Extracellular Signal-regulated Kinase and Glycogen Synthase Kinase 3β Regulate Gephyrin Postsynaptic Aggregation and GABAergic Synaptic Function in a Calpain-dependent Mechanism*

Background: Molecular mechanisms of plasticity at GABAergic synapses are presently unclear. Results: ERK phosphorylates gephyrin at Ser-268 to regulate size of gephyrin postsynaptic scaffold and strength of GABAergic transmission. Ser-268 phosphorylation by ERK is functionally coupled to Ser-270 phosphorylation by GSK3β to determine calpain action on gephyrin. Conclusion: Multiple signaling cascades regulate gephyrin postsynaptic clustering. Significance: Dynamic modulation of gephyrin clustering by phosphorylation regulates GABAergic synaptic transmission. Molecular mechanisms of plasticity at GABAergic synapses are currently poorly understood. To identify signaling cascades that converge onto GABAergic postsynaptic density proteins, we performed MS analysis using gephyrin isolated from rat brain and identified multiple novel phosphorylation and acetylation residues on gephyrin. Here, we report the characterization of one of these phosphoresidues, Ser-268, which when dephosphorylated leads to the formation of larger postsynaptic scaffolds. Using a combination of mutagenesis, pharmacological treatment, and biochemical assays, we identify ERK as the kinase phosphorylating Ser-268 and describe a functional interaction between residues Ser-268 and Ser-270. We further demonstrate that alterations in gephyrin clustering via ERK modulation are reflected by amplitude and frequency changes in miniature GABAergic postsynaptic currents. We unravel novel mechanisms for activity- and ERK-dependent calpain action on gephyrin, which are likely relevant in the context of cellular signaling affecting GABAergic transmission and homeostatic synaptic plasticity in pathology.

Long term plasticity of glutamatergic synapses, such as Long Term Potentiation (LTP) and Long Term Depression (LTD), is widely accepted as a cellular mechanism underlying learning and memory. These forms of plasticity involve synaptic scaling and structural remodeling, enabling stable changes in information processing in specific neuronal circuits. Proteins of the postsynaptic density play a key role in these adaptive changes, especially by regulating trafficking and function of structural and signaling molecules (1,3). Two major signaling cascades involved in long lasting synaptic plasticity are the mTOR (mammalian target of rapamycin) and the MAP kinase pathways (4), with ERK1/2 being essential for early structural adaptations and changes in gene expression underlying multiple forms of learning and memory.
At GABAergic synapses, studies of plasticity mechanisms have focused mainly on the modulation of GABA A receptors (GABA A R) (5,6). 3 Several protein kinases are known to modulate GABAergic transmission by affecting GABA A R channel gating properties or cell surface trafficking (7,8). Similar mechanisms also operate at glycinergic synapses, where, for instance, phosphorylation of a specific residue in the glycine receptor ␤ subunit affects surface mobility and synaptic function (9). With a focus on the scaffolding molecule gephyrin, we have recently demonstrated that regulation of a single gephyrin residue (Ser-270) via GSK3␤-mediated phosphorylation induces the formation of new, functional GABAergic synapses (2). Although several studies identified gephyrin as a target for serine/threonine-directed phosphorylation (10 -12), a recent study detailed 18 different phosphorylation residues on gephyrin (13), demonstrating the importance of gephyrin phosphorylation for its varied functions. Structural changes affecting the postsynaptic scaffold organized by gephyrin have started gaining attention, with studies demonstrating neuronal activity-dependent dynamics of gephyrin clustering in vivo (14 -16). Given that phosphorylation and intracellular Ca 2ϩ rises make gephyrin susceptible to proteolysis by calpain (2), neuronal activitydriven gephyrin dynamics could very likely be phosphorylation-dependent.
Hence, to uncover novel signaling pathways converging on gephyrin and regulating calpain 1 action on gephyrin, we performed MS analysis of gephyrin immunoprecipitated from rat brain homogenate and identified novel phosphorylation and acetylation sites on gephyrin. Site-directed mutagenesis and screening of different gephyrin mutants for alterations in postsynaptic clustering phenotypes allowed us to identify the Ser-268 residue as a novel phosphorylation site on gephyrin important for scaling (up or down) GABAergic transmission. Using a multidisciplinary approach, we demonstrate that ERK modulates the Ser-268 phosphorylation status, thereby influencing GSK3␤-mediated phosphorylation of the closely related Ser-270 residue to regulate gephyrin cluster size and density, respectively. Furthermore, our data shows how neuronal activity-dependent down-regulation of gephyrin clusters is orchestrated via ERK and calpain activation, providing mechanistic insights into plasticity-related changes at GABAergic synapses.
