AMPA receptor pHluorin-GluA2 reports NMDA receptor-induced intracellular acidification in hippocampal neurons

Abstract

NMDA receptor activation promotes endocytosis of AMPA receptors, which is an important mechanism underlying long-term synaptic depression. The pH-sensitive GFP variant pHluorin fused to the N terminus of GluA2 (pH-GluA2) has been used to assay NMDA-mediated AMPA receptor endocytosis and recycling. Here, we demonstrate that in somatic and dendritic regions of hippocampal neurons a large fraction of the fluorescent signal originates from intracellular pH-GluA2, and that the decline in fluorescence in response to NMDA and AMPA primarily describes an intracellular acidification, which quenches the pHluorin signal from intracellular receptor pools. Neurons expressing an endoplasmic reticulum-retained mutant of GluA2 (pH-GluA2 ΔC49) displayed a larger response to NMDA than neurons expressing wild-type pH-GluA2. A similar NMDA-elicited decline in pHluorin signal was observed by expressing cytosolic pHluorin alone without fusion to GluA2 (cyto-pHluorin). Intracellular acidification in response to NMDA was further confirmed by using the ratiometric pH indicator carboxy-SNARF-1. The NMDA-induced decline was followed by rapid recovery of the fluorescent signal from both cyto-pHluorin and pH-GluA2. The recovery was sodium-dependent and sensitive to Na⁺/H⁺-exchanger (NHE) inhibitors. Moreover, recovery was more rapid after shRNA-mediated knockdown of the GluA2 binding PDZ domain-containing protein interacting with C kinase 1 (PICK1). Interestingly, the accelerating effect of PICK1 knockdown on the fluorescence recovery was eliminated in the presence of the NHE1 inhibitor zoniporide. Our results indicate that the pH-GluA2 recycling assay is an unreliable assay for studying AMPA receptor trafficking and also suggest a role for PICK1 in regulating intracellular pH via modulation of NHE activity.

The NMDA receptor (NMDAR) and the AMPA receptor (AMPAR) are ionotropic glutamate receptors mediating excitatory neurotransmission in the brain. NMDARs are critical for the induction of synaptic plasticity and are heteromeric complexes of the homologous subunits NR1, NR2A–D, and NR3A–B (1, 2). AMPARs are responsible for fast excitatory synaptic transmission and are heteromeric complexes of the homologous subunits GluA1–4 (3). Regulation of AMPAR subunit composition and synaptic targeting is important for the expression of certain forms of synaptic plasticity such as long-term potentiation and long-term depression (LTD) (3). NMDAR activation promotes internalization of GluA2-containing AMPARs, which is believed to be critical for LTD expression (4, 5).

Ecliptic pHluorin is a pH-sensitive variant of GFP, which is essentially nonfluorescent at pH values <6.0 but brightly fluorescent at neutral pH (6). By fusing pHluorin to surface receptors it has been possible to assay both receptor internalization into acidic endocytic vesicles and reinsertion of the receptors into the plasma membrane (7–10). To study trafficking of GluA2-containing AMPARs, Ashby et al. (7) fused pHluorin to the extracellular N terminus of GluA2 (pH-GluA2). In pH-GluA2–expressing hippocampal neurons, a marked decline in fluorescence signal was observed upon NMDA stimulation followed by rapid recovery of the signal (7). This was suggested to reflect internalization of surface-expressed receptors followed by fast recycling back to the plasma membrane (7). Subsequently, pH-GluA2 has been used in several studies aimed at investigating mechanisms controlling GluA2 trafficking (7, 8, 11–13).

Protein interacting with C kinase 1 (PICK1) is a dimeric cytosolic PDZ (PSD-95/Discs-large/ZO-1 homology) domain protein that binds the extreme C terminus of GluA2/3 AMPAR subunits as well as the C termini of several other receptors, various ion channels, and neurotransmitter transporters (14–18). The functional role of PICK1 has mainly been described in relation to the regulation of AMPAR trafficking in which PICK1 is thought to regulate receptor surface levels by restricting AMPAR recycling and/or by promoting receptor internalization (19–21). Notably, the interaction of GluA2 with PICK1, as well as with other intracellular scaffolding proteins, is known to be important for NMDAR-induced LTD, and in agreement LTD is severely compromised in knockout mice lacking these proteins (20, 22–26). However, the specific mechanisms underlying regulation of GluA2 trafficking remain unclear.

