Surface Expression of GABAA Receptors Is Transcriptionally Controlled by the Interplay of cAMP-response Element-binding Protein and Its Binding Partner Inducible cAMP Early Repressor*

The regulated expression of type A γ-aminobutyric acid (GABA) receptor (GABAAR) subunit genes plays a critical role in neuronal maturation and synaptogenesis. It is also associated with a variety of neurological diseases. Changes in GABAA receptor α1 subunit gene (GABRA1) expression have been reported in animal models of epilepsy, alcohol abuse, withdrawal, and stress. Understanding the genetic mechanism behind such changes in α subunit expression will lead to a better understanding of the role that signal transduction plays in control over GABAAR function and brings with it the promise of providing new therapeutic tools for the prevention or cure of a variety of neurological disorders. Here we show that activation of protein kinase C increases α1 subunit levels via phosphorylation of CREB (pCREB) that is bound to the GABRA1 promoter (GABRA1p). In contrast, activation of protein kinase A decreases levels of α1 even in the presence of pCREB. Decrease of α1 is dependent upon the inducible cAMP early repressor (ICER) as directly demonstrated by ICER-induced down-regulation of endogenous α1-containing GABAARs at the cell surface of cortical neurons. Taken together with the fact that there are less α1γ2-containing GABAARs in neurons after protein kinase A stimulation and that activation of endogenous dopamine receptors down-regulates α1 subunit mRNA levels subsequent to induction of ICER, our studies identify a transcriptional mechanism for regulating the cell surface expression of α1-containing GABAARs that is dependent upon the formation of CREB heterodimers.

The regulated expression of type A ␥-aminobutyric acid (GABA) receptor (GABA A R) subunit genes plays a critical role in neuronal maturation and synaptogenesis. It is also associated with a variety of neurological diseases. Changes in GABA A receptor ␣1 subunit gene (GABRA1) expression have been reported in animal models of epilepsy, alcohol abuse, withdrawal, and stress. Understanding the genetic mechanism behind such changes in ␣ subunit expression will lead to a better understanding of the role that signal transduction plays in control over GABA A R function and brings with it the promise of providing new therapeutic tools for the prevention or cure of a variety of neurological disorders. Here we show that activation of protein kinase C increases ␣1 subunit levels via phosphorylation of CREB (pCREB) that is bound to the GABRA1 promoter (GABRA1p). In contrast, activation of protein kinase A decreases levels of ␣1 even in the presence of pCREB. Decrease of ␣1 is dependent upon the inducible cAMP early repressor (ICER) as directly demonstrated by ICER-induced down-regulation of endogenous ␣1-containing GABA A Rs at the cell surface of cortical neurons. Taken together with the fact that there are less ␣1␥2-containing GABA A Rs in neurons after protein kinase A stimulation and that activation of endogenous dopamine receptors down-regulates ␣1 subunit mRNA levels subsequent to induction of ICER, our studies identify a transcriptional mechanism for regulating the cell surface expression of ␣1-containing GABA A Rs that is dependent upon the formation of CREB heterodimers.
␥-Aminobutyric acid (GABA) 4 type A receptors (GABA A Rs) are ligand-gated chloride ion channels that mediate the major-ity of fast synaptic inhibition in the mammalian brain (1). They are pentameric in structure and are composed of multiple subunit isoforms coming from eight distinct classes: ␣, ␤, ␥, ␦, , ⑀, , and . Diversity in receptor subtypes is controlled by the regulated expression of 19 different subunit genes and the alternative splicing of individual subunit transcripts (2)(3)(4). Most importantly, differential receptor subunit composition produces functionally and pharmacologically distinct GABA A Rs at certain times during development and in certain regions of the brain (5)(6)(7)(8).
The transcription of different GABA A R subunit genes (GABRs) is likely to involve a complex system of regulatory controls that remain to be identified. Several lines of evidence suggest that ␣1 subunit expression is activity-dependent (9 -11). Treatment with N-methyl-D-aspartate, a selective activator of an important class of excitatory ligand-gated ion channels, stimulates ␣1 subunit expression in cultured cerebellar granule cells (12,13). In contrast, chronic treatment of cortical neurons with GABA decreases levels of GABA A R ␣1 subunit mRNAs (14 -16), which is dependent on voltage-gated calcium channel activity and most likely an alteration in transcription (15). In addition, prolonged benzodiazepine (BZ) treatment decreases ␣1 mRNA levels in the rat hippocampal CA1 region. This decrease is associated with reduced GABA-mediated inhibition and a decrease in BZ potency to inhibit CA1 pyramidal cell-evoked responses (17).
Changes in the expression of ␣1 subunits have been observed in several animal models, including temporal lobe epilepsy, alcohol dependence and withdrawal, swim stress, and neonatal handling with maternal separation (18 -21). Decreased levels of ␣1 subunit mRNAs in individual dentate granule cells have been observed in both patients and animals with temporal lobe epilepsy and are associated with functionally altered GABA A Rs that display decreased zolpidem potentiation and increased sensitivity to zinc blockade (18,(22)(23)(24)(25)(26).
As a foundation for understanding changes in GABA A R function that occur in vivo as a response to seizure activity, we determined whether the human and rat GABRA1 promoters contain consensus regulatory sites for activity-dependent transcription factors such as the cAMP-response element-binding protein (CREB). Phosphorylation-induced activation of CREB is regulated by multiple signal transduction pathways, including cAMP-dependent protein kinase (PKA), Ca 2ϩ -calmodulin dependent kinases, and mitogenactivated protein kinases (MAPKs), that are critical for the transcription of important neural specific genes (41,45). Once phosphorylated on serine 133, CREB bound to its DNA recognition site (CRE) can interact with the coactivator protein CBP (27,28) leading to the recruitment of additional histone acetyltransferases to the promoter region (29,30). The formation of a multiprotein complex containing CREB, CBP, and the basal transcriptional machinery initiates gene transcription (31,32).
In addition to these proteins, there are bZIP repressors, including a class of truncated isoforms that can bind to CRE and CRE-like elements (33). One well characterized repressor isoform of the CREM family is inducible cAMP early repressor (ICER) (34,35). ICER is expressed as a family of four isoforms that are produced from an internal promoter, p2, located in an intron of the CREM gene (34). Alternative splicing of the ␥-exon and the ICER DNA-binding domains (DBDs I and II) generates ICERI, I␥, II, and II␥. The ICER proteins contain DNA binding/leucine zipper domains that make them endogenous inhibitors of transcription driven by CREB and its cognates, CREM and ATF1. ICER expression is inducible in the brain and in neuronal culture by a variety of stimuli. As an antagonist of CREB transcriptional activation, ICER appears to be of pivotal importance in gene regulation of the nervous system (36).