Primary Neuron Culture Treatments and Staining-All animal experiments were approved by the cantonal veterinary office of Zurich. Primary hippocampal mixed cultures were prepared from embryonic day 17 rat embryos and maintained in medium containing minimum essential medium, Nu serum (15%), B27 supplement, HEPES (pH 7.1; 15 mM), glucose (0.45%), sodium pyruvate, and L-glutamine. The cells were transfected after 11 DIV with 1 g of total plasmid concentration using a combination of Lipofectamine 2000 and Combi-Mag (OZ Biosciences) as described (19). The transfected cells were fixed in 4% paraformaldehyde for 10 min at room temperature, permeabilized with 0.1% Triton X-100 in PBS containing 10% normal goat serum, and stained using desired primary antibodies for 60 min at room temperature, followed by three 5-min 1-PBS washes; goat anti-Alexa 488 (Molecular Probes), goat anti-Cy3, and goat anti-Cy5 (Jackson Immunoresearch) were used to label the proteins for immunofluorescence either 4 or 7 days post-transfection (referred to as 11 ϩ 4 or 11 ϩ 7 DIV). ␣2-GABA A R subunit staining was done live for 90 min before proceeding with fixation and permeabilization of cells and proceeding with staining for gephyrin and/or synapsin-1. Pharmacological treatment using 25 M ERK1/2 inhibitor PD98059 (Calbiochem) or vehicle control Me 2 SO (equal volume) were performed overnight, or for 3 h, followed by fixation and staining. GSK3␤ inhibition was done using 25 M inhibitor GSK3␤IX (Calbiochem) overnight, and calpain was inhibited using 30 M inhibitor MDL28170 (Calbiochem) overnight. Neurons were stimulated using 40 mM KCl for 4 min prior to fixation and staining.
HEK-293 Cell Cultures-HEK293-T cells were cultured at 37°C in 5% CO 2 in DMEM supplemented with 10% FBS. The HEK cells were transfected 16 h after plating with 1 g of each plasmid using the transfection agent polyethylenimine (Polysciences Inc.) as suggested by the vendor. The transfected cells were lysed for biochemical analysis 24 h post-transfection using EBC buffer.
To increase the levels of phosphoERK1/2 in HEK cells, transfection of FLAG-gephyrin S268A was performed as described above, and 16 h post-transfection the cells were serum-starved for 4 h and stimulated with 20% FBS-containing medium. The phospho-ERK1/2 levels and FLAG-gephyrin S268A phosphorylation were analyzed at the indicated time points.
Immunoprecipitation and Western Blot Analysis-Immunoprecipitation and Western blot (WB) protocols were performed using whole cell lysate from HEK cells or rat brain in EBC buffer (50 mM Tris, pH 8.0, 120 mM NaCl,0.5% Nonidet P-40; cOmplete mini protease inhibitor mixture) (Roche) and phosphatase inhibitor mixture 1 and 3 (Sigma). 1 g of desired antibody (mouse anti-FLAG, Sigma; mouse IgG, Jackson Immunoresearch) was used for immunoprecipitation, and the complexes were collected using 25 l of protein A-agarose (Calbiochem). The complexes were washed three times in icecold EBC buffer to remove nonspecific interactions before denaturation in 2ϫ SDS loading buffer at 90°C for 3 min. The immunoprecipitated complexes were analyzed using WB assay, with rabbit anti-HA (Santa Cruz, 1:1000) and HRP-conjugated mouse or rabbit secondary.
In Vitro Kinase Assay-In vitro kinase assay was performed using purified full-length STREP-gephyrin expressed in bacteria (21). The purified gephyrin was phosphorylated in kinase buffer (50 mM MOPS, pH 6.5, 100 M ATP, 10 mM MgCl 2 , 1 mM EGTA, and H 2 O to a final volume of 50 l) and purified with activated ERK 0.5 l (Calbiochem) at 30°C for 30 min. The reaction was stopped with the addition of 2ϫ SDS loading buffer and boiling the samples at 90°C for 3 min. WB to detect gephyrin phosphorylation was performed using either mouse monoclonal antibody 3B11 to detect total gephyrin or affinitypurified rabbit anti-phospho-gephyrin Ser-268 (2 g/ml).
MS Analysis-Gephyrin immunoprecipitated from rat whole brain lysate using the monoclonal antibody 3B11 was separated on SDS-PAGE. Coomassie-stained protein bands were subjected to reduction, alkylation, and in-gel tryptic digestion as described previously (22). The digests were analyzed by nanoflow LC-MS/MS using an LTQ Orbitrap Velos mass spectrometer coupled to an EASY-nLC II liquid chromatograph (Thermo Fisher Scientific). A database search was performed against the Swiss-Prot (Rattus norvegicus) using Mascot 2.4 (Matrix Science) via Proteome Discoverer 1.3 (Thermo Fisher Scientific).
Image Analysis and Quantification-Confocal laser scanning microscope (LSM 510 Meta; Carl Zeiss) was used for acquiring images in sequential mode over the full dynamic range of the photodetectors with a 100ϫ objective lens (NA 1.4). The pinhole was set at 1 Airy unit, and pixel size was set at 90 nm. At least nine cells from three independent batches per condition were used to acquire images as a z-stack (three optical sections, 0.5-m step size) for analysis. Image processing and analysis was done using the software ImageJ, for which maximum intensity projections were created from the z-stacks for analysis.
