synaptic receptors in a constitutive manner dependent on interactions between GluR2 with N-ethylmaleimide-sensitive factor (NSF) and group II PDZ domain proteins (Shi et al., 2001Go; Malinow and Malenka, 2002Go).



View larger version (38K):
[in this window]
[in a new window]
  		FIG. 4. 3D reconstruction of confocal image stack showing an area of spiny dendrite from a cultured pyramidal hippocampal neuron expressing a variant of GFP. The surface rendering of this reconstruction allows clear visualization of the varied structure of different spines (reproduced courtesy of Dr. M.C. Ashby, MRC Centre for Synaptic Plasticity, University of Bristol). Scale bar 4 µm.

 

Direct visualization of the intracellular transport of AMPARs using dispersed cultures of hippocampal neurons expressing GFP-GluR1 and GFP-GluR2 showed intracellular GFP-tagged AMPARs are widely distributed throughout the somatodendritic compartment (Perestenko and Henley, 2003Go). No evidence for intracellular clusters of receptors either in the soma or the dendrites was observed, suggesting the relatively free translocation of AMPARs in the soma and dendrites of neurons. AMPARs appear to be transported at rates comparable with fast axonal transport and move in a predominantly, but not exclusively, proximal to distal anterograde direction. Further data suggest that the intracellular transport of GFP-GluR1-containing AMPARs is not activity regulated but that various subunit combinations of the receptor complex are likely to be translocated throughout the soma and dendrites. Synapses may then "capture" these passing receptors as and when required (Perestenko and Henley, 2003Go). Furthermore, live imaging of GluR2 tagged with a pH-sensitive version of GFP that allows visualization of surface receptors only at the working (high) pH (pHlorin) demonstrated that extrasynaptic AMPARs containing the modified subunit react differently to the application of NMDA to synaptic receptors. The extrasynaptic receptors were internalized more quickly than those contained within the synapse (Ashby et al., 2004Go). This data confirms the model whereby extrasynaptic (or juxtasynaptic) receptors are more mobile than their synaptic counterparts, also demonstrated by particle tracking experiments (see below).

C. Lateral Diffusion of AMPA Receptors in the Membrane

Although the majority of interest in AMPAR dynamics has focused upon mechanisms of endo- and exocytosis, particularly with regard to plasticity, an increasing body of evidence suggests that diffusion within the postsynaptic membrane must account at least in part for some AMPAR movement (reviewed in Choquet and Triller, 2003Go). First, upon activation, AMPARs dissociate from their membrane anchors and diffuse away from the synapse prior to entering the constitutive endocytotic pathway, which is initiated either at the edge of the PSD or further afield in the extrasynaptic membrane (Zhou et al., 2001aGo). Second, studies involving stargazin suggest that the protein may recruit AMPARs from extrasynaptic to synaptic sites by binding synaptic PSD-95, then trapping freely diffusing AMPARs from outside the PSD (Chen et al., 2000Go; Schnell et al., 2002Go).

It is unsurprising that new research has centered on the kinetics of receptor movement during lateral diffusion in the neuronal plasma membrane. Interacting molecules outside, inside, and within the membrane can modulate receptor movement, and the subsynaptic cytoskeleton appears to be spatially organized into "corrals", for example, rafts that limit diffusion of receptors within certain boundaries (Choquet and Triller, 2003Go).

Recently, innovative imaging methods involving single-particle tracking have permitted the study of the movement of a single AMPA receptor in the plasma membrane by the attachment of a specific fluorophore-conjugated antibody (Borgdorff and Choquet, 2002Go). GluR2-containing receptors appear confined to the regions surrounding the synapse, and their movement is controlled by basal neuronal activity. GluR2 diffusion ceased after influx of calcium, such as that observed after membrane depolarization, and the receptor appeared tethered. Increased Ca2+ levels may therefore change receptor binding to scaffolding proteins, stabilizing the receptor. This method was used to demonstrate that a proportion of AMPARs are able to exchange rapidly between synaptic and juxtasynaptic sites and that this diffusion is regulated (Tardin et al., 2003Go). This process may also be involved in synaptic plasticity mechanisms (Groc et al., 2004Go). While this study concentrated on the GluR2 subunit, tracking of other AMPAR species would be invaluable for establishing if differential movement took place in line with evidence gained from other studies (Passafaro et al., 2001Go; Shi et al., 2001Go).

