Protein Kinase C (cid:1) Regulates (cid:2) -Aminobutyrate Type A Receptor Sensitivity to Ethanol and Benzodiazepines through Phosphorylation of (cid:2) 2 Subunits

Ethanol enhances (cid:2) -aminobutyrate (GABA) signaling in the brain, but its actions are inconsistent at GABA A receptors, espe-cially at low concentrations achieved during social drinking. We postulated that the (cid:1) isoform of protein kinase C (PKC (cid:1) ) regu-latestheethanolsensitivityofGABA A receptors,asmicelacking PKC (cid:1) show an increased behavioral response to ethanol. Here we developed an ATP analog-sensitive PKC (cid:1) mutant to selectively inhibit the catalytic activity of PKC (cid:1) . We used this mutant and PKC (cid:1) (cid:3) / (cid:3) mice to determine that PKC (cid:1) phosphorylates (cid:2) 2 subunits at serine 327 and that reduced phosphorylation of this site enhances the actions of ethanol and benzodiazepines at (cid:4) 1 (cid:5) 2 (cid:2) 2receptors,whichisthemostabundantGABA A receptor subtype in the brain. Our findings indicate that PKC

Ethanol enhances ␥-aminobutyrate (GABA) signaling in the brain, but its actions are inconsistent at GABA A receptors, especially at low concentrations achieved during social drinking. We postulated that the ⑀ isoform of protein kinase C (PKC⑀) regulates the ethanol sensitivity of GABA A receptors, as mice lacking PKC⑀ show an increased behavioral response to ethanol. Here we developed an ATP analog-sensitive PKC⑀ mutant to selectively inhibit the catalytic activity of PKC⑀. We used this mutant and PKC⑀ ؊/؊ mice to determine that PKC⑀ phosphorylates ␥2 subunits at serine 327 and that reduced phosphorylation of this site enhances the actions of ethanol and benzodiazepines at ␣1␤2␥2 receptors, which is the most abundant GABA A receptor subtype in the brain. Our findings indicate that PKC⑀ phosphorylation of ␥2 regulates the response of GABA A receptors to specific allosteric modulators, and, in particular, PKC⑀ inhibition renders these receptors sensitive to low intoxicating concentrations of ethanol.
␥-Aminobutyrate type A (GABA A ) 3 receptors mediate the majority of rapid inhibitory neurotransmission in the brain and are an important target for ethanol, the most widely abused drug (1). Ethanol modulation of GABA A receptors was first identified in synaptosomal preparations where intoxicating concentrations (10 -30 mM) enhanced receptor function as measured by 36 Cl uptake assays (2,3). However, after 30 years of investigation, it is apparent that ethanol enhancement of synaptic GABA A receptors is variable and in some preparations cannot be detected even at anesthetic concentrations (1,4).
GABA A receptors are pentameric protein complexes of subunits from eight classes (␣1-6, ␤1-3, ␥1-3, ␦, ⑀, , , and 1-3) (5). Most receptors are composed of two ␣ subunits and two ␤ subunits that co-assemble with one ␥2 subunit, which anchors these receptors at synapses where they mediate phasic inhibition (6). A minority contain a ␦ subunit instead of ␥2; these receptors are extrasynaptic and mediate tonic inhibition in neurons (7). Because earlier ethanol studies that focused on GABA currents carried by ␥2-containing receptors produced variable results, recent attention has turned to ethanol effects at receptors containing ␦ subunits. Reports from three laboratories found these receptors enhanced by low (Յ30 mM) intoxicating concentrations of ethanol (8 -10). However, two recent studies were unable to demonstrate low dose ethanol sensitivity of ␦-containing GABA A receptors (11,12), indicating that, like synaptic GABA A receptors, ethanol modulation of extrasynaptic receptors is also variable.
The reasons for this high degree of variability are unknown. One possibility is that intracellular signaling pathways may regulate ethanol sensitivity of GABA A receptors, and the activity of such pathways was not controlled for in these studies. This hypothesis is consistent with our findings in mice lacking protein kinase C⑀ (PKC⑀), which show an increased behavioral response to ethanol (13). Ethanol modulation of GABA A receptors is also increased in PKC⑀ Ϫ/Ϫ mice compared with wild-type mice (13,14). These results suggest that activation of PKC⑀ suppresses potentiation of GABA A receptors by ethanol.
The mechanism by which PKC⑀ regulates the ethanol sensitivity of GABA A receptors is unknown. Here, we examined this mechanism using receptors composed of ␣1, ␤2, and ␥2 subunits, which is the most abundant receptor subtype in the brain (15). Because there are no selective inhibitors of PKC⑀ catalytic activity available, we used a chemical-genetics approach to generate an ATP analog-sensitive mutant, as-PKC⑀, that could be potently and selectively inhibited by a cell-permeant, small molecule inhibitor. Our findings indicate that PKC⑀ regulates the sensitivity of ␣1␤2␥2 receptors to ethanol and benzodiazepines through phosphorylation of serine 327 in the large intracellular loop of ␥2.

EXPERIMENTAL PROCEDURES
Animal Care and Behavior-PKC⑀ Ϫ/Ϫ mice were generated by homologous recombination in J1 embryonic stem cells as described (16). Male and female PKC⑀ ϩ/Ϫ mice were maintained on inbred 129S4/SvJae and C57BL6/J backgrounds and were crossed to produce PKC⑀ ϩ/Ϫ C57BL6/J ϫ129S4/SvJae F1 hybrid mice. These mice were intercrossed to generate F2 hybrid PKC⑀ ϩ/ϩ and PKC⑀ Ϫ/Ϫ littermates for experiments. Mice were genotyped by PCR of tail biopsies (5-10 mm fragment) obtained from non-anesthetized, 10-day-old pups. All animals were between 2 and 4 months old at the time of experimentation. All procedures were conducted in accordance with Ernest Gallo Clinic and Research Center Institutional Animal Care and Use Committee policies. Zolpidem-induced ataxia was measured as described previously (17). The "up and down" method (17) was used to measure initial sensitivity to zolpidem with a starting dose of 0.25 mg/kg and a log dose interval of 0.0276.
