AMPA receptor (AMPAR) abundance and positioning at excitatory synapses regulates the strength of transmission. Changes in AMPAR localisation can enact synaptic plasticity, allowing long-term information storage, and is therefore tightly controlled. Multiple mechanisms regulating AMPAR synaptic anchoring have been described, but with limited coherence or comparison between reports, our understanding of this process is unclear. Here, combining synaptic recordings from mouse hippocampal slices and super-resolution imaging in dissociated cultures, we compare the contributions of three AMPAR interaction domains controlling transmission at hippocampal CA1 synapses. We show that the AMPAR C-termini play only a modulatory role, whereas the extracellular N-terminal domain (NTD) and PDZ interactions of the auxiliary subunit TARP γ8 are both crucial, and each is sufficient to maintain transmission. Our data support a model in which γ8 accumulates AMPARs at the postsynaptic density, where the NTD further tunes their positioning. This interplay between cytosolic (TARP γ8) and synaptic cleft (NTD) interactions provides versatility to regulate synaptic transmission and plasticity.
Excitatory synaptic transmission is primarily mediated by AMPARs1,2. These glutamate-gated cation channels are concentrated at the postsynaptic density (PSD) to mediate fast neuronal communication. This receptor is also central to synaptic plasticity: activity-dependent changes in the abundance of synaptic AMPARs can bi-directionally modify the strength of transmission, leading to either long-term potentiation (LTP) or long-term depression (LTD). For this reason, the mechanisms controlling the synaptic localization of AMPARs have been intensively investigated for decades3,4.
AMPARs are exocytosed to the cell surface and enter the synapse through lateral diffusion, where they are trapped in the PSD5,6,7. There, they concentrate into subsynaptic ‘nanoclusters’ of higher density8,9, which have been suggested to align with presynaptic vesicle release sites for efficient transmission10,11,12.
AMPAR complexes are assembled from four core subunits and various auxiliary proteins13. At hippocampal CA1 synapses, GluA1/2 heteromers predominate and principally associate with the auxiliary subunit TARP γ8 (Transmembrane AMPAR Regulatory Protein, Fig. 1a)14,15,16. TARPs bind to the major PSD components PSD-93/95, through PDZ (PSD-95, Dlg, ZO-1) interactions of their extreme C-terminus17,18,19,20. This interaction is currently the best-characterised mechanism of AMPAR synaptic localisation.
The sequence-diverse AMPAR C-terminal domains (CTDs) have multiple phosphorylation and interaction sites for cytosolic proteins (Figs. 1b, 2a)21, and were central to previous synaptic anchoring models22,23. The GluA2 CTD interacts with NSF (N-ethylmaleimide sensitive fusion protein) and the scaffolding proteins GRIP1/ABP and PICK1, with suggested roles in both surface delivery and synaptic anchoring, while the GluA1 CTD interacts with SAP9722,23. It was proposed that the GluA1 CTD mediates activity-dependent AMPAR trafficking in LTP, while the GluA2 CTD drives constitutive synaptic delivery24,25. This model has since been debated, and the role of AMPAR CTDs remains unclear26,27.
Recently, we and others demonstrated a role for the extracellular AMPAR N-terminal domain (NTD; Fig. 1a) in synaptic transmission and plasticity28,29. We observed an essential requirement for the GluA1 NTD in anchoring at the CA1 synapse for transmission, while GluA2 could contribute to synaptic transmission in the absence of NTD interactions; however, the magnitude of transmission was strongly reduced. The GluA1 NTD was also essential for the maintenance of synaptic potentiation. In addition, using NTD-swap constructs, we demonstrated that subunit-specific interactions of the NTD cause a greater contribution of GluA2 than GluA1 receptors to synaptic transmission28. These results have been independently reported by others29,30. The NTD, sequence-diverse like the CTD, offers subunit-specific synaptic anchoring through interactions with synaptic cleft proteins and has been implicated previously in receptor clustering by neuronal pentraxins31,32.
However, the plethora of proposed interactions and absence of comparison between reports has obscured the core mechanisms of AMPAR synaptic anchoring. Here, we perform a comprehensive comparative analysis of the three major AMPAR interactions, CTD, NTD and TARP γ8 PDZ, to clarify their roles in CA1 synaptic transmission. We demonstrate that the AMPAR CTDs are not required for synaptic anchoring but play a modulatory role in this process. AMPAR transmission is primarily reliant on intracellular interactions of γ8 with the PSD to accumulate receptors at synaptic sites, and on subunit-specific NTD interactions in the synaptic cleft. An interplay of these two core mechanisms defines the strength of transmission and facilitates the changes that underlie synaptic plasticity.
Mutations in the GluA2 CTD do not affect synaptic anchoring
We first set out to clarify the role of the GluA1 and GuA2 CTDs, using single-cell electroporation to express exogenous AMPAR subunits in hippocampal organotypic slices, and dual recordings from transfected and untransfected CA1 pyramidal neurons (Fig. 1c; see Supplementary Figure 1a for experimental details)28. We previously demonstrated that NTD deletion from either GluA1 or GluA2 causes substantial deficits in synaptic transmission using this assay.
Expression of GluA2, unedited at the Q/R site in the channel pore (GluA2Q), provides a read-out for synaptic localisation, where a change in the rectification index (RI) indicates that exogenous (rectifying homomeric) receptors have replaced endogenous (nonrectifying heteromeric) receptors24,28. In dual recordings, comparison of the EPSC amplitude detects changes in the number of synaptic receptors, while changes in the RI determines the proportion of exogenously expressed synaptic receptors. Exogenous GluA2Q reaches the Schaffer collateral to CA1 synapse, causing a strong reduction in synaptic RI and an increase in AMPAR EPSC amplitudes28 (Fig. 1d–e; Supplementary Figure 1f). We assessed whether CTD interactions were necessary for this effect by mutating known CTD interaction sites on the GluA2Q construct (Fig. 1b; Supplementary Figure 1b)22. These include mutating the NSF interaction site (NP830AA25), blocking PDZ-interactions with ABP/GRIP, or PICK1 (addition of Y86325), exchanging the intracellular domains (ICD - CTD and loop 1) of GluA2 with that of GluA1 to prevent any interactions that could mediate constitutive GluA2 delivery25, and simultaneously mutating major protein interaction and phosphorylation sites in the GluA2 CTD22 (CTD-null). All constructs trafficked to the cell surface, as measured by RI changes in somatic outside-out patches (Supplementary Figure 1c–d). However, none of these mutations altered or prevented synaptic expression of GluA2Q. All constructs reduced the synaptic RI (Fig. 1d) and increased EPSC amplitudes similarly to full-length GluA2Q (Fig. 1e), suggesting that GluA2 CTD interactions are not essential for AMPAR synaptic delivery.
CTD interactions modulate synaptic localisation of GluA2 ΔNTD
We previously reported that removal of the GluA2 NTD (GluA2 ΔNTD) limited AMPAR anchoring at the synapse28. GluA2Q ΔNTD receptors (unedited at the Q/R site) were present at synapses, as measured by a change in the synaptic RI; however, the amplitude of evoked EPSCs was decreased by ~50% relative to untransfected cells (Fig. 1f–g; Supplementary Figure 1g). This phenotype suggests that while AMPAR anchoring is impaired by NTD removal, GluA2Q ΔNTD also prevents endogenous receptors from maintaining full transmission, likely by competing for limiting interactors. Mutating the GluA2 ΔNTD at important protein interaction sites could alleviate this competition and the corresponding depression of synaptic transmission.
As exogenous receptors showed no change in auxiliary subunit association in comparison to endogenous receptors28, we assessed whether CTD-interactions cause this effect. Loss of critical protein interactions should alleviate competition with endogenous AMPARs, rescuing both rectification and EPSC depression. Surface trafficking of all GluA2 ΔNTD CTD mutants was unimpaired (Supplementary Figure 1e), demonstrating functional tetrameric receptor formation. Neither mutation of the NSF interaction site, nor blocking PDZ-interactions completely prevented GluA2Q ΔNTD detection at synapses, yet NSF mutation partially alleviated the magnitude of both the RI and EPSC amplitude changes (Fig. 1f–g), implicating a role for this interaction in synaptic delivery. Replacing the ICD of GluA2Q ΔNTD with that of GluA1 also reduced the effect of GluA2Q ΔNTD on synaptic transmission (Fig. 1f–g), further supporting a modulatory role for the GluA2 CTD in receptor synaptic delivery. However, GluA2Q ΔNTD CTD-null caused a decrease in both synaptic RI and EPSC amplitudes, equivalent to GluA2 ΔNTD (Fig. 1f–g), indicating that we are yet to fully understand the ways by which GluA2 CTD interacting proteins exert their function.
While receptor overexpression has the advantage of allowing comparison between the anchoring of exogenous and endogenous receptors, total AMPAR levels will be increased. To ensure that overexpression does not confound our interpretation of CTD interactions, we sought to corroborate these findings using a knockout and rescue approach. By neonatal injection of AAV expressing Cre-EGFP in conditional AMPAR knockout mice (Gria1-3[fl/fl 14), we produced organotypic slices with a mosaic of unmodified and AMPAR-null CA1 neurons. Fourteen to twenty days after Cre transduction, EGFP + cells showed virtually no excitatory synaptic transmission at −60 mV holding potential, demonstrating successful knockout of all AMPAR subunits expressed in these neurons (Supplementary Figure 2a). The remaining ~5–10% of current was NMDAR-meditated29, and NMDAR currents were unaffected by AMPAR deletion when measured at +40 mV holding potential (Supplementary Figure 2a). On this AMPAR-null background, we restored specific AMPAR complexes by single-cell electroporation. AMPAR rescue could be achieved by expressing either GluA2Q or GluA2Q CTD-null (Supplementary Figure 2b–c, f). In addition, both GluA2 ΔNTD and GluA2 ΔNTD CTD-null also showed similar levels of synaptic current rescue (Supplementary Figure 2d–f). Together these data support previous knockout and rescue studies26,33, and further demonstrate that CTD interactions do not play a critical role in synaptic anchoring of GluA2.
