Address correspondence and reprint requests to Dr Severn B. Churn, Department of Neurology, Virginia Commonwealth University, PO Box 980599 MCV Station, Richmond, VA 23298, USA. E-mail: SCHURN@HSC.VCU.EDU
γ-Aminobutyric acid (GABA) is the primary neurotransmitter that is responsible for the fast inhibitory synaptic transmission in the central nervous system. A major post-translational mechanism that can rapidly regulate GABAAR function is receptor phosphorylation. This study was designed to test the effect of endogenous calcium and calmodulin-dependent kinase II (CaM kinase II) activation on both allosteric modulator binding and GABAA receptor subunit phosphorylation. Endogenous CaM kinase II activity was stimulated, and GABAA receptors were subsequently analyzed for bothallosteric modulator binding properties and immunoprecipitated and analyzed for subunit phosphorylation levels. A significant increase in allosteric-modulator binding of the GABAAR was observed under conditions maximal for CaM kinase II activation. In addition, CaM kinase II activation resulted in a direct increase in phosphorylation of the GABAA receptor α1 subunit. The data suggest that the CaM kinase II-dependent phosphorylation of the GABAA receptor α1 subunit modulated allosteric modulator binding to the GABAA receptor.
Several studies have suggested a role for calcium and calmodulin-dependent kinase II (EC 220.127.116.11, CaM kinase II) in modulating GABAAR function. Machu et al. (1993) demonstrated that CaM kinase II will phosphorylate synthetic peptides corresponding to the amino acid sequence of the intracellular loop of the γ2L subunit of the GABAAR. Our laboratory (Churn et al. 2000) and other investigators (Wang et al. 1995) have shown that injection of thiophosphorylated CaM kinase II enhances Cl– current in acutely isolated spinal cord neurons and hippocampal neurons in culture. Additional studies by Aguayo et al. (1998) using patch-clamp techniques in cortical neuronal cultures showed that increasing intracellular calcium levels caused a transient augmentation of the GABAAR Cl– current. This augmentation could be blocked by KN-62, a CaM kinase II inhibitor, indicating the specific actions of calmodulin-dependent phosphorylation in neuronal excitability.
Endogenous CaM kinase II modulation of GABAAR function was first demonstrated by studies showing that CaM kinase II-dependent phosphorylation augmented agonist binding in rat forebrain tissue (Churn and DeLorenzo 1998b). However, the effect of CaM kinase II phosphorylation on specific GABAAR subunits has not been fully characterized. This study was designed to determine the effect of endogenous CaM kinase II activation on benzodiazepine binding to GABAARs and on the phosphorylation of the GABAAR α1 subunit. The GABAAR α1 subunit was chosen due to its high level of expression in the adult brain (Fritschy and Möhler 1995) and its importance for benzodiazepine modulation of GABAAR activation (Dunn et al. 1999; Rudolph et al. 1999; Möhler et al. 2001). The findings demonstrate that CaM kinase II activation results in both positive regulation of allosteric modulator binding and increased phosphate incorporation into at least the GABAAR α1 subunit.
Materials and methods
All materials were reagent grade and purchased from Sigma Chemical Company (St Louis, MO, USA) unless otherwise stated. Isotopes [3H]flunitrazepam (85 Ci/mmol) and [γ-32P]ATP(800Ci/mmol) were purchased from NEN-DuPont (Boston, MA, USA) or ICN Biomedicals, Inc. (Costa Mesa, CA, USA). Adult male Sprague–Dawley rats (150–250 g) were purchased from Harlan Laboratories (Indianapolis, IN, USA). Rabbit GABAAR α1N-terminus subunit-specific antibody (#9I-2B) was supplied by Dr Angel DeBlas (Department of Physiology and Neurobiology, University of Connecticut, Storrs, CT, USA). A second source ofGABAAR subunit antibody, a goat-polyclonal GABAAR α1 N-terminus subunit-specific antibody, protein A agarose, and normal immunoglobulin subclass G (IgG) were purchased from Santa Cruz Laboratories (Santa Cruz, CA, USA). KN-93 was purchased from Calbiochem (La Jolla, CA, USA). IPG strips, Mini Protean II Electrophoresis apparatus, Protean IEF Cell apparatus, Trans Blot apparatus, carrier ampholytes (Biolytes 3–10), sodium dodecyl sulfate (SDS), polyacrylamide, Precision Prestained Broad Range Protein Standards, and Silver Stain Plus kit were purchased from Bio-Rad Laboratories (Richmond, CA, USA). The Vectastain ABC Elite kit was purchased from Vector (Burlingame, CA, USA). Slide-A-Lyzer dialysis membranes were purchased from Pierce (Rockford, IL, USA).
