Over the last years, several AMPAR interacting proteins have been identified. Most of them are cytosolic proteins binding GluR2 C-terminal tail. ABP, GRIP, and PICK1 are PDZ-containing proteins that interact with the last four amino acids of GluR2 subunit.
Schematically, ABP/ GRIP is concentrated at synaptic plasma membrane or in intracellular compartments, and could retain AMPA receptors at these sites. GluR2 phosphorylation by PKC uncouples the receptor from ABP/GRIP anchors. Phosphorylated AMPARs still bind PICK1 and could be trafficked between synapses and intracellular compartments changing synaptic transmission efficacy.
Expression of Stargazin lacking the PDZ binding site rescues surface delivery but not synaptic clustering of AMPAR.
Stargazin has a PDZ binding site at its C terminus that associates with SAP102, and PSD-95/93 MAGUKs. TARPS are associated with AMPARs early in the synthetic pathway and control their maturation, trafficking, and biophysical properties. First, TARPs are involved in folding and assembly of AMPAR, stabilizing and facilitating their export from the ER. Second, Stargazin promotes AMPAR surface expression. Third, TARPs are critical for clustering AMPAR at excitatory synapses through their interaction with PSD-95 (and other MAGUKs), a major component of the postsynaptic scaffold.
PSD-95 over-expression in hippocampal slices enhances specifically synaptic AMPAR-mediated response without changing the number of surface AMPAR. Conversely, Stargazin overexpression increases selectively the number of extrasynaptic AMPAR without changing AMPAR-mediated synaptic currents. These observations indicate that the Stargazin/PSD-95 interaction is involved in the stabilization of AMPARs at synapses.
AMPAR Surface Diffusion Is Decreased on PSD-95 Clusters
PSD95 colocalizes with vGlut1 and Homer
We generally observed that rapidly diffusing (GluR1-containing) AMPARs located in the extrasynaptic membrane (outside PSD-95 clusters) became less mobile when they reached and colocalized with a PSD-95 cluster (Movie S1)
The fraction of immobile AMPARs was 4-fold higher inside compared to outside PSD-95 clusters
AMPAR Clustering Requires the PDZ Binding Site of Stargazin
Used Stargazin-GFP constructs in which the last C-terminal four amino acids corresponding to the PDZ binding site were removed.
When expressed in COS-7 cells, Stargazin WT, but not Stargazin DC, allowed PSD-95-induced GluR2 surface clustering
This indicates that the PDZ binding site of Stargazin is required to cluster AMPAR with PSD-95 in heterologous cells.
we measured miniature synaptic currents in neurons transfected for 24–48 hr either with Stargazin WT::GFP or Stargazin DC::GFP constructs
the mEPSC frequency of Stargazin DC neurons was greatly decreased compared to untransfected and WT
the mEPSC amplitude was significantly decreased in comparison to untransfected neurons
by performing an immunostaining of surface AMPARs in neurons expressing Stargazin DC, we observed a large decrease in receptor clustering at synaptic sites
All together, these results indicate that the PDZ motif of Stargazin that binds PSD-95 is important for the accumulation of surface AMPARs at synapses
AMPAR Diffusion Is Increased at the Surface of Stargazin DC-Expressing Neurons
Diffusing surface AMPARs are stabilized on PSD-95 clusters and the binding of Stargazin to its PDZ-containing partners, such as PSD-95, is critical to cluster AMPARs within synapses.
we compared the diffusion coefficient distributions of GluR1-containing and GluR2-containing AMPARs from control neurons, WT, and Stargazin DC::GFP expressing neurons. The distributions of the diffusion coefficient from GluR1- containing and GluR2-containing AMPARs were similar
the fraction of immobile GluR1-containing and GluR2-containing AMPARs in Stargazin DC-expressing neurons significantly decreased when compared to controls, but there was no change in diffusion rate of already mobile AMPARs.
These results indicate that Stargazin regulates mainly the immobilization of surface AMPARs rather than their mobility per se
Moreover, the relative percentage of time spent by each AMPAR in a state of confined diffusion dropped in Stargazin DC-expressing neurons when compared to control indicating that Stargazin participates in the confinement of AMPAR in restricted area.