We now extend the importance of ICER to include the gene regulation of an important ␣ subunit that is found in the majority of synaptic GABA A Rs. In addition, we establish a potential link between the number of ␣1␥2-containing GABA A Rs and the number of ␣1-containing GABA A Rs at the cell surface. Finally, we show that like forskolin-stimulated PKA activity, activation of D 1 -like dopamine receptors also induces ICER expression and subsequent down-regulation of ␣1 subunit levels. These studies lay the foundation for identifying the genomic programs that link the CREB/CREM family of transcription factors to long term processes of synaptic inhibition and diseases of the nervous system.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfections-Primary rat neocortical cultures were prepared from E18 embryos (Charles River Breeding Laboratories) as described previously (37). Cultures were transfected 6 -8 days after dissociation using a modified calcium phosphate precipitation method (38).
Drug Treatments-Drugs were dissolved in dimethyl sulfoxide (Me 2 SO). Final vehicle concentration was 0.5% or less for all experiments. Drugs were diluted in 20 l of warm conditioned media and added to each dish. Cultures were treated with signaling inhibitors and returned to the incubator for 1 h. Cultures were then treated with signaling activators. Following activation, cultures were harvested at variable time intervals ranging from 15 min to 24 h after treatment. For treatments with the MAPK inhibitor (MEK1/MEK2), the following final concentration was utilized: U0126 (Calbiochem, 20 M) and PD98059 (Calbiochem, 50 M). The general PKC inhibitor staurosporine (Calbiochem, 500 nM), and a conventional PKC isoform inhibitor Gö6976 (Calbiochem, 1 M) were used. The PKA inhibitor H-89 was used at a final concentration of 10 M. Treatment with the signaling activators PMA (Sigma, 1 M) and FSK (Sigma, 20 M) was also utilized. Sister control dishes received vehicle (Me 2 SO) during the pretreatment and treatment phases. Cells were also treated for 6 h with 100 M dopamine (DA) in the presence and absence of the D 1 -like antagonist SCH23390 (10 M). Cells were pretreated with the antagonist for 1 h.
Reporter Assays-GABRA1 promoter fragments (Ϫ894/ϩ70) were cloned upstream of luciferase gene in pGL2 vector (Promega) and were a generous gift of the Farb laboratory. This promoter fragment confers full promoter activity in primary neocortical neurons (Farb laboratory; ϩ70 corresponds to exonic sequence 70 bp downstream from the first nucleotide of the first exon in the human gene). 5 The GABRA1 promoter/ reporter (GABRA1p/reporter) containing the 2-bp mutation at the CRE site (mCRE-GABRA1) was made by site-directed mutagenesis of the wild type-GABRA1p/reporter construct. The ICER expression vector was generated from the CREM cDNA. ICER II␥ cDNA was amplified from an IMAGE human cDNA clone (GenBank TM accession number BC090051, Open Biosystems, Huntsville, AL) using PCR. The sequences of the primers were as follows: ICER II␥ forward 5Ј-CGGGATCCA-TGGCTGTAACTGGAGATGACACAGCTGCCACTGGTG-ACATGCCAAC-3Ј; ICER II␥ reverse 5Ј-GCTCTAGACTAG-TAATCTGTTTTGGGAGAACAAATG-3Ј. The amplified fragments were cloned into the BamHI (5Ј end) and XbaI (3Ј end) sites of the expression vector pcDNA3 (Invitrogen). The construct was sequenced to confirm identity. A Western blot analysis was performed after overexpression of the ICER II␥ cDNA construct in HEK293 cells to confirm expression levels and identity. The dominant-negative M-CREB, K-CREB expression vector was a generous gift of the M. E. Greenberg laboratory. Wild type CREB was generated using M-CREB as a template for site-directed mutagenesis re-establishing the serine 133 phosphorylation site. Eight micrograms of total DNA was transfected into each well of a 6-well dish for studies that accessed reporter activity (GABRA1) after drug treatment. For coexpression studies, 8 g of reporter and 1 g of CREB, M-CREB, K-CRE, or 250 ng of ICER or same amount of control vectors (pRC or pcDNA3 empty vectors) were utilized for each well. 24 h after transfection, cells were assayed for luciferase (Promega) using the Victor 1420 detection system (E. G. Wallace). Luciferase counts were normalized to protein within each dish.
Real Time RT-PCR-PCR primers were designed using primer express software (PE Biosystems). Primer sets for ␣1 subunit and ICER were the following sequences: ␣1 subunit forward 5Ј-CCCCGGCTTGGCAACTA-3Ј; a1 subunit reverse 5Ј-CGGTTTTGTCTCAGGCTTGAC-3Ј; ␣1 subunit Taqman 5Ј-TGCTAAAAGTGCGACCATAGAACCG-AAAGA-3Ј. ICER-specific primers and probe were purchased from Applied Biosystems (Rn00569145_m1). RNA extraction from cells in culture was performed from single wells of a 6-well dish. Cells were harvested using RNeasy Micro RNA extraction kit (Qiagen). Control probes for relative abundance of rRNA or cyclophilin were used in multiplex assays. Reactions were performed using the ABI PRISM 7900HT. Thermocycling was done in a final volume of 20 l containing 40 ng of total RNA. Data were normalized to rRNA or cyclophilin expression.
Chromatin Immunoprecipitation (ChIP) Assay-ChIP was performed as described previously (39). Genomic DNA and protein complex were collected from primary cultured neurons after drug treatment. The DNA-protein complex was immunoprecipitated with an anti-CREB antibody or anti-pCREB antibody. Immunoprecipitated GABRA1 genomic DNA fragments were detected by real time PCR using specific primers that flank the CRE site in the GABRA1 gene. Primer sets for the CRE site were the following sequences: forward 5Ј-TGGTACCACCTTCCT-TTCTAAAATAAA-3Ј; reverse 5Ј-ATACGTCCCAGCGCA-AACC-3Ј; Taqman probe 5Ј-TCTCTCTGGCATGAAGTCA-CCGCCT-3Ј. Data were normalized as percentage of antibody/ input signal and expressed as percent change with respect to vehicle-treated cultures (defined as 100%). Input is the signal from the DNA preparation prior to precipitation. Quality of DNA sonication was verified using PCR primers that amplify a greater than 2-kb fragment of the GABRA1 promoter region. Input gDNA was rejected for precipitation studies if there was efficient amplification of large GABRA1 fragments. Amplification of irrelevant DNA regions was also used to affirm specificity of antibody precipitation.
