The interaction of NSF with the C-terminus of the GluA2 subunit of the AMPA receptor was first uncovered by yeast two-hybrid screening and was quickly shown to be an important factor in the regulation of AMPA receptor surface expression (Nishimune et al 1998, Noel et al 1999) as well as LTD in the hippocampus (Luthi et al 1999) and cerebellum (Steinberg et al 2004). The function of the NSF-GluA2 interaction seems to be to stabilise GluA2-containing AMPA receptors when expressed at the synaptic surface (Braithwaite et al 2002). Interestingly, the binding site for NSF on the GluA2 C-terminus has been shown to overlap with that for AP-2, an adaptor protein that is part of the clathrin-coated internalisation mechanism (Lee et al 2002). Thus a competitive interaction between NSF and AP-2 could provide a mechanism by which internalisation of AMPA receptors is initiated during NMDAR-mediated LTD. Recent work here at the MRC Centre has shown that a member of the Neuronal Calcium Sensor family of proteins, , provides a link between Ca2+ influx via NMDA receptors and clathrin-mediated internalisation of AMPA receptors (Palmer et al 2005).
PICK1 is a protein that was first identified as interacting with PKCalpha via a PDZ domain (protein-protein interaction motifs). It has been shown to interact with a number of glutamate receptors - GluA2 and GluA3 AMPA receptor subunits (Dev et al 1999, Xia et al 1999), the GluK1 and GluK2 subuits of the kainate receptor (Hirbec et al 2003) and the mGlu7 receptor (Dev et al 2000) via their extreme C-terminal PDZ binding motifs.
In addition to clustering AMPA receptors, via a dimerisation domain, PICK1 has been shown to have multiple effects in neurons including roles in the insertion of AMPA receptors (Daw et al 2000), the internalisation of kainate receptors (Hirbec et al 2003) and the regulation of AMPA receptor subunit composition (Terashima et al 2004).
Further evidence of the complex functions of this protein comes from recent work carried out here at the MRC Centre (Hanley & Henley 2005). A Ca2+ binding domain has been identified at the extreme N-terminus that controls the strength of the interaction between PICK1 and GluA2. From this work PICK1 seems to act as a trigger, mediating some of the effects of NMDA receptors on AMPA recptor trafficking, via a localised Ca2+ influx. In addition, it may also play a role in initiating membrane invagination via the BAR domain. This domain forms a curved structure on dimerisation that can deform membranes or stabilise already curved membranes.
GRIP and ABP are multi-PDZ domain proteins that bind to the GluR2 subunit of the AMPA receptor. GRIP binds to the GluA2 subunit but not to GluA1 or 4 (Dong et al 1997), as well as many other proteins (Hirbec et al 2003). Different protein binding partners bind to different PDZ domains within GRIP. For intance, GluA2 binds to PDZ domains 4 and 5 (Dong et al 1997), while ephrin receptors bind to domains 6 and 7 (Torres et al 1998). In addition to receptors, GRIP has also been shown to bind to KIF-5, a kinesin motor protein that is crucial in the transport of AMPA receptors within the neuron (Setou et al 2002).
ABP is a protein closely related to GRIP and exists in two isoforms with 6 and 7 PDZ domains, respectively. The shorter isoform binds to GluA2 via PDZ domains 3, 5 and 6 (Srivastava et al 1998). This isoform and GRIP are functionally indistinguishable from each other. The PDZ domain 2 of ABP can also mediate homodimerisation or heterodimerisation with GRIP.
The functions of ABP/GRIP appear to be many and varied. Given the multiple PDZ domain structure and the ability to interact with a wide range of different proteins, an obvious primary function would appear to be as a scaffolding and targetting protein, taking receptors and other proteins to their final destinations within synapses and anchoring them there. However, the binding of GluA2 to ABP/GRIP is in dynamic equilibrium with PICK1, due to the phosphorylation status of Ser880 (Matsuda et al 1999, Chung et al 2000). When this residue is phosphorylated by PKC, GluA2 binding to GRIP is much weaker, but binding to PICK1 is unaffected. Combined with the intracellular location of the GluA2-ABP/GRIP complex (Daw et al 2000, Braithwaite et al 2002), this allows for a mechanism for GluA2 to be released from GRIP and made available for insertion into the synaptic membrane and for internalised receptors to be captured.
SAP97 is another multi-domain structural protein that interacts with AMPA receptors, but via the GluA1 subunit (Leonard et al 1998). It belongs to the MAGUK family of proteins (membrane-associated guanylate kinase homologues), that also includes SAP90, PSD-95 and many others (for review, see Montgomery et al 2004). These proteins have three PDZ domains, a Src Homolgy 3 (SH3) domain and a guanylate kinase (GUK) domain. The interaction with GluA1 is predominantely is via the PDZ2 domain and unusually requires two binding motifs on the GluA1 C-terminus; a typical PDZ binding sequence on the extreme C-terminus (T-X-L) and a second tri-peptide sequence centred at -10 (SSG; Cai et al 2002, y).