All animal use procedures were in strict accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the Virginia Commonwealth University Institutional Animal Care and Use Committee. Rat forebrains were rapidly removed and homogenized into iced-cold buffer containing 100 mm tris-citrate (pH 7.4), 10 mm EGTA, 10 mm EDTA, 320 mm sucrose, 1 mm dithiothreitol (DTT), 0.3 mm phenylmethylsulfonyl fluoride (PMSF), and 5 nm cypermethrin, a calcineurin inhibitor, as previously described in detail (Churn and DeLorenzo 1998b). Homogenates were subjected to differential centrifugation; 5000 g for 10 min followed by 18 000 g for 30 min. The resultant fraction, P2, was osmotically shocked for 30 min by resuspension into iced-cold deionized H2O to remove endogenous cell constituents. The membrane fraction was then washed by repeated (2 ×) centrifugation and resuspension steps into 50 mm Tris-HCl buffer, pH 7.4. This fraction represented the crude synaptosomic membrane (SPM) fraction (Schulman and Greengard 1978; Churn and DeLorenzo 1998a). The SPM was resuspended into 50 mm Tris-HCl, pH 7.4 and stored at − 80°C until used. In all experiments used for phosphorylation and binding studies the SPMs were thawed and subjected to a 100 000 g spin for 60 min to remove endogenous calmodulin (Schulman and Greengard 1978; Churn and DeLorenzo 1998b).
Endogenous CaM kinase II activity
SPM fractions were normalized for protein concentration and studied for endogenous calcium-dependent phosphorylation. Standard phosphorylation reaction solutions contained SPM fraction (10 µg), 40 mm MgCl2, 7 μm[γ-32P]ATP, 10 mm Tris-HCl (pH 7.4), ± 15 mm CaCl2 and ± 600 nm calmodulin. The increased cation levels were required for freshly prepared and stringently washed membrane fractions (Churn and DeLorenzo 1998b). Reactions were initiated by the addition of calcium, continued for 1 min, and terminated by the addition of 5% SDS solution (Churn et al. 1992a; Churn et al. 1995). Proteins were resolved by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) on a Bio-Rad Mini Protean II apparatus and protein bands visualized with the Biorad Silver Stain Plus kit. Molecular weights of resolved proteins were determined by comparison to Biorad prestained, broad range molecular weight standards. Phosphorylation of GABAAR micropreparation was quantified by computer-assisted autoradiography at a detection efficiency of 25% (Instant Imager, Packard Bioscience, Meriden, CT, USA). To complement the computer-assisted autoradiography, autoradiographs were used as a template for excising radioactive phosphoproteins for quantitation in a liquid scintillation spectrometer (LS 6500, Beckman, Fullerton, CA, USA) (Churn et al. 1995).
Receptor binding assay
Freshly centrifuged synaptoplasmic membrane (SPM) fractions were resuspended in 50 mm Tris-HCl (pH = 7.4). The fractions were then subjected to standard CaM kinase II phosphorylation assays as described above, except that non-radioactive ATP (500 µm) was used and the total reaction volume was increased to 1 mL. Phosphorylation reactions were stopped by dilution of 100 µL of the phosphorylation reaction mixture into 890 µL Tris-citrate containing 5 nm cypermethrin. This final dilution was used for flunitrazepam (FNZ) binding assays. Unless specified, total binding was performed with a final FNZ concentration of 40 nm. For non-specific binding, 40 μm[final] unlabeled FNZ was included in the binding reaction.
For filtration binding assays, reaction mixtures from the above phosphorylation reactions were allowed to equilibrate for 60 min on ice. Reactions were terminated by vacuum filtration using a Millipore manifold (Millipore Corp., Bedford, MA, USA) and Whatman GF/B glass fiber filters (Whatman International, Maidstone, UK). Filters were immediately washed (3 ×) with excess buffer (no drug) to remove non-specific binding. Filters were dissolved into scintillation cocktail (Instagel, Packard Bioscience, Meriden, CT, USA), and quantitated by liquid scintillation counting in a Beckman LS-6500 counter. Specific binding was determined by subtracting non-specific binding from total binding. Each sample was quantitated in triplicate (three total binding plus three non-specific binding).
For estimation of apparent Kd and Bmax for FNZ binding, saturation isotherms were generated. Binding reactions were performed as described above, except that the [3H]FNZ concentration was varied between 0.001 nm and 200 nm. The lower concentrations were used to ensure stable baseline estimation for accurate kinetic determination. All non-specific reactions also contained 200 µm cold FNZ.