siRNA Transfection-siRNAs corresponding to ICER sequence were designed using Ambion siRNA target finder , or PMA and inhibitors for the following 6 h. Cultures were then harvested, and total RNA was extracted. ␣1 subunit mRNA abundance was measured by real time RT-PCR. Data were normalized to rRNA expression. mRNA levels in each treatment group for ␣1 are expressed relative to vehicle (Me 2 SO). Significant changes are as indicated (*, p Ͻ 0.05; **, p Ͻ 0.01; mean Ϯ S.E.). B, PMA produces alteration in GABRA1p activity consistent with changes in endogenous GABA A R ␣1 mRNAs. Rat neocortical neurons were transfected with GABRA1p/reporter construct (Ϫ894/ϩ70). 24 h after transfection, cultures were pretreated with staurosporine as described above in A and harvested for luciferase assays. Results are expressed as percentage of activity measured from control (Me 2 SO). Significant changes are as indicated (*, p Ͻ 0.05; ***, p Ͻ 0.001; mean Ϯ S.E.; n ϭ 5). C, PMA increases ␣1 protein levels in primary cultured neocortical neurons. Cultures were treated with PMA (1 M) for 6 or 24 h. Whole cell extracts were resolved by SDS-PAGE and visualized using enhanced chemiluminescence. ␣1 subunit and ␤-actin levels were quantified by densitometry. Normalized data (␣1/␤actin) are presented as mean Ϯ S.E. and expressed as percent change with respect to vehicle-treated cultures (defined as 100%). Levels of ␤-actin did not vary with PMA treatment and were used as internal control. Significant changes are as indicated (*, p Ͻ 0.05; mean Ϯ S.E.; n ϭ 3).
software. The gene-specific sequences for knocking down ICER expression were targeted to ICER-specific sequences located at the 5Ј end. The siRNAs were chemically synthesized by Ambion. Control negative siRNAs were obtained from Ambion. The following sequences were used: ICER1 sense, 5Ј-CAUGGCUGUAACUGGAGAUUU and antisense, 5Ј-AUCUCCAGUUACAGCCAUGUU; ICER2 sense, 5Ј-CUGGAGAUGAAACUGAUGAUU and antisense, 5Ј-UCAUCAGUUUCAUCUCCAGUU. Transfections of siRNAs were carried out with RNAiFect transfection reagent (Qiagen). Primary cultured neurons were plated in 6-well plates as described above. 5 days after plating, siRNAs (1 g per well) were diluted with 100 l of buffer EC-R and formulated with 12 l of RNAiFect transfection reagent. After 15 min of incubation, the complexes were directly added onto the cells and incubated for 24 h under the normal growth conditions. The cultured media were replaced with Neurobasal medium. The second siRNA transfection was repeated. After additional overnight incubation, FSK was added to stimulate ICER induction, and further analysis was performed.
Nuclear Extract Preparation-Nuclear extracts were prepared as described previously. Neocortical cells in two 100-cm plates were harvested from the culture medium and washed 1ϫ with phosphate-buffered saline. The cell pellet was gently resuspended in 80 l of buffer A (10 mM Tris-Cl, pH 7.9, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 1 mM DTT, 1 mM PMSF, 1ϫ protease inhibitor mixture) and allowed to swell for 15 min, after which 25 l of a 10% Nonidet P-40 solution (Sigma) was added. The cell suspension was vortexed for 5 s and centrifuged at 12,000 ϫ g for 30 s to pellet nuclei. The supernatant was removed, and the nuclear pellet was resuspended in 120 l of buffer C (20 mM Tris-Cl, pH 7.9, 400 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM DTT, 1 mM PMSF, 1ϫ protease inhibitor mixture) and kept on a rocking platform for 1 h at 4°C. The nuclear debris was then removed by centrifugation at 12,000 ϫ g for 5 min at 4°C, and the supernatant transferred to a fresh tube. Glycerol was added at a final concentration of 20%.
DNA Pulldown Assay-DNA pulldown assay was performed as described previously (40). Oligonucleotide duplexes corresponding to the sequence of the CRE site in the ␣1 promoter and its flanking region were covalently linked to a biotin moiety at their 5Ј ends. Streptavidin Dynabeads (DynL A.S.) were washed three times in buffer A (5 mM Tris, pH 8.0, 0.5 mM EDTA, 1 M NaCl). Annealed oligonucleotides were incubated with beads (200 pmol/mg of beads) for 15 min at room temperature in buffer A. Beads were then washed twice with 500 l of buffer A and three times with 500 l of buffer C (20 mM Tris, pH 8.0, 1 mM EDTA, 10% glycerol, 1 mM DTT, 50 mM NaCl). 600 g of nuclear extract from neocortical neurons or from hippocampal tissues in buffer C were incubated with 1 mg of streptavidin Dynabeads bound to the appropriate oligonucleotides. After 15 min of incubation at room temperature, beads were washed three times with buffer C and eluted in 1 M NaCl. Eluted proteins were then analyzed by SDS-PAGE and followed by Western blot with anti-CREB and anti-CREM1 antibodies.
Co-immunoprecipitation-Co-immunoprecipitation experiments were carried out using the Pro-found mammalian coim-munoprecipitation kit form Pierce. Affinity-purified rabbit polyclonal antibodies (50 mg), raised against GABA A Rs ␥2 subunit (Alpha Diagnostics), were coupled to AminoLink Plus gels specified by the manufacturer. Whole cell proteins were extracted with M-PER (mammalian protein extraction reagent) plus protease inhibitor mixture (Roche Applied Science) one time and 1 mM phenylmethylsulfonyl fluoride. Whole cell extracts were centrifuged at 13,000 ϫ g for 30 min to remove cellular debris. Supernatants were incubated with the antibodycoupled gel overnight at 4°C with gentle rocking. Gels were washed five times with buffer, and complexes were recovered with 100 l of the elution buffer provided with the kit. Controls were performed by quenching the same gel before coupling antibodies to test for proteins that may bind nonspecifically to the gel. Approximately 35 l of the eluates were analyzed by Western blot as described above with anti-␣1 and ␥2 polyclonal antibodies. Total cellular proteins were analyzed using Western blot with polyclonal CREB and pCREB antibodies. Proteins were visualized using ECL following incubation with an anti-rabbit horseradish peroxidase-conjugated antibody. Normalized data (P-CREB/CREB) are presented as means Ϯ S.E. and expressed as percent change with respect to vehicle-treated cultures (defined as 100%). Levels of total CREB did not vary with PMA treatment and were used as an internal control. Significant changes are as indicated (***, p Ͻ 0.01; mean Ϯ S.E.; n ϭ 3).