Immunoprecipitation-Frontal cortex was dissected from 8to 10-week-old mice and homogenized with a Teflon-glass homogenizer in ice-cold buffer containing 25 mM Tris-HCl, pH 7.4, 1 mM EDTA, Complete TM protease inhibitor mixture (Roche Applied Science), 2 mM benzamidine chloride, and 0.1 mM phenylmethylsulfonyl fluoride. The homogenate was centrifuged at 200,000 ϫ g for 30 min at 4°C. The pellet was solubilized in ice-cold Tris-buffered saline (10 mM Tris-HCl, pH 7.2, 150 mM NaCl, 0.1% Triton X-100, and Complete TM protease inhibitor mixture), and centrifuged again at 120,000 ϫ g for 15 min at 4°C. The supernatant (500 g of protein) was incubated for 2 h at 27°C with 1 g of rabbit anti-GABA A -␣1 polyclonal antibody 993-2, generated against the C-terminal decapeptide sequence common to rat, mouse, and human GABA A -␣1 (PQLKAPTPHQ), or with purified rabbit IgG (Sigma-Aldrich). For the reverse immunoprecipitation, lysates were incubated with 1 g of rabbit anti-PKC⑀ (18) or purified rabbit IgG. Protein A beads (100 l, Roche Applied Science) were added, and the suspension was incubated overnight at 4°C. Beads were washed four times with ice-cold PBS, and immune complexes were eluted with 100 l of sample buffer. After being heated for 7 min at 90°C, samples were centrifuged at 6,000 ϫ g for 30 s, and proteins were separated in 4 -12% polyacrylamide gels (Invitrogen). Proteins were analyzed using Western blots with goat anti-PKC⑀ (Santa Cruz Biotechnology, Santa Cruz, CA, 1:500 dilution) or goat anti-␣1 subunit antibody (Santa Cruz Biotechnology, 1:500 dilution) followed by peroxidase-conjugated rabbit anti-goat IgG (Roche Applied Science, 1:1000 dilution) and enhanced chemiluminescence (Pierce).
Primary Neuronal Culture-Primary cultures of cortical and hippocampal neurons were prepared from 1-to 3-day-old mice as described (19). Briefly, cells were dissociated from dissected cerebral cortex or hippocampus by treatment with papain (20 units/ml) for 45 min at 37°C, followed by trituration through a 1-ml plastic pipette. For immunostaining, hippocampal neurons were plated at a density of 10 5 cells on 200-mm 2 chamber slides coated with poly-L-lysine in Neurobasal A media containing B27 and Glutamax (Invitrogen) and maintained at 37°C in a humidified atmosphere of 6% CO 2 and 94% air. For studies of cell surface ␥2 subunits, cortical neurons were plated on poly-L-lysine-coated 6-well plates at 1.6 ϫ 10 6 cells/well. For electrophysiology experiments, hippocampal neurons were plated on poly-L-ornithine-coated 35-mm Petri dishes. Half of the medium was changed the next day and changed every 7 days for 3 weeks.
Immunofluorescence Staining of Neurons-Mature (14 -21 days in vitro) neurons were immunostained as described (20). To detect GABA A receptors on the cell surface, living cultures were incubated with rabbit anti-GABA A receptor ␣1 N-terminal peptide (1:100, Chemicon, Temecula, CA) diluted in Ringer's solution containing 0.5 M tetrodotoxin for 90 min at 27°C. After three 10-min washes in Ringer's solution, the cells were fixed with methanol for 10 min at Ϫ20°C. Fixed cells were washed with PBS and incubated for 90 min at 27°C with goat anti-PKC⑀ antibody (1:100, Santa Cruz Biotechnology) diluted in PBS containing 10% normal donkey serum. After three washes with PBS, the cells were then incubated with donkey fluorescein isothiocyanate-conjugated anti-goat and Cy3-conjugated anti-rabbit antibodies (1:200, Jackson Immuno-Research, West Grove, PA) and covered with coverslips in mounting media containing 4Ј,6-diamidino-2-phenylindole (Vector Laboratories, Burlingame, CA). Images (1.5-m optical sections) were acquired using a Zeiss LSM 510 META laser confocal microscope with a Plan-Apochromat 63ϫ/numerical aperture 1.40 oil immersion objective.
Generation of as-PKC⑀-A pCDNA3 plasmid containing N-terminal FLAG epitope-tagged, human PKC⑀ cDNA was provided by Dr. Alex Toker (Harvard University, Boston, MA) and used as a template to replace Met-486 with Ala by site-directed mutagenesis (QuikChange mutagenesis kit, Stratagene, La Jolla, CA) using the following primers: 5Ј-GGACCGCCTCTTTTTCGTCGCCGAATATGTAAATG-GTGGAGACC-3Ј as the forward primer and 5Ј-GGTCTCCA-CCATTTACATATTCGGCGACGAAAAAGAGGCGGTCC-3Ј as the reverse primer. The PCR product was treated with DpnI for 1 h at 37°C and then amplified in Escherichia coli XL 1-Blue. The presence of the M486A mutation was confirmed by DNA sequencing.
Plasmid DNA (10 g) was transfected into COS-7 cells using the SuperFect Transfection Reagent (Qiagen). After culture for 72 h, cells were rinsed twice with 2 ml of ice-cold PBS and incubated for 20 min at 4°C in lysis buffer containing 150 mM NaCl, 1 mM EDTA, 50 mM Tris-HCl (pH 7.4), 1% Triton X-100, and Complete Protease Inhibitor Mixture (Roche Applied Science). Lysates from 50 10-cm plates were pooled and centrifuged at 12,000 ϫ g for 10 min at 4°C. The supernatant was incubated for 3 h at 4°C with 1 ml of anti-FLAG M2-conjugated agarose (Sigma), previously equilibrated with lysis buffer. Agarose beads were transferred to a column and washed sequentially with 15 ml of lysis buffer, 15 ml of buffer B (150 mM NaCl, 50 mM Tris-HCl, pH 7.4) and 5 ml of PKC storage buffer (0.1 mM EGTA, 20 mM HEPES, pH 7.4, 25% glycerol, and 0.03% Triton X-100). To elute FLAG-tagged proteins, the beads were incubated with 1.75 ml of 0.15 mg/ml 3ϫ FLAG peptide solution (Sigma-Aldrich) in PKC storage buffer for 30 min at 4°C and then centrifuged for 10 s at 10,000 ϫ g. Aliquots of the supernatant were stored at Ϫ80°C. The concentration of PKC⑀ in the eluate was measured in triplicate by enzyme-linked immunosorbent assay using recombinant PKC⑀ (Invitrogen) as a standard, rabbit anti-PKC⑀ SN134 (18) diluted 1:1,000 as the primary antibody, and horseradish peroxidase-conjugated anti-rabbit IgG as the secondary antibody (1:3,500 dilution, Chemicon). Immunoreactivity was detected using 3,3Ј,5,5Ј-tetramethylbenzidine as the substrate with absorbance measured at 450 nm.
