Deficits in Phosphorylation of GABAA Receptors by Intimately Associated Protein Kinase C Activity Underlie Compromised Synaptic Inhibition during Status Epilepticus

Status epilepticus (SE) is a progressive and often lethal human disorder characterized by continuous or rapidly repeating seizures. Of major significance in the pathology of SE are deficits in the functional expression of GABAA receptors (GABAARs), the major sites of fast synaptic inhibition in the brain. We demonstrate that SE selectively decreases the phosphorylation of GABAARs on serine residues 408/9 (S408/9) in the β3 subunit by intimately associated protein kinase C isoforms. Dephosphorylation of S408/9 unmasks a basic patch-binding motif for the clathrin adaptor AP2, enhancing the endocytosis of selected GABAAR subtypes from the plasma membrane during SE. In agreement with this, enhancing S408/9 phosphorylation or selectively blocking the binding of the β3 subunit to AP2 increased GABAAR cell surface expression levels and restored the efficacy of synaptic inhibition in SE. Thus, enhancing phosphorylation of GABAARs or selectively blocking their interaction with AP2 may provide novel therapeutic strategies to ameliorate SE.


Introduction
Status epilepticus (SE) is a frequently occurring medical emergency associated with high mortality and morbidity. It is defined by prolonged (Ͼ30 min) unremitting or rapidly repeating seizures (Lowenstein and Alldredge, 1998;Alldredge and Lowenstein, 1999). The fundamental pathophysiology of SE involves a failure of mechanisms that normally abort an isolated seizure, resulting in neuronal hyperexcitability and compromised inhibitory GABAergic synaptic transmission (Treiman, 1990;Lowenstein and Alldredge, 1998;Staley, 2004). Mechanisms underlying these deficits are at least in part postsynaptic and include reduced potency and efficacy of GABA together with altered transmembrane Cl Ϫ gradients (Staley, 2004); accordingly, the frontline treatments for SE are benzodiazepines (Bzs) and barbiturates, which potentiate the activity of GABA A receptors (GABA A Rs), the principal sites of fast synaptic inhibition in the brain.
Although GABA A R receptor allosteric modulators are effective in blocking SE shortly after its induction, they are ineffective at later time periods (Ͼ1 h), and this development of pharmacoresistance is associated with significantly higher rates of mortality. Studies performed in culture have established that GABA A Rs undergo constitutive endocytosis and recycling, a process that can directly regulate the efficacy of synaptic inhibition Blanchet et al., 2003;van Rijnsoever et al., 2005). To date, the relevance of these putative regulatory processes in the pathology of SE remains poorly understood (Goodkin et al., 2005;Naylor et al., 2005).
Here we demonstrate that SE selectively reduces the cell surface stability of distinct GABA A R subtypes and the protein kinase C (PKC)-dependent phosphorylation of 408/9 (S408/9) within the receptor ␤3 subunit. These residues lie within a basic patchbinding motif for the clathrin adaptor protein AP2, a critical regulator of endocytosis. Coincident with decreased phosphory-lation enhanced association of GABA A Rs with AP2 was apparent in SE. Increasing the level of ␤3 subunit phosphorylation via the pharmacological activation of PKC or specifically blocking the binding of GABA A Rs to AP2 enhanced GABA A receptor cell surface expression levels and restored the efficacy of synaptic inhibition during SE. Thus, inhibiting endocytosis or enhancing the phosphorylation of GABA A Rs may provide novel therapeutic strategies to ameliorate SE.

Materials and Methods
Induction of status epilepticus and tissue preparation. All procedures were performed in strict accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the University of Pennsylvania Institutional Animal Care and by The Children's Hospital of Philadelphia Animal Care and Use Committee. Six-to eight-week-old male FVB/N mice (Jackson Laboratories, Bar Harbor, ME) were injected first with scopolamine methyl nitrate (1 mg/kg, i.p., Sigma, St. Louis, MO) to minimize the peripheral cholinergic effects of pilocarpine, and were injected 30 min later with pilocarpine hydrochloride (350 mg/kg, i.p., Sigma) or saline. Vehicle-treated animals were treated with scopolamine alone. After 1 h from the onset of stage 5 seizures, animals were deeply anesthetized with halothane and decapitated at the second cervical vertebrae. The hippocampus was then rapidly removed and rinsed with ice-cold artificial CSF (ACSF) containing 124 mM NaCl, 3 mM KCl, 2 mM CaCl 2 , 25 mM NaHCO 3 , 1.1 mM NaH 2 PO 4 , 2 mM MgSO 4 , and 10 mM D-glucose, equilibrated with 95% O 2 and 5% CO 2 to yield a pH of 7.4. The hippocampus was then cut into 350 m slices using a tissue chopper for slice biotinylation or 400 m slices using a vibratome (VT1000S; Leica, Deerfield, IL) for electrophysiology. Hippocampal slices were incubated in ASCF at 30°C for 1 h recovery before experimentation.
Surface biotinylation. Kinase activators/inhibitors were applied for up to 15 min at 32°C, whereas myristoylated P4 and scrambled control peptides (Tocris, Bristol, UK) were used at a concentration of 90 M in ASCF for 1 h. After treatment, slices were chilled to 4°C and incubated for 30 min with 1 mg/ml NHS-SS-biotin (Pierce, Rockford, IL). Excess biotin was removed by washing three times in cold ACSF and lysed as described previously (Terunuma et al., 2004). After correction for protein content, lysates were incubated with Streptavidin beads for 12 h at 4°C. Bound material was subjected to SDS-PAGE and then immunoblotted with antibodies and visualized by ECL (Pierce). Blots were then quantified using the CCD-based FujiFilm (Tokyo, Japan) LAS 3000 system.