Gephyrin cluster size and density analysis has been described earlier (2). Unless indicated, only gephyrin clusters apposed to synapsin-1 or VGAT-positive puncta were considered for analysis. Total dendritic length in image analysis area was calculated using the line tool in ImageJ, and cluster density per 20-m length of dendrite was calculated from this value. The integrated density values were used to calculate the cluster area in m 2 .
Electrophysiology-Whole cell patch clamp recordings were made from transfected hippocampal neurons (11 ϩ 4 DIV) at room temperature and at a holding potential of Ϫ60 mV. Mock (nontransfected) transfected cells on the same coverslip were used as controls. Recordings were performed as described previously (2) with a HEKA EPC-7 amplifier and Patch Master v2.11 software (HEKA Elektronik). Spontaneous GABAergic miniature postsynaptic currents (mIPSCs) were recorded in the presence of Tetredotoxin (TTX) (0.1 M; Tocris) and isolated pharmacologically using 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) (2 M; Tocris) and AP-5 (50 M; Tocris). At least 175 individual events in a total of three neurons per culture were analyzed for each condition. For the average amplitude calculation, only cells with more than 25-30 events in the recording period were included. Neurons that did not display GABAergic mIPSCs were excluded from the statistical analyses.
Statistical Analyses-Statistical analysis of morphological data (gephyrin cluster size and density) was performed either pair-wise (mutant versus WT) using unpaired Student's t test or Kolmogorov-Smirnov (K/S) test, or using one-way ANOVA followed by Bonferroni post hoc test when multiple groups were compared. Amplitude and interevent interval distributions of mIPSCs were analyzed pair-wise using K/S test. p Յ 0.05 was considered as being statistically significant.

Phosphopeptide Analysis of Gephyrin from Rat Brain
Homogenate-We have reported the identification of Ser-270 as a phosphorylation residue and characterized the functional significance of this residue for GABAergic synapse formation (2). In this report, we also described gephyrin as a substrate for calpain and demonstrated that phosphorylation at Ser-270 promotes gephyrin processing by calpain. Given the importance of gephyrin for GABAergic transmission, we hypothesized that this regulation by calpain might involve activation of more than one signaling pathway. To identify additional kinase pathways and determine their regulation of gephyrin, we immunoprecipitated gephyrin from rat brain homogenate and subjected it to MS phosphopeptide analysis. In this process, we identified 10 phosphorylated and 10 acetylated residues on gephyrin, many of which are novel and have not been described previously (Table 1). Although there is a report identifying gephyrin acety-   Ser-262, Ser-286, Ser-303, Ser-305, and Ser-319 as phosphoresidues in gephyrin from rat brain. Consistent identification of residues phosphorylated across different cell types and organisms suggests that gephyrin phosphorylation is well conserved, and convergence of different signals onto gephyrin might orchestrate its spatial and temporal functional dynamics via the phosphorylation of subset of residues. Among the many phosphorylation residues identified in our own MS analysis, the Ser-268 site drew our attention, because of its proximity to the Ser-270 residue, which we have shown previously to be targeted by GSK3␤, and the fact that it does not appear to be phosphorylated in the C3-gephyrin splice variant, expressed mainly in non-neuronal cells (13). Hence, we proceeded to understand the biological significance of Ser-268 phosphorylation site on gephyrin, focusing on postsynaptic clustering.
Ser-268 Mutation Affects Gephyrin Clustering Phenotype-To investigate the relevance of Ser-268 phosphorylation, we mutated Ser-268 in eGFP-gephyrin to create eGFP-S268A and eGFP-S268E gephyrin mutants for transfection in primary hippocampal neurons. Cultures were transfected by magnetofection after 11 DIV and analyzed 7 days later (11 ϩ 7 DIV) using immunofluorescence staining and confocal microscopy. In this preparation, eGFP-tagged gephyrin forms clusters at postsynaptic sites identified by apposition to axon terminals labeled for a presynaptic marker (synapsin-1 or VGAT) or alternatively by co-localization with the GABA A R ␣2 subunit (2). Therefore, changes in size and density of fluorescent clusters formed by eGFP-gephyrin S268A and S268E mutants were quantified in comparison with WT eGFP-gephyrin ( Fig. 1, A-C). As a control, we verified that eGFP-gephyrin S268A and S268E clusters co-localized with GABA A R, visualized by ␣2 subunit immunofluorescence ( Fig. 1, D and DЈ), confirming their postsynaptic localization.
