Dendritic GluN2A synthesis mediates activity-induced NMDA receptor insertion

Abstract

Long-term synaptic plasticity involves changes in the expression and membrane insertion of cell-surface proteins. Interestingly, the mRNAs encoding many cell-surface proteins are localized to dendrites, but whether dendritic protein synthesis is required for activity-induced surface expression of specific proteins is unknown. Herein, we used microfluidic devices to demonstrate that dendritic protein synthesis is necessary for activity-induced insertion of GluN2A-containing NMDA receptors in rat hippocampal neurons. Furthermore, visualization of activity-induced local translation of GluN2A mRNA and membrane insertion of GluN2A protein in dendrites was directly observed and shown to depend on a 3' untranslated region cytoplasmic polyadenylation element and its associated translation complex. These findings uncover a novel mechanism for cytoplasmic polyadenylation element-mediated posttranscriptional regulation of GluN2A mRNA to control NMDA receptor surface expression during synaptic plasticity.

Figure 1.

Glycine induced a protein synthesis-dependent increase in GluN2A-containing NMDA receptor surface expression. A, Hippocampal neurons were pretreated with anisomycin (aniso) or DMSO for 30 min, treated with glycine or vehicle for 3 min, and then incubated without glycine for 30 min. Anisomycin or DMSO was present in all solutions. Total and biotinylated surface levels of GluN1, GluN2A, and GluN2B levels were measured by Western blotting (repeated-measures ANOVA, Bonferroni t tests, n = 6; B, *p = 0.001, # p = 0.022; C, *p = 0.020, ^#p = 0.015). D–F, Hippocampal neurons were treated with anisomycin, followed by glycine-induced LTP, and then surface GluN1, GluN2A, or GluN2B were immunolabeled. Scale bar, 5 μm. Graphed values are dendritic fluorescence intensities normalized to the control group mean (ANOVA, Bonferroni t tests, n = 45–53; *p = 0.022; ^#p = 0.012). Graphed data are mean ± SEM.

Figure 2.

Dendritic protein synthesis is required for glycine-induced insertion of GluN2A. A, MAP2 staining of 25 DIV neurons cultured in a microfluidic device. CB, Cell bodies; D, dendrites. Scale bar, 20 μm. B, A representative dendritic region immunostained for MAP2 and GluN2A. Scale bar, 3 μm. C, Time-lapse imaging shows microfluidic chambers maintain fluidic isolation throughout the glycine treatment paradigm. Microfluidic devices were filled with PBS (initial), then left-side PBS was replaced with PBS plus Cy3b dye (t₀), and, 30 min later, right-side PBS was replaced with PBS plus Cy5 dye (t₃₀). After 3 min, PBS plus Cy5 dye was replaced with PBS alone (t₃₃) and allowed to sit for an additional 30 min (t₆₀). The Cy3 and Cy5 dye solutions remained restricted to the CB and dendrite (D) sides, respectively, for the duration of the experiment. D, Hippocampal neurons were cultured in microfluidic devices. Anisomycin was applied to either the cell body or the dendrite compartment for 30 min, followed by glycine or vehicle application to the dendrite compartment for 3 min, and an additional 30 min of incubation in solution without glycine. Anisomycin or DMSO was present in all solutions. Then, the neurons were fixed and immunostained for surface GluN2A protein (ANOVA, Bonferroni t tests, n = 65; *p < 0.001, ^#p = 0.008). Scale bar, 5 μm. E, Neurons were treated as in D, except that total GluN2A was immunostained (ANOVA, Bonferroni t tests, n = 65; *p = 0.003, ^#p = 0.011, ^‡p < 0.001, ^φp = 0.004). Scale bar is 3 μm. Data are mean ± SEM.

Figure 3.

CPEB interacts with endogenous GluN2A mRNA and regulates its dendritic localization. A, FLAG-CPEB or FLAG were immunoprecipitated from hippocampal neuron lysates. GluN2A, γ-actin, GluN1, and αCaMKII mRNA levels in input and FLAG immunoprecipitates were quantified by real-time PCR. Precipitated mRNA levels were normalized to input levels, and the graphed values were normalized to the FLAG IP/γ-actin mRNA group mean (ANOVA, Bonferroni t tests, n = 6; *p = 0.009, ^#p = 0.002). B, The GluN2A 3′ UTR CPE sequence was mutated as shown. C, Neuroblastoma cells were transfected with either FLAG-CPEB or FLAG and GFP fused to the GluN2A, ΔCPE-GluN2A, or β-actin 3′ UTR. GFP mRNA levels in FLAG immunoprecipitates and input samples were quantified by real-time PCR. The graphed values were normalized to the FLAG IP/GluN2A mRNA group mean (ANOVA, Bonferroni t tests, n = 6, *p = 0.015). D, Hippocampal neurons were treated with control or CPEB shRNA lentiviruses (expressing GFP) for 4 d and processed for GluN2A mRNA FISH. Dendritic mRNA granules were counted using ImageJ and normalized to the area of the dendritic region. Scale bar, 10 μm. E, Lysates from hippocampal neurons treated with control or CPEB shRNA were Western blotted for CPEB and α-tubulin. F, GFP fused to the vector 3′ UTR, β-actin 3′ UTR, GluN2A 3′ UTR, or Δ CPE-GluN2A 3′ UTR were expressed in neurons for 12 h, then the neurons were fixed and processed for GFP mRNA FISH. Representative images of GFP mRNA FISH and GFP protein fluorescence signals are shown. Scale bar, 10 μm. Cell body and dendritic fluorescence were quantified and graphed as a ratio of dendritic to cell body fluorescence (ANOVA, Bonferroni t tests, n = 18–25 cells, *p = 0.001, ^#p = 0.001, ^‡p = 0.001). Data are mean ± SEM.