GluA1 CTD interactions are not essential for synaptic anchoring
We extended this analysis to GluA1, which is associated with activity-dependent AMPAR trafficking24,34,35,36. When expressed alone, exogenous GluA1 constitutively trafficked to synaptic sites, as evidenced by a change in the synaptic RI (Fig. 2b, see also Supplementary Figure 3b)28,29. Previous work highlighted an essential role for the GluA1 PDZ ligand in synaptic potentiation24,25, therefore we mutated this site in GluA1 (T887A) (Fig. 2a). PDZ mutation did not prevent trafficking to the cell surface (Supplementary Figure 3a) or synapse (Fig. 2b), and co-mutation of the PDZ-ligand and two CTD phosphorylation sites (Fig. 2a)37, also failed to inhibit GluA1 synaptic localisation (Supplementary Figure 3c). Thus, like for GluA2 full-length receptors, we observe no essential role for the GluA1 CTD in constitutive synaptic localisation. Constitutive trafficking is also not dependent on slice activity, as chronic treatment with the Na-channel blocker tetrodotoxin (TTX) did not prevent GluA1-dependent changes to synaptic RI. However, N-terminally tagging GluA1 with GFP prevented it from contributing to transmission (Supplementary Figure 3d), in line with previous observations25,26, possibly due to masking of NTD-interaction sites.
We next assessed the PDZ-dependence of GluA1 in synaptic potentiation, by co-expressing constitutively active CaMKII (truncated [t]CaMKII)24. Similarly to full-length GluA1-expressing cells24,28, we observed robust tCaMKII-induced potentiation of GluA1 ΔPDZ-expressing cells, and a significant reduction in RI (Fig. 2b–c), demonstrating unimpaired incorporation of GluA1 ΔPDZ at synaptic sites upon plasticity. We therefore see no essential requirement for the PDZ-ligand of GluA1 in synaptic transmission or potentiation.
CTD interactions influence GluA1 ΔNTD in synaptic plasticity
Contrary to GluA2, NTD removal prevents constitutive anchoring of GluA1 and also inhibits maintenance of LTP28,29. Moreover, while co-expression of GluA1 full-length with tCaMKII potentiated synaptic EPSCs, expression of GluA1 ΔNTD blocked this potentiation (Fig. 2e; Supplementary Figure 3b)28. This observation is reminiscent of GluA2 ΔNTD’s dominant-negative effect on basal synaptic responses, and it is therefore possible that GluA1 ΔNTD is preventing endogenous receptors from effecting CaMKII potentiation, by competing for GluA1 CTD interactors.
Indeed, while GluA1 ΔΝΤD-expressing cells exhibited no synaptic potentiation on tCaMKII co-expression, PDZ mutation alleviated this inhibition and robust tCaMKII-mediated potentiation of current amplitude was observed (Fig. 2d–e). This potentiation was not accompanied by a change in rectification and therefore was mediated by endogenous receptors. While both receptors reached the cell surface (Supplementary Figure 3a), neither GluA1 ΔNTD or GluA1 ΔNTD ΔPDZ changed synaptic rectification under basal conditions (Supplementary Figure 3e). Together, these data support some role for the GluA1 PDZ ligand in synaptic potentiation. In summary, our comparative analysis of NTD and CTD interactions demonstrates a far more influential requirement for NTD interactions in AMPAR synaptic localisation, and a nonessential, but the modulatory role for the CTD.
The GluA2 NTD is a minimal unit for synaptic transmission
The TARP γ8 PDZ interaction with PSD-95 is currently the principal model for AMPAR incorporation into CA1 synapses. We and others previously demonstrated that NTD removal causes a similar attenuation of EPSC rescue in Gria1-3fl/fl neurons28,29, therefore both the TARP and NTD interactions play major roles in synaptic AMPAR accumulation, even in the absence of endogenous AMPARs. Given that both the components are important for functional transmission, we investigated whether they have redundant or parallel roles.
The ‘TARP-tandem’ configuration, where the AMPAR C-terminus is conjugated to the γ8 N-terminus by in frame expression (termed GluAX_γ8)38, allows control over γ8 PDZ-ligand interactions. While AMPAR CTD interactions will be limited by this modification, we see a little critical requirement for this domain in synaptic anchoring, and therefore will not be of major consequence. Expressing GluA1_γ8 in AMPAR-null cells rescued AMPAR synaptic currents; however, removal of the C-terminal PDZ-ligand of γ8 (TTPV motif) completely prevented this rescue (Fig. 3a–b; Supplementary Figure 4), in agreement with Sheng et al.39.
Subunit-specific effects of the NTD were apparent from our previous work and domain-swap experiments further demonstrated that the GluA2 NTD imparts greater synaptic incorporation than that of GluA128. To shed light on this behaviour, we extended our comparison of TARP and NTD interactions to GluA2. GluA2Q_γ8 robustly rescued evoked AMPAR EPSCs in AMPAR-null cells (Fig. 3c, i). However, in stark contrast to GluA1_γ8 lacking the γ8 PDZ ligand (Fig. 3b), we observed synaptic rescue by GluA2_γ8 ΔPDZ, maintaining around half the relative current amplitude of GluA2Q_γ8 (Fig. 3d, i). NMDAR currents were unchanged by these manipulations (Supplementary Figure 5a).
This result is surprising, given the complete dependence of GluA1_γ8 on γ8 PDZ interactions, so we sought to identify the mechanism supporting this transmission. We expressed GluA2_γ8 ΔNTD ΔPDZ on the AMPAR-null background, which failed to rescue AMPAR EPSCs (Fig. 3e, i), while NMDAR transmission was still intact (Supplementary Figure 5a).
As AMPARs at the CA1 synapse are mainly GluA1/2 heteromers14, we investigated whether these receptors were also able to anchor through their NTD alone. Co-expression of GluA1_γ8 and GluA2R_γ8 for rescue of AMPAR knockout gave nonrectifying responses on the cell surface (Supplementary Figure 5c) demonstrating heteromeric assembly, and synaptic transmission was successfully restored in these cells (Fig. 3f, j). Removal of the γ8 PDZ-ligand from both the subunits (GluA1_γ8 ΔPDZ & GluA2R_γ8 ΔPDZ) did not prevent surface trafficking, heteromeric assembly (Supplementary Figure 5c) or NMDAR mediated synaptic transmission (Supplementary Figure 5b). Critically however, while synaptic AMPAR EPSCs were reduced by this mutation, they were not abolished (Fig. 3g, j), suggesting that the GluA2 NTD can localise AMPARs at the synapse in the absence of other mechanisms. We tested this hypothesis with two experiments.
First, we expressed γ8 PDZ-lacking heteromeric AMPARs with the GluA2 NTD swapped for that of GluA1 (denoted GluA2RA1NTD_γ8 ΔPDZ). Co-expression of GluA1_γ8 ΔPDZ and GluA2RA1NTD_γ8 ΔPDZ in AMPAR-null cells could not rescue synaptic AMPAR currents (Fig. 3h, j), while NMDAR currents remained unimpaired (Supplementary Figure 5b). Secondly, we transferred the GluA2 NTD onto GluA1_γ8 ΔPDZ, creating GluA1A2NTD_γ8 ΔPDZ (Fig. 3k). In contrast to GluA1_γ8 ΔPDZ (Fig. 3b), robust AMPAR transmission was recorded at GluA1A2NTD_γ8 ΔPDZ-expressing synapses, demonstrating functional synaptic localisation (Fig. 3k and Supplementary Figure 5d). Together these data show that, while important for maintaining the full complement of synaptic AMPARs, γ8 PDZ interactions are not essential for CA1 synaptic transmission and work in concert with an anchoring mechanism mediated by the GluA2 NTD.
γ8 PDZ interactions accumulate AMPARs at the PSD
To further investigate whether γ8 PDZ interactions are required to anchor AMPARs at the synapse, we imaged the distribution of surface GluA1_γ8 or GluA1_γ8 ΔPDZ receptors, containing a N-terminal biotin acceptor peptide sequence (AP), in AMPAR-null cells of organotypic slices upon live-labelling with Streptavidin-AF6476 (see Procedure in Fig. 4a). Addition of the small N-terminal AP sequence does not appear alter receptor function6.
GluA1_γ8 was distributed throughout the dendritic arbour of CA1 pyramidal cells, with a strong accumulation in dendritic spines (Fig. 4b–c; Supplementary Figure 6). GluA1_γ8 ΔPDZ, showed a dramatic redistribution, with receptors no longer enriched in spines (Fig. 4b–c; Supplementary Figure 6). The threefold enrichment of GluA1_γ8 in spines relative to the dendritic shaft was lost on deletion of the γ8 PDZ-ligand (Fig. 4c), explaining the dramatic reduction of transmission (Fig. 3b). Therefore, TARP - PSD-95 interactions accumulate receptors in spines, most likely at the PSD.
We next compared the distribution of surface GluA1 and GluA1 ΔNTD. Unexpectedly, both constructs showed a punctate, spine-enriched distribution, suggesting that gross spine accumulation is not dramatically impaired by NTD deletion (Fig. 4d–e; see also Fig. 5d). Therefore, the functional effect of NTD deletion may occur on a subsynaptic scale.
