Rho-Kinase/ROCK Phosphorylates PSD-93 Downstream of NMDARs to Orchestrate Synaptic Plasticity

The N-methyl-D-aspartate receptor (NMDAR)-mediated structural plasticity of dendritic spines plays an important role in synaptic transmission in the brain during learning and memory formation. The Rho family of small GTPase RhoA and its downstream effector Rho-kinase/ROCK are considered as one of the major regulators of synaptic plasticity and dendritic spine formation, including long-term potentiation (LTP). However, the mechanism by which Rho-kinase regulates synaptic plasticity is not yet fully understood. Here, we found that Rho-kinase directly phosphorylated discs large MAGUK scaffold protein 2 (DLG2/PSD-93), a major postsynaptic scaffold protein that connects postsynaptic proteins with NMDARs; an ionotropic glutamate receptor, which plays a critical role in synaptic plasticity. Stimulation of striatal slices with an NMDAR agonist induced Rho-kinase-mediated phosphorylation of PSD-93 at Thr612. We also identified PSD-93-interacting proteins, including DLG4 (PSD-95), NMDARs, synaptic Ras GTPase-activating protein 1 (SynGAP1), ADAM metallopeptidase domain 22 (ADAM22), and leucine-rich glioma-inactivated 1 (LGI1), by liquid chromatography-tandem mass spectrometry (LC-MS/MS). Among them, Rho-kinase increased the binding of PSD-93 to PSD-95 and NMDARs. Furthermore, we found that chemical-LTP induced by glycine, which activates NMDARs, increased PSD-93 phosphorylation at Thr612, spine size, and PSD-93 colocalization with PSD-95, while these events were blocked by pretreatment with a Rho-kinase inhibitor. These results indicate that Rho-kinase phosphorylates PSD-93 downstream of NMDARs, and suggest that Rho-kinase mediated phosphorylation of PSD-93 increases the association with PSD-95 and NMDARs to regulate structural synaptic plasticity.


Introduction
Synaptic plasticity is a key mechanism of learning and memory. During synaptic plasticity, many changes occur between neurons, such as changes in presynaptic vesicles, Ca 2+ concentrations, and neurotransmitter receptors. These alterations regulate the dendritic spine volume, which is increased during long-term potentiation (LTP) and decreased during long-term depression (LTD) [1,2]. Moreover, the enlargement of dendritic spine volume in LTP is crucial for strengthening connections between neurons which is essential for synaptic transmission [3,4]. During LTP, the activation of RhoA regulates the cytoskeleton, which subsequently enlarges the spine volume by increasing actin polymerization and incorporating α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors (AMPARs)  acids. (B) Constitutively active Rho-kinase phosphorylated PSD-93 at the PDZ3-SH3 domain. Purified GST and GST-PSD-93 domain proteins were incubated with [γ 32 P] ATP in the presence or absence of Rho-kinase-cat in vitro. Samples were subjected to SDS-PAGE and silver staining followed by autoradiography. Asterisks denote intake GST-PSD-93 fusion domain proteins. The arrows denote recombinant Rho-kinase-cat and GST. Third bracket represents PSD-93 domain proteins. (C) The bar diagram shows the relative phosphorylation (%) of PSD-93 domain proteins. (D) Schematic presentation of PSD-93 phosphorylation sites. The PSD-93-PDZ3-SH3 domain contains four putative phosphorylation sites based on Rho-kinase consensus motifs of R/KXXpS/T and R/KXpS/T (R, arginine; K, lysine; X, any amino acid; S, serine and T, threonine). P denotes phosphate group. (E) Rho-kinase phosphorylated PSD-93-PDZ3-SH3 domain at four different sites. Purified wild-type and phospho-deficient mutants of PSD-93-PDZ3-SH3 were incubated with Rho-kinase-cat along with [γ 32 P] ATP in vitro. Next, the SDS-boiled samples were subjected to SDS-PAGE and silver staining followed by autoradiography. The upper panel shows autoradiography, and the lower panel shows a silver staining image. (F) Bar diagram showing the relative phosphorylation (%) of PSD-93 domain proteins. The horizontal lines indicate the mean ± SEM of three independent experiments.*, **, *** and **** represent p < 0.05, p < 0.01, p < 0.001 and p < 0.0001, respectively, for Dunnett's multiple comparisons test. "ns" denotes "not significant". (G) The amino acid sequence homology of PSD-93 phosphorylation sites (Thr585 and Thr612) in different species (humans, rats, and mice) is shown in a schematic graph. The numbers represent the amino acid position.