SPM fractions were centrifuged at 18 000 g for 20 min and the pellet resuspended in 50 mm Tris-HCl (pH 7.4), and 200 mm NaCl for covalent [3H]FNZ binding. Binding reactions were performed as described above for filtration binding except that reactions were performed in a total volume of 200 μL. Binding tubes were allowed to equilibrate in the dark at 4°C for 30 min. Covalent linkage was accomplished by exposing the reaction mixture to ultraviolet irradiation (365 nm for 7 min at 5 cm distance) with constant stirring. Reactions were terminated with either 100 μL of SDS solution for standard SDS–PAGE or 20 μL of 22% triton X-100/11 mm EGTA for two-dimensional gel electrophoresis. Specific binding was determined as the difference between total and non-specific binding as described above.
GABAAR subunits were solubilized away from insoluble SPM components by incubation with 1% of the anionic detergent sodium deoxycholate or with 2.5% of the non-ionic detergent triton X-100 plus 1 m KCl (Stauber et al. 1987). Solubilization of GABAAR following endogenous CaM kinase II-dependent phosphorylation required the use of triton X-100 plus 1 m KCl because the divalent cations, Mg2+ and Ca2+, necessary for the phosphorylation reaction interfered with detergent solubilization using deoxycholate. In either case, the detergent mixture was diluted to 1 mL with 50 mm Tris-HCl (pH 7.4), allowed to equilibrate for 60 min at 4°C. The soluble receptor was separated from insoluble proteins by centrifugation at 100 000 g for 60 min (TL-100 Ultracentrifuge, Beckman, Fullerton, CA, USA). Supernatant from triton X-100/1 m KCl solubilization was dialyzed against 10 000-fold excess of 50 mm Tris-HCl (pH 7.4) plus 2.5% triton X-100 + 2 m urea for 60 min using a Slide-A-Lyzer dialysis membrane (Pierce, Rockford, IL, USA) to remove the high concentration of KCl, which interferes with SDS–PAGE resolution. Sample yield from dialysis was approximately 79.6% total protein recovery (range 66.6–86.2% recovery). Supernatant from deoxycholate solubilization did not require dialysis processing.
For protein recovery determination, all protein solutions were quantified with and without detergent present by the method of Bradford (1976) with bovine serum albumin as a standard. Inclusion of 1% deoxycholate or 2.5% triton X-100 did not interfere with the protein determination assay and did not significantly affect the quantification of soluble proteins. Quantification of bovine serum albumin (BSA), the protein utilized to generate a standard curve for protein quantification and balancing, was not significantly affected by the inclusion of 2.5% triton X-100 (8.3% change, p > 0.1, Student's t-test, n = 4). However, inclusion of detergent in crude SPM fractions resulted in a significant increase, approximately 66%, in apparent protein yield when compared with deionized water. The data suggests that the increase in protein yield observed with the inclusion of detergent was due to increased detectable membrane protein and not due to non-specific interference between detergent and the protein chromophore reagent. Therefore, triton X-100 was included in all protein quantifications (Table 1) to control for the apparent increased protein level in membrane fractions in the presence of detergent.
Table 1. Relative solubility of GABAAR protein in deoxycholate and triton X-100 solutions
2.5% Triton X-100
2.5% Triton X-100 + KCl
CaM kinase II α%
Total protein, GABAAR protein and CaM kinase II protein recoveries were determined in three separate detergent solubilization procedures (see Materials and methods). Deoxycholate treatment resulted in greater protein recovery into the soluble fraction when compared with triton X-100, alone, treatment. However, inclusion of the chaotropic agent, 1 m KCl, resulted in increased protein recovery into the detergent-soluble fraction. In addition, triton X-100/1 m KCl mixture produced similar GABAAR protein recoveries when compared with deoxycholate, and did not react with cations necessary for phosphorylation of GABAAR subunits.