The trafficking function of SAP97 appears to be similar for GluA1 as that of ABP/GRIP for GluA2. The interaction occurs principally in the biosynthetic and secretory pathways (Sans et al 2001) and SAP97 provides a linkage between the AMPA receptor subunit and the microtubule-based transport mechanisms, via an interaction with the motor protein myosin VI (Wu et al 2002). More recently, SAP97 has been shown to be directed into spines under the control of CaMKII phosphorylation (Mauceri et al 2004). In this way, SAP97 could potentially deliver GluA1 containing AMPA receptors to dendritic spines.
In addition to trafficking functions, SAP97 may also paly a role in the acute functional regulation of GluA1-containing AMPA receptors. The SH3 and GUK domains provide a binding site for AKAP79 (A-kinase associated protein 79). This protein in turn interacts with both PKA and PP2B (Protein Kinase A and Protein Phosphatase 2B; (College et al 2000, Tavalin et al 2002)) that between them control the phosphorylation state of Ser845 on the GluA1 subunit. This is turn regulates AMPA receptor currents, which are enhanced on phosphorylation of Ser845. Thus the balance of PKA and PP2B activity can determine the functional status of GluA1-containing AMPA receptors.
PSD95 (along with the closely related PSD93) is the best characterised of the MAGUK proteins, with the typical structure of 3 PDZ domains, a SH3 domain and a GUK domain. It performs a scaffolding role in the post-synaptic density, providing anchorage points for many different proteins, including AMPA receptors. However, unlike SAP97 or ABP/GRIP, it does not interact directly with these receptors, rather it binds TARPs (Transmembrane AMPA receptor Regulatory Proteins eg stargazin), which in turn interact with the AMPA receptors (Fukata et al 2005). A recent study has demonstrated that PSD95 and PSD93 jointly control the synaptic targetting of AMPA receptors in mature synapses (Elias 2006). Acute knockdown of either protein using shRNAs resulted in an approximately 50% reduction in AMPA receptor-mediated EPSCs , and ~75% loss if both proteins were knocked out simultaneously. These results contrast with traditional knockout mice, where compensatory mechanisms mean that knockout of either PDS95 or PDS93 have no effect on synaptic transmission (Elias 2006).
SAP102 is another closely related MAGUK with a similar domain structure to PSD95/93. In contrast to these proteins, it is expressed at high levels early in development (Sans et al 2000) and appears to be the principally responsible for the synaptic targetting of AMPARs at this time (Elias 2006). In addition, it may be upregulated following knockout of PSD95 and PSD93, thus compensating for their loss.
TARPs are a family of related proteins that associate with AMPA receptors. The best known and characterised is , the protein that is mutated in the stargazer mouse (Noebels et al 1990). Stargazin, and other TARPs, has been characterised as a gamma-subunit of a voltage-gated calcium channel (Letts et al 1998) and as such has four transmembrane domains with intracellular termini. A PDZ binding site on the extreme C-terminus provides a linkage to PDZ domain proteins such as PDS95 and SAP97 (Chen et al 2000). Interestingly, the microtubule-associated protein light chain 2 is also a strong interactor, though not via a PDZ domain (Ives et al 2004). These interactions are likely to be critical in stargazin's role in the tafficking of AMPA receptors.
The interaction between stargazin and PSD95 is a very important one in determiniung the level of synaptic AMPA receptor surface expression. Over-expression of stargazin leads to increases of extra-synaptic AMPA receptors whilst over-expression of PSD95 leads to increased recruitment of synaptic receptors, but in neither case is synaptic function altered (Schnell et al 2002). Thus stargazin is critical in the trafficking of AMPA receptor subunits to the synapse, via an interaction with SAP97, and is then also critical in determining the level of surface expressed receptor within the synapse via an interaction with PSD95. Indeed, recent work has demonstrated a role for stargazin in the early biosynthetic pathway of the GluA1 subunit (Vandenberghe et al 2005).
In addition to roles in the trafficking and localisation pathways, stargazin is also important as an acute modulator of AMPA receptor function. It has been shown to enhance AMPA receptor mediated currents (Yamakazi et al 2004), affect desensitisation kinetics and agonist responsiveness (Turetsky et al 2005, Priel et al 2005) and mediate AMPA receptor induced NMDA receptor clustering (Mi et al 2005).
Protein 4.1R is a cytoskeletal protein first identified in erythrocytes and the neuronal homologues, 4.1N and 4.1G have been shown to interact with the GluA1 subunit (Shen et al 2000). 4.1R is critical for the organisation and maintenance of the spectrin-sctin cytoskeleton in erythrocytes, via intereactions with intregral membrane proteins. It seems likely that 4.1N/G play a similar role in neurons, linking the GluA1 subunit to the cytoskeleton and helping to stabilise it in the membrane. More recently, these proteins have also been shown to interact with the GluA4 subunit (Coleman et al 2003) in a similar manner.