Rho-Kinase Phosphorylates PSD-93 at Thr612 in Striatal Slices
To monitor the phosphorylation of PSD-93 by Rho-kinase, we generated an antibody that specifically recognized PSD-93 phosphorylation at Thr612. The sensitivity and specificity of this antibody were examined using immunoblotting analysis. In this analysis, in vitro phosphorylation of GST-PSD-93-PDZ3-SH3 by Rho-kinase was identified by using the anti-pT612 PSD-93 phospho-antibody in a dose-dependent manner (Supplementary Materials, Figure S1A,B). The phospho-specific antibody detected phosphorylated GST-PSD-93-PDZ3-SH3 by Rho-kinase without cross-reacting with the nonphosphorylated form of GST-PSD-93-PDZ3-SH3 (p < 0.0001) (Supplementary Materials, Figure S1A,B). Furthermore, the specificity of the antibody was examined using in vitro phosphorylation assays with phospho-deficient PSD-93 mutants. The pT612 PSD-93 antibody recognized the wild-type and the T585A, S590A, and S598A mutants, but did not recognize the T612A and 4A GST-PSD-93-PDZ3-SH3 mutants (p < 0.0001 and p < 0.0001, respectively) (Supplementary Materials, Figure S1C,D). This result indicates that the anti-pT612 PSD-93 antibody specifically recognizes PSD-93 phosphorylation at Thr612.

Rho-Kinase Phosphorylates PSD-93 at Thr612 in Striatal Slices
To monitor the phosphorylation of PSD-93 by Rho-kinase, we generated an antibody that specifically recognized PSD-93 phosphorylation at Thr612. The sensitivity and specificity of this antibody were examined using immunoblotting analysis. In this analysis, in vitro phosphorylation of GST-PSD-93-PDZ3-SH3 by Rho-kinase was identified by using the anti-pT612 PSD-93 phospho-antibody in a dose-dependent manner (Supplementary Materials, Figure S1A,B). The phospho-specific antibody detected phosphorylated GST-PSD-93-PDZ3-SH3 by Rho-kinase without cross-reacting with the nonphosphorylated form of GST-PSD-93-PDZ3-SH3 (p < 0.0001) (Supplementary Materials, Figure  S1A,B). Furthermore, the specificity of the antibody was examined using in vitro phosphorylation assays with phospho-deficient PSD-93 mutants. The pT612 PSD-93 antibody recognized the wild-type and the T585A, S590A, and S598A mutants, but did not recognize the T612A and 4A GST-PSD-93-PDZ3-SH3 mutants (p < 0.0001 and p < 0.0001, respectively) (Supplementary Materials, Figure S1C,D). This result indicates that the anti-pT612 PSD-93 antibody specifically recognizes PSD-93 phosphorylation at Thr612.