7.13 ± 0.35
6.83 ± 0.60
6.25 ± 0.09
3.58 ± 0.22
50.21 ± 2.34
1.55 ± 0.65
22.62 ± 4.77
0.34 ± 0.13
5.46 ± 1.07
55.10 ± 3.56
3.40 ± 0.34
47.61 ± 3.11
5.09 ± 0.22
74.50 ± 1.65
5.79 ± 0.25
92.60 ± 2.01
0.78 ± 0.73
66.00 ± 1.23
Immunoprecipitation of GABAAR subunits
To further characterize the phosphorylation-dependent modulation of FNZ binding, GABAAR subunits were separated from SPM fractions by detergent solubilization and subsequent immunoprecipitation using subunit-specific antibodies. For immunoprecipitation, non-specific binding was reduced by incubation with 50 μL of normal rabbit IgG agarose-conjugated for 1 h, at 4°C, with gentle shaking. This was followed by a 5000 g spin (Centrifuge 5415 C, Eppendorf, Hamburg, Germany) for 5 min to pellet and remove the agarose beads conjugated with the non-specific IgG. Immunoabsorption of specific GABAAR subunits was accomplished by incubating the supernatant with affinity-purified-polyclonal antibody selective for the GABAAR α1 subunit (#9I-2B). The supernatant-antibody mixture was allowed to equilibrate overnight at 4°C with gentle shaking. GABAAR-antibody complex was precipitated by conjugation with 50 μL protein A agarose beads for 60 min, followed by centrifugation at 5000 g for 5 min. The pellet was washed three times with 500 μL of 1% deoxycholate or 500 μL of 2.5% triton X-100/50 mm Tris HCl (pH = 7.4), followed by a 5000 g spin for 5 min for each wash. The washed pellet was solubilized into SDS-stop solution, subjected to SDS–PAGE, and transferred to nitrocellulose as described previously (Churn et al. 1992a; Churn et al. 1995). Confirmation of the GABAAR α1 subunit was performed by western analysis using an antibody developed in a different animal species (SC 7348, Santa Cruz Biotechnology, Santa Cruz, CA, USA).
Two-dimensional gel electrophoresis
Standard two-dimensional gel electrophoresis was employed to separate the α1 subunit of the GABAAR (Mr 48–52 kDa, pI = 4.5–5) away from the subunit of CaM kinase II (Mr 50–52 kDa, pI = 7). GABAAR subunits were solubilized from SPM by 2% triton X-100 and 1 mm EGTA. Crude SPM fractions were centrifuged at 100 000 g and the detergent-soluble fraction was resolved by two-dimensional gel electrophoresis: isoelectric focusing (first dimension, gradient pH 3–10) followed by SDS–PAGE (second dimension). The soluble fraction was diluted into rehydration buffer (final concentration: 8 m urea, 2% triton X-100, 20 mm DTT, 0.5% carrier ampholytes Biolytes 3–10, and bromophenol blue as a marker). The sample was allowed to passively rehydrate into the ReadyStrip IPG dry strip gels (pH 3–10), for 12–14 h and then isoelectric focusing performed for a total of 20 000 volt-hours on a Bio-Rad Laboratories Protean IEF Cell. Strips were then equilibrated for 10 min in equilibration buffer I (0.375 m Tris-HCl (pH 8.8), 6 m urea, 20% glycerol, 2% SDS, and 2% (w/v) (DTT), followed by 10 min in equilibration buffer II (0.375 m Tris-HCl (pH 8.8), 6 m urea, 20% glycerol, 2% SDS, and 2.5% (w/v) iodoacetamide). The second dimension was performed using a Mini-Protean II Electrophoresis Cell. IPG strips were sealed in place using a 0.5% agarose solution pipetted over the strip. Gels were run for approximately 100 volt-hours.
Immunodetection of CaM kinase II α subunit and GABAAR α1 subunit
Western blot analysis was performed to identify both CaM kinase II subunit and GABAAR α1 subunit protein levels as described previously (Churn et al. 1992a; Churn et al. 1995). Proteins were resolved by one- and two-dimensional SDS–PAGE under standard conditions and transferred to nitrocellulose (pore size 0.45 m) with a Bio-Rad Trans Blot apparatus, and subsequently developed with the Vector Vectastain ABC Elite kit. In addition, CaM kinase II and GABAAR subunits were identified by resolving purified kinase or GABAAR fractions in parallel gel lanes as described previously (Churn et al. 1990; Churn et al. 1992a; Churn et al. 1995). The CaM kinase II subunit levels were quantified by either a previously characterized monoclonal antibody directed against the β subunit (Churn et al. 1992b) or the α subunit (Bio Mol, Plymouth Meeting, PA, USA). GABAAR subunit levels were quantitated using two distinct, affinity-purified polyclonal antibodies (see Materials and methods). Linear binding was compared with a linear standard (Churn et al. 1992b; Churn et al. 1995). Equal transfer of protein to nitrocellulose membrane was confirmed by protein staining utilizing a sensitive gold staining procedure (Churn et al. 1992b) of parallel strips of nitrocellulose.