Fluorescence Microscopy-Primary neocortical neurons were cotransfected with pDsRed2-Monomer vector (BD Biosciences) and CMV-ICER construct or control vector pcDNA3. Both ICER or control vector and pDsRed2-Monomer were used at a 1:1 ratio (2 g per 35-mm dish). At 48 h after transfection, cells were washed and fixed by using a standard protocol. Immunocytochemistry was performed by using an ␣1-specific antibody (Upstate) without permeabilizing cells at a 1:200 dilu-tion in 1% bovine serum albumin for 12 h at 4°C, and secondary antibody was conjugated to fluorescein isothiocyanate at a dilution of 1:50 (Jackson ImmunoResearch). DsRed transfected cells were viewed by using an Olympus IX 71 fluorescence microscope equipped with an UPIanApo 60x/0.9 water objective lens. Fluorescence was quantified by using IPLab software (Scanalytics, Inc.) and normalized to the total area of a selected cell. Changes in subunit expression were monitored by double-blind cell-selection procedure with 28 transfected neurons per condition (n ϭ cultures from three different animals).

RESULTS
␣1 Subunit Expression Is Increased after PMA Treatment via PKC Activation-As a first step toward understanding how GABRA1 transcription is regulated by extracellular signaling, primary cultured neurons from E18 embryonic cortex were exposed to PMA, a stimulator of both PKC and MAPK. A 6-h treatment with PMA (1 M) increases ␣1 subunit mRNA levels by 65% as compared with vehicle control when measured by real time RT-PCR (Fig. 1A). Such an increase was completely blocked by the broad spectrum PKC inhibitor, staurosporine (500 nM, 78% of control), and also by the conventional PKC isoform inhibitor Gö6976 (1 M, 114% of control). Increased levels were not blocked by the MEK (MAPK kinase) inhibitors, PD98059 and U0126 (Fig. 1A). To determine whether changes in ␣1 mRNA levels might be because of a transcriptional mechanism, we monitored the activity of a Ϫ894/ϩ70-bp fragment of the human GABRA1 promoter (GABRA1p) in transfected cortical neurons. GABRA1 promoter activity, as measured by the luciferase reporter assay, is up-regulated in response to PMA. 1 M PMA produces a 40% increase in GABRA1p/reporter activity that is dependent upon PKC activation (as determined by specific blockade with the PKC inhibitor, staurosporine) (Fig. 1B). Consistent with changes in ␣1 mRNA levels, levels of ␣1 subunit protein are increased by almost 2-fold 24 h after treatment with PMA (Fig. 1C). Taken together, all observations suggest that neurons possess a mechanism to up-regulate ␣1 subunit levels via activation of PKC and not MAPK.
Activation of CREB after PMA Treatment Is PKC-and Not MAPK-dependent-Given that ␣1 subunit up-regulation is dependent upon PKC and not MAPK, and that activity of GABRA1p is also increased in response to PKC stimulation, we sought to identify the critical transcription factors that mediate this process. Based on the presence of an atypical CRE site in GABRA1 and the fact that CREB is known to be activated by PKC (41)(42)(43), we asked whether CREB might play a role in PMA-induced ␣1 subunit up-regulation. Stimulation of cortical neurons with PMA leads to a time-dependent activation of CREB that reaches a peak level of ϳ3-fold increase when compared with control at 15 min ( Fig. 2A). Like ␣1 subunit mRNA levels, pretreatment with the MEK inhibitor PD98059 fails to block PMA-induced CREB activation ( Fig. 2A, upper panel). To test for the activity of MAPK inhibitors, we also measured MAPK activation of ERK as a control in our studies. As expected, PMA treatment increases MAPK activation in a timedependent manner, and PD98059 inhibits the increase ( Fig. 2A, FIGURE 3. CREB-mediated regulation of GABRA1. A, identity of oligonucleotide sequences (sense strand) used in this study. All 5Ј-nucleotides were biotinylated. The CRE site in the ␣1 promoter is highlighted in blue; and mutations with red. B, CREB binds directly to the CRE site in the GABRA1 promoter as measured by DNA pulldown assay and Western blot. Nuclear extract was obtained from primary cultured neocortical neurons with or without PMA treatment. 600 g of nuclear extract was incubated with 1 mg of Dynabeads Streptavidin-bound to the appropriate oligonucleotides. After 15 min of incubation, beads were washed three times. Eluted proteins were then analyzed by SDS-PAGE and followed by Western blot with anti-CREB antibodies. C, PMA treatment increases association of pCREB with endogenous GABRA1 as measured by ChIP. Primary neocortical neurons were pretreated with the PKC inhibitor staurosporine (500 nM) or vehicle (Me 2 SO) for 1 h. Cultures were then treated with either vehicle (Me 2 SO) or PMA (1 M) for 30 min. Cross-linked genomic DNA and protein complex were collected from primary cultured neurons after drug treatment. CREB-immunoprecipitated GABRA1 genomic DNA fragments were detected by real time PCR using specific primers that flank the CRE site in the GABRA1 gene. (**, p Ͻ 0.001; mean Ϯ S.E.; n ϭ 3.) D, up-regulation of WT-CREB increases GABRA1 promoter activity in transfected primary cultured neurons. WT-CREB and dominant-negative M-CREB and K-CREB expression vectors were cotransfected with GABRA1 promoter/reporter constructs. 24 h after transfection, cultures were harvested for luciferase assays. Results are expressed as a percentage of activity measured from control (cotransfected control Prc vector (without transgene) and GABA A R ␣1p/reporter construct), defined as 100%. (***, p Ͻ 0.001; mean Ϯ S.E.) lower panel). These results confirm that PMA-induced activation of CREB occurs through a MAPK-independent pathway.
To confirm PKC-dependent activation of CREB, we used the PKC inhibitor staurosporine to block PMA-induced CREB phosphorylation (Fig. 2B). Staurosporine also blocks PMA-induced MAPK activation supporting previous observations in the literature that PKC activates ERK as well as its immediate upstream activator MEK (44). Taken together, these data indicate that PMA activates both CREB and MAPK through a PKCdependent pathway. Most importantly, the activation of CREB by PKC is independent of the PKC activation of MAPK, paralleling our PKC-specific change in GABRA1 expression as reported in Fig. 1.
CREB Regulates GABRA1 Gene Expression through PKC Activation-Because there is a CRE in the proximal GABRA1 promoter region (Fig. 3A), we asked whether PMA-induced increase in GABRA1 expression is dependent upon CREB recognition of the proximal promoter region. DNA pulldown assays were used to detect whether PMA treatment enriches binding of CREB to the GABRA1p CRE-binding site. Oligonucleotide duplexes corresponding to the sequence of the CRE site in GABRA1p and its flanking region, or three copies of a consensus CRE site (Fig. 3A), were covalently linked to a biotin moiety at their 5Ј ends, and such oligonucleotides and their nuclear binding proteins were recovered using magnetic streptavidin Dynabeads. Nuclear binding proteins were prepared from cultured neocortical neurons before and after PMA treatment. Beads were collected and washed, and the bound proteins were eluted, separated by SDS-PAGE, and then analyzed by Western blot. As shown in Fig. 3B, 2nd and 3rd lanes, CREB binds to the CRE site in GABRA1p with or without PMA treatment. Moreover, binding is sequence-specific because the presence of a 2-bp mutation in the CRE site removes CREB recognition (Fig. 3B, 4th, 5th, and 7th lanes). CREB binding also occurs at the control consensus oligonucleotides that contain three copies of a classical CRE site.