Identification of PSD-93-Binding Proteins by Liquid Chromatography-Tandem Mass Spectrometry (LC-MS/MS) Analysis
PSD-93 is an adaptor protein that can interact with receptors on the postsynaptic membrane, such as NMDARs and AMPARs, contributing to receptor stabilization [21]. To investigate the effect of PSD-93 phosphorylation by Rho-kinase on its interactions with synaptic proteins and receptors, we performed an immunoprecipitation assay using calyculin Aand/or Y-27632-stimulated striatal slices followed by LC-MS/MS analysis ( Figure 4A). We confirmed MYPT1 phosphorylation by immunoblotting after the stimulation of striatal slices with calyculin A and/or Y-27632. Calyculin A increased the phosphorylation of MYPT1 at Thr853 approximately 60-fold, which was inhibited approximately 40-fold by Y-27632 ( Figure 4B,C). We categorized LC-MS/MS data based on two criteria: first, the ion intensity of the sample-control had to be more than 10 times higher than that of the IgGcontrol; and second, the ion intensity of the calyculin A-treated sample had to be lower or higher than that of the sample-control as well as the Y-27632-treated sample. We identified 28 proteins as PSD-93-binding proteins. We further classified these 28 proteins into two categories. The first category contained 12 phosphorylation-dependent positively regulated proteins (whose binding was increased by calyculin A and repressed by Y-27632), namely Agap3, Dlg1, Dlg4, Dlgap1, Dlgap4, Grin1, Grin2a, Grin2b, Griks, Iqsec2, Nrcam and Rtn3 (Table 1). The second contained 16 phosphorylation-dependent negatively regulated proteins (whose binding was reduced by calyculin A and restored by Y-27632), namely ADAM22, Begain, Cacna2d1, Cit, Cnksr2, Dclk1, Dlgap2, Dlgap3, Kcnj4, LGI1, Lrrc7, Map4, Prph, Shisa7, Shank3 and SynGAP1 (Table 2).

List of Phosphorylation Dependent Positively Regulated PSD-93 Binding Partners
Nrcam Rtn3 Table 2. Phosphorylation dependent negatively regulated PSD-93 binding partners. LC-MS/MS data were sorted according to two criteria: 1. the sample-control vs. IgG-control ratio was greater than 10:1, and 2. binding with PSD-93 upon calyculin A stimulation was lower than that under the sample-control condition or after Y-27632 stimulation. The proteins in the box were negatively regulated upon PSD-93 phosphorylation by Rho-kinase.

List of Phosphorylation Dependent Negatively Regulated PSD-93 Binding Partners
Shisa7 SynGAP1

Chemically Induced-LTP Increases the Rho-Kinase-Mediated Phosphorylation of PSD-93 and the Colocalization of PSD-93 with PSD-95
Synaptic plasticity occurs due to LTP induction in which dendritic spine volume has been changed and the synapses strengthened [3,4]. It has been reported that chemical LTP increases the dendritic spine volume through NMDAR-mediated manners. NMDAR antagonists (D-APV and MK-801) decrease the spine size that was increased by chemical LTP in cultured neurons [24]. To examine whether LTP induces the phosphorylation of PSD-93 in primary striatal neurons, we employed a chemically (glycine)-induced LTP method. Along with the continuous release of glutamate from axonal terminals, in this method, glycine can selectively activate synaptic NMDARs while withdrawing Mg 2+ from the medium using perfusion with glycine containing nACSF buffer [25]. Primary striatal neurons were cultured until the in vitro day (DIV21) and then treated with glycine to induce chemical LTP via NMDAR activation. Chemical LTP induction stimulated the phosphorylation of MYPT1 Thr853 at 10 min (approximately 1.5-fold, ns) and 60 min (approximately 2-fold, p < 0.05), an effect that was significantly inhibited (p < 0.05 and p < 0.01, respectively) by pretreatment of neurons with Y-27632, as described previously [11]. We also found that chemical-LTP induced the phosphorylation of PSD-93 at 10 min (approximately 3-fold, p < 0.01) and 60 min (approximately 4-fold, p < 0.001), and this phosphorylation was significantly diminished by pretreatment with Y-27632 (p < 0.05 and p < 0.05, respectively) ( Figure 6A-C), indicating that Rho-kinase phosphorylates PSD-93 downstream of NMDARs during LTP induction.
Next, we examined the localization of PSD-93 and PSD-95 in dendritic spines during LTP induction. Primary striatal neurons were cultured for DIV14 and infected with the AAV-CAGGS-Flex-EGFP virus along with the AAV-CaMKII-Cre virus to visualize the dendritic spines. The neurons at DIV21 were pretreated with dimethyl sulfoxide (DMSO) or Y-27632 for 20 min and incubated with glycine for 60 min to induce chemical-LTP. We found that chemical-LTP increased the dendritic spine volume by approximately 50% (p < 0.0001), an effect that was significantly suppressed (p < 0.001) with Y-27632 pretreatment ( Figure 6D,E), as previously described [11]. The colocalization of PSD-93 with PSD-95 increased approximately 30% (p < 0.01) after chemical LTP induction in the dendritic spines, but was significantly inhibited by pretreatment with Y-27632 (p < 0.05) ( Figure 6D,F). Revealing the role of Rho-kinase in colocalization of PSD-93 with PSD-95 in dendritic spines helps to better understand the structural synaptic plasticity in neurons. The horizontal lines represent the mean ± SEM of three independent experiments. *, ** and *** represent p < 0.05, p < 0.01, p < 0.001, respectively, and "ns" denotes "not significant", for Tukey's multiple comparisons test. (D) Colocalization of PSD-93 with PSD-95 during chemical-LTP in cultured primary striatal neurons. The neurons were cultured until DIV14 and infected with AAV-EGFP and AAV-Cre virus. After DIV21, the neurons were treated with glycine to induce chemical-LTP and the immunostaining was performed with anti-GFP (green), anti-PSD-93 (red) and anti-PSD-95 (white) antibodies. The scale bar is 10 μM. (E,F) The horizontal lines represent the mean ± SEM of five independent experiments. *, **, *** and **** represent p < 0.05, p < 0.01, p < 0.001 and p < 0.001, respectively, for Tukey's multiple comparisons test.