GST protein "pull-down" assay. GST-␤3, which encodes residues 303-425 of the GABA A R ␤3 subunit major intracellular domain, was purified by affinity chromatography from Escherichia coli extracts and exposed to detergent-soluble extracts prepared from hippocampus as described previously (Brandon et al., 1999;Terunuma et al., 2004). Material was then immunoblotted or subjected to in vitro kinase assays in the presence of 10 Ci 32 P-␥ATP (final ATP concentration 200 M). Kinase assays were performed to 30 min in the presence and absence of 1 M KT5720 (PKA inhibitor), 15 M GF 109203X (PKC inhibitor), or 10 M KN-62 (CamKII inhibitor). Phosphorylation was then assessed by SDS-PAGE and quantified using a phospho-imager.
Immunoprecipitation. Hippocampal slices were lysed in 50 mM Tris-HCl, pH 7.5, 0.5 mM EDTA, 2 mM EGTA, and 1% Triton X-100. Soluble material was then immunoprecipitated with either anti-␤3 antibody or control IgG coupled to protein A Sepharose (Brandon et al., 1999(Brandon et al., , 2000(Brandon et al., , 2002b(Brandon et al., , 2003. Bound material was then either subjected to immunoblotting or to an in vitro kinase assay as outlined above.
Protein phosphatase activity assays. Protein phosphatase activity in hippocampal extracts was assayed by release of 32 P-phosphate from phosphorylated GST-␤3. GST-␤3 immobilized on glutathione agarose beads was phosphorylated with 2 Ci [␥-32 P]-ATP and 100 ng of PKC, followed by washing with dephosphorylation buffer containing 50 mM Tris-HCl, pH 7.0, 10 mM ␤-mercaptoethanol, 1 mM EDTA, 0.1 mM phenylsulfonyl fluoride, 2 mM benzamidine, 1 g/ml leupeptin, and 1 g/ml aprotinin. Hippocampal slices, homogenized in dephosphorylation buffer, were mixed with 32 P-labeled GST-␤3 and incubated at 30°C for 10 min followed by the addition of 200 l of trichloroacetic acid. The mixture was kept for 10 min at room temperature and centrifuged at 13,000 rpm for 5 min, and the pellet was counted for radioactivity.
Measuring basal PKA and PKC activity. PKC activity in total lysates was measured using the SignaTECT (Promega, Madison, WI) kit in the presence and absence of 10 M PKCI18 -36 inhibitor peptide (Brandon et al., 1999). PKA activity was measured using Kemptide as a substrate (Terunuma et al., 2004).

SE selectively decreases GABA A R cell surface stability
To induce SE, FVB/N mice were injected with scopolamine/pilocarpine, and animals with pronounced behavioral seizures were killed 1 h later. Hippocampal slices were prepared from these animals and control animals injected with scopolamine alone and allowed to recover at 30°C for 1 h before experimentation. Slices were labeled with NHS-biotin and lysed, and biotinylated proteins were purified on avidin. The levels of GABA A R ␣, ␤, ␥, and ␦ subunit isoforms were then measured in the resulting cell surface and total fractions using immunoblotting. SE significantly decreased the cell surface expression levels of the GABA A R ␣1 (52 kDa), ␣2 (53 kDa), and ␣4 (67 kDa) subunits to 68.7 Ϯ 4.3% (␣1), 81.7 Ϯ 4.1% (␣2), and 77.5 Ϯ 4.7% (␣4) compared with control, but their total cellular levels were unaltered ( Fig. 1 A, B). Likewise reductions in the cell surface expression levels of the ␤3 (73.3 Ϯ 7.7%) and ␥2 (74.3 Ϯ 7.8%) subunits were also apparent in SE (Fig. 1 A, B). In contrast SE significantly increased the cell surface expression cellular levels of the GABA A R receptor ␣5 (62 kDa) and ␦ (48 kDa) subunits (180 Ϯ 10.65 and 152.3 Ϯ 16.0% of control, respectively). However, the cell surface expression levels of the GABA B R R1 and R2 subunits remained unaltered in SE ( Fig. 1 A, B). To control for the integrity of our cell surface fractions, they were blotted with antibodies against 14-3-3 and actin, both of which are exclusively cytoplasmic proteins. Although 14-3-3 and actin could be detected in total extracts, they were absent from biotinylated cell surface fractions ( Fig. 1 A).
Therefore, SE can differentially modulate the cell surface stability of individual receptor subunits, which suggests that the cell surface expression levels of receptor subtypes containing ␣1-4, ␤3, and ␥2 subunits will be decreased, whereas those containing ␣5 and ␦ subunits will be increased in SE.