Here, we analyze the rapid NMDA-induced decline and recovery of pHluorin signal from pH-GluA2 in hippocampal neurons. Our data provide evidence that the changes in fluorescence signal from pH-GluA2 cannot be attributed to internalization and recycling of the receptor, although they do not dispute that such trafficking is indeed taking place. Instead, our data suggest that in hippocampal neurons a significant portion of the NMDA-induced decline in fluorescence is caused by NMDA-induced changes in intracellular pH (27–29) and the subsequent quenching of intracellular endoplasmic reticulum (ER)-localized pH-GluA2 receptors. Interestingly, our data further suggest that the acidification rapidly recovers in a manner sensitive to Na⁺/H⁺-exchanger (NHE) inhibition and PICK1 expression.

Results

Synaptic, but Not Somatic, pH-GluA2 Is Accessible to Extracellular Thrombin Cleavage.

In agreement with previous findings (7, 26), NMDA stimulation of pH-GluA2–expressing hippocampal neurons induced in the soma and dendrites a rapid decrease in pH-GluA2 fluorescence, which subsequently recovered to the original baseline after ∼25–30 min (Fig. 1, control). The amplitude of the response was smaller in dendrites than in soma, whereas no significant response was seen in spines (Fig. 1 B–D). This distinct pattern of rapid NMDAR-mediated fluorescence decline has been interpreted as representing AMPAR internalization at extrasynaptic and somatodendritic sites (7, 8). We decided to assess the contribution of surface-expressed pH-GluA2 to the fluorescent signal by exploiting an extracellular thrombin protease cleavage site between pHluorin and GluA2 in pH-GluA2. Thrombin perfusion efficiently removed ∼70% of the pHluorin signal from spines and ∼50% of the signal from branched dendrites (Fig. 1 A–C). However, thrombin removed only ∼10% of the signal from the cell somas (Fig. 1D). Subsequent perfusion with NMDA caused an additional ∼10% decrease in pH-GluA2 fluorescence from spines, ∼30% decrease in signal from branched dendrites, and ∼70% decrease in signal from the cell somas, and the fluorescence signal only recovered to postthrombin levels (Fig. 1 B–D). Interestingly, after thrombin cleavage the magnitude of the pHluorin response to NMDA treatment was comparable to that seen without thrombin cleavage. Because thrombin cleavage should only occur extracellularly, these results suggest that the majority of pH-GluA2 in spines and dendrites is on the surface of the neurons. In contrast, only a small fraction of pH-GluA2 in the soma could be cleaved, indicating that the receptors are inaccessible and therefore likely located intracellularly, despite a bright fluorescent signal.

Fig. 1.

Thrombin cleavage of extracellular pHluorin revealed a significant fraction of intracellular pH-GluA2 receptors responsive to NMDA treatment in soma and dendrites. Hippocampal neurons were transfected with pH-GluA2 and perfused for 20 min with thrombin to cleave extracellular pHluorin, followed by stimulation with 20 µM NMDA for 5 min. (A) Representative confocal images of a hippocampal neuron before thrombin cleavage (1), after cleavage (2), and after NMDA stimulation (3). (Scale bar, 20 µm.) (B) Representative time courses of the normalized fluorescence intensity dF/F₀ from a control neuron (black line) and a thrombin-treated neuron (grey line) in spines (means of five spine regions), (C) a dendritic region, and (D) the cell soma. Thrombin, NMDA, or buffer alone was applied at the indicated time points. Black dashed lines indicate baseline levels. The thrombin cleavage curve was fitted to a standard exponential decay equation (red dashed line). The experiments shown are representative of three independent experiments.