Discussion
RhoA and its effector, Rho-kinase, are considered to be important for synaptic functions, but the mechanism by which Rho-kinase regulates synaptic functions is not yet fully understood. In this study, we examined whether Rho-kinase phosphorylates PSD-93, one of the major postsynaptic regulatory proteins, and found that Rho-kinase indeed phosphorylates PSD-93 downstream of NMDARs. The phosphorylation of PSD-93 appears to increase the interaction and colocalization of PSD-93 with PSD-95 in dendritic spines to regulate structural plasticity. The horizontal lines represent the mean ± SEM of three independent experiments. *, ** and *** represent p < 0.05, p < 0.01, p < 0.001, respectively, and "ns" denotes "not significant", for Tukey's multiple comparisons test. (D) Colocalization of PSD-93 with PSD-95 during chemical-LTP in cultured primary striatal neurons. The neurons were cultured until DIV14 and infected with AAV-EGFP and AAV-Cre virus. After DIV21, the neurons were treated with glycine to induce chemical-LTP and the immunostaining was performed with anti-GFP (green), anti-PSD-93 (red) and anti-PSD-95 (white) antibodies. The scale bar is 10 µM. (E,F) The horizontal lines represent the mean ± SEM of five independent experiments. *, **, *** and **** represent p < 0.05, p < 0.01, p < 0.001 and p < 0.001, respectively, for Tukey's multiple comparisons test.

Discussion
RhoA and its effector, Rho-kinase, are considered to be important for synaptic functions, but the mechanism by which Rho-kinase regulates synaptic functions is not yet fully understood. In this study, we examined whether Rho-kinase phosphorylates PSD-93, one of the major postsynaptic regulatory proteins, and found that Rho-kinase indeed phosphorylates PSD-93 downstream of NMDARs. The phosphorylation of PSD-93 appears to increase the interaction and colocalization of PSD-93 with PSD-95 in dendritic spines to regulate structural plasticity.

