NMDA receptor GluRepsilon/NR2 subunits are essential for postsynaptic localization and protein stability of GluRzeta1/NR1 subunit

Abstract

In NMDA receptors, GluRepsilon/NR2 subunits strictly require the GluRzeta1/NR1 subunit to exit from endoplasmic reticulum (ER) to the cell surface in vitro and to the postsynapse in vivo, whereas C terminus-dependent self-surface delivery has been demonstrated for the GluRzeta1 subunit in vitro. To test whether this leads to C terminus-dependent self-postsynaptic expression in neurons in vivo, we investigated the GluRzeta1 subunit in cerebellar granule cells lacking two major GluRepsilon subunits, GluRepsilon1/NR2A and GluRepsilon3/NR2C. In the mutant cerebellum, synaptic labeling for the GluRzeta1 subunit containing the C2 (GluRzeta1-C2) or C2' (GluRzeta1-C2') cassette was reduced at mossy fiber-granule cell synapses to the extrasynaptic level. The loss was not accompanied by decreased transcription and translation levels, increased extrasynaptic labeling, or ER accumulation. Quantitative immunoblot revealed substantial reductions in the mutant cerebellum of GluRzeta1-C2 and GluRzeta1-C2'. The most severe deficit was observed in the postsynaptic density (PSD) fraction: mutant levels relative to the wild-type level were 12.3 +/- 3.3% for GluRzeta1-C2 and 17.0 +/- 4.6% for GluRzeta1-C2'. The GluRzeta1 subunit carrying the C1 cassette (GluRzeta1-C1) was, although low in cerebellar content, also reduced to 12.7 +/- 3.5% in the mutant PSD fraction. Considering a trace amount of other GluRepsilon subunits in the mutant cerebellum, the severe reductions thus represent that the GluRzeta1 subunit, by itself, is virtually unable to accumulate at postsynaptic sites, regardless of C-terminal forms. By protein turnover analysis, the degradation of the GluRzeta1 subunit was accelerated in the mutant cerebellum, being particularly rapid for that carrying the C2 cassette. Therefore, accompanying expression of GluRepsilon subunits is essential for postsynaptic localization and protein stability of the GluRzeta1 subunit.

Figure 1.

Production of mutant mice defective in NMDA receptor GluRϵ1 and/or GluRϵ3 subunit genes. A, Schematic representation of genomic DNA, targeting vector, and disrupted gene of the GluRϵ3 subunit. B, Southern blot analysis of EcoRI-digested genomic DNA using three different probes indicated in A. C, In situ hybridization analysis for GluRϵ1, GluRϵ3, and GluRζ1 mRNAs in the wild-type, GluRϵ1-KO, GluRϵ3-KO, and GluRϵ1/ϵ3-KO brains. D, Semiquantitative evaluation of transcription levels for GluRϵ1 (left), GluRϵ3 (center), and GluRζ1 (right) mRNAs in the granular layer of the cerebellum (mean ± SD). The asterisks indicate statistically significant differences between wild-type and mutant mice (Student's t test; p < 0.001). BSK, Plasmid pBluescript; Cb, cerebellum; Cx, cerebral cortex; DT, diphtheria toxin gene; EI, EcoRI; EV, EcoRV; H, HindIII; Hi, hippocampus; N, NotI; S, SalI. Scale bar, (in C) 1 mm.

Figure 2.

Specificity of GluRζ1-C2, GluRζ1-C2′, and GluRζ1-C1 antibodies. A, Immunoblot using untransfected HEK293 cells (U), cells transfected with GluRζ1 containing the C1 and C2 cassettes (NR1-1), cells transfected with GluRζ1 lacking the C1 but containing the C2′ cassette (NR1-4), and cerebellar proteins (Cb). Cerebellar homogenates (30 μg) were loaded for GluRζ1-C2 and GluRζ1-C2′, whereas cerebellar proteins in the PSD fraction (10 μg) were loaded for GluRζ1-C1. B-G, Immunoperoxidase staining for GluRζ1-C2 (B-D) and GluRζ1-C2′ (E-G) subunits. Antibodies were preincubated without (B, E) or with C2 (C, F) and C2′ (D, G) antigen peptides. Scale bar, 1 mm.

