The δ2 glutamate receptor gates long-term depression by coordinating interactions between two AMPA receptor phosphorylation sites

Abstract

Long-term depression (LTD) commonly affects learning and memory in various brain regions. Although cerebellar LTD absolutely requires the δ2 glutamate receptor (GluD2) that is expressed in Purkinje cells, LTD in other brain regions does not; why and how cerebellar LTD is regulated by GluD2 remains unelucidated. Here, we show that the activity-dependent phosphorylation of serine 880 (S880) in GluA2 AMPA receptor subunit, which is an essential step for AMPA receptor endocytosis during LTD induction, was impaired in GluD2-null cerebellum. In contrast, the basal phosphorylation levels of tyrosine 876 (Y876) in GluA2 were increased in GluD2-null cerebellum. An in vitro phosphorylation assay revealed that Y876 phosphorylation inhibited subsequent S880 phosphorylation. Conversely, Y876 dephosphorylation was sufficient to restore S880 phosphorylation and LTD induction in GluD2-null Purkinje cells. Furthermore, megakaryocyte protein tyrosine phosphatase (PTPMEG), which binds to the C terminus of GluD2, directly dephosphorylated Y876. These data indicate that GluD2 gates LTD by coordinating interactions between the two phosphorylation sites of the GluA2.

Fig. 1.

Increased tyrosine phosphorylation in GluD2-null mice. (A) A schematic diagram illustrating the positions of Y876 and S880 at the C terminus of GluA2. LTD-inducing stimulation activates the mGluR1 and PKC to phosphorylate S880, which is an essential step for AMPA receptor endocytosis during LTD. mGluR1 activation also induces _slowEPSCs through the transient receptor potential channel (TRPC). In this paper, we have discussed whether and how GluD2 regulates these processes. (B) Basal-state phosphorylation of Y876 and S880 of GluA2 subunits of AMPA receptors in GluD2-null mice. Representative immunoblot images of the synaptosomal fraction of WT and GluD2-null cerebellum using an antibody against phosphorylated Y876 (pY876) or S880 (pS880) are shown in Upper. Lower shows intensities of pY876 (Left) or pS880 (Right) bands in GluD2-null cerebellum normalized by intensities of pY876 or pS880 bands in WT cerebellum. Basal-state phosphorylation of Y876, but not S880, was significantly increased in GluD2-null cerebellum. The bar represents mean and SEM. *P < 0.05 (n = 11 for Y876; n = 7 for S880). (C) Reduced _slowEPSC amplitudes in GluD2-null Purkinje cells. By adjusting PF stimulus intensities, similar sizes of PF-evoked _AMPAEPSCs were obtained in GluD2-null and WT Purkinje cells (Top). Then, PFs were stimulated 2–10 times at 100 Hz in the presence of AMPA receptor blockers to evoke _slowEPSCs (Middle). Amplitudes of _slowEPSCs were normalized by amplitudes of _AMPAEPSCs and plotted against the number of stimulations in Bottom. The bar represents mean and SEM. P = 0.001 (n = 20 each). (D) _slowEPSC amplitudes are enhanced by an SFK inhibitor PP1 analog in both WT and GluD2-null Purkinje cells. PFs were stimulated four times at 100 Hz in the presence of AMPA receptor blockers to evoke _slowEPSCs. When a PP1 analog was not included in the patch pipette [Top; PP1 analog (−)], there were no changes in amplitudes of _slowEPSCs between 0 and 20 min, whereas when a PP1 analog was included (Middle; +), amplitudes of _slowEPSCs at 20 min were significantly increased in both WT and GluD2-null Purkinje cells. Lower shows potentiating effects by a PP1 analog on _slowEPSC amplitudes, which were normalized by amplitudes of _AMPAEPSCs, in WT and GluD2-null Purkinje cells. The bar represents mean and SEM. ***P < 0.005 (n = 11 for each group).

Fig. 2.