Phosphorylation of PSD-93
Similar to other MAGUK family proteins, PSD-93 has the potential to interact with other postsynaptic proteins, and this interaction is suggested to be phosphorylation dependent [26]. Several studies have been conducted on the phosphorylation dependent regulation of PSD-93, and one study has shown that extracellular signal-regulated kinases (ERKs) phosphorylate PSD-93 at Ser323 in striatal neurons, but the role of this phosphorylation remains unknown [27]. Another study has demonstrated that Fyn kinase phosphorylates Thr384 of PSD-93 and thus upregulates NMDAR function [28,29]. Other publications have revealed that PSD-93 deletion leads to the mislocalization of Fyn from the synaptosomal membrane. As a result, tyrosine phosphorylation of NR2A and NR2B is depleted [30]. These findings indicate that PSD-93 acts as a membrane-anchored substrate of Fyn and plays a major role in the regulation of Fyn-mediated upregulation of NMDAR function [31]. In the present study, we found that Rho-kinase directly phosphorylated PSD-93 in the PDZ3-SH3 domain in vitro ( Figure 1A-C). In contrast, Rho-kinase did not phosphorylate PSD-95 in vitro. We also found that Rho-kinase phosphorylated PSD-93 at four different sites (Thr585, Ser590, Ser598, and Thr612) in vitro ( Figure 1D-F). Among these four phosphorylation sites, the major phosphorylation sites were Thr585 and Thr612. Stimulation of striatal slices with an NMDAR agonist induced Rho-kinasemediated phosphorylation of PSD-93 at Thr612 (Figure 3A-D). Chemical-LTP increased the phosphorylation of PSD-93 at Thr612 in a Rho-kinase-dependent manner ( Figure 6A-C).
These results indicate that Rho-kinase phosphorylates PSD-93 at Thr612 downstream of NMDARs during synaptic plasticity.

Roles of PSD-93 Phosphorylation in PSD Complex Formation
MAGUK family proteins can bind to AMPARs and NMDARs via their PDZ domains [32]. PSD-93 is a MAGUK family protein that is closely related to PSD-95 in terms of expression, amino acid sequence, domain organization and functions [23]. In this study, PSD-93 was immunoprecipitated with PSD-95, the NR1 subunit of NMDARs, and the GluR1 subunit of AMPARs. Treatment with calyculin A enhanced this effect, whereas pretreatment with Y-27632 inhibited this effect ( Figure 5A-D). Ectopic expression of PSD-93 and PSD-95 in COS7 cells resulted in an interaction between PSD-93 and PSD-95. This interaction was facilitated by the phosphorylation of PSD-93 at Thr612 and Thr585 of PSD-93 by Rho-kinase ( Figure 5E,F). These results suggest that PSD-93 and PSD-95 form heterodimers in a phosphorylation-dependent manner and that the interaction contributes to LTP induction by stabilizing NMDAR localization at the plasma membrane [14,33]. In contrast, Rho-kinase negatively regulates the interaction of PSD-93 with SynGAP1, ADAM22, and LGI1 (Supplementary Materials, Figure S3A-D). PSD-93 can interact with SynGAP1 via the RasGAP domain (amino acids 670-685) [34]. In our previous study, we found that Rhokinase phosphorylates SynGAP1 [11]. Thus, Rho-kinase phosphorylates both SynGAP1 and PSD-93, thereby dissociating SynGAP1 from the heterodimer of PSD-93 and PSD-95. The dissociation of SynGAP1 from the PSD region leads to the activation of the Ras-ERK pathway and promotes spine enlargement [11,35]. ADAM22, LGI1, and MAGUKs (including PSD-95 and PSD-93) form protein complexes in vivo and play an important role in synaptic transmission via AMPARs and NMDARs [36]. In the present study, we found that Rho-kinase phosphorylates the PDZ3-SH3 domain of PSD-93, which is the binding region of ADAM22, and negatively regulates the binding of PSD-93 to ADAM22 and LGI1 in a phosphorylation-dependent manner (Supplementary Materials, Figure S3A,C,D). However, the functional role of the phosphorylation-dependent regulation of PSD-93 binding to ADAM22 and LGI1 by Rho-kinase requires further elucidation.