To determine whether activated CREB is bound to the endogenous GABRA1 in primary cortical neurons, ChIPs were performed. The endogenous GABRA1p DNA-protein complex was immunoprecipitated with either anti-CREB or anti-phospho-CREB (Ser-133) antibodies after a 30-min 1 M PMA treatment, and precipitated GABRA1 genomic fragments were detected by real time PCR. Although binding of CREB to the endogenous GABRA1p is detected by ChIP, levels of total CREB do not change with PMA stimulation (data not shown). However, when DNA is precipitated with the phospho-CREB (pCREB) antibody, there is an ϳ2-fold increase of pCREB association with the promoter suggesting that PMA increases CREB phosphorylation while bound to GABRA1p, consistent with the classical model of CREB-mediated gene regulation. Increase in pCREB association at GABRA1p is also blocked by the PKC inhibitor staurosporine (Fig. 3C).
To further investigate if increased association of pCREB is necessary for GABRA1 up-regulation, wild type CREB (WT-CREB) and serine 133 mutated dominant-negative CREB (M-CREB) were cotransfected with the human GABRA1p/reporter construct (894/ϩ70, see "Experimental Procedures") into neocortical neurons in culture, and promoter activity was measured 24 h after transfection. Overexpression of WT-CREB increases GABRA1p activity ϳ2-fold in transfected primary cultured neurons; however, overexpression of M-CREB is without effect (Fig. 3D). These data strongly suggest that although both CREB and activated CREB bind to the GABRA1 promoter, only the activated CREB mediates PMA-induced GABRA1 expression.
FSK Stimulation Decreases ␣1 Subunit Expression-It is well known that multiple pathways can lead to CREB activation and the regulation of gene expression (41,42,45). Evidence has shown that CREB phosphorylation occurs through the activation of distinct signaling pathways that differentially regulate the expression of multiple target genes (42). To investigate whether CREB phosphorylation via an additional signaling pathway also regulates GABRA1 expression, forskolin (FSK) was used to activate PKA in primary cortical neurons. Although FSK treatment leads to robust and long lasting CREB activation (data not shown), unlike PMA, a 20 M FSK treatment decreases ␣1 subunit mRNA levels in a time-dependent manner (Fig. 4A). 1 h of FSK treatment decreases ␣1 mRNA levels by ϳ30%. This decrease peaks at 2-4 h of treatment (ϳ50%). To determine whether decreased mRNA levels reflect a change in GABRA1 transcription, functional promoter/reporter assays were used to confirm that 4 h of FSK (20 M) decreases GABRA1p activity by ϳ40% (Fig. 4B). The decrease of ␣1 mRNA levels and GABRA1p activity are paralleled by changes in ␣1 protein levels. As shown in Fig. 4C, ␣1 subunit levels show no change after 4 h of FSK but decrease by 32% at 6 h and further decrease by 51% after 24 h. A representative Western blot is shown in Fig. 4C, upper panel.
Activation of cAMP Pathway Induces ICER Expression-As shown in Figs. 2 and 4, PMA and FSK both activate CREB, but ␣1 mRNA levels differ in polarity (PMA increases levels although FSK decreases levels). One possible explanation is that phosphorylated CREB recruits different binding partners via activation of distinct signaling pathways. Like all bZIP transcription factors, CREB family members contain a C-terminal bZIP domain that mediates DNA binding and a leucine zipper domain that facilitates dimerization. One feature common to all bZIP transcription factors is the high degree of homology in the DNA binding domain (DBD) of these proteins. Different members of the family can form homo-and heterodimers that bind to similar DNA-binding sites, CRE or CRElike elements (46,47). In particular, CREB/ICER heterodimers form in a cell-and signal-specific manner (34,35). To determine whether induction of ICER might occur in response to FSK treatment and contribute to GABRA1 repression, RT-PCR was used to monitor the presence of ICER mRNAs after selective drug treatment. As shown in Fig.  5A, ICER mRNA is not induced by PMA, but after a 1-h exposure to FSK, ICER mRNA levels rise reaching a peak at ϳ4 h and then decreasing at 6 h.
ICER gene products are remarkably small proteins of either 108 or 120 amino acids with the predicted sizes of 12 and 13.5 kDa, excluding or including the ␥ domain, respectively (ICER␥ and ICER). To analyze the presence of ICER proteins, Western blot analysis was performed to analyze extracts of FSKtreated primary cortical neurons. FSK induces ICER proteins with kinetics similar to ICER transcripts, rapid and transient, a characteristic of immediate early genes (Fig. 5B). In contrast, ICER is not induced after 4 h of PMA treatment of neocortical neurons (Fig. 5B, 8th lane). The time course for ICER induction in response to FSK is similar to the time course for ␣1 mRNA down-regulation, suggesting that ICER may be a repressor of GABRA1 transcription.
To investigate whether a physiological signal that activates the cAMP/PKA pathway would also regulate ␣1 gene expression, neocortical neurons were treated with dopamine (DA, 100 M), and ICER and ␣1 mRNA levels were detected using real time RT-PCR. Four-hour DA treatment dramatically increases ICER mRNA levels (4.7-fold; Fig. 5C). This increase precedes the decrease in the levels of ␣1 mRNAs and is attenuated by cotreatment with the D 1 -like receptor antagonist SCH23390 and the PKA inhibitor H-89, suggesting that D 1 receptor-mediated activation of the cAMP/PKA pathway is responsible for induction of ICER synthesis. As shown in Fig. 5D, 6-h DA treatment decreases ␣1 mRNA levels by 33%. Fig. 3, B and C, CREB can bind to the CRE site in GABRA1. Activation of the PKC pathway increases pCREB association with GABRA1 through the CRE site. To investigate whether ICER and CREB can both bind to the CRE site in GABRA1 after FSK treatment, DNA pulldown assays were performed as described for Fig. 5. Nuclear extracts prepared from cultured neocortical neurons, either treated with or without FSK for 4 h, are incubated with beads to which ␣1 CRE-containing oligonucleotide duplexes have been attached. After incubation with lysates, bead-purified CRE-binding proteins were separated by SDS-PAGE and then analyzed by Western blot using both CREB and ICER antibodies. Consistent with the results of Fig. 3B, CREB binds to the GABRA1-CRE site before FSK treatment (Fig. 6A, 1st lane,  upper panel). Presence of CREB at the CRE site is not altered after a 4-h FSK treatment (Fig. 6A, 2nd lane, upper panel). As ICER is not expressed in neocortical neurons under control conditions (Fig. 5), ICER is absent at the GABRA1-CRE after binding of control nuclear extracts (Fig. 6A, 1st lane, lower  panel). In contrast, after 4 h of FSK stimulation, there is robust binding of ICER to the GABRA1-CRE (Fig. 6A, 2nd lane, lower  panel).