In our previous study, we had demonstrated that expression of eGFP-gephyrin S270A in cultured neurons was independent of endogenous gephyrin expression (2). Hence, we wanted to make sure that the phenotypes of eGFP-gephyrin S268A and S268E were not being influenced by the presence of endogenous gephyrin. Neurons were co-transfected with an shRNA targeting the GPHN mRNA 3Ј-UTR to silence endogenous gephyrin without affecting the expression of eGFP-constructs (which lack the 3Ј-UTR). Efficiency of this construct to selectively deplete endogenous gephyrin has been documented earlier (2). The cells were analyzed for eGFP-gephyrin postsynaptic clustering after 11 ϩ 7 DIV (Fig. 1E). Consistent with our earlier observations, depletion of endogenous gephyrin did not affect the WT eGFP-gephyrin phenotype; however, silencing endogenous gephyrin expression changed the phenotypes of eGFP-S268A and S268E mutants (Fig. 1, EЈ and EЉ), as confirmed by quantitative analysis (one-way ANOVA; density: F 2,66 ϭ 18.06; p Ͻ 0.0001; size: F 2,1368 ϭ 45.15; p Ͻ 0.001). The density of eGFP-S268A clusters was reduced (Bonferroni post hoc test, p Ͻ 0.01; Fig. 1F), but their size was markedly larger compared with WT gephyrin (p Ͻ 0.001; Fig. 1G). Expression of eGFP-S268E in the absence of endogenous gephyrin resulted in increased density compared with eGFP-gephyrin (11 Ϯ 1.2 versus 7.2 Ϯ 0.4 per 20-m dendrite; p Ͻ 0.05; Fig. 1F), whereas their size was decreased (p Ͻ 0.001; Fig. 1G). The observation that Ser-268 mutant phenotypes are affected by the presence or absence of endogenous gephyrin underscores the fact that sequence alteration at Ser-268 was not the cause for the S268A and S268E phenotypes. Furthermore, these results suggest a primary role of Ser-268 phosphorylation in regulating gephyrin cluster size, whereas the effect of S268A or S268E mutation on cluster density reflects the interaction with another post-translational modification of gephyrin (the difference in S268A and S268E cluster density in the presence or absence of endogenous gephyrin is characterized better in Fig. 5). ERK Regulates Ser-268 Site on Gephyrin-To identify a suitable kinase to phosphorylate the Ser-268 residue, we performed a bioinformatics search for potential conserved kinase domains around Ser-268 using a published algorithm (24) and identified ERK as a potential kinase that might phosphorylate this residue. We therefore tested the ERK inhibitor PD98059 (25 M) in transfected neurons. After overnight treatment we observed increased eGFP-gephyrin cluster density (15.4 Ϯ 1.0 versus 9.4 Ϯ 0.5; Student's t test, t 61 ϭ 8.63; p Ͻ 0.0001) and size (Mann Whitney test, p Ͻ 0.0001; Fig. 2, A-C), mimicking the phenotype of the S268A mutant. As a control, we treated cells expressing eGFP-S268E construct with PD98059 and did not observe any increase in the cluster size (Fig. 2, AЈ and C; Student's t test, t 1552 ϭ 0.52; p ϭ 0.6); however, we saw a significant increase in cluster density (17.3 Ϯ 0.9 versus 6.8 Ϯ 0.4; Student's t test, t 52 ϭ 6.98; p Ͻ 0.0001; Fig. 2B), like the change seen upon depletion of endogenous gephyrin (Fig. 1F). These findings suggested that ERK might be the kinase modulating Ser-268 phosphorylation status to modulate gephyrin cluster size; further, they confirm that the regulation of gephyrin cluster density is not primary mediated by phosphorylation of Ser-268.
Gephyrin Is a Direct Substrate for ERK Phosphorylation-To confirm the role of ERK in phosphorylating gephyrin at Ser-268, we generated an antibody against the Ser(P)-268 peptide Loading controls IP Ab IgG FLAG WB: HA Gephyrin is a direct substrate for ERK phosphorylation. A, detection of Ser(P)-268 in gephyrin pulled down from rat whole brain extract. Cyto and Memb represent the soluble cytosolic fraction and insoluble membrane-associated fraction. The nitro-cellulose membrane was separated into two, and each of the gephyrin immunoprecipitation lanes were probed with either Ser(P)-268 antibody (S268, left panel) or Ser(P)-268 antibody blocked with Ser(P)-268 peptide to demonstrate selective epitope recognition (right panel). Ab, antibody. Each of the two blots were stripped and probed again with 3B11 mouse anti-gephyrin to show that equal levels of gephyrin was precipitated in all the lanes (lower panels). B, in vitro phosphorylation of gephyrin using purified ERK1 and detected using a polyclonal antibody against gephyrin Ser(P)-268. The gephyrin band was detected only in the presence of ERK1 and ATP (fourth lane) and not in the presence of protein phosphatase (PP) or PKA (fifth and sixth lanes). Staining the blot with