Roles of PSD-93 Phosphorylation in Synaptic Plasticity
NMDAR stimulation induces Ca 2+ influx into neurons, activates CaMKII and further activates the RhoA-Rho-kinase pathway, leading to dendritic spine enlargement and LTP [7,11] (Supplementary Materials, Figure S3). However, the mechanism by which Rho-kinase regulates spine enlargement and LTP remains poorly understood. To clarify the function of Rho-kinase in synaptic plasticity, we focused on PSD-93 among the Rho-kinase candidate substrates. MAGUK family proteins including PSD-93 directly interact with NMDARs and indirectly interact with AMPARs via stargazin/TARP, contributing to the membrane localization and stabilization of these receptors [37]. In fact, targeted disruption of the PSD-93 gene not only reduces surface NR2A and NR2B expression, but also NMDAR-mediated excitatory postsynaptic currents and potentials [38]. Simultaneous knockdown/knockout of three MAGUK members-PSD-93, PSD-95 and SAP102-reduces almost all AMPAR-mediated synaptic transmission in rat hippocampal slice cultures [23,39,40]. In the present study, we found that PSD-93 phosphorylation by Rho-kinase promoted the interaction of PSD-93 with PSD-95 ( Figure 5A,B,E,F). Chemical LTP induction increased dendritic spine volume and the colocalization of PSD-93 with PSD-95 in cultured striatal neurons ( Figure 6D,E). On other hand, PSD-93 phosphorylation by Rho-kinase decreased the interaction with SynGAP1 (Supplementary Materials, Figure  S3A,B,E,F). The delocalization of SynGAP1 from postsynaptic density region increases dendritic spine size [11]. Taken together, these results suggest that Rho-kinase positively regulates the interaction of PSD-93 with PSD-95, NMDARs and AMPARs, and negatively regulates the interaction of PSD-93 with SynGAP1, ADAM22 and LGI1 for orchestrating synaptic functions as depicted in Supplementary Materials, Figure S3. Nonetheless, further electrophysiological and behavior studies should be done not only with conditional knockout but also total knockout mice to elucidate a clearer understanding.

Animals
Male C57BL/6J (RRID: MSR_JAX:000664) and pregnant female ICR mice (RRID: IMSR_JAX:009122) were purchased from Japan SLC (Hamamatsu, Shizuoka, Japan). The four mice were kept in a cage (17 cm wide × 28 cm long × 13 cm high) in an animal facility that was pathogen free under standard conditions (23 ± 1 • C, 50 ± 5% humidity) and a 12-h light/dark cycle (light phase 9:00-21:00). All mice had free access to food and water. The mice were carefully handled by laboratory personnel in order to reduce their suffering. Between control and experiment, the mice were randomly chosen. All animal experiments were approved and performed in accordance with the guidelines for the care and use of laboratory animals established by the Animal Experiments Committee of Nagoya University Graduate School of Medicine (approval number: 20094) and Fujita Health University (approval number: AP20037). All experiments were conducted in compliance with the ARRIVE guidelines.

Striatal Neuronal Culture
Primary striatal neurons were collected and isolated from E15-E16 mouse embryos using papain as previously described [43]. In this method, pregnant ICR mice were sacrificed by cervical dislocation. Then the brains were dissected humanly from the mouse embryos and kept in 1 × HBSS (Thermo Fisher Scientific) solution. Later, the striatal neurons were prepared from the collected embryo striatum using neuron dissociation solutions (FUJIFILM Wako, Tokyo, Japan) according to manufacturer's protocol. The neurons were seeded on 60 mm dishes that had been previously coated with poly-Dlysine (PDL; Merck). After 2 h of incubation with Neurobasal Medium TM (Thermo Fisher Scientific) along with 10% fetal bovine serum (FBS; Merck), the medium was replaced with Neurobasal Medium TM (Thermo Fisher Scientific) supplemented with B-27 (Thermo Fisher Scientific) and 1 mM GlutaMAX (Thermo Fisher Scientific). The neurons were cultured until DIV21 in a humidified atmosphere with 5% CO 2 at 37 • C so that they could mature and develop the dendritic spines. After maturation, the neurons were used for experiments.