Increased Y876 phosphorylation inhibits S880 phosphorylation. (A and B) Chemical LTD stimulus decreases Y876 phosphorylation but increases S880 phosphorylation in WT cerebellum (A), whereas it induces no changes at both sites in GluD2-null cerebellum (B). WT and GluD2-null cerebellar slices were treated with 50 mM KCl plus 10 μM L-glutamate for 5 min (K-glu), and the cell lysates were subjected to immunoblot analyses using phosphorylation-specific antibodies against Y876 (pY876) and S880 (pS880). Representative immunoblot images are shown in Upper. In Lower, intensities of pY876 and pS880 bands in K-glu–treated slices were normalized to those band intensities in untreated slices. The bar represents mean and SEM. *P < 0.05; **P < 0.01, respectively (n = 8 each for WT and n = 13 each for GluD2-null cerebellum). (C and D) Y876 phosphorylation inhibits subsequent S880 phosphorylation in the in vitro phosphorylation assay. The C-terminal region of GluA2 was conjugated with GST (GST-GluA2-CT). Y876 was replaced with phenylalanine to produce GST-GluA2^YF-CT. (C) GST-GluA2-CT or (D) GST-GluA2^Y876F-CT was incubated with Src or boiled Src followed by PKC. Phosphorylation of GluA2 was detected by antiphospho-Y876– and antiphospho-S880–specific antibodies. Representative immunoblot images are shown in Upper. The diagrams show intensities of S880 phosphorylation levels (pS880) with prior Src treatment (+Src) normalized by intensities of pS880 with boiled Src (Ctl). Intensities of Y876 phosphorylation levels (pY876) with subsequent PKC treatment (+PKC) normalized by intensities of pY876 without Src (Ctl) are also shown. The bar represents mean and SEM. **P < 0.01 (n = 11 for pY876 and n = 13 for pS880). (E) Inhibition of tyrosine phosphorylation restores K-glu–evoked changes in Y876 and S880 phosphorylation in GluD2-null cerebellum. Cerebellar slices from GluD2-null mice were preincubated with a PP1 analog for 15 min and then treated with K-glu in the presence of a PP1 analog. K-glu treatment significantly decreased Y876 phosphorylation and increased S880 phosphorylation in GluD2-null cerebellum. Representative immunoblot images are shown in Upper. In Lower, intensities of pY876 and pS880 bands in K-glu–treated slices were normalized to those band intensities in untreated slices. The bar represents mean and SEM. *P < 0.05 (n = 16 for pY876 and n = 17 for pS880).

Fig. 3.

Dephosphorylation of tyrosine of GluA2 is sufficient to restore LTD induction in GluD2-null Purkinje cells. (A–F) Inclusion of a PP1 analog in a patch pipette restored LTD in GluD2-null Purkinje cells, but it did not affect LTD in WT Purkinje cells. Application of CJ-stim (consisting of 30 cycles of PF stimulation plus Purkinje cell depolarization at 1 Hz) induced LTD in WT Purkinje cells with (+) or without (−) a PP1 analog (10 μM) in patch pipettes (A and B). CJ-stim successfully induced LTD in GluD2-null Purkinje cells with (E; +) but not without (D; −) a PP1 analog. Collective results (mean and SEM) are shown in C and F. Insets sweeps in A, B, D, and E show PF-EPSCs just before (black traces) and 30 min after (gray traces) CJ-stim. (G–J) Overexpression of unphosphorable GluA2 subunits restored LTD induction in GluD2-null Purkinje cells. Y869, Y873, and Y876 phosphorylation sites (shown on the right in J) were replaced with phenylalanine to produce unphosphorable GluA2 mutants termed as GluA2^Y876F and GluA2^Y869F,Y873F. GluA2^WT, GluA2^Y876F, and GluA2^Y869F,Y873F were overexpressed in GluD2-null Purkinje cells by a Sindbis virus vector. CJ-stim–induced LTD was restored in GluD2-null Purkinje cells expressing GluA2^Y876F (H) but not Purkinje cells expressing GluA2^WT (G) or GluA2^Y869F,Y873F (I). Collective data (mean and SEM) are shown in J. Insets sweeps in G–I show PF-EPSCs just before (black traces) and 30 min after (gray traces) CJ-stim.