SE decreases phosphorylation of the ␤3 subunit
Phosphorylation of GABA A Rs is emerging as a critical process in regulating their accumulation on the neuronal plasma membrane. Central to this modulation are phosphorylation sites for a range of protein kinases that are conserved in the intracellular domains of ␤1-3 subunits; serines 408/9 (S408/9) in the case of ␤3 (Brandon et al., 2002a;Hinkle and Macdonald, 2003;Wang et al., 2003;Jovanovic et al., 2004). These serine residues lie within a "basic patch" binding motif for AP2, a critical regulator of protein entry into the endocytic pathway (Haucke et al., 2000(Haucke et al., , 2006. Phosphorylation of S408/9 decreases the affinity of AP2 for the receptor ␤3 subunit (Kittler et al., 2005). Therefore, we analyzed whether SE induces altered GABA A Rs phosphorylation by immunoblotting both cell surface and total fractions with a phospho-specific antibody against S408/9 and a phospho-independent epitope in the ␤3 subunit (Brandon et al., 2002b;Jovanovic et al., 2004). SE induced a significant decrease in the level of S408/9 phosphorylation for ␤3 subunits on the plasma membrane and in total extracts to 64.2 Ϯ 4.5% and 54.8 Ϯ 5.6% of control, respectively (Fig. 1C).
Pilocarpine is a muscarinic acetylcholine receptor (mAChR) agonist, and activation of these receptors has been established to modulate S408/9 phosphorylation in cultured cortical neurons (Brandon et al., 2002b). Therefore, we examined the effects of low doses of pilocarpine that do not induce SE on ␤3 subunit phosphorylation. Injection of animals with 35 mg/kg pilocarpine, which is sufficient to activate mAChRs (Nathanson, 1987;Wess, 2004), did not significantly modify 408/9 phosphorylation or GABA A R cell surface stability (Fig. 1D). Under the same conditions, 350 mg/kg pilocarpine decreased both 408/9 phosphorylation and GABA A R cell surface expression levels (Fig. 1D). Therefore, modified GABA A R phosphorylation in SE induced by pilocarpine does not simply result from the activation of mAChRs by this agent.
Thus, in addition to decreasing GABA A R cell surface stability, SE clearly also reduces the phosphorylation of S408/9 in the ␤3 subunit, key residues for regulating the endocytosis of these proteins (Kittler et al., 2005;Chen et al., 2006).

SE decreases PKC activity and its association with the receptor ␤3 subunit
Basal phosphorylation of S408/9 in neurons is largely dependent on PKC activity; however, it is also dependent on substrates of cAMP-dependent protein kinase (PKA) (Brandon et al., 1999(Brandon et al., , 2000(Brandon et al., , 2002b(Brandon et al., , 2003. We thus compared the activity of these kinases in hippocampal extracts prepared from control and SE. As measured using the neurogranin peptide in the presence and absence of GF 10923, a specific PKC inhibitor, PKC activity was significantly reduced to 70.8 Ϯ 2.5% of control in SE (Fig. 2 A). However, the activity of PKA as measured by the phosphorylation of Kemptide was reduced by only 5% (Fig. 2 B).
Previous studies have revealed that the intimate association of PKC isoforms with the intracellular domains of GABA A R ␤1-3 subunits is important in regulating GABA A R phosphorylation, and for the ␤3 subunit, amino acids 404 -415 are critical in mediating PKC binding (Brandon et al., 2002a,b). Therefore, we examined whether the deficit in PKC activity evident in SE alters the targeting of this kinase to GABA A Rs. To initiate these studies, detergent-soluble hippocampal lysates were exposed to a glutathione S-transferase fusion protein encoding the intracellular domain of the ␤3 subunit (GST-␤3) and bound material subjected to in vitro kinase assays (Brandon et al., 2002b). GST-␤3 was phosphorylated under these conditions, but the level of phosphorylation by SE extracts was significantly Hippocampal slices from control (Con) or SE were labeled with NHS-biotin and lysed, and biotinylated proteins were purified on avidin. A, Cell surface and total fractions (1/10 of input; In) were immunoblotted with antibodies against GABA A receptor ␣1-2, ␣4 -5, ␤3, ␥2, and ␦ subunits, GABA B R1 and R2 subunits, 14-3-3, and actin. B, Data in SE (open bars) were normalized to the cell surface levels evident in control slices (filled bars; 100%). C, SE decreases the phosphorylation of S408/9 in the GABA A R ␤3 subunit. Top, Hippocampal slices from control and SE were subjected to biotinylation, and cell surface fractions were blotted with a phospho-specific antibody against S408/9 in the ␤3 subunit (p-␤3) and anti-␤3 antibodies (T-␤3). Bottom, The ratio p-␤3/T-␤3 signals were then measured in control (filled bars) and SE (open bars) with values normalized to those in control (100%). In all panels, asterisks indicate significant difference from control ( p Ͻ 0.01; Student's t test; n ϭ 6 -8). D, Hippocampal slices were prepared from Con, and animals were injected with 35 (p35) and 350 mg/kg (p350) pilocarpine, respectively. Slices were biotinylated, and cell surface and total fractions were isolated and immunoblotted with the respective antibodies as indicated.
lower than that seen in control (Fig. 2C) (67.5 Ϯ 5.0% at 30 min). Phosphorylation by either extract was inhibited by GF 10923, but not by inhibitors of PKA or Ca/calmodulin type 2-dependent protein kinases (Fig. 2D). Moreover, consistent with published studies (Brandon et al., 2002b) phosphorylation of GST-␤3 by both control and SE occurred principally on S408/9 (Fig. 2E), whereas this was not the case for GST alone (data not shown).