PKA-induced ICER Synthesis Is Responsible for GABRA1 Transcriptional Repression-As shown in
Whether ICER induction is responsible for the decrease of ␣1 mRNA levels after FSK treatment still remains to be identified. To begin to address this possibility, we performed promoter assays in neocortical neurons in culture in the presence and absence of overexpressed CREB and ICER along with GABRA1p/reporter constructs. Similar molar amounts of CREB and ICER were transfected in each promoter study. As shown in Fig. 6B, CREB overexpression alone increases GABRA1 activity, whereas overexpression of ICER alone or in combination with CREB reduces GABRA1p activity.
We next determined whether ICER induction is PKA-dependent consistent with FSK-induced down-regulation of endogenous ␣1 subunit expression. The PKA inhibitor H-89 (10 M) partially blocks FSK-induced ICER expression (53% blockade, Fig. 6C) as well as CREB phosphorylation (Fig. 6D ) by activation of the PKA pathway in primary neocortical neurons. Because it has been reported that FSK can also activate the MAPK pathway (48), the MEK inhibitor U0126 was also used to determine whether an additional signaling pathway contributes to FSK-induced pCREB and ICER synthesis. In contrast to H-89, pretreatment with U0126 (10 M) has no effect on FSKinduced ICER expression, suggesting that the induction of ICER under these conditions is MAPK-independent (Fig. 6C,  4th lane). The MAPK pathway may, however, be involved in FSK-induced CREB phosphorylation. As shown in Fig. 6D, 3rd  and 4th  These results suggest that PKA and MAPK pathways synergistically contribute to FSK-induced pCREB levels. Such an effect was not observed in FSK-induced ICER synthesis (Fig. 6C, 3rd  and 5th lanes) further supporting the conclusion that FSK induction of ICER is MAPK-independent.
Although H-89 partially inhibits ICER expression, it fails to reverse the down-regulation of ␣1 subunit mRNA levels that is induced by FSK (Fig. 7A). This result might occur if activation of PKA is also necessary for maintaining basal levels of ␣1 sub- FIGURE 6. PKA activation of CREB and ICER differentially regulates ␣1 promoter activity in neocortical neurons. A, CREB and ICER both bind to the CRE site in GABRA1p as measured by DNA pulldown assay and Western blot. Nuclear extract was prepared from primary cultured neocortical neurons with or without 4 h FSK. 600 g of nuclear extract was incubated with 1 mg of Dynabeads Streptavidin-bound to A1 CRE oligonucleotides as described under "Experimental Procedures." Eluted proteins were analyzed by Western blot with anti-CREB and anti-CREM antibodies. A representative gel is shown. Experiment was performed three times with similar results. B, ICER overexpression or CREB and ICER coexpression decreases GABRA1p activity. CREB and ICER expression vectors were cotransfected with GABRA1p/reporter constructs. 24 h after transfection, cultures were harvested for luciferase assays. units. In this scenario, inhibiting PKA may inhibit the function of a yet to be identified critical activator. To further determine the importance of ICER for FSK-induced down-regulation of GABRA1 transcription, two pairs of ICER siRNAs were transfected into primary cultured neurons prior to FSK treatment. Transfection of ICER siRNAs was found to successfully inhibit ICER induction (ϳ50%) after a 4-h FSK treatment (Fig. 7B). The percentage of inhibition is consistent with the transfectional efficiency of siRNAs in the assay. Most importantly, specific inhibition of ICER induction by siRNAs attenuates the decrease of ␣1 mRNA levels induced by FSK. ␣1 subunit mRNA levels decrease by 48% in scramble siRNA-transfected neurons after 4 h of FSK treatment, similar to the decrease in 4-h FSK-treated cultures in the absence of siRNA transfection (Fig. 5A). In ICER siRNA-transfected neurons, 4-h FSK treatment causes an increase in ␣1 mRNA levels to 69 and 71% of control levels, again consistent with transfectional efficiency of siRNAs in these studies (30 -50%) (Fig. 7, B and C, see insets). Taken together, our results strongly suggest that induction of ICER is responsible for PKA-induced down-regulation of GABRA1 transcription.
The CRE Site in GABRA1 Is Responsible for Bi-directional Regulation of ␣1 Subunit Expression-To determine whether CREB and ICER recognition of the GABRA1-CRE site is necessary for their regulation of ␣1 promoter activity in cultured cortical neurons, a mutant form of CREB (K-CREB), which does not bind DNA but can dimerize to endogenous CREB, was cotransfected with GABRA1p/reporter constructs. Unlike wild type CREB, overexpression of K-CREB fails to increase GABRA1p activity (Fig. 3D, 4th column). Interestingly, a 2-bp mutation in the GABRA1-CRE site (Fig. 8A) does not change promoter activity in transfected neurons suggesting that CREB is not responsible for basal levels of ␣1 gene expression (Fig. 8B). However, the PMA-induced increase in GABRA1 promoter activity is abolished in promoter constructs that contain the CRE mutation. Similar to the PMA dependence on an intact CRE site, FSK down-regulation of ␣1 promoter activity is also dependent on the CRE site (Fig. 8B). Finally, the role of the CRE site in GABRA1p was determined using overexpression of CREB/ICER constructs to show loss of CRE-mediated bidirectional regulation (Fig.  8C) and determined directly using a DNA pulldown purification assay with nuclear extracts from treated and untreated primary cultured neocortical neurons. CREB binding to the GABRA1-CRE site is lost when the oligonucleotides contain a 2-bp mutation (Fig.  3B, 4th, 5th, and 7th lanes). Taken together, these results show that CREB regulates GABRA1 gene expression through the CRE site of the GABRA1p and that binding of CREB and/or CREB and ICER is necessary for bi-directional changes in GABRA1 transcription.