Coomassie showed equal amounts of full-length gephyrin in all lanes (lower panel). C, FLAG-gephyrin (geph) interaction with HA-ERK1 and HA-ERK2 in HEK293 cells. Immunoprecipitation (IP) for FLAG-gephyrin using mouse anti-FLAG antibody followed by WB against HA showed that HA-ERK1 interacts with FLAG-gephyrin (top blot, second lane) and FLAG-GC domain (fifth lane). Loading controls demonstrate the presence of FLAG and HA constructs as indicated. The scheme depicts gephyrin domain organization. D, bacterial purified STREP-gephyrin immobilized using STREP-tactin beads and incubated with HEK293 cell extracts overexpressing HA-ERK1 and HA-ERK2 shows an interaction between STREP-gephyrin and ERK1, as well as a weak interaction with HA-ERK2. and performed in vitro kinase assay using purified ERK1 and PKA. First, we tested the presence of native gephyrin phosphorylated at Ser-268 in the rat brain by WB, using whole brain homogenate fractionated into soluble cytosolic fraction and insoluble membrane-bound fraction (Fig. 3A). The protein levels were controlled using the monoclonal antibody 3B11 (lower panels). The Ser(P)-268 antibody showed a more intense signal in the membrane-bound fraction of gephyrin compared with the cytosolic fraction. The specificity of the antibody was confirmed by preincubation with excess phosphopeptide prior to WB (right top panel). Next, we proceeded with the in vitro kinase assay using bacterially expressed STREP-gephyrin and purified kinase (Fig. 3B). Our Ser(P)-268 antibody detected phosphorylated gephyrin only in the presence of ERK1 and ATP (fourth lane), but not when we performed the same assay in the absence of ATP or presence of protein phosphatase or purified PKA (fifth and sixth lanes). The amount of STREPgephyrin used in our assay is shown (bottom panel) using simple Coomassie staining of the membrane after the WB with Ser(P)-268 antibody.
Next, to confirm that gephyrin is a direct substrate for ERK1, we looked for potential interactions between HA-ERK1 or HA-ERK2 (25) and FLAG-gephyrin overexpressed in HEK293 cells (Fig. 3C). Pulldown for FLAG-gephyrin followed by WB for HA-ERK showed an interaction between HA-ERK1 and FLAGgephyrin (Fig. 3C, top panel, second lane). We also tested individual gephyrin domains FLAG-G, FLAG-GC, and FLAG-E for ERK-1 interaction (Fig. 3C, top panel, fourth, fifth, and sixth lanes) and found a robust interaction with the N-terminal GC domain of gephyrin but not the E-domain.
We additionally confirmed this interaction using a bacterially overexpressed STREP-gephyrin. We pulled down HA-ERK from lysates of HEK293 overexpressing HA-ERK1 or HA-ERK2 (Fig. 3D, bottom panel) using bacterial STREP-gephyrin. WB for HA confirmed the interaction between gephyrin and HA-ERK1; in addition we could also see a weak interaction with HA-ERK2. Collectively, these results confirmed ERK as a candidate enzyme phosphorylating gephyrin at Ser-268.
Functional Characterization of Gephyrin Ser-268 Mutants-To assess the functional relevance of Ser-268 phosphorylation, whole cell patch clamp recordings of pharmacologically isolated GABAergic mIPSCs were performed in 11 ϩ 4 DIV hippocampal neurons. Compared with mock transfected cells present in the same cultures, overexpression of WT eGFPgephyrin did not influence mIPSC amplitudes or interevent intervals ( Fig. 4 and Table 2). In contrast, the average amplitude of mIPSC recorded in neurons expressing S268A was 34% larger than control (Fig. 4, B and C; K/S test, p Ͻ 0.01), and the interevent intervals were shortened by 45% (Fig. 4, D and E, and Table 2; K/S test, p Ͻ 0.001), suggesting increased size and density of functional GABAergic synapses in S268A-expressing neurons. Cells expressing eGFP-gephyrin S268E mutant showed mIPSC similar to WT or mock transfected cells (Fig. 4,  A-E). Analyses of rise and decay kinetics of GABAergic mIPSC showed only minor differences in S268A or S268E mutants (Fig.  4, F and G). Thus, constitutive blockade of gephyrin Ser-268 phosphorylation facilitates GABAergic synaptic transmission by increasing GABAergic mIPSC amplitude and by the formation of additional GABAergic release sites.
Phosphorylation at Ser-268 and Ser-270 Regulates Distinct Aspects of Gephyrin Clustering-A potential cross-talk between two phosphorylation residues on gephyrin became apparent, as PD98059 treatment of S268E caused an increase in cluster density and the down-regulation of endogenous gephyrin in S268Eexpressing neurons using GPHN shRNA also influenced postsynaptic cluster density. Given the identification of GSK3␤ as  Table 2 for statistical analysis; ***, p Ͻ 0.001).