All statistical analysis of data was performed using GraphPad Prism 3.0 (GraphPad Software, San Diego, CA, USA), or Sigma Stat 2.03 (SPSS Inc, Richmond, CA, USA). Levels of phosphorylation activity were compared with control using the Student's t-test (two-tailed distribution) or one-way anova with Tukey post-hoc analysis to control for type-1 errors in multiple comparisons. Computer assisted densitometric analysis was performed using Inquiry (Loats Associates Inc, Westminster, MD, USA). Apparent Kd and Bmax values for FNZ binding were determined by scatchard analysis using the Equilibrium Binding Data Analysis Program (EBDA, GA McPherson, Milltown, NJ, USA).
Effect of kinase activation on allosteric modulator binding
To elucidate the effect of endogenous CaM kinase II activation on GABAAR affinity for allosteric modulators, specific FNZ binding was determined in crude SPM fractions subjected to phosphorylation under conditions basal and maximal for CaM kinase II activation. The phosphorylation conditions employed were optimal for CaM kinase II activation in SPM fractions (Churn and DeLorenzo 1998a). Under conditions basal for CaM kinase II activation, specific FNZ binding was approximately 0.95 ± 9 pmol/mg (Fig. 1a). The specific FNZ binding observed following basal CaM kinase II activation was not significantly different from FNZ binding observed in naive SPM fractions (no phosphorylation reaction). Under conditions for maximal CaM kinase II activation, a significant 47% increase in specific FNZ binding was observed when compared with basal phosphorylation conditions (1.40 ± 0.10 pmol/mg, p < 0.001 Student's t-test, n = 9). The increase in allosteric binding was not observed if calmodulin was omitted from the reaction mixture (0.99 ± 0.05 pmol/mg). In addition, the increased binding observed under conditions for maximal CaM kinase II activity could be blocked with inclusion of KN-93, a selective calmodulin kinase inhibitor. The FNZ binding under maximal conditions for CaM kinase II activation plus inhibitor (maximal + KN-93) was not significantly different from basal phosphorylation binding (0.90 ± 0.11 pmol/mg, p > 0.05 different from control, Student's t-test, n = 9). KN-93 had no effect on FNZ binding under basal conditions for CaM kinase II activation (basal + KN-93). In addition, KN-92, an inactive but chemically similar compound, did not block the kinase activation-induced increase in FNZ binding (data not shown). Thus, under conditions optimal for CaM kinase II activity (Churn and DeLorenzo 1998a), a significant calmodulin-dependent and KN-93-sensitive increase in specific FNZ binding was observed.
To characterize the effect of CaM kinase II activation on FNZ binding kinetics, FNZ binding isotherms were generated under basal and maximal phosphorylation conditions (Fig. 1b). Under basal phosphorylation conditions, FNZ binding affinity (Kd) and maximal binding (Bmax) were similar to that observed for other investigators (Stephenson et al. 1982; Sigel et al. 1983; Sigel and Barnard 1984). Binding affinity was approximately 4–6 nm and specific binding saturated at approximately 50–60 nm as defined by the EBDA program (see Materials and methods). As observed above, maximal CaM kinase II activation resulted in a significant increase in apparent maximal FNZ binding (Bmax). However, there was no significant alteration of binding affinity (Kd approximately 5 nm). Specific binding saturated at FNZ concentration of approximately 50 nm which was not different than drug concentration saturation levels observed under basal phosphorylation. Thus the CaM kinase II-dependent modulation of benzodiazepine binding was selective for maximal benzodiazepine (BZ) binding and did not alter receptor affinity.
Separation of GABAAR α1 and CaM kinase II α subunit via detergent solubilization
CaM kinase II subunit, a neuronally enriched enzyme, has a similar molecular weight (Mr 50 kDa) as the GABAAR α1 subunit (Mr 52 kDa), and is capable of autophosphorylation. Therefore, to effectively study the phosphorylation of the GABAAR α1 subunit, it must be separated from the CaM kinase II subunit. Detergent solubilization and subsequent centrifugation was used to separate crude SPM (SPM) into a detergent-solubilized supernatant (Sup) and detergent-insoluble pellet (Pel) (Fig. 2). Standard receptor solubilization procedures were performed, except that deoxycholate was replaced with triton X-100/1 m KCl. This substitution was necessary since deoxycholate, a slightly anionic detergent (Ersson et al. 1989) precipitated in the presence of the cations utilized for the activation of CaM kinase II. Inclusion of 1 m KCl with the triton X-100 solution resulted in equivalent solubilization of GABAAR protein when compared with deoxycholate (Table 1). Triton X-100 treatment of crude SPM fractions alone resulted in approximately 50% protein recovery into the soluble fraction. However, inclusion of the chaotropic agent, 1 m KCl, resulted in an increased recovery of soluble protein to almost 95% of the total recoverable protein. The combination of triton X-100/KCl was significantly greater in solubilizing SPM proteins when compared with triton X-100, alone. The triton X-100/1 m KCl resulted in approximately 66% recovery of GABAAR protein into the detergent-soluble fraction, which was comparable to receptor protein recoveries using deoxycholate (data not shown).