Fig. 4.

Dephosphorylation of tyrosine of GluA2 is necessary for LTD induction in WT Purkinje cells. (A) Diagram indicating three peptides used in this study. Synthetic peptides correspond to the GluA2 C terminus between positions 869 and 877. Alanine replaced three tyrosine residues in the original peptide (pep-3Y) to produce unphosphorable peptide (pep-3A). All tyrosine residues were phosphorylated in pep-3pY. (B–E) LTD is specifically inhibited by pep-3pY, which could serve as a pseudosubstrate for tyrosine phosphatase, included in Purkinje cells. Inclusion of pep-3Y (B) or pep-3A (D) in patch pipettes did not affect CJ-stim–induced LTD in WT Purkinje cells, but pep-3pY (C) inhibited LTD induction. Collective data (mean and SEM) are shown in E. Inset sweeps show PF-EPSCs before and 30 min after CJ-stim.

Fig. 5.

PTPMEG regulates tyrosine dephosphorylation levels at PF–Purkinje cell synapses. (A) Seven C-terminal amino acids of GluD2 determine _slowEPSC amplitudes in Purkinje cells. _slowEPSCs were evoked in GluD2-null Purkinje cells expressing WT GluD2 transgenes (GluD2-null/TgWT) or mutant GluD2 transgenes lacking the seven C-terminal amino acids (GluD2-null/TgCTΔ7) as described in Fig. 1C. Representative traces are shown in Left. Amplitudes and charges of _slowEPSCs were normalized by amplitudes and charges of _AMPAEPSCs and plotted against the number of stimulations in Right. The bar represents mean and SEM. (B and C) Basal phosphorylation of Y876 and S880 of GluA2 subunits in PTPMEG-null mice. Representative immunoblot images of the synaptosomal fraction of WT and PTPMEG-null cerebellum using an antibody against pY876 or pS880 are shown in Upper. Lower shows intensities of pY876 bands (B) or pS880 bands (C) in PTPMEG-null cerebellum normalized by those band intensities in WT cerebellum. Basal phosphorylation of Y876, but not S880, was significantly increased in PTPMEG-null cerebellum. The bar represents mean and SEM. *P < 0.05 (n = 10 for Y876; n = 7 for S880). (D) Chemical LTD stimulus induces no changes at Y876 and S880 sites in PTPMEG-null cerebellum. PTPMEG-null cerebellar slices were treated with 50 mM KCl plus 10 μM L-glutamate for 5 min (K-glu), and the cell lysates were subjected to immunoblot analyses using phosphorylation-specific antibodies against pY876 and pS880. Representative immunoblot images are shown in Upper. In Lower, intensities of pY876 and pS880 bands in K-glu–treated slices were normalized to those band intensities in untreated slices. The bar represents mean and SEM (n = 15 each).

Fig. 6.

