Fig 2: Phosphomimetic mutations of the MPR regulate the affinity of the AMPA receptor–TARP complex to AP-3.A, amino acid sequences of the C terminus of AMPAR subunits and GluA1 mutants. Serine residues that can be phosphorylated by PKC, CaMKII, and PKA are indicated. These residues were replaced with aspartate and alanine to mimic phosphorylation (blue) and dephosphorylation (red). Although Ser816 is not directly phosphorylated, it enhances the effect of the Ser818 mutation. B, schematic drawing of the coimmunoprecipitation assay. Lysates of HEK293 cells expressing STG, GluA1 mutants, and FLAG-tagged µ2 or µ3 were immunoprecipitated using the anti-GluA1 antibody. C and D, the effect of mutation of all serine residues of GluA1 on the interaction with µ2 or µ3. While µ2 was similarly coimmunoprecipitated with GluA1AAAA and GluA1DDDD (C), µ3 was preferentially coimmunoprecipitated with GluA1AAAA than GluA1DDDD (D). Top, the intensity of the band corresponding to µ2 or µ3 that was coimmunoprecipitated was normalized to the intensity of the respective molecule in the input lysate. Data are presented as the mean + SEM and individual data points (yellow circles) (Mann–Whitney U test, *p < 0.05; n = 4). Bottom, the intensity of the band corresponding to GluA1 (left) or STG (right) coimmunoprecipitated was normalized to the intensity of the respective molecule in the input lysate. Data are presented as the mean + SEM and individual data points. E, the effect of the position of the mutations on the interaction with µ3. µ3 was preferentially coimmunoprecipitated with GluA1AADD than with GluA1DDAA. The intensity of µ3 in the immunoprecipitated fraction was normalized to that of the input lysate. Data are presented as the mean + SEM and individual data points (Mann–Whitney U-test, *p < 0.05; n = 6). F, the effect of mutations in the MPR on the interaction with µ3. µ3 was preferentially coimmunoprecipitated with GluA1AA than with GluA1DD. The intensity of µ3 in the immunoprecipitated fraction was normalized to that of the input lysate. Data are presented as the mean + SEM and individual data points (Mann–Whitney U-test, **p < 0.01; n = 5). µ2, µ subunit of AP-2; µ3, µ subunit of AP-3; AMPARs, AMPA-type glutamate receptors; AP, adaptor protein; CaMKII, calmodulin-dependent protein kinase II; HEK293, human embryonic kidney 293; MPR, membrane-proximal region; n.s., not significant; STG, stargazin; TARPs, transmembrane AMPAR regulatory proteins.

Fig 3: MPR regulates NMDA-induced AMPAR internalization.A, C, and E, immunocytochemical analysis of the effects of the MPR on NMDA-induced trafficking of cell surface GluA1. Cultured hippocampal neurons expressing enhanced green fluorescent protein and HA-tagged WT (GluA1wt) (A) or phospho-deficient GluA1 (GluA1AA) (C) or phosphomimetic GluA1 (GluA1DD) (E) were treated with 50 µM NMDA for 10 min. Cell surface HA-GluA1 were stained (red) after fixation, and neurons were immunostained for total HA-GluA1 (blue) after treatment with Triton-X. The dendritic regions marked by squares were enlarged in the panels to the right. The scale bars represent 10 µm. B, D, and F, quantification of NMDA-induced reduction in the ratio of the surface to total GluA1 fluorescence intensities with NMDA treatment. Data are represented as the ratio of surface HA-GluA1 immunoreactivity normalized by total HA-GluA1 immunoreactivity. The ratio in control neurons was defined as 100% (n = 8–14). Data are presented as the mean + SEM and individual data points. **p < 0.01 and *p < 0.05; n.s. by two-tailed Student’s t test. AMPARs, AMPA-type glutamate receptors; HA, hemagglutinin; MPR, membrane-proximal region; n.s., not significant; NMDA, N-methyl-d-aspartate.

Fig 4: Phosphomimetic mutations of the MPR increased recycling of GluA1 to the cell surface.A, schematic drawing of antibody feeding assay. Living neurons expressing HA-GluA1 mutants were labeled with an anti-HA antibody. After NMDA treatment, remaining cell surface antibodies were removed by acid treatment. After a 30-min incubation to allow the recycling of HA-GluA1, neurons were fixed and recycled, and internal HA-GluA1 was visualized by Alexa 546– and Alexa 350–conjugated secondary antibodies, respectively. B, immunocytochemical analysis of the effects of the MPR phosphorylation on the recycling of GluA1 after NMDA treatment. Cultured living hippocampal neurons expressing HA-GluA1wt or HA-GluA1AA or HA-GluA1DD were subjected to the antibody feeding assay. The dendritic regions marked by white rectangles are enlarged in the panels to the right. The scale bars represent 10 µm. C, quantification of the recycled GluA1. Data are represented as the ratio of recycled HA-GluA1 staining/total HA-GluA1 staining intensity. The ratio of HA-GluA1wt was defined as 100% (n = 12 cells). Data are presented as the mean + SEM and individual data points. **p < 0.01 and *p < 0.05 by one-way ANOVA and Student–Newman–Keuls post hoc test. HA, hemagglutinin; MPR, membrane-proximal region; NMDA, N-methyl-d-aspartate.

Fig 5: Phosphomimetic mutations of the MPR inhibit the transport of GluA1 to late endosomes and lysosomes.A, colocalization of HA-tagged mutant GluA1 with an early endosome marker, enhanced green fluorescent protein-tagged Rab4 at 0 min, and 3 min after NMDA treatment. The scale bar represents 10 µm. B, colocalization of HA-tagged mutant GluA1 with a late endosome/lysosome marker, enhanced green fluorescent protein-tagged Rab7 at 0 min, and 10 min after NMDA treatment. C and D, quantification of the colocalization of HA-tagged WT or mutant GluA1 with Rab proteins. Data are represented as the ratio of colocalized HA-GluA1 staining/total HA-GluA1 staining intensity. The ratio in the neurons without NMDA stimulation (0 min) was defined as 100% (n = 9–12 cells). Data are presented as the mean + SEM and individual data points. **p < 0.01 and *p < 0.05; n.s. by the Kruskal–Wallis test and Steel–Dwass post hoc test. HA, hemagglutinin; MPR, membrane-proximal region; n.s., not significant; NMDA, N-methyl-d-aspartate.