ICER Induction Is Associated with Decrease in ␣1␥2-Containing GABA A Rs and Decrease in ␣1 Subunits Detected at the Cell Surface-The majority of GABA A R subtypes that contain ␣1 subunits are composed of ␣1␤2␥2 (49). To determine whether the decreased expression of GABRA1 after FSK treatment may result in an alteration in the subunit composition of GABA A Rs, we performed coimmunoprecipitation of GABA A Rs. Neocortical neurons were treated with FSK or vehicle for 24 h, and whole cell protein extracts were prepared and then immunoprecipitated with anti-␥2 subunit antibodies that were chemically linked to AminoLink Plus gel. After several stringent washes, the precipitate was eluted and separated by SDS-PAGE. Western blot was performed using anti-␣1 or -␥2 antibodies. The ratio of ␣1 to ␥2 subunits was used to reflect the abundance of ␣1 subunits in ␥2 containing GABA A Rs. 24 h of FSK treatment decreases ϳ40% the ratio of ␣1 to ␥2 subunits compared with vehicletreated cultures (Fig. 9A). This decrease is consistent with ; n ϭ 6.) C, ICER siRNA treatment attenuates FSK-induced decreases in ␣1 mRNA levels (**, p Ͻ 0.01; ***, p Ͻ 0.001; mean Ϯ S.E.; n ϭ 6.) ICER siRNAs (ICER1 or ICER2) and scramble siRNAs (control) were transfected into primary cultured neocortical neurons prior to FSK treatment. After 4 h of FSK, RNA was isolated and mRNA levels of ICER (B) or ␣1 (C) were determined by real time RT-PCR. Data were normalized to cyclophilin mRNA levels. B and C, inset histograms display the relationship between % reduction in ICER and % recovery of ␣1 mRNA levels in the presence of ICER siRNAs after treatment with FSK. Note that % recovery of ␣1 mRNAs is consistent with % reduction of ICER mRNAs after siRNA treatment.
the decrease in expression of ␣1 subunits after FSK stimulation. ICER induction as mimicked by overexpression in individual neurons was also shown to regulate the levels of endogenous ␣1 subunits to the same degree as FSK-induced down-regulation when assayed using nonpermeabilized transfected neurons (Fig. 9B). This result strongly suggests that ICER can regulate the number of ␣1-containing GABA A Rs at the plasma membrane.

DISCUSSION
We have shown that two distinct signal transduction pathways regulate GABRA1 through CREB binding and activa-tion. It is known that CREB can be phosphorylated through a PKC-dependent pathway (41). Results of our studies show that PMA up-regulates GABRA1 expression by activating CREB via the PKC pathway. Although PKC can directly phosphorylate CREB in vitro at multiple sites, including Ser-133 (43), most of the results in the literature suggest that CREB phosphorylation occurs downstream of PKC after activation of the ERK pathway (41,44,48). Because PMA, in addition to PKC stimulation, also activates the MAPK pathway, we used MEK (MAPK kinase) inhibitors in our studies to determine whether PMA-induced CREB phosphorylation and GABRA1 up-regulation are MAPK-dependent, as suggested by studies in the literature. Results of our studies suggest that activation of PKC up-regulates GABRA1 expression in a MAPK/ERK-independent manner that is mediated by CREB phosphorylation. Gö6976, an inhibitor of the conventional PKC isoforms, blocks both CREB phosphorylation and PMA-induced up-regulation of GABRA1, whereas MAPK inhibitors do not. These results when taken together suggest that PMA may directly activate the conventional form of PKC to phosphorylate CREB. Recent findings that PKC␣, one isoform of conventional PKCs, can translocate to the nucleus and directly phosphorylate CREB support our observations of a direct mechanism (50).
Increased GABA A R ␣1 subunit mRNA levels in response to PMA treatment are accompanied by increased binding of pCREB to the endogenous GABRA1 promoter in cultured neocortical neurons as measured by chromatin precipitation. Overexpression of WT-CREB increases GABRA1 promoter activity, whereas a Ser-133 mutated dominant-negative form of CREB (M-CREB) is without effect. Taken together, these findings suggest that CREB is a stimulus-inducible transcriptional activator controlled by phosphorylation that up-regulates endogenous GABRA1 transcription.
In addition to PKC activation, we show that activation of the cAMP pathway, using FSK, decreases ␣1 subunit mRNA levels in neocortical cultures in the presence of pCREB. One feature common to bZIP transcription factors is the high degree of homology in their DBDs. Different members of this family form homo-and/or heterodimers binding to similar DNA recognition sites. It is known that when a CRE is occupied by a homodimer or heterodimer specific to the activator isoforms, transcription of a target gene is augmented. However, when the same DNA site is occupied by a heterodimer containing an activator and a repressor isoform or by repressor homodimers, transcription is repressed, presumably because of impaired interaction with CBP or TFIID (46,47). Control over the occupancy of the CRE site by activating or repressing dimers is believed dependent upon the relative abundance and affinities of the three classes of dimers for the target CRE.
We show that ICER, a powerful repressor among the CREB family of transcription factors, is induced in primary cultured neocortical neurons through the activation of the cAMP/PKA pathway by forskolin and, most importantly, by the neurotransmitter DA (Fig. 5, C and D). The fact that DA induces ICER strongly suggests that the activity of endogenous neurotransmitter receptor systems use the CREB signaling pathway to regulate inhibition in the brain. Given the important role of DA in Results are expressed as percentage of activity measured from wild type GABRA1p. Data are shown as mean Ϯ S.E., n ϭ 5. **, p Ͻ 0.01. C, cotransfection of WT-CREB or ICER with mCRE-promoter has no effect on promoter activity. WT-CREB and/or ICER were cotransfected with wild type or CRE-mutated GABRA1p/ reporter constructs. 24 h after transfection, cultures were harvested, and luciferase assays were performed. Results are expressed as percentage of activity measured from control (cotransfection of empty Prc and Pcneo vectors and mCRE-promoter/reporter constructs), defined as 100%. (Data are shown as mean Ϯ S.E., n ϭ 5.) **, p Ͻ 0.01. multiple brain structures and its potential relationship to multiple disorders, including Parkinson disease and drug abuse, it will be important in the future to determine whether ICER induction is a key contributor to altered inhibitory processes of the injured brain.
Once ICER is induced, like other bZIP transcription factors, it forms either homodimers or heterodimers with other CREB/CREM isoforms and inhibits target gene expression (34,35,51,52). Our results show that activation of the cAMP/PKA pathway causes both CREB phosphorylation and ICER induction. We detected CREB along with ICER at the GABRA1-CRE upon activation of the cAMP pathway. Overexpression of ICER alone or with CREB represses GABRA1 promoter activity consistent with our understanding that repressor homodimers, or activator and repressor heterodimers, repress target gene expression in other systems. In addition, inhibition of ICER induction using cognate ICER siRNAs attenuates the decrease in ␣1 mRNA levels in response to PKA activation, providing the first direct evidence that ICER mediates down-regulation of GABRA1.