Characterization of as-PKC⑀-Enzyme activity was measured by fluorescence polarization using the Protein Kinase C Assay Kit from Invitrogen. PKC⑀ (0.5 ng) or PKC⑀-M486A (3 ng) was added to kinase buffer (20 mM HEPES, pH 7.4, 10 mM MgCl 2 , 0.02% Nonidet P-40, 0.03% Triton X-100, 0.1 g phosphatidylserine, and 0.05 mM sodium vanadate) with 250 nM substrate peptide (RFARKGSLRQKNV) and 12.5 nM phorbol 12-myristate 13-acetate in a final volume of 40 l. The mixture was incubated at 22°C for 5 min in the dark. The kinase reaction was initiated by addition of ATP (1-100 M) in 10 l and was stopped after 0 -20 min by addition of 1 l of 0.5 mM staurosporine (Calbiochem) in Me 2 SO. A 50-l aliquot of the reaction mixture was transferred to a well of a dark-sided 96-well plate and mixed with 50 l of a solution containing 2ϫ fluorescein-labeled phosphopeptide and 2ϫ anti-phosphoserine antibody, according to the manufacturer's instructions. After incubation for 30 min at 22°C in the dark, plane-polarized fluorescence was measured using an Analyst HT (Molecular Devices, Sunnyvale, CA) with excitation at 485 nm and emission at 530 nm. Polarization (P) was calculated by the equation, P ϭ (I para Ϫ I perp )/(I para ϩ I perp ), where I para is the parallel emission fluorescence intensity and I perp is perpendicular emission fluorescence intensity. The Michaelis-Menten constant (K m ) and maximum velocity (V max ) were calculated from the equa- Stable Transfection of HEK293 Cells with as-PKC⑀-HEK293 cells were obtained from the American Type Culture Collection (Manassas, VA). HEK293 cells that stably express rat ␣1␤2␥2S GABA A receptors (21) were provided by G. H. Dillon, (University of N. Texas Health Science Center, Fort Worth, TX). Human PKC⑀ M486A was amplified by PCR and then subcloned without the FLAG-epitope tag into pIRESpuro2 (Clontech, Palo Alto, CA). Cells were plated on poly-L-ornithine-coated 6-well plates at a density of 10 5 cells per well in 2 ml of growth medium (minimal essential medium with 10% fetal bovine serum, 0.2 mM l-␣-glutamine, 100 g/ml streptomycin, and 100 units/ml penicillin). The cells were incubated overnight at 37°C in a humidified atmosphere of 5% CO 2 :95% air before transfection. Transfection complexes were prepared by mixing 2 g of plasmid DNA with 6 l of TransIT-293 reagent (Mirus, Madison, WI) in 100 l of serum-free medium. Complexes were added dropwise to each well, and the cells were incubated for 24 h at 37°C in 5% CO 2 :95% air. Clones were selected 48 h later in 5 g/ml puromycin. Medium was replaced daily, and the concentration of puromycin was reduced to 3 g/ ml after 1 week. Expression of PKC⑀ or PKC⑀ M486A in surviving cells was confirmed by Western blot analysis.
To test the effect of inhibiting PKC⑀ on GABA A receptor function, cells were incubated with external solution containing 1 M 1-naphthyl-PP1 (1Na) or vehicle (0.1% Me 2 SO) for 30 min prior to stimulation with GABA. Cells were then voltageclamped at Ϫ75 mV, and the GABA dose-response relationship was determined in the presence or absence of 1 M 1Na. The response to allosteric modulators was then assessed by co-application of the modulator with an EC 20 concentration of GABA. Flunitrazepam was applied 5 s prior to co-application with GABA, and ethanol was co-applied with GABA by local perfusion using a Perfusion Fast-Step SF-77B system (Warner Instruments, Inc., Hamden, CT) driven by pClamp 9 software (Axon Instruments, Union City, CA). Unless otherwise stated, there was a 1-min washout period between each drug application. Whole cell currents were recorded using an Axopatch 200B patch amplifier (Axon CNS-Molecular Devices, Union City, CA), filtered at 2 kHz, and digitized at 5 kHz with a Digi-data 1322A interface and pClamp 9 software (Axon CNS). The serial resistance was monitored continuously during each experiment, and data from cells that showed a Ͼ30% change in resistance were discarded. All recordings were obtained at 27°C.
Phosphorylation of GST-␥2 Intracellular Loop Fusion Proteins-A plasmid (provided by R. Olsen, UCLA) containing the cDNA sequence encoding the mouse ␥2S large intracellular loop (amino acids 318 -404) in pGEX-2T (GE Healthcare, Piscataway, NJ) was used to generate an alanine mutation at Ser-327 with a QuikChange site-directed mutagenesis kit (Stratagene) and the primers: 5Ј-GTCAGCAACCGGAAGC-CAGCCAAGGATAAAGACAAAAAGAAGAAAAACCC-3Ј as the forward primer and 5Ј-GGGTTTTTCTTCTTTTTGTCTT-TATCCTTGGCTGGCTTCCGGTTGCTGAC-3Ј as the reverse primer. BamHI and EcoRI fragments encoding the native and S327A mutant loops were subcloned into pGEX-6P-2 (GE Healthcare).