The NMDA-Induced Decrease in Signal Derives Primarily from Intracellular pH-GluA2.

Most cellular organelles are acidic (pH <∼6.5). However, the ER has a neutral pH, allowing pHluorin to be fluorescent (pH ∼7.2) if present in this compartment (30). Truncation of the intracellular C terminus (ΔC49) of GluA2 results in ER retention of the receptor (31). To test whether ER-localized pH-GluA2 responds to NMDA stimulation, we expressed pH-GluA2 ΔC49 in the hippocampal neurons, where it displayed diffuse localization without synaptic clustering, consistent with ER retention (Fig. 2 A and B). Despite comparable pHluorin signal intensity, immunolabeling demonstrated a dramatic reduction in surface expression of pH-GluA2 ΔC49 compared with pH-GluA2 (Fig. S1). Nevertheless, NMDA stimulation of pH-GluA2 ΔC49 caused a larger fluorescence decline compared with that of pH-GluA2 in both somatic and dendritic areas (Fig. 2 C and D). This suggests that the pH-GluA2 signal and the NMDA-induced decrease in both soma and dendrites are derived primarily from intracellular receptors that most likely reside in the ER. To exclude the possibility that the NMDA-induced decline in fluorescence reflected trafficking of pH-GluA2 from the ER to the Golgi (pH ∼6.5), we blocked the vacuolar H⁺-ATPase with bafilomycin A1 to increase pH inside the Golgi. Importantly, this treatment had no apparent effect on the NMDA-induced fluorescence decline (Fig. S2).

Fig. 2.

A truncated, ER-retained pH-GluA2 receptor showed a larger response to NMDA. (A) Representative confocal image of a hippocampal neuron expressing pH-GluA2 ΔC49. (Scale bar, 20 µm.) The construct has a more diffuse localization compared pH-GluA2 (B). The pHluorin signal during 5-min NMDA stimulation was quantified from soma (C) and dendrites (D) of pH-GluA2 ΔC49–expressing neurons (grey) and pH-GluA2–expressing neurons (black). The curves show the time course of the normalized average pHluorin fluorescence intensity dF/F₀ ± SE, n = 10 neurons and n = 12 neurons, respectively.

Both NMDA and AMPA Induce Intracellular Acidification.

Our data suggest that pH-GluA2 and pH-GluA2 ΔC49 report an intracellular acidification upon NMDA receptor activation, which has previously been described in dissociated neurons as well as in acute and cultured slices from hippocampus (27–29). To test this we expressed cytosolic pHluorin (cyto-pHluorin). The construct showed an even distribution in the cell soma, nucleus, dendritic areas, and spines without synaptic clustering (Fig. 3A). Consistent with intracellular acidification, stimulation of cyto-pHluorin–expressing neurons with NMDA led to fast quenching of the fluorescent signal both in the soma and in dendrites (Fig. 3 B–D). The amplitude of the decline was larger than that for pH-GluA2 and more similar to that observed for ER-retained pH-GluA2 ΔC49. We also tested the effect of 100 µM AMPA in the presence of the NMDAR blocker dl-2-amino-5-phosphonovaleric acid (DL-APV). Interestingly, this also produced a response from cyto-pHluorin–expressing neurons, as well as from pH-GluA2–expressing neurons (Fig. 3 E and F). The response was similar to that observed upon NMDA treatment (Fig. 3 C and D) and, thus, intracellular acidification can be induced also by direct activation of AMPARs.

Fig. 3.