Striatal Slice Culture
Samples from 7-to 8-week-old C57BL/6J male mice were used for striatal slice culture. In this method, the mice were sacrificed humanly by beheading. Then the brain of each mouse was immediately collected in Krebs-HCO 3 -buffer (pH 7.4, 124 mM NaCl, 26 mM NaHCO 3 , 10 mM D-glucose, 4 mM KCl, 1.25 mM KH 2 PO 4 and 1.5 mM CaCl 2 ). According to previously described methods [45,46], the mouse brain was coronally sliced (350 µM) using a VT1200S vibratome (Leica Microsystems). Then, the striatal slices were incubated in the Krebs-HCO 3 -buffer containing 10 µg/mL adenosine deaminase (Roche) at 30 • C for 30 min with oxygenation (95% O 2 /5% CO 2 ). The adenosine deaminase-containing Krebs-HCO 3 -buffer was changed to fresh Krebs-HCO 3 -buffer, and the slices were incubated at 30 • C for 30 min. The striatal slices were treated with calyculin A (250 nM) for 60 min with or without Y-27632 (20 µM). After drug treatment, the striatal slices were frozen in liquid nitrogen and stored at −80 • C. The proteins were extracted from striatal slices using 1% SDS buffer. A BCA (Fujifil Wako) assay was carried out to quantify the protein concentration in each sample.

Immunoprecipitation Assay
The calyculin A (250 nM for 60 min) and Y-27632 (20 µM for 60 min) treated striatal slices were sonicated immediately with 320 µL of RIPA buffer (pH 7.5, 150 mM NaCl, 50 mM Tris-HCl, 1 mM EDTA, 1% NP-40, 0.1% SDS, 0.5% sodium deoxycholate, protease inhibitor cocktail [Roche], PhosStop [Roche] and calyculin A [50 nM]). The solution was centrifuged at 16,000× g at 4 • C for 10 min. The soluble supernatant was transferred to a new tube and incubated with an anti-PSD-93 antibody (1-2 µg) along with a control rabbit anti-IgG antibody (1-2 µg). The samples were gently rotated on a rotator at 4 • C for 1 h. Next, 25 µL of Protein A Sepharose beads (20% ethanol) were added to each tube, and the tubes were rotated for 1 h. The proteins unbound to beads were washed out with a wash buffer (150 mM NaCl, 50 mM Tris-HCl, 1 mM EDTA, pH 7.5). Immunoblotting and LC-MS/MS were then performed on the immunoprecipitated samples.

Mass Spectrometry
The mass spectrometry (MS) was performed as previously described by [45] with some modifications. The mouse striatal slices were treated with calyculin A (250 nM) with or without Y-27632 (20 µM). Subsequently, an immunoprecipitation assay with an anti-PSD-93 antibody was performed as described previously. The PSD-93-interacting proteins were later isolated from the beads by rotating with guanidine solution (7 M guanidine and 50 mM Tris) for 1 h. The disulfide bond of proteins was then reduced with 5 mM dithiothreitol for 30 min. The hydroxyl group of amino acids were alkylated with 10 mM iodoacetamide for 1 h in the dark. A trypsin solution (50 mM NH 4 HCO 3 , 1.2 M urea, and 0.5 µg of Trypsin/Lys-C) was used to make small peptides from proteins. Desalting was performed by using SPEC tips (Nikkyo Technos, Tokyo, Japan) according to the manufacturer's protocol. The peptides were analyzed by LC-MS using an Orbitrap Fusion Mass Spectrometer (Thermo Fisher Scientific Inc., MA, USA) coupled to an UltiMate 3000 RSLCnano LC system (Dionex Co., Amsterdam, The Netherlands) using a nano HPLC capillary column (150 mm × 75 m id, Nikkyo Technos Co., Tokyo, Japan) via a nanoelectrospray ion source.
Reversed-phase chromatography was performed with a linear gradient (0 min, 5% B; 100 min, 40% B) of solvent A (2% acetonitrile with 0.1% formic acid) and solvent B (95% acetonitrile with 0.1% formic acid) at an estimated flow rate of 300 nl/min. A precursor ion scan was carried out using a 400-1600 mass-to-charge ratio (m/z) prior to tandem MS (MS/MS) analysis. Tandem MS was performed by isolation at 0.8 Th with the quadrupole, HCD fragmentation with a normalized collision energy of 30%, and rapid scan MS analysis in the ion trap. Only those precursors with a charge state of 2-6 were sampled for MS2. The dynamic exclusion duration was set to 10 sec with a 10-ppm tolerance. The instrument was run in top speed mode with 3 sec cycles. A peak list was generated and calibrated using MaxQuant software [47]. Database searches against the reference proteome of Mus musculus obtained from UniProtKB were performed using MaxQuant software. False discovery rates (FDRs) at the peptide, protein, and site levels were set to 0.01. The peak ion intensities obtained from two independent experiments were analyzed.