NTD-deletion alters the nanoscale arrangement of synaptic AMPARs
To investigate the effect of NTD deletion on receptor localisation in greater detail, we set up a receptor labelling approach in dissociated hippocampal cultures that allows super-resolution imaging of surface AMPARs. GluA1 and GluA2Q both with and without their NTD were tagged at their N-termini with an HA-epitope, which can be labelled live using a fluorescently-labelled anti-HA single-chain Fv-fragment (scFv-Clasp) (Fig. 5a)40. This small, monovalent probe (~5 nm diameter) prevents receptor clustering by epitope cross-linking and increases localisation precision in single-molecule imaging.
Using confocal microscopy, we observed a moderate decrease in surface expression of GluA2 ΔNTD receptors but not of GluA1 ΔNTD, while intracellular expression was increased for both constructs (Fig. 5b–c; Supplementary Figure 7a–b). As seen in slices, the surface-expressed ΔNTD receptors were enriched at dendritic spines, albeit displaying a slightly decreased spine enrichment in comparison to full-length receptors (Fig. 5d; Supplementary Figure 7c–d).
Since synaptic AMPARs concentrate in ‘nanoclusters’8,9, we used 3D STORM (Stochastic Optical Reconstruction Microscopy) to investigate the impact of the NTD on synaptic and subsynaptic receptor distribution at the nanoscale (Fig. 5e–l, Fig. 6). For this experiment, we only analysed neurons displaying comparable AMPAR surface expression levels, as quantified from widefield images collected prior to STORM imaging. Interestingly, both the GluA1 and GluA2 ΔNTD mutants occupied a larger synaptic area compared to the respective full-length receptors, which coincided with an increase in the corresponding presynaptic area, marked by the active zone protein bassoon (Fig. 5i–l). As the number of observed localisations was only marginally increased for GluA1 ΔNTD but not for GluA2 ΔNTD, this resulted in a significant reduction in the overall density of ΔNTD receptors at the synapse (Fig. 5j and Supplementary Figure 7g).
To quantify subsynaptic receptor distributions (‘AMPAR Nanoclusters’) we used a clustering algorithm (OPTICS) which performs well for datasets with variable densities41,42,43. AMPARs lacking the NTD were still assembled into nanoclusters but exhibited subunit-selective differences (Fig. 6a). Specifically, GluA1 ΔNTD nanoclusters displayed a larger diameter, fewer localisations and a reduced cluster density compared to full-length GluA1, while the properties of GluA2 ΔNTD nanoclusters appeared surprisingly unchanged (Fig. 6c–e, Supplementary Figure 7i–k). In addition, we computed that, on average, one nanocluster is present per synapse for GluA1 and GluA2 full-length as well as for GluA2 ΔNTD receptors whereas the averaged cluster number for GluA1 ΔNTD was slightly increased to 1.5 (Fig. 6b and Supplementary Figure 7i–j; representative subsynaptic AMPAR distributions are shown in Fig. 5g, h, Fig. 6a and Supplementary Figure 7e–f). Together, these results indicate that the NTD is not essential for nanocluster formation but affects synaptic receptor densities and nanocluster properties in a subunit-specific manner.
AMPAR trafficking defines the strength of excitatory synapses, and has therefore been investigated for decades4,44. Many molecular interactions have been reported but there is incomplete synthesis of proposed models, which has muddied our understanding. Here, we demonstrate that while AMPAR CTD interactions play a modulatory role in shaping transmission, both intracellular interactions with the PSD through TARP γ8 and contacts mediated by the extracellular NTD control AMPAR transmission at CA1 synapses. These two interactions have unique roles: γ8 accumulates receptors at the PSD, while the NTD acts in a subunit-selective manner on a synaptic and subsynaptic level to allow functional transmission (Fig. 7).
TARP PDZ interactions have become the principal model for AMPAR anchoring in recent years. First demonstrated at cerebellar granule cell synapses, where the PDZ-ligand of γ2 is essential for transmission45, this model was extended to hippocampal synapses17,46,47. However γ8, not γ2, is the predominant TARP in hippocampal neurons15,16,48,49. Mice with deletion of the γ8 PDZ-ligand show only a 30% reduction in CA1 AMPAR transmission, while LTP remains intact20, questioning the requirement for PDZ interactions in hippocampal transmission. In contrast, Sheng et al.39 demonstrate complete dependence of GluA1_γ8 on PDZTARP interactions for both the synaptic transmission and plasticity, concluding that all AMPAR transmission requires PDZ interactions with the PSD39,50. While we can directly replicate the PDZ-dependence of GluA1_γ8, this does not apply to all AMPARs. The GluA2 NTD acts as a minimal unit for synaptic anchoring even in the absence of γ8 PDZ ligands and could support transmission in γ8 ΔPDZ mice20. While clearly an important mechanism in AMPAR synaptic localisation, γ8 PDZ interactions are not the sole requirement for synaptic transmission. The role of other auxiliary subunits in synaptic anchoring, such as cornichons15, is excluded from our analysis and remains to be determined.
The functional role of extracellular NTD interactions was reported only recently28,29. The dominant role of the GluA2 NTD described here, recapitulates the ‘synapto-sticky’ phenotype that we reported previously28, and will allow strict control over the contribution of GluA2-lacking, calcium-permeable AMPARs51,52.
The mechanisms of NTD action are so far unclear. In contrast to γ8’s PDZ, NTD removal does not prevent synaptic accumulation of receptors. The largely unaltered gross distribution of ΔΝΤD receptors contrasts starkly with their functional deficits, and suggests the domain has a subsynaptic influence on receptor organisation at the synapse. Correspondingly, we observe changes in the synaptic distribution of receptors on NTD removal, with ΔΝΤD receptors forming larger, less dense arrangements at the PSD. Previous modelling studies comparing postsynaptic receptor distributions between transsynaptically aligned clusters and uniformly distributed arrangements demonstrated at least a 30% reduction in transmission can be caused by these subtle changes in subsynaptic positionings8. This effect therefore likely makes a significant contribution to the decrease in currents on NTD deletion, however other factors such as spine-surface receptor availability may also contribute to the observed effects.
Within the PSD, AMPARs form nanoclusters8,9, and are positioned opposite vesicle release sites11,53. Surprisingly, nanocluster formation is not dependent on the NTD of either GluA1 or GluA2, suggesting that intracellular interactions with PSD-95 are sufficient for their formation8. PSD-95 forms nanoclusters that are transsynaptically aligned with presynaptic release sites8,54.
It is intriguing that GluA1 ΔNTD receptors, which do not appear to contribute to functional transmission, are still localised and clustered at synapses. This surprising observation will require further analysis. NTD removal does alter the properties of these clusters suggesting a role for synaptic cleft interactors in AMPAR organisation. There is emerging evidence that even modest changes in subsynaptic receptor distributions, such as clustering or transsynaptic alignment, can strongly influence receptor activation8,10, or give rise to synaptic plasticity55,56,57. Therefore, the subtle rearrangements observed on NTD deletion may cause substantial deficits in functional transmission. Given the nonuniform, protein-dense nature of the synaptic cleft58,59, a plethora of pre- or postsynaptic protein interaction partners would be able to tune AMPAR transmission through the NTD. Multiple reports demonstrate AMPAR-specific effects on deletion of transsynaptic synaptic adhesion molecules, such as Neurexins and LRRTMs60,61, and Pentraxin-dependent AMPAR clustering via the NTD has a clear role at other synapses in the brain62,63,64. A recent report further confirms the requirement for GluA1 NTD interactions in the maintenance of LTP, demonstrating that LTP can occur with GluA2 ΔNTD receptors alone30. The authors propose that neuroplastin-65 interacts with the GluA1 NTD to maintain AMPAR-dependent synaptic potentiation. These data fully support the subunit-specific AMPAR NTD roles, which we previously proposed28, with the GluA2 and GluA1 NTDs acting specifically to maintain basal transmission and synaptic potentiation, respectively.
Like others, we observe no essential role for the AMPAR CTD in constitutive receptor anchoring26,27,46, and no requirement for the GluA1 CTD PDZ ligand for synaptic transmission or potentiation26,65,66. Further investigation is required to assess their requirement for synaptic plasticity27, including the parallel role of direct receptor modification for potentiation, rather than alteration of receptor trafficking alone67. While not essential, we do observe some influence of CTDs, conferring dominant-negative effects on NTD-deleted receptors. Specifically, the GluA2 CTD acts in basal transmission, and the GluA1 CTD in synaptic potentiation, echoing the subunit-specific AMPAR trafficking rules proposed to rely on CTD interactions25. While the CTD is not a critical player in AMPAR synaptic localization, it likely influences the trafficking and recycling of receptors68.
Changing the synaptic AMPAR content can control synaptic strength69, as demonstrated in vivo during behavioural tasks70. Our work proposes that γ8 PDZ and AMPAR NTD interactions are the major determinants of the postsynaptic receptor complement, with their interplay acting to trap and localise the required AMPAR content at each connection6. Plasticity-induced changes in the arrangement of the PSD are far better documented than those in the synaptic cleft11,66,71. Our data suggest that changes in the abundance or distribution of extracellular AMPAR binding slots may occur within the synaptic cleft, with the potential to serve as a mechanism for LTP. The identity and dynamics of these machineries, and the mechanisms of their pre-, post- or transsynaptic control is now of great importance to understanding information storage at synapses throughout the brain.