Figure 6.

Immunoblot analysis. A, GluRζ1 subunit in wild-type cerebral and cerebellar homogenates. The proteins loaded in lanes for total GluRζ1 (ζ1-pan), GluRζ1-C2, and GluRζ1-C2′ are 10 μg for the cerebrum and 30 μg for the cerebellum, and those for GluRζ1-C1 are 30 and 50 μg, respectively. B, Cerebellar contents of GluRζ1, GluRϵ1, GluRϵ3, PSD-95, and synaptophysin in the wild-type, GluRϵ1-KO, GluRϵ3-KO, and GluRϵ1/ϵ3-KO mice. Each lane contains 30 μg of proteins. C, GluRζ1-C2, GluRζ1-C2′, or calreticulin levels in homogenates (Hom) and P3 fraction (P3) prepared from the wild-type and GluRϵ1/ϵ3-KO cerebella. Each lane contains 30 μg of proteins. D, GluRζ1-C2, GluRζ1-C2′, PSD-95, or synaptophysin levels in homogenates (H), synaptosomal fraction (S), and PSD fraction (P) in the wild-type, GluRϵ1-KO, GluRϵ3-KO, and GluRϵ1/ϵ3-KO mice. The proteins loaded in each lane are 30 μg for homogenates, 10 μg for the synaptosomal fraction, and 3 μg for the PSD fraction. E, GluRζ1-C1 levels in the PSD fraction. Each lane contains 10 μg of PSD fraction proteins. F, GluRϵ2 levels in homogenates prepared from the wild-type cerebral cortex (Cc, Wild) and from the cerebellum of the wild-type (Cb, Wild) or GluRϵ1/ϵ3-KO (Cb, ϵ1/ϵ3-KO) mouse. The proteins loaded in each lane are 10 μg for the cerebral cortex and 30 μg for the cerebellum. G, Histograms showing progressive loss of GluRζ1 proteins in mutant cerebella. The left histograms show the percentage of GluRζ1-C2 and GluRζ1-C2′ levels in the mutant cerebellar homogenates (see left ordinate) and PSD fraction (see right ordinate) relative to the levels in wild-type cerebellar homogenates (mean ± SEM). The right histograms show the percentage of GluRζ1-C1 levels in the PSD fraction of mutant cerebella relative to the level in wild-type PSD fraction (mean ± SEM). The statistical difference relative to the wild-type levels in each fraction was determined by a Student's unpaired t test (^*p < 0.05; ^**p < 0.01).

Figure 3.

Immunohistochemical alterations of GluRζ1-C2 (A-C, E-H) and GluRζ1-C2′ (D, I-L) in mutant mice lacking GluRϵ subunits. A-D, Pairs of wild-type (top) and mutant (bottom; A, GluRϵ1-KO; B, GluRϵ3-KO; C, D, GluRϵ1/ϵ3-KO) brains. Photographs are negative images printed directly from preparates stained by immunoperoxidase using diaminobenzidine as a chromogen. E, F, I, J, Double fluorescence for the GluRζ1 subunit (green) and nuclear staining with propidium iodide (PI; red) in the cerebellar granular layer of the wild-type (E, I) and GluRϵ1/ϵ3-KO (F, J) mice. G, H, K, L, Double immunofluorescence for the GluRζ1 subunit (green) and PSD-95 (red) in the cerebellar granular layer of the wild-type (G, K) and GluRϵ1/ϵ3-KO (H, L) mice. CA1 and CA3, CA1 and CA3 regions of the hippocampus; Cb, cerebellum; CP, caudate-putamen; Cx, cerebral cortex; DG, dentate gyrus; MO, medulla oblongata; OB, olfactory bulb; Th, thalamus. Scale bars: A, 1 mm; E, F, 5 μm.