Fusion proteins were produced in E. coli BL21(DE3)pLysS cells (Invitrogen) and purified by affinity chromatography using glutathione-Sepharose 4B (GE Healthcare). Phosphorylation was performed using 0.1 M human recombinant PKC⑀ (Invitrogen) in 10 l of kinase buffer containing 20 mM HEPES, pH 7.4, 0.1 mM EGTA, 0.03% Triton X-100, 10 mM MgCl 2 , 0.48 g/l L-␣-phosphatidylserine (Avanti Polar Lipids, Alabaster, AL), 1 M phorbol 12-myristate 13-acetate, and 0.5 mM ATP. PKC⑀ was preincubated with kinase buffer for 3 h at 27°C to stimulate autophosphorylation and thereby maximize the activity of PKC⑀. Then 1.25 pmol of purified GST-␥2S loop or GST-␥2S S327A loop and 10 Ci of [␥-32 P]ATP were added to the reaction mixture, which was incubated at 37°C for the indicated times. At each time point, 10 l of reaction mixture were removed and mixed with 2.5 l of 5ϫ SDS sample buffer to stop the reaction. Proteins were separated by SDS-PAGE on 4 -12% gradient gels (Invitrogen), and the amount of radioactivity incorporated was quantified by phosphorimaging (Typhoon 9410, GE Healthcare). Specific radioactivity per mole of [␥-32 P]ATP was determined by counting diluted aliquots of the stock [␥-32 P]ATP solution to obtain a conversion value for calculating the molar ratio of 32 P incorporated into each fusion protein substrate.
Detection of ␥2 S327 Phosphorylation by Western Analysis-GST-␥2S loop proteins were phosphorylated in vitro in the presence of non-radioactive ATP. Samples of frontal cortex and hippocampus were homogenized in lysis buffer containing 25 mM Tris-HCl, pH 7.4, 1 mM EDTA, and Complete TM protease inhibitor mixture (Roche Applied Science). Samples were adjusted to a concentration of 2 mg/ml and some were incu-bated with 8,000 units of lambda protein phosphatase (Upstate Biotechnology, Lake Placid, NY) at 30°C for 30 min. Control samples not treated with phosphatase were incubated with Phosphatase Inhibitor Mixture 1 (Sigma-Aldrich).
Statistical Analysis-Data shown are mean Ϯ S.E. values. Data were first tested for normality using the Kolmogorov-Smirnov test. Because all data were normally distributed, mean values were compared by either two-tailed, unpaired t-tests or ANOVA with indicated post-hoc tests. Differences between pairs of means were considered significant where p Ͻ 0.05. Except where indicated, dose-response relationships were analyzed by non-linear regression analysis using Prism 4 (Graph-Pad Software, Inc., San Diego, CA). Potentiation of the GABA response by allosteric modulators was calculated as the ratio of the current elicited by modulator plus GABA divided by the current elicited by GABA alone.

RESULTS
Increased Response to Zolpidem in PKC⑀ Ϫ/Ϫ Mice-To determine whether PKC⑀ regulates GABA A receptors that contain ␣1 and ␥2 subunits, we first examined behavioral responses in mice that lack PKC⑀. Because behavioral responses to benzodiazepines are increased in these mice (13), we tested their response to zolpidem, an imidazopyridine that selectively acts at the benzodiazepine site on GABA A receptors that contain ␣1 and ␥2 subunits (22). We found that PKC⑀ Ϫ/Ϫ mice showed greater rotarod ataxia in response to 1 mg/kg zolpidem than wild-type littermates (Fig. 1A). Two-way repeated measures ANOVA showed main effects of genotype (F 1,48 ϭ 5.66, p ϭ 0.044) and time (F 6,48 ϭ 11.19, p Ͻ 0.001) with an interaction between these factors (F 6,48 ϭ 2,87, p ϭ 0.018). The threshold dose of zolpidem for producing ataxia was also lower in PKC⑀ Ϫ/Ϫ mice compared with wild-type controls (Fig. 1B). These differences in behavior were not associated with altered distribution of ␣1 subunits in the brain (supplemental Fig.  S1A), abundance of ␣1 or ␥2 subunits in frontal cortex, cerebellum, striatum, or amygdala (supplemental Fig. S1B), or altered cell surface abundance of ␥2 subunits (supplemental Fig. S1C) in PKC⑀ Ϫ/Ϫ mice when compared with wild-type littermates. These findings suggest that the function, rather than the abundance, of zolpidem-sensitive GABA A receptors is increased in PKC⑀ Ϫ/Ϫ mice. Consistent with this conclusion, PKC⑀ and ␣1 subunits could be co-immunoprecipitated from frontal cortex of wild-type mice (Fig. 1, C and D) and were co-localized at the cell membrane in cultured hippocampal neurons from wildtype animals (Fig. 1E), indicating that PKC⑀ and ␣1-containing GABA A receptors physically interact in a complex.
Generation of an ATP Analog-sensitive Mutant of PKC⑀-PKC⑀ can transduce signals by phosphorylation of substrates or by protein-protein interaction. The latter has been demonstrated for PKC⑀-stimulated neurite outgrowth in human neuroblastoma cells (23). The ⑀V1-2 peptide (EAVSLKPT), which was designed as an inhibitor of PKC⑀ translocation (24), can be used to selectively inhibit PKC⑀. However, because ⑀V1-2 does not bind the catalytic domain or inhibit the catalytic activity of PKC⑀, this peptide cannot be used to distinguish between PKC⑀-mediated effects due to protein-protein interactions or phosphorylation. Furthermore, no small molecules are known that specifically inhibit the catalytic activity of PKC⑀ over other PKC isozymes, due to the highly conserved ATP binding sites of protein kinases. Therefore, to examine whether PKC⑀-mediated phosphorylation is required for modulation of GABA A receptors, we used a chemical-genetics approach to develop a selective small molecule inhibitor. We generated a single amino acid substitution at methionine 486 in the ATP binding pocket of PKC⑀, which we predicted would permit binding of ATP analogs with bulky side groups attached to the purine base. Such a strategy has been used to successfully generate ATP analog-sensitive kinase mutants, which can also be selectively inhibited by analogs of the kinase inhibitor 4-amino-5-(4-methylphenyl)-7-(t-butyl)pyrazolo [3,4-d]pyrimidine (PP1) at concentrations that do not affect native kinases (25).
We generated human PKC⑀ M486A (as-PKC⑀) containing a FLAG-epitope tag at the amino terminus using site-directed mutagenesis. Kinetic analysis using different concentrations of ATP revealed the K m for as-PKC⑀ (241.3 Ϯ 39.1 M) to be 12-fold greater than the K m for native PKC⑀ (20.4 Ϯ 4.0 M). However, both enzymes had similar turnover (k cat ) values (170,600 Ϯ 16,500 min Ϫ1 for as-PKC⑀ and 156,700 Ϯ 3,700 min Ϫ1 for PKC⑀). The similarity in k cat values indicates that as-PKC⑀ can utilize ATP as efficiently as native PKC⑀ at the high ATP concentrations (2-10 mM) found in cells.