NMDA- and AMPA-induced intracellular acidification is sufficient for fluorescence quenching of cytosolic pHluorin. (A) Representative image of a hippocampal neuron expressing cyto-pHluorin. The pHluorin signal was almost completely quenched during NMDA treatment. (Scale bar, 20 µm.) (B) The pHluorin signal intensity during 5-min NMDA stimulation was quantified from soma (C) and dendrites (D) of cyto-pHluorin–expressing neurons (grey) and pH-GluA2–expressing neurons (black). The curves show the time course of the normalized average pHluorin fluorescence intensity dF/F₀ ± SE, n = 16 neurons and n = 12 neurons, respectively. Cyto-pHluorin showed a larger response to NMDA compared with pH-GluA2. The pHluorin signal intensity during 2-min AMPA stimulation (100 µM) was quantified from soma (E) and dendrites (F) of cyto-pHluorin–expressing neurons (grey) and pH-GluA2–expressing neurons (black). The curves show the time course of the normalized average pHluorin fluorescence intensity dF/F₀ ± SE, n = 3 neurons for all conditions. (G) Representative confocal images of hippocampal neurons loaded with the ratiometric pH-sensitive cell dye SNARF-1 at different intracellular pH. The SNARF-1 signal was calibrated using cell-permeating nigericin and extracellular buffer with pH values as indicated. (H) The average time course of the ratiometric SNARF-1 fluorescence at the indicated wavelengths is shown as means ± SE, n = 3 experiments. (I) The average time course of the ratiometric SNARF-1 fluorescence during 5-min NMDA stimulation using the pH-calibrated microscope settings is shown as means ± SE, n = 3 neurons. The amplitude of the NMDA-induced fluorescence response indicates an intracellular acidification of pH <6.5.

Membrane-permeable dynamin inhibitors are being used to demonstrate dynamin-dependent endocytosis of receptors. We decided to test whether an inhibitor of dynamin might also affect changes in cyto-pHluorin signal. Dynasore (80 µM) attenuated the NMDA-induced pH-GluA2 response. However, the NMDA-induced cyto-pHluorin response was also significantly affected by Dynasore (80 µM) (Fig. S3). Thus, dynamin inhibition by Dynasore is not a reliable control to validate trafficking of pHluorin-tagged receptors.

To measure the absolute pH shift induced by NMDA in hippocampal neurons, we used the ratiometric long-wavelength seminaphthorhodafluor carboxy-SNARF-1. We first performed an in situ calibration experiment in which we loaded carboxy-SNARF-1 into neurons and recorded the dual emission from the dye while superfusing the cells with pH-controlled imaging buffers. To equilibrate the intracellular pH with the extracellular buffer, we used the ionophore nigericin (20 µM) in the presence of 100 mM KCl (Fig. 3 G and H). Using identical microscopic settings, we observed NMDA-induced intracellular acidosis corresponding to pH values of 6.0–6.5, sufficient for almost complete quenching of the pHluorin signal (Fig. 3I and ref. 6).

Surface exposure of pHluorin is commonly confirmed by application of extracellular acidic MES buffer, pH 6.0 (7). Consistent with quenching of surface pHluorin signal, application of acidic buffer caused in dendritic areas a rapid decrease in pH-GluA2 fluorescence followed by rapid recovery upon perfusion with artificial cerebrospinal fluid (ACSF), pH 7.4 (Fig. S4 A and B). In the soma, where most receptors are likely localized to the ER, perfusion with acidic buffer also led to a decrease in signal followed by recovery in ACSF, although the time course seemed somewhat slower (Fig. S4 A and B). Interestingly, application of acidic MES buffer also decreased the fluorescence from cyto-pHluorin and ER-retained pH-GluA2 ΔC49 with a time course similar to that observed in the somas for pH-GluA2 (Fig. S4 C and D). These data indicate a high permeability of the cell membrane to protons, in agreement with previous results (32). They also call into question the use of acidic MES buffer as a means to evaluate surface expression of pHluorin-tagged receptors.

PICK1 Regulates the Na⁺/H⁺ Exchanger-Mediated Recovery from NMDA-Induced Intracellular Acidosis.