Figure 4.

Immunoperoxidase for the cerebellar cortex using antibodies specific to GluRζ1-C2 (A-D), GluRζ1-C2′ (E-H), GluRϵ1 (I-L), GluRϵ3 (M-P), and PSD-95 (Q-T). Note a gradual reduction of immunoreactivities for the GluRζ1 subunit from A to D or from E to H. All sections are counterstained with hematoxylin. Cell bodies of Purkinje cells are indicated by asterisks. GL, Granular layer; ML, molecular layer. Scale bar, 20 μm.

Figure 5.

Postembedding immunogold for GluRζ1-C2 (A, C, E) and GluRζ1-C2′ (B, D, F) at mossy fiber-granule cell synapses. Note that postsynaptic labeling for both variants is severely reduced in the GluRϵ1/ϵ3-KO mouse (C, D), in contrast to dense labeling in the wild-type mouse (A, B, arrows). E, F, The mean number of immunogold particles per micrometer of synaptic (white columns) and extrasynaptic (black columns) membranes of asymmetrical synapses in the wild-type and GluRϵ1/ϵ3-KO mice. Bars on the columns represent the SEM. The asterisks indicate statistically significant differences (Student's t test; p < 0.05). Gr, Granule cell dendrite; MF, mossy fiber terminal. Scale bar, 100 nm.

Figure 7.

Transcription, translation, and protein turnover analyses for the GluRζ1 subunit. A, Northern blot analysis for GluRζ1 mRNA in the wild-type and GluRϵ1-KO cerebellar cultures. The total RNA loaded in each lane is 5 μg. B, Immunoblot with GluRζ1-pan and PSD-95 antibodies for cell lysates from wild-type and GluRϵ1-KO cerebellar cultures. Each lane contains 30 μg of proteins. C, Pulse-labeling experiment. After pulse-labeling with [³⁵S]Met/Cys for 30 min, cell lysates were immunoprecipitated with GluRζ1-pan antibody. Immunoprecipitates were then analyzed to compare the amount of precipitated GluRζ1 subunit by immunoblot (Blot: ζ1-pan) and to compare the radioactivity of pulse-labeled GluRζ1 subunit (³⁵S-GluRζ1). D, Protein stability experiment at 4 hr after anisomysin injection to the wild-type and GluRϵ1/ϵ3-KO mice. Cerebellar homogenates from vehicle-injected (-) and anisomycin-injected (+) mice were immunoreacted with GluRζ1-pan, GluRζ1-C2, GluRζ1-C2′, PSD-95, or synaptophysin antibody. E, Histograms showing different stabilities among total GluRζ1 subunit (pan), GluRζ1-C2, and GluRζ1-C2′ in anysomicin-injected cerebellar homogenates. Values are expressed as the percentage of the protein levels relative to those from vehicle-injected homogenates in each mouse type (mean ± SEM). The statistical difference in the reduction between the wild-type and GluRϵ1/ϵ3-KO mice was determined by a Student's unpaired t test (^*p < 0.05).

Figure 8.

Anatomical and synaptic organization of the wild-type (A, C, E) and GluRϵ1/ϵ3-KO (B, D, F) cerebella. A, B, Normal cytoarchitecture of the cerebellar cortex is shown by hematoxylin-stained Epon semithin sections. Normal cerebellar histoarchitecture is also shown by hematoxylin-stained microslicer sections (insets). C, D, Electron micrographs showing huge mossy fiber terminals (MF) forming asymmetrical synapses with granule cell dendrites (asterisk). E, F, Double immunofluorescence for vesicular glutamate transporter VGluT1 (red) and vesicular GABA transporter VGAT (green) in the cerebellar cortex. GL, Granular layer; Go, Golgi cell; ML, molecular layer; PC, Purkinje cell. Scale bars: A, 10 μm (inset, 1 mm); C, 0.2 μm; E, 10 μm.