In conclusion, AMPAR surface diffusion is modulated by the binding of Stargazin to PDZ-containing scaffold proteins
AMPAR Mobility Is Increased at Synaptic Sites by Stargazin DC Overexpression
First, we found that the fraction of immobile (GluR1) receptors was decreased at both extrasynaptic and synaptic sites in neurons expressing Stargazin DC as compared to Stargazin WT
Second, the median diffusion coefficients of the mobile receptors remained unchanged in all conditions and compartments
Third, the amount of time spent by receptors at synapses was strongly decreased in cells expressing Stargazin DC
Finally, we extended our analysis to older neurons (15–20 DIV). In these neurons, surface AMPARs can be trapped reversibly at spiny synapses
On the one hand, the median diffusion coefficient of GluR1 containing synaptic receptors was significantly lower in 15–20 DIV neurons than in 8–10 DIV neurons (as we previously showed for GluR2)
On the other hand, Stargazin DC overexpression increased GluR1 mobility specifically at synaptic and not extrasynaptic sites
Altogether, these results indicate that Stargazin interaction with proteins containing PDZ domains is involved in (1) the immobilization of GluR1 AMPAR within the synaptic membrane and (2) the developmental increase in GluR1 AMPARs trapping at synapses, in agreement with the rise in Stargazin and PSD93/95 expression during development
The PDZ-Binding Site of GluR2 Controls Its Surface Expression but Not Its Lateral Mobility
Given the striking role of Stargazin C terminus in controlling AMPAR surface diffusion, we wondered if AMPAR subunits C termini had any role in controlling surface movements. The direct interaction of GluR2 C terminus with the PDZ-containing proteins ABP/GRIP and PICK1 has been shown to play an important role in the regulation of AMPARs expression at synaptic sites. Whether these proteins are involved solely in modulating the surface expression of the AMPARs or whether they also anchor surface AMPARs at synapse, however, remains unclear.
We first used a mutant GluR2, GluR2-DC, in which the last C-terminal four amino acids corresponding to the PDZ binding site were removed.
We compared the surface expression of GluR2 DC::GFP and wild-type GluR2 in cultured hippocampal neurons. Since the GFP tag is coupled to the extracellular N terminus of GluR2, the surface receptors could be specifically immunolabeled with an anti-GFP. The signal coming from this surface staining was normalized to that of the signal of the GFP, which corresponds to the total intracellular and surface expression of the recombinant protein. Note: Interesting! Never heard of this methodology for determining the proportion of receptors expressed at the membrane vs intracellular
GluR2 surface expression was reduced by half when its PDZ binding site was deleted.
GluR2 DC still colocalized with Homer1c so, while GluR2 DC is less expressed at the neuronal membrane, it's still clustered at excitatory synapses.
To investigate the role of GluR2 PDZ interactors in controlling GluR2 lateral mobility, we tracked in real time the movement of GluR2:WT:GFP or GluR2:DC:GFP at the neuronal surface using QDots coupled to anti-GFP. Note: This thing is a monster... GluR2:DC ↔ GFP ↔ anti-GFP ↔ Qdot Remember from their 2004 paper, they were already reporting a decrease in diffusion rate, specifically at the synapse, when they compared Qdots and at the synapse compared to a fluorescent marker
diffusion of GluR2 were not changed by the deletion of the PDZ binding site. Indeed, the fraction of immobile receptors, percentage time in confined sites, and the median diffusion coefficients of mobile receptors were similar for GluR2:WT:GFP and GluR2:DC:GFP
We analyzed receptor movements according to their synaptic or extrasynaptic location, and did not detect any difference between GluR2 DC and control diffusion (neither the fraction of immobile receptors nor median diffusion rate of mobile receptors)
Furthermore, the mean time spent within synapses was unchanged by the deletion of the PDZ-binding motif (results confirmed with SVKI manipulation)
Altogether, these results show that, in resting conditions, PDZ proteins interacting with GluR2 C terminus are mainly involved in the regulation of GluR2 surface expression but not in its trapping at synaptic sites
AMPAR Surface Diffusion Is Modulated by PSD-95/ Stargazin Interaction
To specifically investigate whether the PSD-95/Stargazin interaction modulates AMPAR surface diffusion, we used PSD-95/Stagazin compensatory mutants where the interaction between the PDZ domain and its ligand is converted from class I to class II (Schnell et al., 2002). Schematically, the Stargazin mutant (StargazinT321F) can only interact with the compensatory mutant of PSD- 95 (PSD-95H225V) and not with the native PSD-95.
StargazinT321F::GFP alone displayed a uniform distribution and did not coaggregate with v-Glut1 clusters. However, expression of both StargazinT321F::GFP and PSD-95H225V relocated StargazinT321F::GFP clusters to synaptic sites
Thus, as previously shown (Schnell et al., 2002), the synaptic targeting of Stargazin is dependent on the presence of synaptic PSD-95
Regarding GluR2-AMPAR surface trafficking, there was far less immobile GluR2 in StargazinT321F neurons compared to controls and StargazinT321F/PSD-95H225V expressing neurons.
The diffusion coefficient of the mobile GluR2s were not affected in all of the conditions, consistent with a role of the Stargazin/PSD-95 interaction in the immobilization of surface GluR2-containing AMPARs rather than in the receptor mobility
The diffusion coefficient of the mobile GluR2-containing AMPARs was not significantly affected in all of the conditions, consistent with a role of the Stargazin/PSD-95 interaction in the immobilization of surface GluR2-containing AMPARs rather than in the receptor mobility
Similar results for the surface trafficking were obtained for GluR1-containing AMPARs (data not shown).