We next examined whether commercially available kinase inhibitors inhibit as-PKC⑀ and whether analogs of PP1 selectively inhibit as-PKC⑀. The IC 50 for the nonspecific kinase inhibitor staurosporine was 14-fold higher and for the general PKC inhibitor bisindolylmaleimide-I was 2-fold higher for as-PKC⑀ compared with native PKC⑀ (Table 1). In contrast, 1Na and 1-naphthylmethyl-PP1 inhibited as-PKC⑀ but not native PKC⑀, with 1Na being 65-fold more potent than 1-naphthylmethyl-PP1 against as-PKC⑀. 1Na did not inhibit the activity of a commercially available mixture of native PKC isozymes. In addition, 1Na did not inhibit representative members of the conventional (PKC␥), novel (PKC␦), or atypical (PKC) PKC subfamilies (Table 1).
␥2 S327 Is Required for PKC⑀ Regulation of GABA A Receptors-Next, we examined the role of subunit phosphorylation in the regulation of receptor function by PKC⑀. Prior in vitro studies have suggested that there is one phosphorylation site for serinethreonine kinases on ␤2 (serine 410) and two on the long splice variant of ␥2, ␥2L (serine 327 and 343), with only one of these (serine 327) present in the short splice variant ␥2S (26). To investigate whether any of these sites are required for GABA A receptor modulation by PKC⑀, we transiently expressed ␣1, ␤2, and ␥2L receptor subunits carrying alanine substitutions at these residues into HEK293 cells that stably express as-PKC⑀ at a level that is 5-fold higher than endogenous PKC⑀ ( Fig. 2A). In non-transfected HEK293 cells, levels of PKC⑀ were low ( Fig.  2A), allowing as-PKC⑀ to dominate over endogenous PKC⑀ in these stably transfected cells.
As expected, 1Na did not alter the response to GABA alone in cells expressing native receptors or any of the mutant receptor  ). B, the zolpidem ED 50 for inducing rotarod ataxia was decreased in PKC⑀ Ϫ/Ϫ mice (Ϫ/Ϫ) compared with wild-type (ϩ/ϩ) littermates. Data are mean values Ϯ 95% confidence intervals from six mice per genotype. C, ␣1 subunit immunoreactivity was detected by Western analysis of whole brain lysates from PKC⑀ ϩ/ϩ mice immunoprecipitated with anti-PKC⑀ antibody. No ␣1 subunit immunoreactivity was detected in immunoprecipitates from PKC⑀ Ϫ/Ϫ brain lysates or when rabbit IgG was used instead of anti-PKC⑀ antibody. D, PKC⑀ immunoreactivity was detected by Western analysis of whole brain lysates from PKC⑀ ϩ/ϩ mice immunoprecipitated with anti-␣1 antibody. No PKC⑀ immunoreactivity was seen when lysates were immunoprecipitated with rabbit IgG. E, PKC⑀ (left, green) and ␣1 subunits (middle, red) co-localize (right, merge) at the cell surface of cultured hippocampal neurons from wild-type mice. Boxes refer to enlarged regions shown in the bottom panels. Scale bar ϭ 20 m.   NOVEMBER 9, 2007 • VOLUME 282 • NUMBER 45 JOURNAL OF BIOLOGICAL CHEMISTRY 33057 combinations (Fig. 2, B and C, and Table 2). This result is consistent with our prior chloride flux experiments, which showed that PKC⑀ does not affect the response to the direct GABA A receptor agonist muscimol (13). We next examined the response to a benzodiazepine, because PKC⑀ Ϫ/Ϫ mice showed an increased behavioral response to zolpidem (Fig. 1, A and B). For receptors containing native ␣1, ␤2, and ␥2L subunits, fluni-trazepam increased the GABAstimulated current, which was completely blocked by 30 M bicuculline (Fig. 2D). Treatment with 1 M 1Na increased the potency of flunitrazepam 4.9-fold, decreasing its EC 50 from 98 nM to 17 nM, without affecting efficacy (Fig. 2, E and F, and Table 2). Interestingly, all subunit combinations that included the ␥2L S327A mutation showed no response to 1Na, and this mutation alone was sufficient to completely block the effect of 1Na ( Fig. 2G and Table 2). In addition, without 1Na treatment, flunitrazepam potency was greater in cells expressing ␥2L S327A when compared with cells expressing native ␥2L (Fig. 2H and Table 2). By contrast, receptors containing the ␤2 S410A mutant responded to flunitrazepam and 1Na in a manner indistinguishable from native receptors ( Table 2). Like currents in cells expressing native receptors, currents elicited with mutant receptors or in the presence of 1Na were completely blocked by 30 M bicuculline (Fig.  2D). These results indicate that the ␥2L S327A mutation mimics the effect of PKC⑀ inhibition, suggesting that PKC⑀ decreases benzodiazepine potency through phosphorylation of ␥2 S327 . We next examined whether ␥2L S327 phosphorylation regulates ethanol sensitivity of GABA A receptors. In cells containing native ␥2L or mutant ␥2L S327A subunits, ethanol enhanced the GABAstimulated current, which was completely inhibited by 30 M bicuculline (Fig. 3A). Treatment with 1Na increased the effect of ethanol (Fig. 3, A-C) in cells expressing native receptors. Two-way, repeated measures ANOVA showed a main effect of ethanol concentration (F 4,48 ϭ 38.87, p Ͻ 0.001) and 1Na treatment (F 1,48 ϭ 20.26, p Ͻ 0.001) with a significant interaction  between these factors (F 4,48 ϭ 4.59, p Ͻ 0.003). By contrast, 1Na did not increase the response to ethanol in cells containing ␥2L S327A (Fig. 3D). Moreover, without 1Na treatment, the response to ethanol was greater in cells expressing ␥2L S327A when compared with cells expressing native ␥2L (Fig. 3E). Twoway repeated measures ANOVA showed main effects of ethanol concentration (F 4,60 ϭ 35.98, p Ͻ 0.001) and subunit type (F 1,60 ϭ 8.16, p Ͻ 0.012) with a significant interaction between these