Previously, it was shown, using hippocampal cultures from PICK1 knockout mice or a small molecule inhibitor of PICK1, that the absence or inhibition of PICK1, respectively, accelerates the recovery of the pH-GluA2 fluorescent signal upon NMDA stimulation (8, 11). To test whether this effect was specific for pH-GluA2, we used shRNA to knock down (KD) PICK1 in hippocampal neurons expressing cyto-pHluorin (Fig. 4 A and B and Fig. S5). As for pH-GluA2, we observed a small but statistically significant decrease in amplitude (Fig. 4C). Moreover, the recovery half-time (t_1/2) was significantly decreased in PICK1 KD neurons compared with control (Fig. 4D).

Fig. 4.

PICK1 KD affects the NHE-mediated recovery from intracellular acidosis after NMDA stimulation in hippocampal neurons. Hippocampal neurons were cotransfected with cyto-pHluorin and either control mCherry or PICK1 shRNA (KD) with an mCherry reporter. (A) Representative confocal images showing the cyto-pHluorin fluorescence from control and PICK1 KD neurons at the indicated time points during 5-min NMDA stimulation. (Scale bar, 20 µm.) (B) The time course of the normalized average pHluorin fluorescence intensity dF/F₀ is shown from somas of control neurons (black) and PICK1 KD neurons (purple). (C) Quantification of the amplitude of the pHluorin signal response. (D) Quantification of the pHluorin signal recovery half-time t_1/2. Data are means ± SE, n = 11 neurons from each condition. PICK1 KD significantly decreased the response amplitude and accelerated the pHluorin signal recovery after NMDA stimulation compared with controls, *P = 0.03 and *P = 0.01, respectively, unpaired t tests. (E) Neurons were treated with 10 µM or 100 µM amiloride for 10 min during the recording period. Data are means ± SE, n = 5 neurons for each condition. A concentration of 10 µM amiloride significantly decelerated the recovery compared with controls, *P = 0.04. No significant effect of PICK1 KD was observed when the neurons were treated with 10 µM amiloride, P > 0.05, one-way ANOVA and Tukey’s post hoc test. A concentration of 100 µM amiloride almost eliminated the recovery within the recording period. (F) Neurons were treated with 100 nM of the selective NHE1 inhibitor zoniporide. Data are means ± SE, n = 8 neurons. Zoniporide also significantly decelerated the recovery t_1/2 compared with controls, *P = 0.04. Again, no significant effect of PICK1 KD was observed when the neurons were treated with zoniporide, P > 0.05, one-way ANOVA and Tukey’s post hoc test.

To test whether the recovery from NMDA-induced acidosis involved plasma membrane NHEs, we first studied the effect of extracellular sodium depletion and found that the recovery was absent in choline-substituted ACSF (Fig. S6). Next, we tested the NHE blocker amiloride, which at 100 µM should block not only the NHE1 and NHE2 isoforms but also other isoforms, including NHE3 (33, 34). Here, the recovery was essentially eliminated without changing the amplitude of the response (Fig. 4E). At 10 µM amiloride, which should primarily block NHE1, the recovery was significantly prolonged but not abolished (Fig. 4E). This suggests the involvement of more NHE subtypes in the fluorescence recovery, which was further supported by using the NHE inhibitor 5-ethylisopropyl amiloride (EIPA), which at 10 µM would be expected to block NHE1–3 (33, 34) (Fig. S7). Finally, we tested the inhibitor zoniporide at 100 nM, which should selectively block NHE1 (35), and found that it prolonged but did not eliminate recovery (Fig. 4F). Together, these results suggest that proton extrusion through NHEs is a likely mechanism underlying pHluorin fluorescence recovery after NMDA-induced intracellular acidosis.

Interestingly, the accelerating effect of PICK1 KD on the fluorescence recovery was decreased to nonsignificant levels in the presence of 10 µM amiloride (Fig. 4E). Furthermore, the accelerating effect of PICK1 KD on the fluorescence recovery was abolished in the presence of the NHE1 inhibitor zoniporide (Fig. 4F). Thus, PICK1 might regulate recovery after NMDA-induced acidosis via NHEs, including in particular NHE1. The effect of PICK1 on the recovery is unlikely to involve acid-sensing ion channels (ASICs), which bind PICK1 and also are inhibited by amiloride, because the ASIC inhibitor psalmotoxin (PcTX) had no effect on the fluorescence recovery (Fig. S8).