To characterize the ability of triton X-100 to separate the GABAAR receptor away from CaM kinase II, crude SPM, detergent-soluble supernatant and detergent-insoluble pellet were normalized for protein concentration, resolved on SDS–PAGE and subjected to western analysis. Computer-assisted densitometry was performed on immunostained western blots to estimate the relative distribution of CaM kinase II α subunit and GABAAR α1 subunit protein. Balanced fractions (20 µg/lane) were resolved by SDS–PAGE and reacted with a monoclonal antibody directed against the CaM kinase II α subunit (Fig. 2a). Significant immunoreactivity was observed in the SPM fraction and the detergent-insoluble pellet fraction. Recovery of CaM kinase II α subunit protein, from the SPM fraction, was approximately 55% in the pellet fraction and 1% in the supernatant fraction. These findings are consistent with previous observations that CaM kinase II is enriched in the synaptic region and consistent with studies demonstrating detergent-insoluble CaM kinase II in the SPM (Goldenring et al. 1984b; Kelly et al. 1984, 1985).
To estimate the relative distribution of GABAAR α1 subunit protein, balanced fractions (200 µg/lane) were resolved, and probed with a goat-polyclonal GABAAR α1-subunit-specific antibody (Fig. 2b). In contrast to CaM kinase II, GABAAR immunoreactivity was observed in the SPM fraction as well as the detergent-soluble supernatant, with little recovery in the detergent-insoluble pellet. Recovery of GABAAR 1 from the SPM fraction was 66% in the supernatant fraction with an undetectable recovery in the detergent-insoluble pellet. The data demonstrate that detergent solubilization and subsequent centrifugation significantly separated the GABAAR α1 subunit into the detergent-soluble supernatant away from the CaM kinase II subunit that remained in the detergent-insoluble pellet.
Isolation of GABAAR α1 subunit via immunoprecipitation
To further separate GABAAR protein from CaM kinase II, and to study the isolated GABAAR α1 subunit, solubilized GABAAR subunits were immunoprecipitated using GABAAR-subunit-selective antibodies (see Materials and methods) (Fig. 3). The antibody-specific immunoprecipitate was resolved on SDS–PAGE, and transferred to nitrocellulose for western analysis. The resolved proteins were tested for immunoreactivity to determine the relative abundance of either CaM kinase II α subunit protein or GABAAR α1 subunit protein.
Figure 3(a) shows a silver stain of the protein resolution pattern of the detergent-soluble supernatant (Sup), the non-specific immunoprecipitation pellet (Pel), and the specific immunoprecipitation pellet (Sp). The major 50 kDa band observed in the specific immunoprecipitation pellet lane was the heavy chain of the polyclonal antibody used to precipitate the GABAAR protein. The GABAAR α1 subunit resolved below the IgG heavy chain (arrows). Figure 3(b) shows the relative distribution of GABAAR α1 subunit protein and CaM kinase II subunit protein. The detergent-solubilized and immunoprecipitated sample contains only the GABAAR α1 subunit without significant contamination with the CaM kinase II subunit. Thus the data show an effective separation and isolation of GABAAR α1 subunit from CaM kinase II via detergent solubilization and immunoprecipitation, allowing efficient analysis of GABAAR α1 subunit.