factors (F 4,60 ϭ 4.54, p Ͻ 0.003). Ethanol-enhanced GABA-stimulated currents were completely blocked by 30 M bicuculline when 1 Na was present in both native receptors and receptors with ␥2L S327A (Fig. 3A). These results demonstrate that the ␥2 S327A mutation increases the response of GABA A receptors to ethanol and occludes further enhancement by a PKC⑀ inhibitor, strongly suggesting that PKC⑀ modulates the response to ethanol through phosphorylation of ␥2 S327 . The ␥2 S327A Mutation Reduces Rates of Inactivation and Desensitization in Cells Treated with Flunitrazepam-By fitting the current traces activated by 100 M GABA to a onecomponent exponential function and comparing the time constants, we found that currents activated by GABA alone showed similar rates of activation (261 Ϯ 32 versus 252 Ϯ 47 ms), inactivation (292 Ϯ 50 versus 338 Ϯ 50 ms), and desensitization (7128 Ϯ 927 versus 7026 Ϯ 1107 ms) in receptors comprised of native ␥2L (n ϭ 5) or ␥2L S327A (n ϭ 6). In contrast, we found differences in channel kinetics when currents were stimulated by GABA plus flunitrazepam. For currents activated by EC 20 GABA and 300 nM flunitrazepam, receptors containing ␥2L or ␥2L S327A produced currents with similar activation rates (776 Ϯ 26, n ϭ 8, versus 790 Ϯ 38 ms, n ϭ 9), but ␥2L containing receptors had a significantly faster inactivation rate (117 Ϯ 19, n ϭ 8, versus 211 Ϯ 16 ms, n ϭ 9; p Ͻ 0.05 by t test). Moreover, whereas 6 of 8 ␥2L-transfected cells showed desensitization, only 3 of 9 ␥2L S327A -transfected cells desensitized. For those cells that showed desensitization, the desensitization rate was faster (p Ͻ 0.05 by two-tailed, t test) in ␥2L-tranfected cells (1847 Ϯ 122 ms) when compared with ␥2L S327A -transfected cells (2892 Ϯ 589 ms). For currents activated by EC 20 GABA and 100 mM ethanol, the activation rates (567 Ϯ 164, n ϭ 8 versus 459 Ϯ 134 ms, n ϭ 13) and deactivation rates (215 Ϯ 111 versus 212 Ϯ 97 ms) were similar for ␥2Land ␥2L S327A -containing receptors. The majority of the cells tested did not show desensitization (7 out of 8 ␥2L cells and 11 out of 13 ␥2L S327A cells). These findings indicate that the ␥2 S237A mutation does not alter channel kinetics in response to GABA alone or with ethanol but instead selectively reduces inactivation and desensitization rates when benzodiazepines are present.

PKC⑀ Phosphorylation of GABA A Receptors
PKC⑀ Phosphorylates ␥2 S327 in Vitro-To investigate whether PKC⑀ phosphorylates ␥2 S327 , we generated two GST fusion proteins, one containing the large intracellular loop of ␥2S, which includes Ser-327, and another containing the same sequence but with the S327A mutation. PKC⑀ phosphorylated the native sequence to a maximal stoichiometry of 1.00 Ϯ 0.06, whereas the fusion protein containing the ␥2 S327A mutation was a comparatively poor substrate for PKC⑀ (Fig. 4A). Twofactor ANOVA showed main effects of substrate (F 1,30 ϭ 119.5, p Ͻ 0.0001) and time (F 4,30 ϭ 60.55, p Ͻ 0.0001) with an inter-action between these factors (F 4,30 ϭ 13.03, p Ͻ 0.0001). These findings indicate that PKC⑀ phosphorylates ␥2 S327 in vitro.
To examine whether PKC⑀ phosphorylates ␥2 S327 in intact tissues, we used an affinity-purified antibody against a peptide derived from the large intracellular loop of ␥2 that contains phospho-␥2 S327 . This anti-␥2-S(P)327 antibody detected a GST-␥2 loop fusion protein phosphorylated by PKC⑀ in vitro, but not the non-phosphorylated fusion protein (Fig. 4B).
PKC⑀ Regulates Allosteric Responses in HEK Cells Containing ␥2S and Regulates ␥2S Phosphorylation at Ser-327-To examine receptor phosphorylation in intact cells we transfected a HEK293 cell line (HEKGR) that stably expresses ␣1, ␤2, and ␥2S subunits (21) with as-PKC⑀ and selected a subclone (HEKGR-as-PKC⑀) that expresses as-PKC⑀ at a level 3.3-fold greater than endogenous PKC⑀. Because some previous studies had suggested that the long splice variant of ␥2 (␥2L) is required to confer ethanol sensitivity to recombinant GABA A receptors (27)(28)(29)(30), we first examined whether ethanol and PKC⑀ could regulate responses in these cells. We then used our anti-␥2-S(P)327 antibody to detect ␥2 S327 phosphorylation in these cells before and after treatment with 1Na. As expected, 1Na did not alter the response to GABA (Fig.  5A) in HEKGR-as-PKC⑀ cells. However, 1Na increased flunitrazepam potency (Fig. 5B), decreasing the log EC 50 value from 2.18 Ϯ 0.09 nM in control cells (n ϭ 10) to 1.4 Ϯ 0.13 nM in 1Na-treated cells (n ϭ 10). Ethanol modestly enhanced GABAstimulated currents in these cells (Fig. 5C) (F 5,40 ϭ 6.67, p Ͻ 0.0001 by ANOVA). The effect of ethanol was further increased by treatment with 1Na (Fig. 5C), similar to what we observed in HEK293 cells that express ␣1␤2␥2L receptors (Fig. 3). Twoway repeated measures ANOVA showed main effects of 1Na treatment (F 1,64 ϭ 26.08, p ϭ 0.0001) and ethanol concentration (F 4,64 ϭ 40.48, p Ͻ 0.0001), with an interaction between these factors (F 4,64 ϭ 4.85, p ϭ 0.0018). The effect of 1Na required the presence of as-PKC⑀, because in the parent HEKGR cell line, 1Na had no effect on potentiation of GABA A current by ethanol or by the benzodiazepine diazepam (Fig.  5D). These results indicate that receptors containing ␥2L or ␥2S subunits are both responsive to ethanol when expressed in HEK293 cells and that receptors containing either ␥2L or ␥2S show PKC⑀ regulation of sensitivity to flunitrazepam and ethanol.