To further analyze PKC association with GABA A Rs, we used an antibody specific to phospho-S657, a residue conserved in all classical PKC isoforms (␣, ␤, and ␥), the phosphorylation of which reflects the catalytic activity of these enzymes (Parker, 2003). The level of phospho-S657 binding to GST-␤3 from SE extracts was significantly reduced to 61.2 Ϯ 1.8% of control (Fig. 2F). In agreement with this, the levels of the ␣, ␤, and ␥ isoforms of PKC bound to GST-␤3 were also significantly reduced from those seen in SE ( Fig. 2F) (33.1 Ϯ 1%, 67.1 Ϯ 2%, and 44.9 Ϯ 1.3% of control, respectively). As a control for these experiments, we assessed binding of the receptor for activated-C kinase (RACK-1) to GST-␤3. RACK-1 binds to GST-␤3 but at a site distinct from that for PKC; they bind at residues 395-404, respectively (Brandon et al., 2002b). In contrast to our experiments with PKC, the level of RACK-1 binding to GST-␤3 from SE extracts was unaltered relative to control (Fig.  2F) (101.6 Ϯ 3% of control).
Immunoprecipitation was further used to examine the association of PKC with whole GABA A Rs. To do so, detergentsoluble hippocampal extracts were immunoprecipitated with antibodies against the ␤3 subunit, and precipitated material was subjected to kinase assays. Phosphorylation of a major band of 57 kDa corresponding to the receptor ␤3 subunit (Brandon et al., 1999) was seen with material immunoprecipitating with anti-␤3 antibody, but not control IgG. However, phosphorylation of this band was significantly reduced to 15 Ϯ 5% of control in SE (Fig. 3A). Significantly phosphorylation by both extracts was abolished in GF 10923 (Fig. 3B), and the level of the ␣, ␤, and ␥ PKC isoforms coimmunoprecipitating from SE extracts was reduced to 67.5 Ϯ 8.9% of control (Fig. 3C). Together, our results demonstrate that the level and activity of PKC isoforms associated with the GABA A Rs in the hippocampus are reduced in SE.

SE increases the association of protein phosphatase 2A with the ␤3 subunit
In addition to analyzing ␤3 subunit phosphorylation, we compared its dephosphorylation in SE, which is mediated largely by PP1 and PP2A Terunuma et al., 2004). For these experiments, the ability of material bound to GST-␤3 to dephosphorylate exogenous PNPP was measured (Terunuma et al., 2004). A highly significant increase ( p Ͻ 0.01) in dephosphorylation of PNPP by SE to 297.9 Ϯ 8.9% of control was evident (Fig. 4A). This effect was completely blocked by 20 nM okadaic acid, a concentration that specifically inhibits PP2A. Consistent with this, increased binding of the PP2A catalytic subunit to GST-␤3 was also observed (Fig. 4B). However, the level of total phosphatase activity in SE and control remained unaltered (Fig.  4C,D).

Blocking endocytosis or activating PKC enhances GABA A R cell surface stability in SE.
Our biochemical studies demonstrate that S408/9 phosphorylation is reduced during SE together with the level of PKC associated with GABA A Rs. The pool of PP2A associated with GABA A Rs is increased in SE, which may further promote de- Data were normalized to level of activity seen with control extracts (100%). B, Similar experiments were also performed using 50 M Kemptide, a PKA substrate, in the absence (filled bars) and presence (open bars) of 20 nM Walsh peptide, a specific PKA inhibitor, and data were normalized to the level of activity evident in control (100%). C, Phosphorylation of GST-␤3 by hippocampal extracts. Top, Fusion proteins were incubated with hippocampal extracts, and bound material was incubated with 10 Ci of 32 P-␥ATP at 30°C followed by SDS-PAGE. Bottom, The level of phosphorylation was then calculated, and data were normalized to levels seen with control extracts (100%). D, PKC activity mediates the phosphorylation of GST-␤3 by hippocampal extracts. Material bound to GST-␤3 was incubated for 30 min with 10 Ci 32 P ␥ATP at 30°C under control conditions (1) or in the presence of 1 M KT5720 (2; PKA inhibitor), 15 M GF 109203 (3; PKC inhibitor), or 10 M KN-62 (4; Ca/calmodulin type 2-dependent protein kinase), and phosphorylation was examined by SDS-PAGE. E, Hippocampal extracts phosphorylated GST-␤3 on S408/9. GST-␤3 or a mutant in which S408/9 had been converted to alanine residues were incubated with hippocampal extracts, bound material was incubated for 30 min with 10 Ci 32 P-␥ATP at 30°C, and phosphorylation was measured via SDS-PAGE. Top, Autoradiogram. Bottom, Coomassie stain of the same gel. Left, Control; right, SE. F, PKC binding to GST-␤3 is reduced from SE extracts. Top, Material bound to GST-␤3 was immunoblotted with antibodies against p-S657, antibodies specific to individual PKC isoforms, and against RACK-1. Bottom, The level of binding for each protein was then determined for control (filled bars) and SE (open bars) extracts, with data being normalized to the levels seen from control (100%). In all panels, asterisks indicate significant difference from control ( p Ͻ 0.01; Student's t test; n ϭ 7-10). phosphorylation of S408/9. Residues 401-412 of the ␤3 subunit bind directly to the 2 subunit of AP2, a critical regulator of endocytosis. This interaction is decreased by phosphorylation of S408/9 within this basic patch 2-binding motif (Brandon et al., 2002a;Kittler et al., 2005;Haucke, 