Agonist-induced PKC phosphorylation regulates GluK2 SUMOylation and kainate receptor endocytosis

Abstract

The surface expression and regulated endocytosis of kainate (KA) receptors (KARs) plays a critical role in neuronal function. PKC can modulate KAR trafficking, but the sites of action and molecular consequences have not been fully characterized. Small ubiquitin-like modifier (SUMO) modification of the KAR subunit GluK2 mediates agonist-evoked internalization, but how KAR activation leads to GluK2 SUMOylation is unclear. Here we show that KA stimulation causes rapid phosphorylation of GluK2 by PKC, and that PKC activation increases GluK2 SUMOylation both in vitro and in neurons. The intracellular C-terminal domain of GluK2 contains two predicted PKC phosphorylation sites, S846 and S868, both of which are phosphorylated in response to KA. Phosphomimetic mutagenesis of S868 increased GluK2 SUMOylation, and mutation of S868 to a nonphosphorylatable alanine prevented KA-induced SUMOylation and endocytosis in neurons. Infusion of SUMO-1 dramatically reduced KAR-mediated currents in HEK293 cells expressing WT GluK2 or nonphosphorylatable S846A mutant, but had no effect on currents mediated by the S868A mutant. These data demonstrate that agonist activation of GluK2 promotes PKC-dependent phosphorylation of S846 and S868, but that only S868 phosphorylation is required to enhance GluK2 SUMOylation and promote endocytosis. Thus, direct phosphorylation by PKC and GluK2 SUMOylation are intimately linked in regulating the surface expression and function of GluK2-containing KARs.

Fig. 1.

KA stimulation leads to PKC phosphorylation of GluK2 and PKC-dependent SUMOylation of GluK2 in neurons. (A) Representative blot showing the levels of GluK2 phosphorylation after KA stimulation. Neurons expressing YFP-myc-GluK2 were stimulated with 20 μM KA for 5 min in the presence or absence of the PKC inhibitor chelerythrine, immunoprecipitated with anti-myc antibody (sheep), and probed with anti–phospho-Ser/Thr and then with anti-myc antibody (mouse). The graph shows the normalized YFP-myc-GluK2 phosphorylation data. *P < 0.05; n = 4. (B) Mapping of GluK2 phosphorylation sites using S846A and S868A nonphosphorylatable mutants. The experiment was performed identically to that shown in A. The graph compares GluK2 WT and phospho-null mutants, either control or KA-stimulated, normalized to YFP-myc-GluK2 WT. *P < 0.05; n = 5. (C) (Left) representative anti-GluK2 Western blot showing non-SUMOylated (lower band) and SUMO-modified (upper band) GluK2 in 18 DIV neurons after stimulation with 20 μM KA, 1 μM PMA, or both KA and PMA. (Right) Representative blot showing the effects of 20 μM KA, 5 μM chelerythrine, or KA after preincubation with chelerythrine. The graphs show the ratio of SUMOylated to non-SUMOylated GluK2 normalized to the nontreated control. *P < 0.05; (Left) n = 15; (Right) n = 7.

Fig. 2.

Colocalization between SUMO-1 and GluK2 increases after PKC activation. Effects of NMDA, KA, KA plus chelerythrine, chelerythrine, and PMA on the colocalization of SUMO-1 and GluK2 in hippocampal neurons. Note the increased colocalization (yellow in the composite images) after PKC activation. The KA-induced increase in colocalization is abolished by the PKC inhibitor chelerythrine. The boxes in the composite images indicate the magnified areas of dendrites (shown in the lower panels). Green represents GluK2; red, SUMO-1; yellow, colocalization. The graph shows colocalization of GluK2 and SUMO-1 in dendrites as the percentage of GluK2 immunoreactivity colocalizing with the SUMO-1 signal. **P < 0.01; ***P < 0.001. Data are from three independent experiments each analyzing 10–15 cells with at least three regions of interest per cell. (Scale bar: 20 μm.)

Fig. 3.

SUMOylation of GluK2 PKC phospho-mutants. (A) SUMOylation of nonphosphorylated and PKC-phosphorylated GST-CT-GluK2. Purified proteins were phosphorylated in vitro with PKC and then subjected to in vitro SUMOylation. The graph shows GST-CT-GluK2 SUMOylation after phosphorylation by PKC. In each lane, the intensity of the upper (SUMO-modified) band was divided by the intensity of the lower (nonmodified) band. *P < 0.05, n = 5. (B) SUMOylation of GluK2 in COS-7 cells. (Left) GluK2(WT) SUMOylation in cells coexpressing GluK2, Ubc9, and SUMO-1. (Middle and Right) SUMOylation levels for serine 868 and 846 mutants, respectively. The graph shows mutant GluK2 SUMOylation in COS-7 cells, normalized to the WT control. ***P < 0.001; n = 8–10 per condition.

Fig. 4.

Increased YFP-myc-GluK2 and SUMO-1 colocalization after KA stimulation is prevented by blocking S868 phosphorylation. Colocalization of virally expressed YFP-myc-GluK2 with endogenous SUMO-1 in control or KA-treated (20 μM, 20 min) cultured hippocampal neurons. Note that the KA-induced increase in colocalization (indicated in yellow in the composite images) is prevented by the K886R (non-SUMOylatable) and S868A (phospho-null) mutations, but not by the S846A mutation. The boxes in the composite images indicate the magnified areas (shown in the lower panels). Green indicates YFP-myc-GluK2; red, SUMO-1; yellow, colocalization. The graphs show the percentage of YFP-myc-GluK2 signal colocalizing with SUMO-1 immunoreactivity, with data for each mutant normalized to its respective control. *P < 0.05; n = 3 independent experiments, with 8–12 cells analyzed per condition. (Scale bar: 20 μm.)

Fig. 5.

KA-induced GluK2 endocytosis is prevented by blocking S868 phosphorylation. (A) Whole-cell patch clamp recordings of GluK2-expressing HEK293 cells showing a SUMO-1–dependent decrease in KA-evoked currents. Comparison of GluK2 WT, S846A, and S868A using patch solution containing either active (open circles) or inactive (filled circles) SUMO-1. Traces were recorded for 15 min and plotted on the graph normalized to the first minute. Example traces are the average of the first minute (black) and 10–15 min (gray). The graph shows data obtained at 10 min after the first response. Note that SUMO-1 infusion significantly reduced KA-evoked responses mediated by GluK2 WT and S846A, but did not significantly reduce responses mediated by the S868A mutant. (Scale bars: 300 pA and 500 ms). *P < 0.05, ***P < 0.001; n = 5–7. (B) Surface biotinylation of control or KA-treated (20 μM, 20 min) cortical neurons virally expressing YFP-myc-GluK2 WT or mutants. Total (12%) and surface-expressed receptors were detected with anti-GluK2 antibody. The lower panel shows the same blots probed for β-actin. The graph shows quantification of these experiments, with each mutant shown normalized to its control. *P < 0.05; ***P < 0.001; n = 6–10.