PKC⑀ Phosphorylates ␥2 S327 in Vivo-We next used the anti-␥2-S(P)327 antibody to investigate whether ␥2 S327 phosphorylation is decreased in PKC⑀ Ϫ/Ϫ mice. Anti-␥2-S(P)327 immunoreactivity was decreased in the hippocampus and frontal cortex of PKC⑀ Ϫ/Ϫ mice compared with wild-type littermates (Fig. 6A). The presence of residual immunoreactivity in PKC⑀ Ϫ/Ϫ brain tissue suggested that ␥2 S327 might also be phosphorylated by another kinase besides PKC⑀. To investigate this possibility, we treated samples from frontal cortex with lambda protein phosphatase for 30 min before performing Western analysis. Phosphatase treatment of wild-type samples decreased anti-␥2-S(P)327 immunoreactivity to levels found in PKC⑀ Ϫ/Ϫ samples, whereas treatment of samples from PKC⑀ Ϫ/Ϫ mice did not decrease anti-␥2-S(P)327 immunoreactivity further (Fig. 6B). Analysis of these data by two-factor ANOVA showed main effects of genotype (F 1,18 ϭ 13.20, p ϭ 0.0019) and treatment (F 1,18 ϭ 11.40, p ϭ 0.0034) with an interaction between genotype and treatment (F 1,18 ϭ 5.51, p ϭ 0.030). These findings suggest that the residual immunoreactivity present in samples from PKC⑀ Ϫ/Ϫ mice is not due to ␥2 phosphorylated at Ser-327 but, rather, is due to cross-reactivity of the antibody with non-phosphorylated ␥2 subunits. Consistent with this conclusion, we found that overnight incubation of the anti-␥2-S(P)327 antibody with a dephosphorylated version of the immunizing peptide prevented its ability to detect a protein band at 48 kDa (data not shown). Taken together, these results indicate that PKC⑀ is the major kinase that phosphorylates ␥2 S327 in vivo.

Ethanol Potentiation of GABA A Currents in Hippocampal
Neurons-Using cultured hippocampal neurons from wildtype and PKC⑀ Ϫ/Ϫ mice we examined whether absence of PKC⑀ would increase sensitivity to ethanol. In initial experiments we were unable to demonstrate any response to ethanol (Fig. 7A). However, a portion of whole cell current in cultured hippocampal neurons is carried by receptors comprised of ␣␤ subunits, which can be inhibited by 5 M Zn 2ϩ (31) and are insensitive to ethanol (28). Therefore, to eliminate currents carried by receptors containing only ␣ and ␤ subunits, we measured ethanol responses in the presence of 5 M ZnCl 2 . In cells from wild-type mice (n ϭ 12), the residual GABA current was increased 75 Ϯ 11% by 30 nM diazepam indicating that a substantial portion of the current was carried by receptors that contain ␥2 subunits. In neurons from PKC⑀ Ϫ/Ϫ mice (n ϭ 14) diazepam produced a greater effect (p ϭ 0.0092) enhancing the GABA-stimulated current by 127 Ϯ 14%. In wild-type neurons, treatment with ethanol (Fig. 7, B and D) produced a small but significant increase in the Zn 2ϩ -resistant current (F 5,30 ϭ 6.94, p Ͻ 0.0001 by repeated measures ANOVA). In contrast, neurons from PKC⑀ Ϫ/Ϫ mice showed a much greater response to ethanol (Fig.  7, C and D). Analysis of these data by two-way repeated measures ANOVA showed main effects of genotype (F 1,53 ϭ 6.87, p ϭ 0.02) and treatment (F 4,53 ϭ 12.01, p Ͻ 0.001) with an interaction between these factors (F 4,53 ϭ 2.72, p ϭ 0.039). These findings indicate that Zn 2ϩ -resistant GABA currents are enhanced to a greater extent by diazepam and ethanol in PKC⑀ Ϫ/Ϫ hippocampal neurons when compared with wildtype neurons.
Ethanol-induced Phosphorylation of ␥2 S327 in Vivo Requires PKC⑀-In mice, exposure to 4.0 g/kg ethanol in vivo produces GABA A receptor resistance to ethanol when measured 1 h later in cerebellar microsacs (3,17). This acute tolerance of GABA A receptors to ethanol occurs in wild-type mice but not in PKC⑀ Ϫ/Ϫ mice (17). Because increased phosphorylation at ␥2 S327 is associated with diminished receptor sensitivity to ethanol, we investigated whether in vivo exposure to ethanol increases phosphorylation of ␥2 S327 . Mice were injected with saline or with 4 g/kg ethanol intraperitoneally and then the cerebellum was removed and analyzed by Western blot analysis 1 h later. After saline injection, ␥2S(P)327 immunoreactivity was increased in the cerebellum of wildtype (n ϭ 6) relative to PKC⑀ Ϫ/Ϫ mice (n ϭ 6). In wild-type mice immunoreactivity was increased after injection of 4g/kg ethanol (Fig.  8). In contrast, ethanol did not alter ␥2S(P)327 immunoreactivity in cerebellar tissue from PKC⑀ Ϫ/Ϫ mice. Analysis of these data by two-factor ANOVA showed main effects of genotype (F 1,20 ϭ 47.64, p Ͻ 0.0001) and treatment (F 1,20 ϭ 6.46, P 0.0194) with a significant interaction between these factors (F 1,20 ϭ 4.93, p ϭ 0.0381).