2006;Chen et al., 2006). Therefore, we assessed whether deficits in S408/9 phosphorylation during SE led to enhanced association of GABA A Rs with AP2. Hippocampal extracts were immunoprecipitated with anti-␤3 and then immunoblotted with antibodies against the receptor ␤3 subunit or the ␣ subunit of AP2. Both proteins immunoprecipitated with anti-␤3 but not control IgG. The ratios of AP2/␤3 signals were then determined after controlling for the respective inputs. The ratio showed a significant increase of 217.2 Ϯ 10.9% in SE compared with control (Fig. 5A). To examine the significance of this enhanced association with AP2, the effects of blocking clathrindependent endocytosis on GABA A R cell surface expression levels in SE were examined. Hippocampal slices were incubated at 30°C for 1 h and then for up to an additional 60 min in the presence of myristoylated P4 (P4) peptide, a potent inhibitor of GABA A R endocytosis Bogdanov et al., 2006). The cell surface expression levels of GABA A Rs incorporating ␤3 were then measured via immunoblotting after biotinylation. P4 peptide produced a significant increase in the cell surface expression levels of receptors incorporating ␤3 subunits to 142 Ϯ 5.4% of control ( p Ͻ 0.01) after 30 min, an effect that was sustained for up to 60 min. This effect was not replicated with the control scrambled (Sc) peptide (Fig. 5B). The effects of pharmacologically activating PKC on the phosphorylation and cell surface stability of receptors in SE were examined. Treatment of SE slices with PDBU increased the phosphorylation of S408/9 for both cell surface and total populations of GABA A Rs over 15 min (Fig. 5C) (141 Ϯ 4.2% and 175 Ϯ 5.2% of control at 10 min, respectively). Coincident with this, activation of PKC produced a sustained increase in the cell surface levels of GABA A R receptors over the same time course (145.9 Ϯ 3.5% of control at 10 min). In contrast, under the same conditions the cell surface expression levels of GABA A Rs were not altered by exposure to PDBU.
These results demonstrate that blocking endocytosis or increasing the phosphorylation of S408/9 in the ␤3 subunit enhances GABA A R cell surface stability during SE.

Phospho-dependent inhibition of GABA A R AP2 binding modifies mIPSCs in SE
To analyze the functional significance of our biochemical studies, we compared the properties of mIPSCs in CA1 neurons from  SE. A, B, Phosphorylation of the ␤3 by intimately associated PKC activity. A, Top, Detergent-solubilized hippocampal extracts were immunoprecipitated with IgG or anti-␤3 antibody, and precipitated material was incubated with 10 Ci 32 P-␥ATP at 30°C for 15 min followed by SDS-PAGE. A, Bottom, The level of ␤3 subunit phosphorylation was then determined for control and SE extracts, and the level of ␤3 subunit phosphorylation was normalized to that seen with control extracts (100%). B, Phosphorylation was also assessed in the absence (Ϫ) or presence (ϩ) of 15 M GF 109203. C, The levels of the ␣, ␤, and ␥ PKC isoforms associated with GABA A R are reduced in SE. Material immunoprecipitating with IgG or anti-␤3 from control and SE extracts was immunoblotted with an antibody specific to the PKC ␣, ␤, and ␥ isoforms. In, 10% of the input used for each immunoprecipitation. GST-␤3 was exposed to lysates from control (filled bars) and SE (open bars) hippocampus, and the ability of bound material bound to dephosphorylated 14.2 M PNPP in the absence (Ϫ) and presence (ϩ) of 20 nM okadaic acid (OA) was measured. Data were then normalized to the level of dephosphorylation seen with control extracts (100%). B, Increased binding of PP2A to GST-␤3 from SE extracts. Top, Material bound to GST-␤3 or 10% of the input used for each immunoprecipitation (In) was immunoblotted with an antibody against the PP2A catalytic subunit. Bottom, The level of PP2A bound from control (filled bar) and SE (open bar) extracts was then determined, and values were normalized to those evident in control (100%). C, D, Dephosphorylation of 32 P-GST-␤3 by hippocampal extracts. Hippocampal lysates were prepared from control (filled bars and circles) and SE (open bars and circles). C, The ability of 200 -50 g of total detergent-soluble protein from these extracts to dephosphorylate 32 P-GST-␤3 was measured over a 10 min time course. D, The ability of 50 g of total protein from control (filled circles) and SE (open circles) to dephosphorylate 32 P-GST-␤3 over a time course of 40 min was also examined. In all panels, asterisks indicate significant difference from control ( p Ͻ 0.01; Student's t test; n ϭ 3-4).
control and SE slices (Kapur and Coulter, 1995). mIPSCs were isolated using bath application of tetrodotoxin and glutamate receptor antagonists. Under these conditions, remaining events were blocked by the GABA A R antagonist bicuculline (data not shown), confirming their identity as mIPSCs. We first compared the properties of mIPSCs in CA1 pyramidal neurons from control slices and those from animals undergoing SE using standard patch pipette electrolyte. mIPSC peak amplitude was significantly reduced from 48.5 Ϯ 3.8 pA in control compared with 39.8 Ϯ 1.9 pA in SE (Fig. 6 A, Table 1). In contrast, mIPSC frequency and rise and decay times were very similar in controls and SE (Fig. 6 A, Table 1).