Cell Culture
From a 1 mL stock that was stored at −80 • C, the COS7 (ATCC, Manassas, VA, USA) cells were cultured in 100-mm plates. The cells were cultured overnight at 37 • C with Dulbecco's modified Eagle's medium (DMEM) (Merck) containing 10% FBS (Merck) supplements and 5% CO 2 in a humidified atmosphere. The cells were incubated until they reached 70-80% confluence. Then, the culture plates were randomly chosen for control and experiments. The pEF-BOS-GST and pCAGGS-Myc plasmids were transformed with Lipofectamine 2000 (Thermo Fisher Scientific) into cells according to the manufacturer's protocols. Later the cells were treated with calyculin A (50 nM for 12 min) with or without Y-27632 (20 µM for 30 min). The stimulated cells were collected with lysis buffer (pH 7.5, 150 mM NaCl, 20 mM Tris-HCl, 1 mM EDTA, 1% NP-40, PhosStop [Roche], protease inhibitor cocktail [Roche] and 50 nM calyculin A) for the GST pull-down assay.

GST Pull-Down Assay
The pEF-BOS-GST and pCAGGS-Myc plasmids were transfected with lipofectamine 2000 (Thermo Fisher Scientific) into COS7 cells according to the manufacturer's protocols. Next, the COS7 cells were treated with 50 nM calyculin A with or without 20 µM Y-27632 as previously described [48]. The stimulated cells were sonicated with 700 µL of lysis buffer (pH 7.5, 150 mM NaCl, 20 mM Tris-HCl, 1 mM EDTA, 1% NP-40, protease inhibitor cocktail [Roche], and PhosStop [Roche], 50 nM calyculin A). The solution was then centrifuged at 16,000× g for 10 min at 4 • C. The supernatant was transferred to a fresh tube and incubated in a rotor with 20 µL of glutathione-Sepharose 4B beads (GE Healthcare) for 1 h at 4 • C. After rotation, the unbound GST-proteins were removed from the beads by washing with wash buffer (pH 7.5, 20 mM Tris-HCl, 1 mM EDTA, 150 mM NaCl). The beads were then boiled for 10 min with a 1× SDS buffer. Finally, the boiled samples were subjected to immunoblotting with the indicated antibodies.

Immunoblotting
Protein (5-20 µg) from each sample was subjected to SDS-PAGE on 7-8% acrylamide gels. The individual proteins were separated using SDS-PAGE. After SDS-PAGE, the proteins were transferred from the gel onto polyvinylidene difluoride membranes (Immobilon-FL, Millipore, Bedford, MA, USA) using a Trans-Blot Turbo system. Next, the Blocking One buffer (Nacalai Tesque, Kyoto, Japan) was used to block the membranes at room temperature for 30 min. Depending on the primary antibody, the membranes were incubated overnight at 4 • C or at room temperature for 1 h. Anti  :1000), and anti-LGI1 (1:1000) were used. The unbound primary antibodies were washed away, and the membranes were incubated with goat anti-rabbit Alexa Fluor 680 and/or goat anti-mouse IRDye 800CW for 1 h at room temperature. The total protein or phospho-proteins were detected by infrared imaging (LI-COR Biosciences Lincoln, NE). The band intensities were quantified using ImageStudio software (LI-COR Biosciences).

Chemical LTP Assay
Primary striatal neurons were dissected from E16 embryonic mice and cultured according to the protocol described in striatal neuronal culture. After DIV21, the neurons were used for chemical LTP induction as previously described [11]. In this method, the neurons