Discussion

We and others have previously used pHluorin fused to GluA2 to study trafficking of AMPARs in response to activation of NMDARs (5, 10, 11, 26, 36, 37). In agreement with data obtained using other assays, the results suggest that NMDAR activation leads to rapid internalization followed by recycling of GluA2-containing AMPARs in the soma and dendrites of hippocampal neurons (5, 10, 11, 26, 36, 37). Nonetheless, we provide here strong evidence that the majority of the NMDA-induced changes in pH-GluA2 signal do not reflect AMPAR trafficking, but rather report an intracellular decrease in pH, which is detected by intracellular pH-GluA2 receptors likely localized to the ER. Our data do not challenge the well-established activity-dependent trafficking of GluA2 but rather reveal that the use of pHluorin fluorescence in studies of GluA2 trafficking, as well as trafficking of other membrane proteins, which are partially ER-retained, is complex and that intracellular pH changes induced by NMDA and AMPA, and possibly other manipulations, make data interpretation difficult.

Several different experimental approaches support our conclusion. First, we showed that thrombin, which cleaves pHluorin from surface pH-GluA2, did not affect the decrease in pHluorin signal in response to NMDA in the soma of hippocampal neurons and only by ∼50% in the dendrites. Second, neurons expressing an ER-retained mutant GluA2 (pH-GluA2 ΔC49) displayed a slightly larger response to NMDA than neurons expressing wild-type pH-GluA2. Third, NMDA-induced fluorescence quenching and subsequent recovery was observed by expressing cytosolic pHluorin alone. Taken together, our data suggest that the pH change in response to NMDA stimulation is almost sufficient to explain the reduction in pHluorin signal. However, we cannot presently exclude that other environmental variables could contribute, such as, for instance, changes in the intracellular chloride concentration, which also may affect fluorescence of GFP variants (38).

Consistent with previous studies (7), we observed that perfusion with acidic MES buffer (pH 6.0) decreased the pH-GluA2 signal in both the soma and dendrites. However, this was also observed for the ER-retained pH-GluA2 ΔC49 and for cyto-pHluorin. Likely, the very rapid decrease in pH-GluA2 signal observed in spines and dendrites reflects quenching of surface-exposed pHluorin, whereas the slower decrease observed in the soma reflects intracellular acidification; hence, these data further support the idea that the somatic pH-GluA2 signal is derived primarily from intracellular receptors.

Lin and Huganir (8) showed that the NMDA-induced decline in pH-GluA2 signal was inhibited by the NMDAR blocker DL-APV and by removal of extracellular Ca²⁺, which indicated that the response was mediated by calcium influx through the NMDARs. However, we observed that AMPA in the presence of DL-APV also caused a substantial decrease in fluorescence from both pH-GluA2 and cyto-pHluorin. Thus, intracellular acidification might be induced directly by AMPAR activation and membrane depolarization. This is consistent with previous data showing that depolarizing concentrations of extracellular K⁺ also promote intracellular acidification (27). The calcium dependency might therefore be explained by membrane depolarization and subsequent activation of voltage-gated calcium channels.

To quantify the NMDA-induced decrease in intracellular pH, we used the pH-sensitive dye carboxy-SNARF-1 and obtained direct evidence that NMDAR activation shifts the intracellular pH down to 6.0–6.5. These values are lower than reported in hippocampal slices but within the same range observed for NMDA-induced acidosis and glutamate excitotoxicity in dissociated hippocampal neurons (27–29). Interestingly, our data suggest a major role of NHEs in mediating the recovery of this large decrease in pH upon NMDA stimulation. NHEs are activated mainly at low intracellular pH and our results with different NHE inhibitors are consistent with involvement of several isoforms, including NHE1–3. Our evidence for an important, but not exclusive, role of NHE1 is based on deceleration, but not elimination, of recovery by the selective NHE1 inhibitor zoniporide. We also provide evidence for a role of the GluA2-binding PDZ-domain protein PICK1 in regulating intracellular pH. Previously, PICK1 was shown to reduce the fluorescence recovery rate of the pH-GluA2 signal after NMDA stimulation (8). Here, we show that PICK1 also reduces the recovery rate of signal from cytosolic pHluorin. Because the effect of PICK1 KD was absent both in the presence of 10 µM amiloride and of 100 nM of zoniporide, it is conceivable that PICK1 exerts its action via negative regulation of NHE1 function. The precise underlying mechanism should be further explored in future experimental efforts.