the kinase targeting Ser-270 (2) and the proximity of the Ser-268 site, we speculated that these two sites are functionally coupled and influence each other's phosphorylation status. Hence, we wanted to test whether gephyrin cluster size was under the control of Ser-268 residue and cluster density under the control of Ser-270 residue. We generated S268A/S270A, S268A/S270E, and S268E/S270E eGFP-gephyrin double mutants and co-expressed them in neurons along with GPHN 3Ј-UTR shRNA to silence endogenous gephyrin (Fig. 5, A-E). Quantification of the mutant phenotypes confirmed the idea that these two sites are functionally coupled (Fig. 5, F and G; one-way ANOVA, cluster density: F 3,72 ϭ 19.31, p Ͻ 0.0001; cluster size: F 3,3269 ϭ 5.44; p ϭ 0.001). As predicted, S268A/ S270A mutant showed an increase in both density (Bonferroni post-test, p Ͻ 0.01; Fig. 5F) and size (p Ͻ 0.01; Fig. 5G) of postsynaptic gephyrin clusters compared with WT gephyrin, whereas expression of S268A/S270E showed increased size (p Ͻ 0.01; Fig. 5G) but no change in density of gephyrin clusters (Fig.  5F). Finally, the S268E/S270E mutant showed a phenotype similar to the WT in terms of cluster density and size (Fig. 5, E-G). These findings indicate that dephosphorylation at Ser-268 promotes cluster size increase, whereas dephosphorylation at Ser-270 promotes increase in cluster density. However, they do not yet explain why treating neurons expressing eGFP-gephyrin S268E with PD98059 results in increased cluster density. Therefore, we tested the respective roles for the Ser-268 and Ser-270 phosphorylation status on gephyrin clustering using pharmacological inhibitors. First, we treated neurons co-transfected with eGFP-gephyrin S268E and 3Ј-UTR-gephyrin shRNA with the GSK3␤ inhibitor, GSK3␤ IX, or vehicle control (Fig. 5H). As per our initial expectations, we found an increase in the density of clusters ( Fig. 5I; Student's t test, t 13 ϭ 2.289, p ϭ 0.0394) but no change in size (Fig. 5J) compared with cells treated with vehicle. To conclusively demonstrate the interaction between the Ser-268 and Ser-270 sites, we treated cultures expressing eGFP-gephyrin S268E/S270E mutant with PD98059 or GSK3␤ IX (Fig. 5, K and N). As expected, both inhibitors had no effect on the density (PD98059: t 14 ϭ 1.128 p ϭ 0.269; GSK3␤ IX: t 14 ϭ 0.3337, p ϭ 0.7436) (Fig. 5, L and O) or size (Fig. 5, M and P) (26). Our data showed that long term pharmacological block in ERK signaling using PD98059 ( Fig. 2A) increases the size and density of eGFP-gephyrin clusters at GABAergic synapses, producing a phenotype similar to S268A/ S270A double mutant (Fig. 5C). Hence, we wanted to investigate the short term effects of ERK signaling at GABAergic synapses and evaluate its significance compared with long term effects. To this end, we blocked ERK signaling using PD98059 for 3 h in cultures of eGFP-gephyrin-transfected neurons (DIV 11 ϩ 4) (Fig. 6, A and AЈ). Quantitative image analysis did not reveal a significant reduction in cluster density compared with vehicle (Me 2 SO)-treated controls (4.9 Ϯ 0.4 versus 3.7 Ϯ 0.4; Student's t test, t 18 ϭ 1.265, p ϭ 0.222), but these clusters showed a significant increase in size (0.29 Ϯ 0.01 versus 0.38 Ϯ 0.01 m 2 ; t 906 ϭ 3.393, p ϭ 0.0007) (Fig. 6, B and C). To check whether 3 h of exposure to PD98059 affected endogenous gephyrin clustering in a similar way, we treated DIV 15 neurons and stained for endogenous gephyrin using the gephyrin monoclonal antibody mAb7a (Fig. 6, D and DЈ). This treatment caused a significant reduction in cluster density (7 Ϯ 0.5 versus 4.9 Ϯ 0.4 clusters/20-m dendrite; t 16 ϭ 2.972, p ϭ 0.009); however, an increase in size was not evident (Fig. 6, E and F).
To determine the functional implications of short term blockade of ERK signaling on GABAergic transmission, we measured the amplitude and frequency of GABAergic mIPSC in nontransfected 15 DIV hippocampal neurons after a 3-h ERK inhibition with PD98059 or vehicle. This short treatment with PD98059 resulted in increased interevent intervals by 57% (Fig.  6, G and H, and Table 2), in line with the morphological data (Fig. 6, E and F). Interestingly, this loss of functional synapses was compensated by a modest increase in mIPSC amplitude by 27% (Fig. 6, I and J, and Table 2). These alterations in GABAergic signaling reflected the morphology of S268A/S270E gephyrin mutant. Hence, it is conceivable that the S268A/ S270E mutant depicts one possible physiological state wherein gephyrin phosphorylation at Ser-270 facilitates down-regulation of synaptic number to balance the excitability and/or network activity of the cell.