All experimental procedures were performed under procedure project license (PPL) P81248620 in accordance with UK Home Office regulations and were licensed under the Animals (Scientific Procedures) Act of 1986 following local ethical approval (Animal Welfare and Ethical Review Body (AWERB), Institute: MRC Laboratory of Molecular Biology). Tissue for a slice or dissociated hippocampal cultures was harvested from an unascertained mixture of sexes at postnatal ages up to P8 either from C57BL/6JOla wild-type (RRID:MGI:3691859) or Gria1-3fl/fl mice (reported in28), as indicated. To generate the Gria1-3fl/fl line, mice with floxed loci at Gria1, 2 and 3 genes [Gria1lox/lox (RRID:IMSR_JAX:019012), Gria2lox/lox (RRID:IMSR_EM:09212), Gria3lox/lox (RRID:IMSR_EM:09215)] were interbred to produce mice homozygous for all floxed alleles (Gria1lox/lox; Gria2lox/lox; Gria3lox/lox, denoted Gria1-3fl/fl). All animals were housed with unlimited access to food and water on a 12:12 h light:dark cycle at normal room temperature (20–22 oC) at 45–65% humidity.
Dissociated hippocampal culture
Hippocampal neurons from postnatal (P0-P1) C57BL/6JOla wild-type mice were prepared following Beaudoin et al.72. Briefly, mouse hippocampi were dissected in ice-cold HBSS (Ca2 + and Mg2 + free, Gibco, Cat.# 14175095) containing 0.11 mg ml−1 sodium pyruvate (Gibco, Cat.# 12539059), 0.1% glucose and 10 mM HEPES (Gibco, Cat.# 15630056) and cells were dissociated with trypsin (0.25% wt/vol, Gibco, Cat.# 15090-046). Neurons were plated onto poly-L-lysine coated glass coverslips (12 mm round coverslips, Corning, Cat.# 354085 for confocal imaging or 24 mm round coverslips #1.5, Glaswarenfabrik Karl Hecht GmbH & Co KG, Cat.# 1001/24_15 92100105080 for STORM imaging) in equilibrated culture medium, which contains 86.55% Minimum Essential Medium (MEM) (Gibco, Cat.# 21090022), 10% heat-inactivated foetal bovine serum (Gibco Cat.# 11573397), 0.45% glucose, 1 mM sodium pyruvate and 2 mM GlutaMax (Gibco, Cat.# 35050038). Cultures were kept at 37 °C and 5% CO2 in equilibrated maintenance medium containing 96% Neurobasal medium (Gibco, Cat.# 21103049), 2% B-27 plus Supplement (Gibco, Cat.# A3582801) and 2 mM GlutaMax, until 21 days in vitro (DIV). Half of the medium was replaced every 4–7 days.
Organotypic slice culture
Hippocampal organotypic slice cultures were prepared following Stoppini et al.73. Hippocampi from P6-8 mice were isolated in ice-cold high-sucrose Gey’s balanced salt solution containing (in mM): 175 Sucrose, 50 NaCl, 2.5 KCl, 0.85 NaH2PO4, 0.66 KH2PO4, 2.7 NaHCO3, 0.28 MgSO4, 2 MgCl2, 0.5 CaCl2 and 25 glucose at pH 7.3. Hippocampi were cut into 300 μm thick slices using a McIlwain tissue chopper and cultured on Millicell cell culture inserts (Millipore Ltd) in an equilibrated slice culture medium (37 °C/5% CO2). Culture medium contained 78.5% MEM, 15% heat-inactivated horse serum, 2% B27 supplement, 2.5% 1 M HEPES, 1.5% 0.2 M GlutaMax supplement, 0.5% 0.05 M ascorbic acid, with additional 1 mM CaCl2 and 1 mM MgSO4 (all from Thermo Fisher Scientific; Waltham, MA). The medium was refreshed every 3–4 days. Cultures were transfected at 5–10 DIV by single-cell electroporation (SCE) and recordings were performed 4–5 days after transfection.
All constructs were created by in vivo assembly cloning unless otherwise stated74 using the Rattus norvegicus coding sequence. GluA1 (flip isoform; Uniprot P19490) or GluA2 (Q/R edited where specified, R/G edited, flip isoform; Uniprot P19491) variants used for electrophysiology and slice tissue imaging were expressed from the pRK5 vector. ∆NTD constructs were as reported previously28. The NTD coding sequence (GluA1: Ala1 - Ala373, GluA2: Val1 - Thr377) was replaced by a c-myc epitope sequence, leaving the entire NTD-LBD linker sequence intact. GluA2 NSF mutation was achieved by Asn830Ala and Pro831Ala mutations, PDZ by addition of a C-terminal Tyr residue (Tyr863) and CTD-null by all aforementioned mutations plus Ser842Ala. For GluA1/2 intracellular domain swap, GluA2 Thr553 to Thr568 (loop 1) and Ala820 to Iso862 (CTD) were replaced by GluA1 sequence from Ser549 to Ser564 (loop 1) and Ser816 to Leu889 (CTD). GluA1 PDZ mutation was achieved by Thr887Ala mutation, and was combined with Ser831Ala and Ser845Ala mutations to generate the CTD-null construct. N-terminal GFP tagging was achieved by insertion of the EGFP coding sequence between GluA1 Phe3 and Pro4 residues, flanked by Ala-Arg and Ala-Ser linker residues respectively (for replication of25).
TARP-tandem constructs were produced by splicing of the TARP γ8 coding sequence (Uniprot Q8VHW2) in frame with AMPAR subunits, separated by a Gly-Ser-Gly-Ser-Gly linker sequence (see14). PDZ deletion was achieved by removal of the γ8 terminal Tyr-Tyr-Pro-Val sequence (Tyr418 - Val421). GluA1/2 NTD swaps were performed as previously reported28, and involved exchanging both NTD and NTD-LBD linker sequence between the subunits (corresponding to, GluA1: Ala1 - Thr390, GluA2: Val1 - Thr394). pN1-EGFP (Clontech, 6085-1) was used for cell visualisation.
For surface biotinylation, GluA1 was tagged at the N-terminus with the acceptor peptide sequence prior to Ala1 of GluA1 coding sequence (Insertion of Ala-“Gly-Leu-Asn-Asp-Iso-Phe-Glu-Ala-Gln-Lys-Iso-Glu-Trp-His-Glu”-Gly sequence, AP-tag in quotation marks). GluA1 ΔNTD was tagged by replacing the c-myc epitope with the AP-tag sequence. The Lys residue in this sequence is modified by biotin conjugation (see75). ER-localised biotin ligase (BirA), and pC1-tdTomato were gifts from Andrew Penn, and were used to create the pC1-tdTomato-IRES-BirAER construct.
For confocal and 3D STORM imaging of cultured hippocampal neurons, N-terminally HA-tagged AMPAR constructs were expressed from the doxycycline-inducible pcDNA4/TO vector (Invitrogen Cat. No: V102020, including pcDNA™6/TR for expression of the tetracycline repressor (TR) protein). The HA-epitope was inserted after Ala1 for GluA1 full-length (flip isoform) and after Val1 for GluA2 full-length (Q/R unedited, R/G edited, flip isoform) (Insertion for GluA1: Ala-Tyr-Pro-Tyr-Asp-Val-Pro-Asp-Tyr-Ala, Insertion for GluA2: Val-Tyr-Pro-Tyr-Asp-Val-Pro-Asp-Ala, HA-tag in italics). HA-tagged GluA1 and GluA2 ΔNTD constructs are equivalent to those previously reported, with the c-myc epitope replaces by the HA-tag sequence. pN1-EGFP (Clontech, 6085-1) was used for cell visualisation.
Constructs for 12CA5 scFv-Clasp anti-HA (CA5-VH(X3)-Mst1_pET11c (#5019), heavy chain, and CA5-VL-Mst1(Y3)_pET11c (#5020), light chain) were kindly provided by Junichi Takagi40. The constructs were subcloned using restriction digestion into the pHL-sec vector containing a C-terminal 6xHis-tag (Addgene # 99845). Restriction sites for subcloning were introduced by PCR at the 5’ and 3’ ends of scFv-Clasp heavy and light chains (see Supplementary Table 3 for primer sequences). The light chain inset was cloned into the pHL-sec vector using AgeI and KpnI restriction sites, keeping a stop codon before the His-tag. The heavy chain sequence was introduced using AgeI and Bsp1407I (inset)/Acc65I (vector) sites, in frame with the His-tag.
Transfection of neuronal cultures
Cultured hippocampal neurons were transfected at DIV 6 with plasmids expressing doxycycline-inducible, HA-tagged GluA1 or GluA2 receptors. AMPAR-expressing constructs were co-transfected with pcDNA6/TR and EGFP-pN1 at a 1:1.57:0.29 ratio using Ca2PO4 precipitation. 2 µg (12 mm coverslips) or 4 µg (24 mm coverslips) of DNA were added to 250 mM CaCl2 in nuclease-free water, mixed at a 1:1 ratio with 2x HEPES buffered saline solution (50 mM HEPES, 280 mM NaCl, 1.5 mM Na2HPO4 in nuclease-free water, pH 7.05) to a final volume of 100 (12 mm coverslips) or 200 µl (24 mm coverslips) and incubated for 20 min at room temperature. Coverslips containing neurons were moved to fresh plates with conditioned maintenance medium supplemented with 1 mM Kynurenic acid (Sigma, Cat.# K3375, 0.5 ml/12-well for 12 mm and 1 ml/6-well for 24 mm coverslips). 100-200 µl transfection mix were added to each 12- or 6-well, respectively, and incubated for 1.5–2 h at 37 °C and 5% CO2. Subsequently, coverslips were washed in maintenance medium conditioned at 37 °C and 10% CO2 (1 ml/12-well and 2 ml/6-well) for 20 min at 37 °C and 5% CO2 to dissolve precipitates and eventually moved back to their original culture wells. Exogenous AMPAR expression was induced at DIV 18 using 7.5 µg/ml doxycycline (Merck, Cat.# D3447) and neurons were harvested at DIV 21 for imaging.
