NMDAR‐dependent Argonaute 2 phosphorylation regulates miRNA activity and dendritic spine plasticity

Abstract MicroRNAs (miRNAs) repress translation of target mRNAs by associating with Argonaute (Ago) proteins to form the RNA‐induced silencing complex (RISC), underpinning a powerful mechanism for fine‐tuning protein expression. Specific miRNAs are required for NMDA receptor (NMDAR)‐dependent synaptic plasticity by modulating the translation of proteins involved in dendritic spine morphogenesis or synaptic transmission. However, it is unknown how NMDAR stimulation stimulates RISC activity to rapidly repress translation of synaptic proteins. We show that NMDAR stimulation transiently increases Akt‐dependent phosphorylation of Ago2 at S387, which causes an increase in binding to GW182 and a rapid increase in translational repression of LIMK1 via miR‐134. Furthermore, NMDAR‐dependent down‐regulation of endogenous LIMK1 translation in dendrites and dendritic spine shrinkage requires phospho‐regulation of Ago2 at S387. AMPAR trafficking and hippocampal LTD do not involve S387 phosphorylation, defining this mechanism as a specific pathway for structural plasticity. This work defines a novel mechanism for the rapid transduction of NMDAR stimulation into miRNA‐mediated translational repression to control dendritic spine morphology.

A-E The effects on luciferase reporters incorporating LIMK1, APT1 and LIN41 3 0 UTRs seen in Fig 5 are fully explained by regulation via miR-134, miR-138 and Let-7, respectively. Cultured cortical neurons transfected with Ago2 molecular replacement constructs as well as Renilla luciferase and Firefly luciferase reporters containing LIMK1 (A, D), APT1 (B, E) or LIN41 (C) 3 0 UTRs containing mutations in the seed regions for miR-134, miR-138 and Let-7, respectively, were treated with NMDA or vehicle for 3 min. Ten minutes after NMDA washout, lysates were prepared for dual-luciferase assays. For (D, E), cultures were also treated with kinase inhibitors as shown 20 min before NMDA or vehicle. Error bars are SEM. F, G Translation of luciferase reporters incorporating PUM2 or CREB1 is regulated by other miRNAs in addition to miR-134. Experiment was performed as in (A), except Firefly luciferase reporters contained PUM2 (F), CREB1 (G) 3 0 UTRs carrying mutations in the seed regions for miR-134. n = 5. ***P < 0.001, two-way ANOVA, Bonferroni post hoc test. Error bars are SEM.
A B C Figure EV4. NMDAR-dependent decrease in APT1 expression is insensitive to Akt inhibition or Ago2 phosphorylation at S387 (related to Fig 7).
A Endogenous APT1 protein levels are rapidly reduced in response to NMDAR stimulation. Cortical neuronal cultures were exposed to NMDA or vehicle for 3 min, and lysates were prepared 10, 20 or 40 min after NMDA washout and analysed by Western blotting. Graphs show quantification of APT1 expression normalised to vehicle control; n = 6. **P < 0.01; ***P < 0.001 one-way ANOVA, Bonferroni post hoc test. Error bars are SEM. B NMDAR-dependent decrease in APT1 is Akt-independent. Cortical neuronal cultures were treated with Akti-1/2 20 min before NMDA or vehicle application, and lysates were prepared 40 min after NMDA washout and analysed by Western blotting. Graphs show quantification of APT1 expression normalised to vehicle control; n = 6. *P < 0.05; two-way ANOVA, Bonferroni post hoc test. Error bars are SEM. C NMDAR-dependent decrease in dendritic APT1 expression is unaffected by Ago2 phosphorylation at S387. Cortical neurons were transfected with molecular replacement constructs expressing Ago2 shRNA plus shRNA-resistant GFP-Ago2 (WT, S387A or S387D), fixed 40 min after NMDA washout, permeabilised and stained with APT1 antibodies (red channel). GFP signal was maximised at acquisition so that dendrites could be effectively visualised. Graph shows APT1 staining intensity in dendrites normalised to vehicle control. Scale bar = 20 lm; n = 10 cells from five independent experiments. ***P < 0.001 two-way ANOVA, Bonferroni post hoc test. Error bars are SEM.
Source data are available online for this figure.
The EMBO Journal Ago2 phosphorylation and spine plasticity Dipen Rajgor et al

EV3
The EMBO Journal e97943 | 2018 ª 2018 The Authors A B C Figure EV5. Ago2 phosphorylation at S387 is not required for NMDA-induced AMPAR internalisation (related to Fig 8).
A NMDA-induced AMPAR internalisation is insensitive to Ago2 S387 mutations. Cortical neurons were transfected with molecular replacement constructs expressing Ago2 shRNA plus shRNA-resistant GFP-Ago2 (WT, S387A or S387D). Twenty minutes after NMDA or vehicle application, live cells were stained with GluA2 antibodies (red channel). GFP signal was maximised at acquisition so that dendrites could be effectively visualised. Graph shows surface GluA2 staining intensity in dendrites normalised to vehicle control. Scale bar = 20 lm; n = 15 cells from three independent experiments. **P < 0.01; ***P < 0.001, two-way ANOVA, Bonferroni post hoc test. Error bars are SEM. B Endogenous GluA1, but not GluA2, subunit levels are reduced in response to NMDAR stimulation. Cortical neuronal cultures were exposed to NMDA or vehicle for 3 min, and lysates were prepared 10, 20 or 40 min after NMDA washout and analysed by Western blotting. Graphs show quantification of GluA1 and GluA2 expression normalised to vehicle control; n = 6. *P < 0.05; one-way ANOVA, Bonferroni post hoc test. Error bars are SEM. C NMDAR-dependent decrease in GluA1 is Akt-independent. Cortical neuronal cultures were treated with Akti-1/2 20 min before NMDA or vehicle application, and lysates were prepared 40 min after NMDA washout and analysed by Western blotting. Graph shows quantification of GluA1 expression normalised to vehicle control; n = 5. *P < 0.05; one-way ANOVA, Bonferroni post hoc test. Error bars are SEM.
Source data are available online for this figure.