Interaction of GluD2 with enzymatically active PTPMEG is necessary for LTD induction. (A–C) The catalytic domain of PTPMEG was sufficient to restore LTD in GluD2-null mice. The catalytic phosphatase domain of PTPMEG was directly fused to the C terminus of a mutant GluD2 lacking the seven C-terminal amino acids (GluD2^ΔCT7) to produce GluD2^ΔCT7-PTP. GluD2^ΔCT7 and GluD2^ΔCT7-PTP were expressed in GluD2-null Purkinje cells by the Sindbis virus vector. CJ-stim induced LTD in GluD2-null Purkinje cells expressing GluD2^ΔCT7-PTP (B) but not Purkinje cells expressing GluD2^ΔCT7 (A). Collective data (mean and SEM) are shown in C. Insets sweeps show PF-EPSCs just before (black traces) and 30 min after (gray traces) CJ-stim. (D–F) Interaction with endogenous PTPMEG is necessary for LTD in WT Purkinje cells. A phosphatase-inactive mutant PTPMEG (PTPMEG^DA) was produced by replacing aspartate in the catalytic domain with alanine. The PDZ domain, by which PTPMEG binds to the C terminus of GluD2, was further deleted to produce PTPMEG^DA-ΔPDZ. PTPMEG^DA or PTPMEG^DA-ΔPDZ was transduced into WT Purkinje cells using a Sindbis virus vector. CJ-stim–evoked LTD was inhibited in Purkinje cells expressing PTPMEG^DA (D) but not Purkinje cells expressing PTPMEG^DA-ΔPDZ (E), indicating that PTPMEG^DA bound to GluD2 and replaced endogenous PTPMEG as a dominant-negative molecule. Collective data (mean and SEM) are shown in F. Insets sweeps show PF-EPSCs just before (black traces) and 30 min after (gray traces) CJ-stim.

Fig. 7.

GluA2 is a substrate of PTPMEG. (A) A substrate-trap mutant of PTPMEG interacts with GluA2. The cell lysate of HEK293 cells transfected with GluA2 was pulled down with the catalytic domain of PTPMEG fused with GST (GST-PTP). Aspartate in the catalytic center was replaced with alanine to produce a substrate-trap mutant (GST-PTP^DA). As reported for a related phosphatase PTPH1, endogenous VCP was trapped by GST-PTP^DA but not GST-PTP. N-ethylmaleimide–sensitive factor (NSF), a protein related to VCP, was not pulled down by GST-PTP^DA or GST-PTP. Similarly, GluA2 was more effectively pulled down by GST-PTP^DA. (B) GluA2 C-terminal peptides are directly dephosphorylated by PTPMEG in vitro. Cysteine in the catalytic center was replaced with serine to produce another substrate trap mutant (GST-PTP^CS). A tyrosine-phosphorylated synthetic peptide, pep-3pY, derived from the C terminus (869–877) of GluA2 was incubated with GST-PTP, GST-PTP^DA, GST-PTP^CS, or GST alone and subjected to the isobaric tag-based quantitative mass spectrometric analyses. The diagram indicates phosphorylation levels of peptides treated with each GST-fusion protein normalized by phosphorylation levels of peptides treated with GST alone. Bars represent mean and SEM. *P < 0.05 (n = 6 for each group). (C) PTPMEG dephosphorylates Y876 of GluA2 in HEK293 cells. GluA2 and WT or a phosphatase-inactive mutant (DA) PTPMEG were coexpressed in HEK293 cells, and the cell lysates were subjected to immunoblot analyses using a phosphorylation-specific antibody against pY876. Representative immunoblot images are shown in Left. Right shows intensities of pY876 bands normalized by those band intensities in cells expressing WT PTPMEG. Phosphorylation levels of Y876 of GluA2 were significantly higher in cells coexpressing a phosphatase-inactive mutant PTPMEG. The bar represents mean and SEM. *P < 0.05 (n = 8 for each group).

Fig. 8.

A proposed role of GluD2 in the regulation of inducibility of cerebellar LTD. (A) In WT cerebellum, GluD2 maintains low levels of phosphorylation at Y876 of the GluA2 subunit of AMPA receptors through PTPMEG that binds to the C terminus. LTD-inducing stimuli further dephosphorylate Y876 by unknown mechanisms, including activation of PTPMEG by conformational changes of GluD2 and inactivation of SFKs. Y876 dephosphorylation allows S880 phosphorylation by PKC, leading to the replacement of anchoring proteins from glutamate receptor interacting protein (GRIP) to PICK1 and allowing AMPA receptor endocytosis during LTD. (B) In GluD2-null and PTPMEG-null Purkinje cells, high basal-state phosphorylation at Y876 prevents subsequent phosphorylation at S880 of GluA2 during LTD-inducing stimulus, thus inhibiting GluA2 endocytosis and LTD.