Neonatal viral injection
In vivo, the viral injection was performed according to76. P1 neonatal mice were anaesthetised using isoflurane, bilaterally injected using a pulled glass needle in the hippocampal area with 0.5 μL AAV-hSyn-Cre-EGFP at 3 × 1012 GC ml−1 (Addgene #105540, virus produced by U-Penn Vector Core), and returned to the parent cage until P7, when tissue was harvested for organotypic culture.
Single-cell electroporation (SCE)
CA1 cells of hippocampal organotypic slices were transfected using an adapted version of the single-cell electroporation method described in77. DNA plasmids were diluted to 33 ng/μl with potassium-based intracellular solution and the mixture was back-filled into borosilicate microelectrode pipettes. Slices were submerged in HEPES-based artificial cerebrospinal fluid (aCSF) containing (in mM): 140 NaCl, 3.5 KCl, one MgCl2, 2.5 CaCl2, 10 HEPES, 10 Glucose, 1 sodium pyruvate, 2 NaHCO3, at pH 7.3. Plasmids were introduced into individual cells by the application of a short burst of current pulses (60 pulses at 200 Hz) while in cell-attached mode. To visualise transfected cells, pN1-EGFP or pC1-tdTomato (Clontech; Mountain View, CA) were mixed with AMPAR-expressing plasmids at a base pair ratio of 1:7. For heteromeric receptor transfection, GluA1 and GluA2 plasmids were transfected at a 1:2 ratio. In tCaMKII experiments, the ratio between tCaMKII-EGFP and AMPAR-expressing plasmids was 1:1. For rescue of AMPAR-null cells, SCE was used to target neurons strongly labelled with nuclear EGFP fluorescence from in vivo viral transduction. For slice imaging experiments, AP-AMPAR and tdTomato-IRES-BirAER plasmids were transfected at a ratio of 3:1.
Transfected hippocampal slice cultures were submerged in aCSF containing (in mM): 125 NaCl, 2.5 KCl, 1.25 NaH2PO4, 25 NaHCO3, 10 glucose, 1 sodium pyruvate, 4 CaCl2, 4 MgCl2 and 0.001 SR-95531 (Tocris Cat# 1262) at pH 7.3 and saturated with 95% O2/5% CO2. 100 mM D-APV (Tocris Cat# 0106, HelloBio Cat# HB0225) was used to isolate AMPAR currents for rectification index recordings. 2 mM 2-chloroadenosine (Tocris Cat#3136) was added to aCSF to dampen epileptiform activity. Three to six milliohm borosilicate pipettes were filled with intracellular solution containing (in mM): 135 CH3SO3H, 135 CsOH, 4 NaCl, 2 MgCl2, 10 HEPES, 4 Na2-ATP, 0.4 Na-GTP, 0.15 spermine, 0.6 EGTA, 0.1 CaCl2, at pH 7.25. All synaptic recordings were performed using a dual whole-cell patch clamp, involving simultaneous recording from a neighbouring pair of GFP positive and negative cells for exogenous expression experiments. For AMPAR rescue experiments, dual recordings were performed from one EGFP (Cre) and tdTomato (SCE rescue) positive cell, and a neighbouring nonfluorescent cell. EPSCs were evoked by single-pulse simulations of Schaffer collaterals in the stratum radiatum of CA1 using a monopolar glass electrode, filled with aCSF at 0.2 Hz. Cells were held in the voltage-clamp configuration for at least 3 min after achieving whole-cell configuration and starting axon stimulation before initiating data collection. This ensures reliable synaptic current stimulation and sufficient equilibration of spermine from the intracellular solution. Recordings were collected using a Multiclamp 700B amplifier (Axon Instruments), digitised using a Digidata 1440 A interface (Axon Instruments) and recorded using Clampex (pClamp10).
Somatic AMPAR responses were recorded by pulling outside-out patches from fluorescence positive or negative CA1 cell bodies and patches were subjected to fast-exchange perfusion using a two-barrel theta glass tube controlled by a piezoelectric translator (Physik Instrumente). Theta barrels solutions were HEPES-based aCSF (see SCE) containing 100 μM cyclothiazide (Tocris Cat# 0713) and 100 μM D-APV, either with or without 1 mM L-glutamate. In voltage-clamp mode, a 500 ms holding potential ramp from −100 mV to +100 mV was applied to patches (see Figure S1).
Electrophysiology data analysis
Data were analysed using Clampfit (pClamp 10—Molecular Devices). Synaptic stimulations across at least 20 sweeps were averaged, and peak amplitudes were measured relative to baseline preceding the stimulation artefact. For rectification index calculation, only recordings with the temporal alignment of peak amplitudes between negative (−60 mV) and positive (+40 mV) holding potentials were included in the analysis. Rectification index was calculated from peak current amplitudes at −60, 0 and +40 mV holding potentials as in Eq. 1:
AMPAR peak amplitudes were measured from −60 mV holding potential recordings. NMDAR synaptic currents were measured from +40 mV holding potential recordings, and represent the amplitude 100 ms after the stimulus artefact, to prevent AMPAR current contamination. The normalisation of dual recording peak amplitudes was performed by division of all values across the cell pair by the mean amplitude of all control cell responses. Where normalised summary graphs are depicted without individual points, all amplitude values are presented as scatter graphs in either the Main Text Figures or Supplementary Information. Somatic response rectification indices are calculated from current amplitudes from membrane potential ramps (−100 to +100 mV). Recordings in the absence of glutamate were subtracted from those in the presence of glutamate for each patch. All recordings during which the series resistance varied by more than 20% or exceeded 20 MΩ were discarded. Membrane potentials are given without correction for liquid junction potentials. All data are pooled recordings from at least two individual preparations, except surface rectification recordings which are from one preparation.
Generation of scFv-clasp against HA
HEK293S GnTI– cells (ATCC, Cat.# CRL-3022, RRID:CVCL_A785) were transiently transfected with heavy (CA5-VH(X3)-His_pHL_sec) and light chains (CA5-VL(X3)_pHL_sec) at a 1:1 ratio using polyethyleneimine (PEI, 1 mg/ml, MW 40,000, Polysciences Cat.# 24765) transfection reagent. A mixture of DNA (0.5 mg) and PEI (1 mg) diluted in 20 ml serum-free culture medium (Dulbecco’s Modified Eagle Medium (DMEM), High glucose, GlutaMax supplement, pyruvate, Gibco, Cat.# 31966-021) was added per 250 ml of culture volume. Cells were maintained in culture medium (DMEM, as described) supplemented with 2% FBS, 2 mM Glutamax, and 1% MEM Nonessential amino acids solution, Gibco, Cat.# 11140050) at 30 °C for 7 days after transfection (see78). Cleared cell supernatant, containing secreted scFv-Clasp, was concentrated using an Akta Flux and stored overnight at 4 °C in 20 mM Tris-HCl pH 8, before the addition of 4 mM imidazole and incubation on a rotor (111 RPM) with TALON metal affinity resin (Takara, Cat.# 635503, 4 ml slurry for ~300 ml concentrated supernatant) at 16 °C, for 4 h the following day. After collection, beads were washed with 20 mM Tris/PBS pH 8, non-specific proteins were eluted with 10 mM imidazole in 20 mM Tris-HCl and 500 mM NaCl, pH 8, followed by protein elution with 250 mM imidazole in 20 mM Tris-HCl, 500 mM NaCl, pH 8. Fractions containing scFv-clasp were pooled, concentrated and further purified by size-exclusion chromatography on a Superdex 200 26/60 column L (GE Healthcare, Cat.# 28-9893-36). Purified scFv-Clasp were flash frozen in aliquots (in 10 mM HEPES, 150 mM NaCl, pH 7.4) and stored at −80 °C until use.
scFv-Clasp fluorescent conjugation
scFv-Clasp (~1 mg/ml) was adjusted to pH 8.3 with 1 M NaHCO3, incubated either with Alexa Fluor (AF) 647 NHS ester (Invitrogen, Cat.# A20106) or CF568 NHS ester (Sigma-Aldrich, Cat.# SCJ4600027), according to manufacturer’s instructions, for 1.5–2 h while shaking at room temperature in the dark. Subsequently, the solution was centrifuged for 10 min at 10,000 g at 4 °C. If necessary, the volume was adjusted to 500 µl with PBS, and loaded on an equilibrated G-25 column (Column PD Minitrap G-25, GE Healthcare, Cat.# 28-9180-07). Once the solution fully entered the column, conjugated scFv-Clasp was eluted with 1 ml PBS. To remove any residual unbound dye, the eluate was washed 3 times with PBS in an Amicon Ultra-4 Centrifugal Filter Unit (NMWL 3 kDa, Millipore, UFC800324) and finally concentrated to <250 µl. The degree of labelling, assessed photometrically, was 5.5 for AF647-scFv-Clasp and 2 for CF568-scFv-Clasp.
Surface labelling of biotinylated receptors in slice tissue
Surface labelling of biotinylated AMPARs in slice tissue was performed as described in6. 20 μΜ biotin (Sigma-Aldrich Cat# B4501) was added to the culture medium after SCE transfection. Four days later, slices were washed in HEPES-based aCSF (see SCE) and excess biotin was removed by dialysis in HEPES-based aCSF against 0.25 ml 2 mg/ml avidin Texas red (Thermo Fisher Scientific Cat# A820) through a Slide-A-Lyzer MINI (3500 molecular weight cut-off, Thermo Fisher) for 45 mins. Slices were then incubated for 45 min at room temperature in HEPES-based aCSF containing 120 nM Streptavidin-AF647 (Invitrogen). Slices were washed three times over 45 mins before fixation in phosphate-buffered saline (PBS) containing 4% paraformaldehyde and 4% sucrose (PFA, Sigma-Aldrich, Cat.# P6148) at 4 °C overnight. Slices were washed and stored in PBS until imaging. For imaging, slices were inverted on a glass-coverslip and imaged in PBS using a 63x/1.4NA oil-immersion objective on a commercial Leica TCS SP8 confocal microscope using LAS X software. Z-stacks of whole cells or whole stratum radiatum dendrites were taken, with EGFP, tdTomato and AF647 excited with 478 nm, 554 nm and 653 nm laser lines, with equivalent power settings across experimental conditions.