Effect of kinase activation on phosphorylation of the GABAAR α1 subunit
To determine whether CaM kinase II activation modulates GABAAR function through phosphorylation of receptor subunits or by modulating receptor-associated proteins, endogenous-CaM kinase II-dependent phosphorylation was performed (Churn and DeLorenzo 1998b). GABAAR α1 subunit protein was then detergent solubilized from crude SPM, immunoprecipitated and analyzed for the degree of phosphorylation. KN-93 was included in additional maximal reactions to determine the specific effects of CaM kinase activation (Fig. 4). Identity of GABAAR subunits was confirmed by western analysis (Fig. 4b) and compared with the silver stain of the protein pattern (Fig. 4c). As expected, GABAAR α1 subunit was minimally phosphorylated under conditions basal for CaM kinase II activation (7.2 ± 0.4 fmol/mm2/min) (Fig. 4a). The basal phosphorylation level demonstrates the viability of phosphorylation systems to maintain GABAAR subunit phosphorylation under control conditions. Subjecting SPM fractions to conditions maximal for CaM kinase II activity resulted in a 92% increase in phosphate incorporation into the GABAAR α1 subunit (13.8 ± 0.7 fmol/mm2/min) compared with basal phosphorylation levels (Fig. 4a). Consistent with agonist binding (Churn and DeLorenzo 1998b) and allosteric modulator binding (Fig. 1), the maximal level of GABAAR α1 subunit phosphate incorporation required the inclusion of both calcium and calmodulin during the phosphorylation reaction. In addition, the CaM kinase II activation-mediated increase in GABAAR α1 subunit phosphorylation could be significantly inhibited by the inclusion of KN-93 during the phosphorylation reactions (7.0 ± 0.2 fmol/mm2/min). This was not significantly different than phosphate incorporation under basal phosphorylation conditions.
In addition to phosphorylation of the GABAAR α1 subunit, activation of endogenous CaM kinase II resulted in the increased phosphate incorporation into other proteins that coprecipitated with the GABAAR α1 subunit (Fig. 4a). While the other proteins were not identified, a protein withmolecular weight similar to the GABAAR γ subunit (approximately 47 kDa) was also phosphorylated in a calcium and calmodulin-dependent fashion. CaM kinase II activation resulted in an approximate two-fold increase in phosphate incorporation into this protein. Inclusion of KN-93 partially blocked the CaM kinase II activation-dependent phosphate incorporation into this protein. Interestingly, no phosphorylated bands were seen that corresponded to the GABAAR β subunit (Mr 56–58) (Fig. 4a). The data suggests that CaM kinase II activation results in GABAAR α1 subunit phosphorylation in a Ca2+/calmodulin-dependent manner. In addition, a protein with molecular weights corresponding to the GABAAR γ but not the GABAAR β subunits that coprecipitate with the GABAAR α1 subunits was phosphorylated by endogenous CaM kinase II activation.
GABAAR two-dimensional electrophoresis
To further characterize CaM kinase II-dependent phosphorylation of GABAAR, SPM fractions were subjected to endogenous CaM kinase II activation, detergent solubilized, and subjected to two-dimensional gel electrophoresis (Fig. 5). This study took advantage of the significantly different isoelectric points of CaM kinase II (∼7.0) (Goldenring et al. 1984a) and GABAAR α1 subunit (∼4.5–5) (Sweetnam et al. 1988). Figure 5(a) shows a two-dimensional gel electrophoresis resolution pattern of a detergent-soluble SPM fraction, following covalent FNZ binding. The detergent-soluble fraction resolved into multiple proteins with a major band of approximately 50 kDa (pI approximately 5.2) and a minor band of molecular weight approximately 52 kDa. The 52 kDa protein band displayed a pI of approximately 4.5–5.0 and comigrated with the major specific, covalent FNZ binding peak. This protein was identified as the major FNZ binding protein (Fig. 5b).
CaM kinase II-dependent phosphorylation reactions were performed and reactions were resolved by two-dimensional gel electrophoresis. Under maximal conditions for CaM kinase II activation, six major phosphoproteins were observed in the corresponding autoradiograph (Fig. 5b). A major phosphoprotein with molecular weight approximately 52 kDa and an isoelectric point of pH = 4.5–5 was observed (arrow). The phosphoprotein aligned with the major FNZ binding protein observed in Fig. 5(a).
The GABAAR has been shown to play an important inhibitory role in neurotransmission and mediates many important processes in the CNS (Macdonald and Olsen 1994). It has also been shown that the function of GABAAR can be modulated by phosphorylation (Stelzer et al. 1988; Wang et al. 1995) and that CaM kinase II will phosphorylate GABAAR subunits in expression systems at specific residues of the major intracellular loop of each of its subunits (Machu et al. 1993). This study characterized the role of endogenous CaM kinase II on allosteric modulator binding and phosphorylation of GABAAR subunits. The results showed that activation of CaM kinase II resulted in a significant augmentation of BZ binding to the GABAAR. In addition, it was shown that at least the GABAAR α1 subunit is phosphorylated by activation of endogenous CaM kinase II. Since the GABAAR subunit is crucial for benzodiazepine binding, the data suggests that CaM kinase II activation modulated benzodiazepine binding in part by phosphorylation of the GABAAR subunits.