As expected the mobility of the receptors was changed on StargazinT321F/ PSD-95H225V clusters, immobilization being increased and and median diffusion being reduced
Thus, these results indicate the critical role of the specific interaction between Stargazin and PSD-95 in stabilizing AMPAR in neuronal membrane
Stargazin and AMPA Receptors Diffuse as Complexes in the Neuronal Membrane
We then investigated the dynamic of AMPAR/Stargazin/ PSD-95 complexes
Using anti-HA-coupled QD, we first followed Stargazin surface movements in neurons coexpressing Stargazin::HA and PSD-95::GFP and measured Stargazin diffusion according to its localization with respect to PSD-95 clusters. Freely diffusing extrasynaptic Stargazin was reversibly stabilized on PSD-95::GFP clusters
Accordingly, on PSD-95 clusters, the fraction of immobile Stargazin was increased and the median diffusion of mobile Stargazin was decreased
It should be noted that the diffusion properties of Stargazin were modified on PSD-95 clusters to the same extent as those of AMPARs.
However, AMPAR could diffuse out of synapses due to unbinding from Stargazin or to unbinding of Stargazin from PSD-95. To distinguish between these alternatives, we studied the effect of crosslinking induced GluR2 immobilization on Stargazin::GFP diffusion using FRAP.
Neurons were cotransfected with Stargazin::GFP and an extracellularly TdimerDsRed-tagged GluR2. We incubated neurons with excess anti-DsRed antibody to specifically crosslink GluR2::TdimerDsRed. Such a treatment immobilizes surface expressed AMPARs.
For FRAP analysis, we selected two types of regions, containing either scattered or clustered Stargazin::GFP. Stargazin clusters are most likely synaptic, 76% colocalized with Homer1c.
We first measured the recovery of the fluorescence signal after the photobleaching of Stargazin::GFP in control condition (without antibody). Consistent with the results obtained with single quantum dots tracking, the fluorescence recovery was slower and occurred to a lower extent
Altogether, these data strongly suggest that AMPAR and Stargazin diffuse as complexes in both synaptic and extrasynaptic plasma membrane
It should be noted that a fraction of immobile AMPARs was not localized on PSD-95 clusters, possibly due to the existence of a small subset of synapses that lack PSD-95 but express the Stargazin interacting protein PSD-93, as seen in vivo (Elias et al., 2006). Consistently, we observed few excitatory terminals not associated with a PSD-95 immunostaining (see Figure S1), but we could not explore this heterogeneity further in our cultured hippocampal neurons since the anti-PSD-95 antibody (clone 7E3-1B8) we used slightly crossreacts with PSD- 93 (Sans et al., 2000).
None of the AMPAR subunits bind directly PSD-95. Among the several postsynaptic proteins that interact with AMPARs and which then may serve a link to PSD- 95, Stargazin and the other members of the TARP family have emerged as key partners for AMPAR trafficking
Stargazin overexpression increases selectively the number of extrasynaptic AMPARs without changing AMPARs mediated synaptic currents, but its interaction with PSD-95 is critical for clustering AMPARs at excitatory synapses
Third, other scaffolding proteins, such as SAP-97 or NSF, interact with specific AMPAR subunits
Fourth, the neuronal pentraxin NARP and NP1 are enriched at excitatory synapses and interact directly with all of the four AMPAR subunits inducing AMPARs surface clustering. NARP and NP1 could thus act as AMPARs stabilizing extracellular factors.
It should be noted that we observed immobile receptors at extrasynaptic sites (outside Homer 1c::TdimerDsRed clusters), some of them being released by Stargazin DC expression. These receptors could be trapped by extrasynaptic clusters of PSD-95 or other MAGUKs interacting with Stargazin such as SAP-102 and PSD-93
the remaining fluorescence recovery observed during our experiments suggests that a small fraction of Stargazin can diffuse alone in the neuronal membrane. In support of this observation, biochemical data have shown that the interaction between TARP proteins and AMPARs can be disrupted by glutamate (Tomita et al., 2004), demonstrating that under certain conditions AMPARs and Stargazin can be trafficked independently
PSD-95 has a rather slow turnover at synapses, in the order of 25% over 5 min (Okabe et al., 2001; Sharma et al., 2006), a value which is much slower than the one we found for Stargazin (25% in 30 s). This suggests that the reversible link that allows AMPARs to traffic in and out synapses is mostly the Stargazin-PSD-95 interaction
This could suggest that the interaction of Stargazin with SAP102 and then with increasing level of PSD-95/93 is involved in the higher trapping efficiency of AMPAR at mature synapses.
Stargazin interaction with PSD- 95 can be modulated by phosphorylation (Chetkovich et al., 2002). The PKA phosphorylation of Stargazin C terminus prevents Stargazin binding to PSD-95 (Chetkovich et al., 2002). Furthermore, Stargazin Cter tail is quantitatively phosphorylated on a set of serine residues. Phosphorylation and dephosphorylation of Stargazin are regulated by NMDAR activity and necessary for LTP and LTD of hippocampal synaptic transmission, respectively (Tomita et al., 2005). It will be of interest to determine how these processes regulate AMPARs surface trafficking to and from synapses.