Cultured hippocampal neurons were live-labelled at DIV 21 with 50 nM scFv-Clasp either conjugated to Alexa647 (confocal imaging) or CF568 (STORM imaging) at room temperature for 20 min in aCSF (in mM: 120 NaCl, 5 KCl, 1.2 MgCl2*6H2O, 2 CaCl2, 25 HEPES, 30 glucose, pH set to 7.3–7.4 with 2 M NaOH). Neurons were briefly washed 5 times over 10 min in aCSF and PBS and fixed in 4% PFA, 4% sucrose in PBS for 10 min at room temperature. Subsequently cells were washed again, permeabilised with 0.1% Triton X-100 (Fisher Bioreagents, Cat.# BP151-500) and treated with blocking solution containing 1% bovine serum albumin (Fisher Bioreagents, Cat.# BP1605-100) and 10% normal goat serum (Sigma-Aldrich, Cat.# G9023) in PBS. Neurons were incubated sequentially with primary and secondary antibody solutions prepared in PBS supplemented with 1% BSA and 10% normal goat serum for 2 h at room temperature, and washed in PBS after each incubation. For 3D STORM imaging a second fixation was performed after incubation with the secondary antibody and coverslips were stored in PBS at 4 °C until imaging. For confocal imaging, coverslips were mounted in ProLong Diamond antifade Mountant (Invitrogen, Cat.# P36961) and left to cure for 48 h at room temperature before imaging.
The following antibodies were used in this study: rabbit anti-HA (Sigma-Aldrich, Cat.# H6908, RRID:AB_260070, 1:200, confocal imaging), rabbit anti bassoon (Synaptic Systems Cat# 141 003, RRID:AB_887697, 1:200, STORM imaging) goat anti rabbit IgG AF568 (Thermo Fisher Scientific Cat# A-11036, RRID:AB_10563566, 1:200, confocal imaging), goat anti rabbit IgG AF647 (Thermo Fisher Scientific Cat# A-21245, RRID:AB_2535813, 1:200, STORM imaging).
Images were acquired with a Leica TCS SP8 confocal inverted microscope controlled by Leica Application Suite X (LAS X) software and equipped with a tunable pulsed white light laser and hybrid detectors. For quantification of AMPAR expression levels and surface distribution in cultured hippocampal neurons, z-stacks of whole dendrite segments were acquired using a 63x/1.4NA oil-immersion objective. EGFP, Alexa Fluor 568 and Alexa Fluor 647 were excited at 488, 561 and 633 nm, respectively. All conditions were acquired with the same settings across different preparations.
Confocal imaging analysis
For AMPAR surface distribution in tissue, images were analysed in FIJI. Dendritic rejoins were z-projected (sum) before median filtering using a 1-pixel radius filter. Line profiles were performed on all spines that were suitably perpendicular to the dendritic shaft within each field of view, from tdTomato fluorescence. Grey values for surface receptor fluorescence were measured along a line of 300 μm thickness from raw images. Line profiles were normalised to the maximal dendritic fluorescence, and spine enrichment is quantified as the peak spine fluorescence after normalisation. Quantifications in Fig. 4c and e are from two and one preparations respectively.
Confocal images of cultured hippocampal neurons were z-projected (maximal intensity) and a median filter (radius = 1) was applied. For quantification of expression levels, masks of regions of interests, generated from thresholded composite images of permeabilized and surface HA stains were applied to each channel and integrated densities were determined. Thresholds and filters were applied to the entire image and the same settings were used across all conditions. For surface distribution, line scans across spines perpendicular to the dendritic shaft were performed based on EGFP fluorescence. Spine enrichment was calculated as peak spine fluorescence over peak dendrite fluorescence. Fluorescence intensity was normalised to maximal dendritic intensity for representative line scans in Supplementary Figure 7c, d. Same datasets were used for quantification of expression levels and surface distribution, two images with the lowest surface expression were excluded for analyzing GluA2 ΔNTD spine enrichment. Images shown in Fig. 5b, c were processed as described above. Where needed, contrast adjustments were performed uniformly across the whole image, with the same settings applied to full-length and respective ΔNTD AMPAR example images. Images were analysed and pseudo-coloured in Fiji (ImageJ).
Samples were incubated with Tetraspeck Microspheres (100 nm, Invitrogen, Cat.# T7279) for image registration and mounted in imaging buffer containing 5% Glucose (w/v), 100 mM Mercaptoethylamine (Sigma-Aldrich, Cat.# 30070), 0.8 mg/ml glucose oxidase (Sigma-Aldrich, Cat.# G2133), 50 µg/ml catalase (Sigma-Aldrich, Cat.# C30) in PBS in an Attofluor Cell Chamber (Invitrogen, Cat. # A7816). The imaging buffer was freshly prepared for each coverslip. The imaging chamber was filled with buffer to full capacity and sealed by an additional coverslip on top to minimise air-induced oxidation of the sample. Three-dimensional STORM images were acquired using a Nikon Eclipse TI-E N-STORM inverted microscope equipped with an Apochromat TIRF 100x/1.49 NA oil-immersion objective, 405, 488, 561 and 647 nm lasers and an EMCCD camera (iXon Ultra DU897, Andor). The illumination intensity at the sample was ~2KW/cm2. 3D localisation was achieved using the astigmatism method79. Conventional fluorescence images for different channels were acquired first. Before STORM image collection, the field of view was illuminated with the activating laser at a low incident angle to deactivate out-of-focus fluorophores. For STORM data acquisition, samples were illuminated in pseudo-TIRF mode with the incident angle close to the critical angle with either the 647 nm (AF647) or the 568 nm (CF568) lasers using 100% intensity. The 405 nm laser was used at 2% to reactivate AF647 fluorophores to maintain adequate localisation density. Fifty thousand frames at 50 Hz, were collected for each channel sequentially, the axial focal plane was maintained by using the Perfect Focus System during image acquisition.
STORM imaging analysis
Single-molecule identification was performed with NIS Elements Advanced Research software (Nikon) using default settings for peak identification with minimum intensity (height) set to 1000. Identical settings were used for every image. Lateral drift correction was performed using in-built cross-correlation between subsets for the frame sequence. Molecules outside the Z calibration range, appearing in more than five consecutive frames, with fewer than 500 photons and a localization precision greater than 20 nm, were discarded. Tetraspeck Microspheres were used for image registration. Individual localisations in the reconstructed images (Fig. 5e–h) are shown as a normalised Gaussian with the width corresponding to the localization uncertainty. Images were thresholded uniformly, with the same settings applied to full-length versus respective ΔNTD AMPAR images. For visualisation of dendrite outlines, conventional fluorescence images of co-transfected EGFP were smoothened, rendered semi-transparent and overlaid onto 3D STORM images using Adobe Photoshop. For density plots (Fig. 6a), synaptic AMPAR clusters were reconstructed as scatter plots in MATLAB (Mathworks) and colour coded based on their local density (number of points (60) within search radius defined as 99th percentile of shortest pair-wise distances). As a control, observed localisations were randomly re-distributed within the original cluster boundaries.
Quantitative analysis on imported molecule lists was performed in MATLAB using custom scripts. STORM images were reconstructed as scatter plots and isolated synapses, defined as opposing postsynaptic AMPAR and presynaptic Bassoon clusters were selected for analysis. Only synaptic clusters with an unambiguous orientation were selected and rotated with the synaptic cleft perpendicular to the z-axis. Noise points (i.e. localisations on the outer edges of synaptic clusters) and nanoclusters were identified using the OPTICS (Ordering points to identify the clustering structure) algorithm which offers superior robustness when working with datasets of different densities compared to other methods, such as DBSCAN41,42,43. OPTICS scripts for calculating the reachability distances and subsequent cluster extraction were adapted from Michal Daszykowski (http://chemometria.us.edu.pl/index.php?goto=downloads) and Alex Kendall (https://github.com/alexgkendall/OPTICS_Clustering), respectively, with minor modifications for increased robustness. To generate ‘reachability plots’ representing the hierarchical structure of clusters, points were linearly ordered based on their spatial relationship with a minimum number of points (MinPts) set to 30 (noise identification) or 25 (nanocluster detection). Each channel was analysed separately. To reduce bias, results from 100 repeated runs, each starting from a different localisation were averaged for individual clusters. Averaged reachability distances greater than the 90th percentile were used as the cut-off to define noise points. Subsequently, synaptic areas were analysed using the in-built MATLAB function ‘boundary’ and density was calculated as the number of localisations per area. Clusters with a synaptic area >0.6 µm2 were excluded from analysis to ensure that only individual synapses were analysed71,80. Overall synaptic volumes were analysed using in-built MATLAB function ‘alpha shape’ using the default alpha radius (criticalAlpha) and property ‘one-region’. For nanocluster quantification, the smallest detected clusters, if present, containing a minimum number of 25 points were extracted from individual reachability plots as described in (41, ξ set to 0.02) using an adaptation of the aforementioned MATLAB script by Alex Kendall. Due to variability in cluster edge detection, only the localisations that were detected as part of a putative cluster in over 70% of the iterations were considered to constitute to a nanocluster. Furthermore, only nanoclusters with a ratio of averaged reachability distance to overall reachability distance greater than 0.8, more than 25 localisations and a density (localisations/area) greater than 3 × 103 were considered for analysis. Nanocluster detection criteria were determined based on randomised datasets, which were generated by subjecting the observed localisations in each AMPAR synaptic cluster to random redistribution within the measured cluster boundaries. 10 different randomisations were performed for each dataset, analysed as described above, giving an estimated nanocluster detection rate below 0.5% within the randomly re-distributed localisations dataset. Nanocluster volumes were determined using MATLAB inbuilt functions ‘alphaShape’ and density was calculated as a number of localisations per volume. Diameters were defined as the longest axis across the nanocluster area, determined by the function ‘boundary’.