Figure 4.

Dendritic GluN2A mRNA translation is mediated by the 3′ UTR CPE sequence. A, Synaptoneurosome fractions were isolated from mouse hippocampus and pretreated with anisomycin or vehicle, and then glutamate and glycine or vehicle were applied for 15 min. GluN2A and tubulin (loading control) protein levels were analyzed by Western blotting (ANOVA, Bonferroni t tests, n = 6; *p = 0.013). B, C, Glycine induced dendritic fluorescence recovery of Dendra2–3′ UTR, but not Dendra2 alone, Dendra2-ΔCPE-3′ UTR, or when anisomycin was applied. In the histograms, an orange bar depicts the glycine application, and photoconversion was performed just after start of glycine application (black arrow; RM-ANOVA, Bonferroni t tests, n = 10–12; *p < 0.001). D, GFP fused to the GluN2A 3′ UTR or ΔCPE-GluN2A 3′ UTR were expressed in neurons for 48 h, then neurons were treated with glycine (or vehicle), fixed, and processed for GFP mRNA FISH. Scale bar, 5 μm). GFP FISH fluorescence was quantified as above (ANOVA, n.s.; n = 20–24 cells). Data are mean ± SEM.

Figure 5.

Gld2 and Ngd regulate glycine-induced GluN2A synthesis and surface expression in dendrites. A, Hippocampal neurons were treated with control, Gld2 shRNA, or Ngd shRNA for 3–4 d, and stimulated with glycine or vehicle. Surface proteins were biotinylated, and total and surface proteins were immunoblotted for GluN2A and α-tubulin (loading control). Protein levels were quantified by densitometry (two-way ANOVA, post hoc Bonferroni t tests, n = 6; B, Gld2 KD effects on basal and glycine-induced labeling of surface: *p = 0.001, ^#p = 0.007; total: *p = 0.015, ^#p = 0.005; C, surface: main: p = 0.001, Ngd KD main: p = 0.001, interaction: p = 0.006, ^φp = 0.001, *p = 0.012, ^#p = 0.001, ^‡p = 0.002; total: cLTP main: *p = 0.005, Ngd KD main: ^#p = 0.001, interaction: 0.940). D, A schematic showing neuron culture and treatment in microfluidic devices. Lentiviral shRNA was applied to the CB compartment at 24 DIV, and the cultures were used 3–4 d later. For glycine treatment, anisomycin was applied to the CB compartment 30 min before glycine treatment (applied to dendrites only). Glycine was applied for 3 min then removed; anisomycin was present in the CB compartment throughout the experiment. E, F, Hippocampal neurons were fixed in the microfluidic devices, and then immunostained for MAP2 and either (E) surface or (F) total GluN2A (two-way ANOVA, Bonferroni t tests; E, n = 40–45, *p = 0.015, ^#p = 0.012; F, n = 40–45, *p = 0.012, ^#p = 0.006). Scale bars, 3 μm. Data are mean ± SEM.

Figure 6.

The CPE sequence controls GluN2A surface expression in dendrites. SEP-GluN2A-3′ UTR or SEP-GluN2A-ΔCPE-3′ UTR along with mCherry (to visualize the entire neuron) were expressed in cultured hippocampal neurons for 18–24 h. Scale bar, 5 μm. Glycine or vehicle was applied for 3 min (orange bar), and the neurons were imaged every 5 min for 30 min (RM-ANOVA, Bonferroni t tests, n = 5–8 cells; *p < 0.009). Data are mean ± SEM.

Figure 7.

Differential regulation of GluN2A surface expression by somatic and dendritic protein synthesis. GluN2A protein levels and surface expression within dendrites are regulated by both somatic and dendritic protein synthesis. (1) Under basal conditions, dendritic GluN2A levels are regulated by constitutive protein synthesis in the soma. Somatic GluN2A is trafficked into dendrites, likely in the form of post-Golgi vesicles, where it is either inserted in the membrane or held in reserve for subsequent regulation (blue arrows). The posttranscriptional mechanisms regulating somatic GluN2A synthesis remain unclear. (2) Following synaptic activation, intracellular signaling (dashed line) induces the local translation of GluN2A mRNA that is regulated by the CPEB-associated complex in dendrites, resulting in the dendritic synthesis and membrane insertion of GluN2A-containing NMDA receptors, likely via components of a dendritic secretory pathway.