To examine the role that altered endocytosis plays in this phenomenon, we used intracellular dialysis with the patch pipette to introduce dominant peptides that have been documented to block the interaction of the AP2 adaptin with the GABA A R ␤3 subunit. For these experiments, we used Pep␤3 (KTHLR-RRSSQLK) or its phosphorylated equivalent Pep␤3P (KTHLR-RRSPSPQLK), which correspond to the basic patch AP2-binding motif residues 401-412 in the ␤3 subunit (Kittler et al., 2005;Haucke, 2006). Pep␤3 has an affinity of 300 nM for AP2, whereas that for Pep␤3P is reduced to 1300 nM (Kittler et al., 2005). In agreement with these differing affinities, Pep␤3P but not Pep␤3P blocks GABA A R endocytosis in cultured neurons (Kittler et al., 2005;Chen et al., 2006). Dialysis of neurons with Pep␤3 significantly increased mIPSC amplitude from 36.2 Ϯ 3.1 at 0 to 53.9 Ϯ 7.5 pA at 20 min in CA1 SE neurons, but their frequency was not altered (Fig. 6 B, Table 1). Importantly, this effect was not replicated with Pep␤3P (Fig. 6C, Table 1).
To further evaluate this observation, we compared mIPSC amplitude at 2 min intervals over a 20 min recording period for cells dialyzed with these distinct agents. Pep␤3 produced a timedependent increase in mIPSC amplitude that reached 139 Ϯ 0.5% of that evident at 0 time after 20 min (Fig. 6 D). In common with standard electrolyte Pep␤3P did not alter mIPSC amplitude over the same time course (Fig. 6 D). The effects of Pep␤3 and Pep␤3P on mIPSC decay and rise times were measured. Data were fitted by single exponentials to determine decay at 0 time and after 20 min (Fig. 6 E, Table 1). For Pep␤3, decay was significantly slowed from 15.9 Ϯ 0.9 ms at 0 time to 21.7 Ϯ 1.3 ms at 20 min, an effect not seen with Pep␤3P (16.3 Ϯ 1.4 ms vs 18.2 Ϯ 1.7 ms, respectively). In contrast, mIPSC rise time was not modulated by either peptide (Table 1). The effect of Pep␤3 on mIPSCs from CA1 neurons in control slices was also assessed. Intriguingly, and in contrast to our results using SE, Pep␤3 was without effect on mIPSC parameters in these neurons (supplemental Fig. 1, available at www.jneurosci.org as supplemental material). Together, our results suggest that SE reduces mIPSC amplitude in hippocampal CA1 neurons compared with control. It is also evident that blocking the interaction of GABA A R with the AP2 adaptin in SE can restore mIPSC amplitude to levels seen in control neurons in addition to slowing their decay.

Activation of PKC modifies modulates mIPSC in SE
It is apparent that SE leads to reduced phosphorylation of S408/9 in the GABA A R ␤3 subunit (Fig. 1). We also illustrated that the activation of PKC in SE increases both S408/9 phosphorylation and the cell surface stability of GABA A Rs (Fig. 5). Therefore, we assessed the effects of PKC activation on mIPSC properties in SE. As illustrated in Figure 7A, 100 nM PDBU introduced into CA1 neurons via the patch pipette increased mIPSC amplitude from 35.5 Ϯ 3.2 pA at 0 min to 51.5 Ϯ 4.6 pA after 20 min, an effect that was blocked by coapplication of 10 M calphostin C, a PKC inhibitor (Cal) (39.2 Ϯ 3.5 vs 40.6 Ϯ 3.9 pA at 0 and 20 min, respectively) (Fig. 7 A, B, Table 1). Cal treatment alone was without effect on mIPSC amplitude (36.0 Ϯ 3.9 vs 37.1 Ϯ 4.1 pA at 0 and 20 min, respectively) (Fig. 7C, Table 1). In contrast to these results, none of these treatments modulated either mIPSC frequency or rise times ( Fig. 7A-C, Table 1).
To further evaluate the effects of PKC activation, we compared mIPSC amplitude at 2 min intervals over a 20 min recording period for neurons exposed to modulators of this kinase. This revealed that PDBU produced a time-dependent increase in mIPSC amplitude that reached 142 Ϯ 5.4% of control after 20 min, an effect not replicated using PDBU/Cal or Cal alone (Fig.  7D). Finally we examined the effects of PKC activation on mIPSC decay. As illustrated in Table 1 PDBU slowed mIPSC decay (15.3 Ϯ 0.9 vs 21.3 Ϯ 1.9 ms at 0 and 20 min, respectively), an effect that was blocked by coapplication of Cal (14.6 Ϯ 2.3 vs 16.2 Ϯ 2.5 ms at 0 and 20 min, respectively).