Statistical testing of rectification index dual recordings was performed using a two-tailed paired t test, as data are normally distributed paired data. Testing of amplitude comparisons from dual recordings was performed using a two-sided Wilcoxon matched-pairs signed-rank test on data prior to normalisation, as peak amplitudes are nonparametric. All bars represent mean value, ± standard error of the mean (SEM) unless stated otherwise. For surface receptor imaging in tissue, box and whiskers represent mean, upper and lower quartiles (box), and min to max values (whiskers). Samples were statistically tested using a Mann–Whitney test as data is nonparametric. The expression levels (Supplementary Figure 7a, b) and synaptic properties, such as areas, densities, localisations and volumes (Fig. 5i, j, Supplementary Figure 7g, h) of ΔNTD AMPARs and corresponding full-length receptors in cultured hippocampal neurons were compared using unpaired ttest as data were normally distributed; corresponding data are shown as mean ± S.E.M. For nonparametric data such as surface distribution (Fig. 5d), and nanocluster properties as number, diameter, localization, densities and volumes (Fig. 6b–e, Supplementary Figure 7k) statistical significance was assessed using Mann–Whitney test and data is shown as median ±95% confidence interval (CI). Relative frequency distributions (Supplementary Figure 7i, j) were analysed using Kolmogorov–Smirnov test. Slopes of linear regressions (Fig. 5k, l) were compared with F test. Sample sizes and biological replicates are given in the figure legends. Statistical analysis was performed in GraphPad Prism 8. Throughout the manuscript, * indicates p < 0.05, **p < 0.01, ***p < 0.001 and ns specifies no significance.
Further information on research design is available in the Nature Research Reporting Summary linked to this article.
All data supporting the findings of this study are provided within the paper and its supplementary information. A source data file is provided with this paper. All additional information will be made available upon reasonable request to the authors. Source data are provided with this paper.
The image analysis code supporting the current study is available from the corresponding author on request, and is derived from original OPTICS MATLAB scripts by Michal Daszykowski and Alex Kendall. These can be found at http://chemometria.us.edu.pl/index.php?goto=downloads, and https://github.com/alexgkendall/OPTICS_Clustering, respectively.
Traynelis, S. F. et al. Glutamate receptor ion channels: structure, regulation, and function. Pharmacol. Rev. 62, 405–496 (2010).
Pinggera, A., Watson, J. F. & Greger, I. H. AMPA Receptors. eLS vol. 1 635–644 (Wiley, 2020).
Malinow, R., Mainen, Z. F. & Hayashi, Y. LTP mechanisms: from silence to four-lane traffic. Curr. Opin. Neurobiol. 10, 352–357 (2000).
Nicoll, R. A. A brief history of long-term potentiation. Neuron 93, 281–290 (2017).
Makino, H. & Malinow, R. AMPA receptor incorporation into synapses during LTP: the role of lateral movement and exocytosis. Neuron. 64, 381–390 (2009).
Penn, A. C. et al. Hippocampal LTP and contextual learning require surface diffusion of AMPA receptors. Nature. 549, 384–388 (2017).
Choquet, D. Linking nanoscale dynamics of AMPA receptor organization to plasticity of excitatory synapses and learning. J. Neurosci. 38, 9318–9329 (2018).
MacGillavry, H. D., Song, Y., Raghavachari, S. & Blanpied, T. A. Nanoscale scaffolding domains within the postsynaptic density concentrate synaptic ampa receptors. Neuron. 78, 615–622 (2013).
Nair, D. et al. Super-resolution imaging reveals that AMPA receptors inside synapses are dynamically organized in nanodomains regulated by PSD95. J. Neurosci. 33, 13204–13224 (2013).
Raghavachari, S. & Lisman, J. E. Properties of quantal transmission at CA1 synapses. J. Neurophysiol. 92, 2456–2467 (2004).
Tang, A. H. et al. A trans-synaptic nanocolumn aligns neurotransmitter release to receptors. Nature. 536, 210–214 (2016).
Xie, X., Liaw, J. S., Baudry, M. & Berger, T. W. Novel expression mechanism for synaptic potentiation: alignment of presynaptic release site and postsynaptic receptor. Proc. Natl. Acad. Sci. USA. 94, 6983–6988 (1997).
Greger, I. H., Watson, J. F. & Cull-Candy, S. G. Structural and functional architecture of AMPA-type glutamate receptors and their auxiliary proteins. Neuron. 94, 713–730 (2017).
Lu, W. et al. Subunit composition of synaptic AMPA receptors revealed by a single-cell genetic approach. Neuron. 62, 254–268 (2009).
Schwenk, J. et al. Regional diversity and developmental dynamics of the AMPA-receptor proteome in the mammalian brain. Neuron 84, 41–54 (2014).
Rouach, N. et al. TARP γ-8 controls hippocampal AMPA receptor number, distribution and synaptic plasticity. Nat. Neurosci. 8, 1525–1533 (2005).
Schnell, E. et al. Direct interactions between PSD-95 and stargazin control synaptic AMPA receptor number. Proc. Natl Acad. Sci. 99, 13902–13907 (2002).
Opazo, P., Sainlos, M. & Choquet, D. Regulation of AMPA receptor surface diffusion by PSD-95 slots. Curr. Opin. Neurobiol. 22, 453–460 (2012).
Sheng, M. & Kim, E. The postsynaptic organization of synapses. Cold Spring Harb. Perspect. Biol. 3, a005678 (2011).
Sumioka, A. et al. PDZ binding of TARPγ-8 controls synaptic transmission but not synaptic plasticity. Nat. Neurosci. 14, 1410–1412 (2011).
Patriarchi, T., Buonarati, O. R. & Hell, J. W. Postsynaptic localization and regulation of AMPA receptors and Cav1.2 by β2 adrenergic receptor/PKA and Ca 2+/CaMKII signaling. EMBO J. 37, e99771 (2018).
Shepherd, J. D. & Huganir, R. L. The cell biology of synaptic plasticity: AMPA receptor trafficking. Annu. Rev. Cell Dev. Biol. 23, 613–643 (2007).
Barry, M. F. & Ziff, E. B. Receptor trafficking and the plasticity of excitatory synapses. Curr. Opin. Neurobiol. 12, 279–286 (2002).
Hayashi, Y. et al. Driving AMPA receptors into synapses by LTP and CaMKII: requirement for GluR1 and PDZ domain interaction. Science. 287, 2262–2267 (2000).
Shi, S. H., Hayashi, Y., Esteban, J. A. & Malinow, R. Subunit-specific rules governing AMPA receptor trafficking to synapses in hippocampal pyramidal neurons. Cell. 105, 331–343 (2001).
Granger, A. J., Shi, Y., Lu, W., Cerpas, M. & Nicoll, R. A. LTP requires a reserve pool of glutamate receptors independent of subunit type. Nature. 493, 495–500 (2013).
Zhou, Z. et al. The C-terminal tails of endogenous GluA1 and GluA2 differentially contribute to hippocampal synaptic plasticity and learning. Nat. Neurosci. 21, 50–65 (2018).
Watson, J. F., Ho, H. & Greger, I. H. Synaptic transmission and plasticity require AMPA receptor anchoring via its N-terminal domain. Elife. 6, e23024 (2017).
Díaz-Alonso, J. et al. Subunit-specific role for the amino-terminal domain of AMPA receptors in synaptic targeting. Proc. Natl Acad. Sci. 114, 7136–7141 (2017).
Jiang, C. H., Wei, M., Zhang, C. & Shi, Y. S. The amino-terminal domain of GluA1 mediates LTP maintenance via interaction with neuroplastin-65. Proc. Natl. Acad. Sci. USA. 118, e2019194118 (2021).
O’Brien, R. J. et al. Synaptic clustering of AMPA receptors by the extracellular immediate- early gene product Narp. Neuron 23, 309–323 (1999).
García-Nafría, J., Herguedas, B., Watson, J. F. & Greger, I. H. The dynamic AMPA receptor extracellular region: a platform for synaptic protein interactions. J. Physiol. 594, 5449–5458 (2016).
Panicker, S., Brown, K. & Nicoll, R. A. Synaptic AMPA receptor subunit trafficking is independent of the C terminus in the GluR2-lacking mouse. Proc. Natl Acad. Sci. 105, 1032–1037 (2008).
Mack, V. et al. Conditional restoration of hippocampal synaptic potentiation in Glur-A-deficient mice. Science. 292, 2501–2504 (2001).
Zamanillo, D. et al. Importance of AMPA receptors for hippocampal synaptic plasticity but not for spatial learning. Science. 284, 1805–1811 (1999).
Park, P. et al. The role of calcium-permeable AMPARs in long-term potentiation at principal neurons in the rodent hippocampus. Front. Synaptic Neurosci. 10, 42 (2018).
Lee, H. K. et al. Phosphorylation of the AMPA receptor GluR1 subunit is required for synaptic plasticity and retention of spatial memory. Cell. 112, 631–643 (2003).