D. AMPA Receptor Delivery to Synapses

The mechanisms by which AMPARs are brought to and inserted at the postsynaptic membrane represent a major research interest for many workers in the field. In particular, the activity-dependent components of this regulation which underlie changes in synaptic plasticity are an intensely studied area of neuroscience. In essence, there are two basic processes by which AMPARs could be delivered to the correct postsynaptic location: direct exocytosis of receptors to the site of action, or insertion into the membrane at a separate location with subsequent diffusion to the PSD. In fact, evidence to date suggests that both mechanisms can occur.

Single-particle tracking and video microscopy of the lateral membrane mobility of native AMPARs containing GluR2 in rat-cultured hippocampal neurons revealed that AMPARs alternate between rapid diffusive and stationary behavior. In older neurons, the stationary periods increased in frequency and length and were usually associated with synaptic sites. Increasing intracellular calcium causes rapid receptor immobilization and local accumulation on the surface, suggesting that calcium influx can inhibit AMPAR diffusion and that lateral receptor diffusion to and from synapses is important for the regulation of receptor numbers at synapses (Borgdorff and Choquet, 2002Go; Choquet and Triller, 2003Go). Using different methods, it has also been reported that surface insertion of GluR1 occurs slowly in basal conditions and is stimulated by NMDA receptor activation, whereas GluR2 exocytosis is constitutively rapid. In addition, GluR1 and GluR2 show different spatial patterns of surface accumulation, consistent with GluR1 being inserted initially at extrasynaptic sites and GluR2 inserted more directly at synapses (Passafaro et al., 2001Go; Gomes et al., 2003Go; Sheng and Hyoung Lee, 2003Go).

E. AMPA Receptor Turnover at Synapses

In the past, it was generally accepted that AMPARs within the postsynaptic membrane are relatively static, at least under basal conditions, with a constitutive turnover of surface-expressed receptors in the order of hours to days (Archibald et al., 1998Go; Huh and Wenthold, 1999Go). However, electrophysiological recordings predicted a more rapid emergence of AMPARs to the cell surface, such as that seen during the acquisition of AMPARs at silent synapses (Luscher et al., 1999Go; Kim and Lisman, 2001Go). Evidence that receptor internalization may also occur with equal rapidity was initially gathered during experiments involving blockade of the GluR2-NSF interaction (Nishimune et al., 1998Go; Song et al., 1998Go). However, as set out below, it was discovered that GluR2-containing AMPARs undergo rapid NSF-dependent cycles of internalization and reinsertion into the postsynaptic membrane with a half-life in the order of a few minutes. The principle of rapid NSF-dependent recycling has also been extended to G-protein-coupled receptors via {beta}-arrestins (Miller and Lefkowitz, 2001Go) and GABAA receptors via GABA receptor-activating protein (Kittler et al., 2001Go), suggesting that this may be an important general regulatory synaptic mechanism.

The involvement of AMPAR interactors in receptor cycling has now been extensively studied (Ehlers, 2000Go; Osten et al., 2000Go; Ehrlich and Malinow, 2004Go; Lee et al., 2004Go; Nakagawa et al., 2004Go) and has had major implications for understanding the cellular processes underlying synaptic plasticity (for review, see Collingridge and Isaac, 2003Go; Collingridge et al., 2004Go; Lee et al., 2004Go). The application of glutamate to cultured hippocampal neurons causes a significant reduction in AMPARs, but not NMDARs from synaptic sites (Lissin et al., 1999Go), supporting the hypothesis that dynamic movement of AMPARs occurs in response to spontaneous synaptic activity. Subsequent work demonstrated that loss of AMPARs from synapses can also be promoted by NMDA, AMPA, insulin, or mGluR receptor activation (Carroll et al., 1999Go; Lin et al., 2000Go; Snyder et al., 2001Go; Ehlers, 2003Go; Lee et al., 2004Go).

Several studies have focused on identifying the molecular basis of AMPAR cycling at the postsynaptic membrane. Increasing synaptic activity has been shown to cause a decrease in the number of surface-expressed AMPARs and size of AMPAR clusters in cultured neurons (Lissin et al., 1999Go). The internalization of GluR2-containing AMPARs and subsequent targeting for lysosomal degradation can be triggered by NMDA receptor activation and is mediated through the formation of clathrin-coated pits (Carroll et al., 1999Go; Ehlers, 2000Go; Lee et al., 2004Go). Furthermore, GluR2-containing receptors are internalized and recycled in response to the application of AMPA (Lee et al., 2004Go), but probably only in a subset of synapses as miniature excitatory postsynaptic currents decrease in frequency but not size (Carroll et al., 1999Go; Lissin et al., 1999