Together, these results demonstrate that activation of PKC enhances the amplitude and slows the decay of mIPSCs in SE. This is consistent with the ability of this kinase activity to enhance Figure 5. Blocking endocytosis and the activation of PKC increases GABA A R cell surface expression levels in SE. A, Increased association of GABA A Rs with AP2 in SE. Detergent-solubilized extracts were immunoprecipitated with IgG or anti-␤3 antibodies, and precipitated material was immunoblotted with antibodies against the GABA A R ␤3 subunit or the ␣ subunit of AP2. In represents 20% of the input used for each immunoprecipitation. After controlling for recovery of the ␤3 subunit, the level of AP2 binding was determined for control (filled bar) and SE (open bar), and data were normalized to levels evident in control (100%). B, Blocking endocytosis increases GABA A R cell surface stability in SE. Hippocampal slices from SE animals were incubated in the presence of 90 M P4 or control Sc peptide. Top, Slices were then subjected to biotinylation with NHS-biotin, and cell surface fractions were immunoblotted with anti-␤3 antibody. Bottom, The levels of cell surface ␤3 in slices exposed to either Sc (filled bars) or P4 (open bars) were then normalized to those evident at 0 time (100%). C, D, PKC activators increase GABA A receptor phosphorylation and cell surface expression levels in SE. Slices were incubated with 100 nM PDBU and labeled with NHS-biotin. Cell surface (C) and total fractions (D) were then immunoblotted with pS408/9 antibody. The level of phosphorylation of S408/9 was normalized to that evident at 0 time (100%). E, F, PKC activators increase GABA A receptor cell surface stability in SE. Cell surface fractions from SE slices were blotted with anti-␤3 (E) and antibodies against the GABA B R2 subunit (F ), and levels of each protein were normalized to those seen at 0 time (100%). In all panels, asterisks indicate significant difference from control ( p Ͻ 0.01; Student's t test; n ϭ 5-7).
the levels of S408/9 phosphorylation and the cell surface stability of GABA A Rs containing ␤3 subunits after this insult.

Discussion
SE arises from deficits in the efficacy of GABAergic inhibition that include altered GABA A R functional expression and pharmacology; however, the mechanisms underlying these deficits remain to be established (Lowenstein and Alldredge, 1998;Alldredge and Lowenstein, 1999). To address this critical question, we have examined the impact of SE on GABA A R functional expression together with the efficacy of synaptic inhibition in hippocampal slices derived from mice injected with pilocarpine. This animal model exhibits many similarities to patients undergoing SE including the development of pharmacoresistance (Kapur and Coulter, 1995;Coulter, 2001).
As measured using biotinylation, the cell surface stability of the GABA A R ␣1, 2, 4, ␤3, and ␥2 subunits was significantly reduced in SE compared with controls, whereas those encoding the GABA B R R1 and R2 subunits were unaltered. In contrast, the cell surface stability of the ␣5 and ␦ subunits was significantly increased. Thus, our results demonstrate that SE can induce alter-ations in the cell surface stability of individual GABA A R subunits. Although the precise subunit compositions of the GABA A R subtypes expressed on the cell surface of hippocampal neurons remains unknown, it is evident that heteromeric receptors that incorporate ␥2 subunits (␣␤␥2) are sensitive to functional modulation by benzodiazepines, whereas those incorporating ␦ subunits (␣␤␦) are not (Farrant and Nusser, 2005;Rudolph and Mohler, 2006). The increase in cell surface expression levels of receptor subtypes containing ␦ subunits at the expense of those containing ␥ in SE may thus be a potential mechanism to explain altered GABA A R functional expression after this insult. In keeping with our results, deficits in the number of ␥2-containing GABA A Rs clustered at inhibitory synapses have been reported in SE (Goodkin et al., 2005;Naylor et al., 2005).
In parallel with measurements on GABA A R cell surface stability, we examined the effects of SE on receptor phosphorylation, a key regulatory mechanism for modulating their functional properties and cell surface stability (Hinkle and Macdonald, 2003;Wang et al., 2003;Jovanovic et al., 2004). Compared with control, deficits in the phosphorylation of S408/9 were evident for cell Figure 6. Blockade of endocytosis increases the amplitude of mIPSCs and slows their decay in CA1 neurons undergoing SE. A, SE reduces mIPSC amplitude. Top, Representative sweeps. Middle, bottom, Cumulative probability analysis for mIPSC amplitude (middle) and frequency (bottom; interevent interval) recorded from CA1 neurons in control (black) and SE (red) slices. B, C, Pep␤3 selectively enhances mIPSC amplitude in SE neurons. Top, Representative sweeps at 0 and 20 min for CA1 neurons exposed to 100 g/ml Pep␤3 (KTHLRRRSSQLK; B)orPep␤3P (KTHLRRRSPSPQLK; C) via intracellular dialysis. Cumulative probability for mIPSC amplitude and frequency are also shown at 0 time (red) and 20 min (black), respectively. D, Time-dependent enhancement of mIPSCs by Pep␤3. mIPSC peak amplitude from CA1 neurons dialyzed with control (blue), Pep␤3 (black), and Pep␤3P (red) over a 20 min time course was normalized to that seen at 0 time (100%). E, Pep␤3 modulates mIPSC decay. Scaled mIPSCs are shown for CA1 neurons from SE slices dialyzed with Pep␤3 or Pep␤3P at 0 (black) and 20 min (red), respectively. In all panels, asterisks indicate significant difference from control ( p Ͻ 0.01; Student's t test; n ϭ 8 -12). The kinetic parameters of mIPSCs of CA1 neurons in the absence and presence of peptides, PKC inhibitors, and activators were calculated at both 0 time and 20 min. In all panels, asterisks indicate significant difference from control (Con; p Ͻ 0.01; n ϭ 8 -12).