DISCUSSION
It has been difficult to identify PKC isozyme-specific targets, because the catalytic domains of PKC isozymes are very similar and there are no isozyme-selective inhibitors of catalytic activity available. Here we used a chemical-genetics approach to generate an ATP analog-sensitive mutant of PKC⑀ and selectively inhibit the catalytic activity of an individual PKC. Using this mutant we identified a specific mechanism by which PKC⑀ regulates the function of ␣1␤2␥2 GABA A receptors. We demonstrated that PKC⑀ regulates the allosteric effect of benzodiazepines and ethanol at these receptors through phosphorylation rather than by a protein-protein interaction. Importantly, we found that phosphorylation at Ser-327 on ␥2 subunits, but not phosphorylation at other predicted PKC phosphorylation sites on ␤2 or ␥2 subunits, is required for PKC⑀ modulation of GABA A receptor function. Finally, we demonstrated that PKC⑀ directly phosphorylates ␥2 S327 in vitro, ␥2 S327 phosphorylation is reduced in mice lacking PKC⑀, and exposure to ethanol in vivo increases ␥2 S327 phosphorylation in the cerebellum of wild-type mice but not PKC⑀ Ϫ/Ϫ mice. These findings establish that PKC⑀ phosphorylation of Ser-327 in ␥2 subunits regulates the response of ␣1␤2␥2 GABA A receptors to specific allosteric modulators. Moreover, our results show that concentrations of ethanol achieved during social drinking can increase the function of ␣1␤2␥2 receptors by 40 -60% when PKC⑀ phosphorylation of ␥2 S327 is prevented.
10 -60 min with phorbol esters, which activate several PKC isozymes and other proteins (32), reduced receptor activation by GABA, and this effect was absent in receptors with alanine mutations at ␤1 S409 , ␥2L S327 , and ␥2 S343 (33). Phosphorylation of the ␤1 subunit was responsible for this effect, because a single ␤1 S409A mutation was sufficient to eliminate phorbol ester-induced down-regulation. Another study using ␣1␤2␥2S receptors expressed in Xenopus oocytes found a very short-lived and modest effect of ␥2S S327A expression; after 1.5 min of phorbol ester treatment the current amplitude decreased by 29% in native receptors but only by 19% in receptors carrying ␥2S S327A , although this 10% difference disappeared after 10 min (34). Our studies with HEK293 cells expressing ␥2S S327A indicate no effect of this mutation on current stimulated for up to 10 min by GABA alone (data not shown). Therefore, in contrast to its clear regulation of allosteric responses shown in our present study, ␥2 S327 phosphorylation does not appear to play a role in regulating the direct response to GABA.
Previous expression studies using Xenopus oocytes or mouse L(tkϪ) cells suggested a requirement for the long splice variant of ␥2 subunits, ␥2L, in conferring PKC-mediated sensitivity to low doses (Ͻ50 mM) of ethanol (27)(28)(29)(30). However, subsequent studies failed to observe a requirement for ␥2L in receptor responses to ethanol (35)(36)(37) and mice expressing ␥2S but not ␥2L showed normal behavioral responses to ethanol (38). We found that receptors containing the long or short splice variant of ␥2 respond similarly to ethanol and that inhibition of PKC⑀ enhances the effect of ethanol in both receptor types. These findings add to the body of evidence indicating that splice variation in ␥2 or phosphorylation at Ser-343 in ␥2L does not modulate GABA A receptor responses to ethanol.
In primary hippocampal neurons, we found that it was not possible to demonstrate ethanol enhancement of whole cell GABA currents. However, in cultured rat hippocampal neurons there is a benzodiazepine-insensitive tonic current that exhibits high sensitivity to Zn 2ϩ (IC 50 ϳ 1.89 M), contributes to the tonic current, and is likely composed of ␣␤ receptors (39). In stably transfected L(tkϪ) mouse fibroblasts, ethanol enhances GABA-stimulated chloride uptake through ␣1␤1␥2 receptors but not ␣1␤1 receptors, suggesting a requirement for ␥2 subunits for response to ethanol (28). Therefore, we used 5 M Zn 2ϩ to inhibit such ethanol-insensitive currents, because  showing increased ␥2-S(P)327 immunoreactivity in wild-type compared with PKC⑀ Ϫ/Ϫ cerebellum after saline injection. Immunoreactivity increased after ethanol injection in wild-type but not in PKC⑀ Ϫ/Ϫ cerebellum (n ϭ 8 for each condition). *, p Ͻ 0.05 when compared with saline-treated, PKC⑀ Ϫ/Ϫ samples, and † , p Ͻ 0.05 compared with saline-treated, wild-type samples by Bonferroni tests.
in HEK293 cells expressing recombinant GABA A receptors, currents carried by ␣1␤3 receptors are inhibited by Ͼ90% by 5 M Zn 2ϩ , whereas ␣1␤3␥2 receptors are inhibited by less than 10% (31). Our finding that 5 M Zn 2ϩ unmasked a response to ethanol indicates the presence of an ethanol-insensitive current that contributes significantly to GABA responses in these cells and underscores the difficulty in using whole cell patch clamp recording to detect ethanol sensitivity in primary neurons that express both ethanol-sensitive and ethanol-insensitive receptors.
Our finding that PKC⑀ inhibition alone was sufficient to increase allosteric sensitivity indicates that there is an active endogenous phosphatase associated with ␥2 subunits. One candidate is calcineurin (protein phosphatase 2B), which can be co-immunoprecipitated with ␥2 subunits and which dephosphorylates ␥2 S327 in mouse hippocampus (40). Therefore, PKC⑀ and calcineurin may act in concert to reciprocally regulate GABA A receptor phosphorylation and function in the brain.
Genetic variation in the ␥2 subunit plays an important role in behavioral responses to ethanol. Quantitative trait locus analysis has revealed two loci associated with severity of acute alcohol withdrawal in mice within a 3-centimorgan interval on mouse chromosome 11 that includes the GABA A receptor gene cluster encoding ␣1, ␣6, ␤2, and ␥2 subunits (41). DBA/2J mice show particularly severe alcohol withdrawal seizures and posses a unique ␥2 allele resulting in an alanine to threonine substitution at residue 11 in the mature ␥2 protein. Moreover, ␥2 ϩ/Ϫ mutant mice show more severe alcohol withdrawal seizures than wild-type littermates (41). In addition to alterations in alcohol withdrawal seizure severity, allelic variation in ␥2 is associated with susceptibility to ethanol-induced ataxia (42). It is therefore interesting that, compared with wild-type mice, PKC⑀ Ϫ/Ϫ mice show less severe alcohol withdrawal seizures (43) and less acute tolerance to alcohol-induced ataxia (17). Together with our present results, these findings suggest that PKC⑀ modulates the intensity and duration of alcohol intoxication and the severity of alcohol withdrawal through phosphorylation of GABA A receptors at ␥2 S327 .