Neuronal Activity-dependent Regulation of Gephyrin Clusters by Calpain-Activity-dependent dynamics of gephyrin clusters at GABAergic synapses in vivo have been reported recently (14). We have shown earlier that gephyrin is a substrate for calpain, a Ca 2ϩ -dependent protein that can be activated by  ERK signaling. Hence, we wanted to test whether neuronal activity-dependent alteration of gephyrin clusters was dependent on ERK and calpain activation. ERK activation was achieved by depolarizing cultured hippocampal neurons transfected with eGFP-gephyrin using 40 mM KCl for 4 min before fixation and staining. Calpain activation was prevented using co-transfection of its endogenous inhibitor, calpastatin. As expected, this mild stimulation paradigm led to a robust increase in the levels of phospho-ERK1/2 immunofluorescence (Fig. 7, A and  AЈ). Despite its brief duration, it caused a significant decrease in eGFP-gephyrin cluster size (0.46 Ϯ 0.01 versus 0.35 Ϯ 0.01 m 2 ) without affecting their density (6.5 Ϯ 1 versus 4.8 Ϯ 0.4 clusters/20-m dendrite) (Fig. 7, B and BЈ). However, KCl treatment had no observable effect on gephyrin clustering density  Table 2 for statistical analysis). Scale bars: A and D, 10 m. Ctrl, control; PD, PD98059.  (Fig. 7, C and CЈ). These changes in morphology were confirmed by statistical analysis (one-way ANOVA; density: F 3,56 ϭ 1.015, p ϭ 0.393; size: F 3,1707 ϭ 6.329, p ϭ 0.0003; Fig. 7, D and E). These findings reveal that activity-dependent gephyrin dynamics at GABAergic synapses involves ERK activation and calpain action on gephyrin.

DISCUSSION
Only little data are available on the functional characterization of the diverse phosphorylation sites that have been identified on gephyrin (9,11,27,28). Given the sheer number of sites potentially modified by diverse kinases, it is likely that multiple functional interactions take place between them, and that some of these sites act as "master switches" controlling the phosphorylation status of downstream phosphoresidues. Our results identify Ser-268 as a phosphorylation site on gephyrin targeted by ERK signaling to modulate GABAergic synaptic function by regulating the size of postsynaptic gephyrin clusters. Further, we unravel a functional coupling between Ser-268 with Ser-270, which are targeted by ERK and GSK3␤, respectively, allowing for changes in gephyrin cluster size and density that are mirrored by changes in amplitude and frequency of GABAergic mIPSCs. Although direct evidence for the phosphorylation state-dependent co-regulation between the sites Ser-268 and Ser-270 has been hard to obtain because of technical difficulties, the indirect evidence provided by the mutational analysis of these sites suggests that this is likely the mechanism of regulation. Therefore, gephyrin phosphorylation status directly impacts on GABAergic synaptic function, presumably by allowing formation of synapses (2) and recruitment (or stabilization) of GABA A R to the postsynaptic density. Further, by extending our earlier work, we also demonstrate that activation of calpain is a general mechanism restricting postsynaptic gephyrin clustering in a phosphorylation-dependent manner. Therefore, multiple signaling cascades converge onto gephyrin to modify its scaffolding properties at GABAergic postsynaptic density and influence synapse function and homeostasis in the CNS.
Our data do not provide any evidence of position hierarchy for Ser-268 or Ser-270 residues within the context of other identified gephyrin phosphorylation sites. However, given the importance of ERK1 and GSK3␤ in neuronal function and synaptic plasticity, Ser-268 and Ser-270 appear more likely to represent downstream sites involved in fine-tuning of gephyrin (and GABAergic) function rather than "upstream switches." The importance of acetylation on serine and threonine residues for protein regulation is still unclear, but the identification of acetylation at Ser-268 suggests that protein acetylation might add another regulatory step to prevent "mistaken" phosphorylation at Ser-268 by ERK and subsequent down-regulation of FIGURE 8. Calpain activity influences eGFP-gephyrin clustering. A and AЈ, co-expression of Myc-Cast (blue) with eGFP-gephyrin S268E construct in cultured hippocampal neurons (11 ϩ 7 DIV) increases postsynaptic gephyrin clustering to the same level as seen in cells expressing S268A (see Fig. 1). Representative panels of transfected dendrites stained with synapsin 1 (red) to confirm that postsynaptic localization of eGFP-gephyrin clusters are shown. B and C, histograms of cluster density distribution (normalized per 20-m dendritic length) and cluster size (means Ϯ S.E.; ***, p Ͻ 0.001), confirming the increased density and size. D and DЈ, direct effect of calpain on S268E mutant (11 ϩ 7 DIV, double-stained for VGAT) was confirmed by blocking calpain with MDL28170 (30 M) overnight. E and F, quantification of cluster density showed no significant change, whereas the cluster area was significantly increased (***, p Ͻ 0.001). G and GЈ, MDL28170 treatment in neurons transfected with WT eGFP-gephyrin and its effect on clustering. H and I, quantification of WT eGFP-gephyrin clusters after MDL28170 treatment showed no change in cluster density but a significant increase in cluster overall size (means Ϯ S.E.; **, p Ͻ 0.01; see main text for statistical analyses). Scale bar, 5 m.