Shi, Y., Lu, W., Milstein, A. D. & Nicoll, R. A. The stoichiometry of AMPA receptors and TARPs varies by. Neuronal Cell Type Neuron. 62, 633–640 (2009).
Sheng, N. et al. LTP requires postsynaptic PDZ-domain interactions with glutamate receptor/auxiliary protein complexes. Proc. Natl Acad. Sci. 115, 3948–3953 (2018).
Arimori, T. et al. Fv-clasp: an artificially designed small antibody fragment with improved production compatibility, stability, and crystallizability. Structure. 25, 1611–1622.e4 (2017).
Ankerst, M., Breunig, M. M., Kriegel, H. P. & Sander, J. OPTICS: ordering points to identify the clustering structure (ACM Press, 1999).
Daszykowski, M., Walczak, B. & Massart, D. L. Looking for natural patterns in data. Part 1. Density-based approach. Chemom. Intell. Lab. Syst. 56, 83–92 (2001).
Daszykowski, M., Walczak, B. & Massart, D. L. Looking for natural patterns in analytical data. 2. Tracing local density with OPTICS. J. Chem. Inf. Comput. Sci. https://doi.org/10.1021/ci010384s (2002).
Malinow, R. & Malenka, R. C. AMPA receptor trafficking and synaptic plasticity. Annu. Rev. Neurosci. 25, 103–126 (2002).
Chen, L. et al. Stargazin regulates synaptic targeting of AMPA receptors by two distinct mechanisms. Nature. 408, 936–943 (2000).
Bats, C., Groc, L. & Choquet, D. The Interaction between stargazin and PSD-95 regulates AMPA receptor surface trafficking. Neuron. 53, 719–734 (2007).
Opazo, P. et al. CaMKII triggers the diffusional trapping of surface AMPARs through phosphorylation of stargazin. Neuron. 67, 239–252 (2010).
Inamura, M. et al. Differential localization and regulation of stargazin-like protein,??-8 and stargazin in the plasma membrane of hippocampal and cortical neurons. Neurosci. Res. 55, 45–53 (2006).
Park, J. et al. CaMKII phosphorylation of TARPγ-8 is a mediator of LTP and learning and memory. Neuron. 92, 75–83 (2016).
Menuz, K., Kerchner, G. A., O’Brien, J. L. & Nicoll, R. A. Critical role for TARPs in early development despite broad functional redundancy. Neuropharmacology. 56, 22–29 (2009).
Park, P. et al. On the Role of Calcium-Permeable AMPARs in Long-Term Potentiation and Synaptic Tagging in the Rodent Hippocampus. Front. Synaptic Neurosci. 11, 4 (2019).
Cull-Candy, S., Kelly, L. & Farrant, M. Regulation of Ca2+-permeable AMPA receptors: synaptic plasticity and beyond. Curr. Opin. Neurobiol. 16, 288–297 (2006).
Li, S. et al. Release sites are positioned to activate NMDA receptors. bioRxiv 2020.05.01.072157. https://doi.org/10.1101/2020.05.01.072157 (2020).
Fukata, Y. et al. Local palmitoylation cycles define activity-regulated postsynaptic subdomains. J. Cell Biol. 202, 145–161 (2013).
Chen, H., Tang, A.-H., Blanpied, T. A., Burrone, J. & Holzbaur, E. Subsynaptic spatial organization as a regulator of synaptic strength and plasticity Transsynaptic alignment can control synaptic function. Curr. Opin. Neurobiol. 51, 147–153 (2018).
Heine, M. & Holcman, D. Asymmetry between pre- and postsynaptic transient nanodomains shapes neuronal communication. Trends Neurosci. 43, 182–196 (2020).
Heine, M. et al. Surface mobility of postsynaptic AMPARs tunes synaptic transmission. Science. 320, 201–205 (2008).
Perez de Arce, K. et al. Topographic mapping of the synaptic cleft into adhesive nanodomains. Neuron. 88, 1165–1172 (2015).
Biederer, T., Kaeser, P. S. & Blanpied, T. A. Transcellular nanoalignment of synaptic function. Neuron. 96, 680–696 (2017).
Bhouri, M. et al. Deletion of LRRTM1 and LRRTM2 in adult mice impairs basal AMPA receptor transmission and LTP in hippocampal CA1 pyramidal neurons. Proc. Natl Acad. Sci. USA. 115, E5382–E5389 (2018).
Dai, J., Aoto, J. & Südhof, T. C. Alternative splicing of presynaptic neurexins differentially controls postsynaptic NMDA and AMPA receptor responses. Neuron. 102, 993–1008.e5 (2019).
Farhy-Tselnicker, I. et al. Astrocyte-secreted glypican 4 regulates release of neuronal pentraxin 1 from axons to induce functional synapse formation. Neuron. 96, 428–445.e13 (2017).
Sia, G. M. et al. Interaction of the N-terminal domain of the AMPA receptor GluR4 subunit with the neuronal pentraxin NP1 mediates GluR4 synaptic recruitment. Neuron. 55, 87–102 (2007).
Lee, S.-J. et al. Presynaptic neuronal pentraxin receptor organizes excitatory and inhibitory synapses. J. Neurosci. https://doi.org/10.1523/JNEUROSCI.2768-16.2016 (2016).
Kim, C. H. et al. Persistent hippocampal CA1 LTP in mice lacking the C-terminal PDZ ligand of GluR1. Nat. Neurosci. 8, 985–987 (2005).
Kerr, J. M. & Blanpied, T. A. Subsynaptic AMPA receptor distribution is acutely regulated by actin-driven reorganization of the postsynaptic density. J. Neurosci. 32, 658–673 (2012).
Kristensen, A. S. et al. Mechanism of Ca2+/calmodulin-dependent kinase II regulation of AMPA receptor gating. Nat. Neurosci. 14, 727–735 (2011).
Buonarati, O. R., Hammes, E. A., Watson, J. F., Greger, I. H. & Hell, J. W. Mechanisms of postsynaptic localization of AMPA-type glutamate receptors and their regulation during long-term potentiation. Sci. Signal. 12, eaar6889 (2019).
Matsuzaki, M., Honkura, N., Ellis-Davies, G. C. R. & Kasai, H. Structural basis of long-term potentiation in single dendritic spines. Nature. 429, 761–766 (2004).
Roth, R. H. et al. Cortical Synaptic AMPA Receptor Plasticity during Motor Learning. Neuron. 105, 895–908.e5 (2020).
Hruska, M. et al. Synaptic nanomodules underlie the organization and plasticity of spine synapses. Nat. Neurosci. 21, 671–682 (2018).
Beaudoin, G. M. J. et al. Culturing pyramidal neurons from the early postnatal mouse hippocampus and cortex. Nat. Protoc. https://doi.org/10.1038/nprot.2012.099 (2012).
Stoppini, L., Buchs, P. A. & Muller, D. A simple method for organotypic cultures of nervous tissue. J. Neurosci. Methods. 37, 173–182 (1991).
García-Nafría, J., Watson, J. F. & Greger, I. H. IVA cloning: a single-tube universal cloning system exploiting bacterial In Vivo Assembly. Sci. Rep. 6, 27459 (2016).
Howarth, M., Takao, K., Hayashi, Y. & Ting, A. Y. Targeting quantum dots to surface proteins in living cells with biotin ligase. Proc. Natl Acad. Sci. USA. 102, 7583–7588 (2005).
Ho, H., Fowle, A., Coetzee, M., Greger, I. H. & Watson, J. F. An inhalation anaesthesia approach for neonatal mice allowing streamlined stereotactic injection in the brain. J. Neurosci. Methods. 342, 108824 (2020).
Rathenberg, J., Nevian, T. & Witzemann, V. High-efficiency transfection of individual neurons using modified electrophysiology techniques. J. Neurosci. Methods. 126, 91–98 (2003).
Elegheert, J. et al. Lentiviral transduction of mammalian cells for fast, scalable and high-level production of soluble and membrane proteins. Nat. Protoc. https://doi.org/10.1038/s41596-018-0075-9 (2018).
Huang, B., Wang, W., Bates, M. & Zhuang, X. Three-dimensional super-resolution imaging by stochastic optical reconstruction microscopy. Science. 319, 810–813 (2008).
Sun, Y., Smirnov, M., Kamasawa, N. & Yasuda, R. Rapid ultrastructural changes of PSD and extrasynaptic axon-spine interface membrane during LTP induced in single dendritic spine. J. Neuro. Sci. https://doi.org/10.1101/840629 (2019).
The authors are very grateful to Andrew Penn for advice and discussions on surface receptor labelling in slice tissue, dissociated culture transfection, and for providing tdTomato and BirAER expression plasmids. This work would not have been possible without support from the Biological Services teams at both the Laboratory of Molecular Biology and Ares facilities. We are also very grateful to Nick Barry and Jerome Boulanger of the LMB Light Microscopy facility for support with confocal and STORM imaging and analysis, Junichi Takagi for providing scFv-Clasp expression constructs, Veronica Chang for assistance with scFv-Clasp protein production, and Nejc Kejzar for assistance with cluster analysis. We would like to thank Teru Nakagawa and Ole Paulsen for critical reading of the manuscript and constructive feedback. This work was supported by grants from the Medical Research Council (MC_U105174197) and BBSRC (BB/N002113/1).
The authors declare no competing interests.
Peer review information Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
About this article
Cite this article
Watson, J.F., Pinggera, A., Ho, H. et al. AMPA receptor anchoring at CA1 synapses is determined by N-terminal domain and TARP γ8 interactions. Nat Commun 12, 5083 (2021). https://doi.org/10.1038/s41467-021-25281-4