surface populations of GABA A Rs containing ␤3 subunits in SE. Basal phosphorylation of S408/9 is largely dependent on the activity of PKC, whereas dephosphorylation is regulated principally by PP1 and PP2A (Brandon et al., 2000;Jovanovic et al., 2004;Terunuma et al., 2004). The specific association of these enzymes and their associated scaffolds with the GABA A R ␤3 subunit is also a critical determinant in regulating the stoichiometry of S408/9 (Brandon et al., 2000(Brandon et al., , 2002b(Brandon et al., , 2003Jovanovic et al., 2004;Terunuma et al., 2004). Consistent with the deficits in both global PKC activity and the phosphorylation of S408/9 in SE, the levels and activity of the ␣, ␤, and ␥2 isoform of PKC associated with GABA A R receptors were significantly reduced after this insult. In contrast to these results, the activity and level of PP2A associated with GABA A R were increased in SE. Therefore, the modified association of GABA A R with these signaling molecules provides a molecular mechanism to explain the decreased steady-state phosphorylation of S408/9 in SE. S408/9 lie within a basic patch-binding motif for AP2 between residues 401-412 of the ␤3 subunit, and phosphorylation of S408/9 decreases the affinity of AP2 for the ␤3 subunit (Kittler et al., 2005;Chen et al., 2006). Concordant with these observations, deficits in S408/9 phosphorylation in SE were coincident with increased association of GABA A Rs with AP2. Activating PKC in SE enhanced the cell surface expression levels of receptors containing ␤3 subunits together with the level of 408/9 phosphorylation. Similar enhancements of GABA A R cell surface number were seen when blocking endocytosis by inhibiting dynamin activity. These experiments thus suggest that the decreased cell surface expression of GABA A Rs during SE results from their reduced phosphorylation, which acts to promote their endocytosis by facilitating their interaction with AP2.
To test the relevance of these biochemical observations for inhibitory synaptic transmission, the properties of mIPSCs were compared in control and SE slices. Consistent with our biochemical studies, mIPSC amplitudes were reduced in CA1 neurons in SE compared with control, whereas their frequencies and rise and decay times were unaltered. The effects of selectively blocking the interaction of GABA A Rs with AP2 were then examined using dominant negative peptides corresponding to residues 404 -412 in the ␤3 subunit, Pep␤3 and its phosphorylated equivalent Pep␤3P, which have affinities of 300 and 1600 nM for AP2, respectively (Kittler et al., 2005;Chen et al., 2006). Pep␤3 increased mIPSC amplitude in SE to values similar to those in control CA1 neurons over a 20 min recording period. Although neither peptide modulated mIPSC frequency or rise time, Pep␤3 specifically slowed their decay.
Pep␤3 did not alter mIPSC properties in CA1 neurons from control slices. In contrast to this, Pep ␤3 has been shown to increase mIPSC amplitude in cultured hippocampal and cortical neurons in addition to medial spiny neurons from the nucleus accumbens (Kittler et al., , 2005Chen et al., 2006). This discrepancy may reflect lower rates of endocytosis or higher rates of basal ␤3 subunit phosphorylation in slices compared with culture. Finally, in addition to these experiments with dominant negative peptides, we also established that activation of PKC increased mIPSC amplitude during SE and slowed their decay but did not alter mIPSC frequency or rise time. In parallel with this functional modulation, we also established that PKC activation increased the phosphorylation of S408/9 together with the number of cell surface GABA A Rs containing ␤3 subunit. Together, our electrophysiological studies demonstrate that blocking the binding of GABA A R ␤3 subunits to the AP2 or increasing phosphorylation of S408/9 in this subunit can enhance the inhibitory synaptic transmission in SE.
In addition to a model of SE, termination of seizures induced by pilocarpine with benzodiazepines generates a spontaneous sei- Figure 7. PKC-dependent activity modulates mIPSCs in SE. A-C, PKC activity enhances mIPSC amplitude in SE. Representative sweeps from CA1 neurons exposed to PKC activators/inhibitors and cumulative amplitude distributions for events at 0 time (red) and 20 min (black) are shown. D, PDBU enhances mIPSC amplitude in SE. Peak amplitudes were measured over a 20 min recording period for SE CA1 neurons treated with PDBU (100 nM; blue), PDBU/calphostin (100 nM and 10 M, respectively; red), and calphostin alone (10 M; black), and data were normalized to those seen at 0 time (100%). zure phenotype, which is used as a model of temporal lobe epilepsy (TLE) (Coulter, 2001). It will therefore be of interest to examine whether altered levels of GABA A receptor phosphorylation persist in TLE, as has been reported for NMDA receptor phosphorylation (Niimura et al., 2005). Moreover, it will be of significance to examine the relationship between modified GABA A receptor phosphorylation and compromised transcriptional regulation of individual receptor subunit genes that are evident in TLE (Brooks-Kayal et al., 1998;Raol et al., 2006).
Collectively, our studies demonstrate that SE decreases PKCdependent phosphorylation of conserved phosphorylation sites within the GABA A R ␤3 subunit. This deficit in phosphorylation enhances GABA A R binding to the AP2 adaptin, promoting their endocytosis and reducing their cell surface expression levels, thus compromising the efficacy of synaptic inhibition. Therefore, the development of agents that selectively decrease GABA A R endocytosis or enhance ␤ subunit phosphorylation may be useful therapeutic agents to increase the strength of synaptic inhibition and thus alleviate SE and other forms of epilepsy.