In our studies, although both CREB and ICER bind to the CRE site of GABRA1 as monitored using DNA pulldown assay, it is difficult to determine whether ICER homodimers, or ICER and CREB heterodimers, or both, drive down-regulation of ␣1 expression in vivo. Based on the literature, when CREB and ICER are coexpressed, both ICER homodimers and ICER/CREB heterodimers have been detected at various CRE sites (35,51,53). The precise sequence of the CRE site has been shown to affect the affinity of CREB and its family members for its DNA element (54,55). Given that the ␣1 CRE site is a noncanonical CRE, it is reasonable to expect that one form of the repressing dimers might have a higher affinity for the asymmetric CRE site. Further studies will be needed to determine whether there is, in fact, a preferential ICER dimer form that recognizes the GABRA1 in response to FSK. In contrast, PMA stimulation fails to induce ICER; therefore, activated CREB most likely forms homodimers to up-regulate GABRA1 transcription.
The GABA A R ␣1 subunit is the most abundant ␣ subunit variant in the brain. At least half of the GABA A Rs are believed to contain the ␣1 subunit that is highly expressed throughout most brain regions (56 -58). The majority of GABA A R subtypes containing the ␣1 subunit are found with a ␤2 and ␥2 subunit, ␣1␤2␥2 (49), at the synapse. In vitro studies have suggested that ␣1 subunit expression confers specific pharmacological properties to the GABA A R. Recombinant expression studies indicate that the presence of the ␣1 subunit contributes to the efficacy of GABA agonists (59) and the maximal response to BZs (60). Knockdown of ␣1 subunit expression using antisense deoxyoligonucleotides results in a decrease in GABA-mediated chloride flux (61). Patch clamp recordings show a reduction in the amplitude of evoked inhibitory postsynaptic currents in slices from visual cortex after treatment with ␣1 antisense deoxyoligonucleotides (62). In addition, a decrease in ␣1 subunit expression has been associated with a decrease in the E max of muscimol-stimulated chloride flux after chronic ethanol administration (19), although there is no change in total receptor number (63).
Recently, two mouse lines in which the ␣1 subunit has been genetically deleted have been developed and characterized FIGURE 9. Effects of ICER induction on receptor levels and cell surface expression. A, FSK stimulation of primary cultured neocortical neurons decreases the abundance of ␣1␥2 containing GABR A Rs. Cultures were treated with either vehicle (Me 2 SO) or FSK (20 M) for 24 h. Total cellular proteins were extracted and applied to ␥2 subunit antibody-coupled AminoLink Plus gel. After overnight incubation, the ␥2 subunit antibody-coupled protein complexes were eluted and separated by SDS-PAGE and then followed by Western blot with anti-␣1 and ␥2 subunit antibodies. Controls were performed by quenching the gel before coupling antibodies to test for proteins that may bind nonspecifically to the gel matrix. ␣1 and ␥2 subunit levels were quantified by densitometry. Normalized data (␣1 subunit/␥2 subunit) are presented as means Ϯ S.E. and expressed as percent change with respect to vehicle (VEH)-treated cultures (defined as 100%). A representative Western blot shows decreases in ␣1 subunit abundance after ␥2 subunit immunoprecipitation in FSK-treated cultures (upper); quantification of data is shown in the lower histogram. (*, p Ͻ 0.05; mean Ϯ S.E.; n ϭ 3.) B, overexpression of ICER decreases the endogenous levels of ␣l subunit detected at the cell membrane. Primary cultured neocortical neurons were cotransfected with pDsRed2-Monomer and ICER expression (CMV-ICER) or control vectors (CMV-empty). At 48 h after transfection, unpermeabilized cells were fixed and stained with an ␣1-specific antibody using a standard protocol. The DsRed transfected cells were viewed by using a Olympus IX71 fluorescence microscope, and the images were analyzed by using IPLab software (see "Experimental Procedures"). Representative images are shown (empty vector, upper panel; ICER construct, lower panel). Quantitation data are presented in the histogram (***, p Ͻ 0.01; mean Ϯ S.E.; n ϭ 3). (64 -66). Both lines exhibit more than a 50% loss of GABA A /BZ receptor number, which is consistent with previous estimates of the percentage of ␣1 subunit-containing receptors (56,67). In ␣1 Ϫ/Ϫ mice, the initial electrophysiological studies revealed loss of the developmental shortening of spontaneous inhibitory postsynaptic currents and diminished miniature inhibitory postsynaptic currents in stellate cells of the cerebellum (66); moreover, the potency and maximal efficacy of muscimol-stimulated 36 Cl Ϫ uptake in cerebral cortical synaptoneurosomes was reduced (65,68). Electrophysiological analyses of GABA A Rs in Purkinje neurons of ␣1 Ϫ/Ϫ mice also showed dramatically reduced GABA currents and zolpidem potentiation (64). Furthermore, knock-out mice exhibit increased bicuculline-induced seizure susceptibility compared with wild type (65).
In our studies, we show the degree of decrease in expression of the ␣1 subunit through cAMP/PKA signaling is consistent with the decrease in abundance of the ␣l␥2-containing GABA A Rs in neocortical cultures. Given that the majority of ␥2-containing GABA A Rs are synaptic and that ␥2 is required for synaptic trafficking of these receptors, decreases in ␣1 subunit expression may lead to a change in the number or kind of GABA A Rs at the synapse. ␣1␥2-containing GABA A Rs are thought to be type I BZ receptors that have high affinity for zolpidem, CL218.872, and some ␤-carboline derivatives (59, 69 -71). In addition, low levels of ␣1 associated with ␥2-containing GABA A Rs result in high sensitivity to zinc blockade, as is seen in dentate granule cells of epileptic brain (18). As a first step to determine whether a change in ␣1 subunit expression as mediated by ICER induction is relevant to a change in processes of inhibition that may be active in disease, we determined that up-regulation of ICER in individual neurons down-regulates the amount of ␣1-containing GABA A Rs that can be detected at the cell surface membrane (Fig. 9B).
Taken together, our results are summarized as a working model (Fig. 10); CREB activation via the PKC pathway leads to the formation of CREB homodimers that bind to the GABRA1p-CRE site to increase transcription. In contrast, CREB activation via the PKA pathway, along with PKA-induced ICER synthesis, leads to the formation of ICER homodimers or CREB/ICER heterodimers to repress transcription. The discovery that ␣1 subunit levels can be differentially regulated by CREB/CREM family members uncovers a new mechanism of GABRA1 gene expression that may be operative in neurological diseases (such as temporal lobe epilepsy, alcohol dependence/ withdrawal, and stress) and may provide novel opportunities for therapeutic intervention.