It is striking that not only PICK1 but also GRIP1 (glutamate-interacting protein 1), another PDZ-domain protein binding the GluA2 C terminus, as well as KIBRA (kidney and brain-expressed protein), which interacts with PICK1, were shown to differentially regulate the recovery of the pH-GluA2 signal (26, 36). In addition, mutations in the GluA2 C terminus that interfere with PICK1 or GRIP1 binding accelerated the recovery rate of the pH-GluA2 signal (8). These findings are interesting because the effect of the three proteins and the GluA2 mutations on the pH-GluA2 recovery rate is paralleled by an effect on LTD expression. It is also striking that cell-permeable peptides targeting dynamin function impaired the NMDA-induced decline in fluorescent signal (8). Furthermore, it is remarkable that synaptic depression has been reported to occur in cultured hippocampal slices in response to intracellular acidosis (39). Taken together, the results suggest a complex relationship in synaptic plasticity between AMPARs, trafficking processes, synaptic scaffolding proteins, and regulation of pH by NHEs.

It has been shown based on both biochemical and immunocytochemical methods that PICK1 and GRIP1-2 can regulate GluA2 recycling (19, 24). Moreover, it was demonstrated that PICK1 reduces recycling of other PDZ-domain binding partners that sort to the RAB11 slow recycling pathway (40). Thus, the present data do not challenge the fact that GluA2-interacting proteins regulate AMPAR recycling but demonstrate limitations of the pH-GluA2 assay. A tempting consideration is that PICK1, together with GluA2 containing AMPARs, as well as other interacting proteins, might affect the overall function of recycling compartments. This might lead to impaired recycling also of the NHEs and thereby decreased acid extrusion capacity after NMDA-induced acidosis. More generally, this could affect trafficking to the plasma membrane also of lipid membrane and other proteins involved in spine growth (41). Indeed, it has been shown that PICK1 KD increases spine size and NMDAR activation results in spine shrinkage, which is blocked by PICK1 KD (21). Thus, PICK1 might play a role in synaptic plasticity that involves not only control of AMPAR trafficking but also overall regulation of spine structure and stability.

Experimental Procedures

Molecular Biology.

The pRK5-pHluorin-GluA2 (pH-GluA2) was generated from the original pcDNA3-vector construct (kindly provided by Jeremy Henley, Bristol, UK) by subcloning into the pRK5-vector. The pH-GluA2 ΔC49 and cyto-pHluorin constructs were generated using PCR amplification and subcloning into the pRK5-pH-GluA2 vector. The FUGW-Sh18 (PICK1 KD)-mCherry construct was generated by PCR amplification of the mCherry-encoding sequence and subcloning into a FUGW-Sh18 (PICK1 KD)-GFP (kindly provided by Robert Malenka, Stanford, CA).

Hippocampal Cultures and pHluorin Assay.

Hippocampal cultures and the pHluorin imaging assay were performed at 25 °C essentially as described (8) (more information is given in SI Experimental Procedures).

Thrombin Assay.

The neurons were continuously perfused with 1 U/mL thrombin protease (Novagen) in imaging buffer for 20 min to cleave extracellular surface pHluorin as described in SI Experimental Procedures.

SNARF-1 pH Calibration Assay.

The 5-(and-6)-carboxy SNARF-1 (10 µM; Molecular Probes) was loaded into hippocampal neurons and the dual emission was imaged as the >585/530–600 nm ratio (more information is given in SI Experimental Procedures).