GABAergic transmission. The observation that the S268E/ S270E gephyrin double mutant exhibits a clustering phenotype similar to WT instead of a complete loss of clusters is a further indication that Ser-268 and Ser-270 likely are downstream effector sites rather than upstream modulators.
A recent report describing a shRNA screen to knock down signaling pathways that affect gephyrin clustering identified ERK, GSK3␤, and mTOR as effectors of gephyrin scaffolds (29). In this report, shRNA-mediated knockdown of ERK resulted in reduced density of gephyrin clusters. Our data also show a reduced gephyrin cluster density when ERK action of gephyrin is blocked either with S268A mutation (Fig. 1F) or using PD98059 (Fig. 6E). In addition, our functional data show that ERK-mediated reduction in gephyrin clusters is coupled to prolongation of GABAergic mIPSC interevent intervals, apparently compensated for by increased mIPSC amplitude (Fig. 6, G-J). Our molecular dissection of this phenomenon shows that phosphorylation at Ser-268 and Ser-270 is controlled by ERK and GSK3␤ signaling pathways regulating distinct aspects of gephyrin clustering. In particular, we show that treating neurons expressing eGFP-S268E with PD98059 does not alter gephyrin cluster size but increases their density, and this increase in density is facilitated by GSK3␤ regulation of the Ser-270 site. It is possible that blocking ERK signaling likely affects GSK3␤ phosphorylation of Ser-270, thereby allowing the formation of small, supernumerary postsynaptic gephyrin clusters. Furthermore, the specificity of the effects seen by mutagenesis and pharmacological treatment argues against the possibility that alterations in gephyrin S268A and S268E clustering are due to changes in its primary sequence.
By investigating the role of calpain in modulation of gephyrin clustering, our study provides new insight into the underlying molecular mechanism. The present observation that co-expression of Cast with eGFP-S268E mutant increases postsynaptic gephyrin cluster size and density to the levels seen in neurons expressing gephyrin S268A unambiguously demonstrates a role for calpain in restricting gephyrin clustering. In contrast, we have shown previously that Cast co-expression with eGFP-S270E has no effect on cluster size of this gephyrin mutant (2), suggesting that regulation of cluster size by calpain is dependent specifically on phosphorylation of Ser-268. Based on this evidence, one can speculate that the Ser-270 residue dominates over the Ser-268 site.
The main insight into the role of calpain for regulating gephyrin clustering comes from our experiments demonstrating activity-dependent regulation of ERK and calpain to reduce the size of gephyrin clusters upon chronic neuronal depolarization by KCl (Fig. 7). The fact that Cast co-expression prevents the cluster size reduction induced by KCl exposure directly demonstrates the involvement of calpain in this process, most likely in concert with ERK-mediated Ser-268 phosphorylation.
A recent report showed that Ser-270 phosphorylation affects the epitope recognized by the monoclonal antibody mAb7a (30), reflecting a change in gephyrin conformation (28). This structural change might explain why the presence of endogenous gephyrin affects the phenotype of Ser-268 phospho-mutants ( Fig. 1) but not that of Ser-270 phospho-mutants (2). Furthermore, this conformational change could be the underlying basis for the Ser-270 site dominating over the Ser-268 residue to determine calpain action on gephyrin. In light of the complex interplay between the phosphorylation status of Ser-268 and Ser-270 on gephyrin cluster size and density, we have summarized our main results schematically (Fig. 9).
In neurons, ERK controls several forms of long term plasticity, including NMDA-dependent and independent forms of LTP, and sustained ERK activity is required for structural remodeling of dendritic spines underlying learning and memory (31). Although it is easy to envision how alterations at glutamatergic synapses have to be paralleled by concomitant changes at GABAergic synapses to prevent hyperexcitability or silencing of the network, our present results suggest a more complex scenario, by which ERK activity constitutively limits the strength of GABAergic synapses, as seen by the marked  Step 2, upon inactivation of ERK and GSK3␤ signaling (or activation of unidentified phosphatases), the Ser-268 and Ser-270 residues allow fresh gephyrin molecule recruitment into the new clusters, as well as formation of new clusters. This dephosphorylated gephyrin is also insensitive to calpain regulation.
Step 3, phosphorylation of either Ser-268 or Ser-270 residue renders gephyrin sensitive to calpain regulation, leading to reduced cluster size or reduced density. This graded response to calpain sensitivity is facilitated by changes in gephyrin conformation, depending on the Ser-270 phosphorylation status.
increase in mIPSC amplitude in hippocampal neurons upon 3 h of PD98059 exposure. This effect of ERK is likely permissive for potentiation of glutamatergic synapses, whereas homeostatic regulation of GABAergic transmission by gephyrin likely involves protein phosphatases or deacetylases to counteract the action of ERK. Hence, further characterizations of such additional signal transduction pathways converging on gephyrin are likely to shed light on the mechanistic adaptability of GABAergic inhibition to fluctuations within an existing neuronal network.