The effect of phosphorylation on the GABAAR α1 subunit was determined under conditions basal and maximal for CaM kinase II activation, followed by the immunoprecipitation of the α1 subunit, and subsequent analysis. The GABAAR α1 subunit is highly expressed in adult brain (Fritschy and Möhler 1995) and has been shown to be important for benzodiazepine modulation of GABAAR activation (Dunn et al. 1999; Rudolph et al. 1999; Möhler et al. 2001). Specific BZ modulation of GABAAR α1 subunit-containing receptors has been shown to be involved in sedation (Rudolph et al. 1999, 2001), amnesia (Rudolph et al. 1999), and anticonvulsant activity (Rudolph et al. 1999; Crestani et al. 2000). Therefore, the GABAAR α1 subunit plays a significant role in modulating function of the CNS.
CaM kinase II is a key regulatory enzyme capable of phosphorylating many disparate neuronal proteins (Churn 1995) and has been shown to positively modulate GABAAR function (Wang et al. 1995; Aguayo et al. 1998; Churn et al. 2000). Injection of thiophosphorylated CaM kinase II has been shown to enhance Cl− current in acutely isolated spinal cord (Wang et al. 1995) and hippocampal neurons in culture (Wang et al. 1995; Churn et al. 2000). NMDA channel activation will initially increase GABA-evoked currents and the observed increase in agonist-evoked current can be blocked by KN-62 (Aguayo et al. 1998). Since NMDA stimulation increases CaM kinase II activity (Soderling et al. 1994), a link between CaM kinase II activation and GABAAR function would be consistent with the effect of NMDA stimulation and modulation of GABAAR current. Conversely, it has been shown that increased intracellular Ca2+ can also decrease GABAAR function (Stelzer and Shi 1994; Jones and Westbrook 1997). The conflicting findings may represent different tissue preparations or modulation of different intracellular enzyme systems. For instance, activation of calcineurin will result in decreased GABAAR currents (Stelzer and Shi 1994; Jones and Westbrook 1997). Since calcineurin activation requires less influx of Ca2+ than does CaM kinase II, qualitative changes in internal free Ca2+ concentration may account for the differences observed.
To characterize the effect of phosphorylation on modulating benzodiazepine binding by the GABAAR, FNZ binding to GABAAR was determined following basal and maximal conditions for CaM kinase II activation. In addition, coincubation with KN-93 was performed to show the effect was due to CaM kinase II activation and not through non-specific calmodulin-dependent effects. Under maximal conditions for CaM kinase II activation, FNZ binding was increased significantly compared with basal binding levels. Co-incubation with KN-93 under maximal kinase conditions blocked the kinase-activation-dependent increase in FNZ binding. In addition, FNZ binding levels in KN-93-treated fractions showed no significant change from basal binding levels. The data indicated that activation of CaM kinase II positively modulated GABAARs binding of allosteric modulators. The CaM kinase II activation-dependent alteration of FNZ binding kinetics demonstrated that the activation effect was selective for maximal binding. Although the apparent Bmax for FNZ binding displayed an increase from basal to maximal phosphorylation conditions, the apparent Kd did not change significantly. The data suggested that the apparent number of receptors increased without a significant alteration of the binding properties of the receptors.
An increase in Bmax without a change in Kd can be explained by multiple mechanisms. It has been demonstrated that GABAAR function is dependent on receptor phosphorylation state (Akaike 1992). Thus, increasing CaM kinase II-dependent phosphorylation can increase apparent receptor binding by increasing net phosphorylation state of membrane receptors. It is possible that a percentage of GABAARs exist in the plasma membrane which lie dormant and inactive. Following CaM kinase II-dependent phosphorylation, the receptors become active and able to bind to allosteric modulators. Direct phosphorylation could activate previously dormant receptors, which would increase the total number of functional receptors as visualized in the data as an increase in Bmax.
Another possibility for CaM kinase II modulation is that a spare pool of non-functional, internalized receptors are inserted into the membrane due to CaM kinase II activation. GABAAR expressing γ2 subunits have been shown to undergo receptor internalization (Connolly et al. 1999b). The internalized receptors are associated with clathrin-coated vesicles and may be recycled to the membrane surface (Barnes 2000). Activation of protein kinase C has been shown to decrease the recycling of GABAAR to the neuronal surface (Connolly et al. 1999a), and thus cause a decrease in functional GABAAR expressed at the surface. It is possible that CaM kinase II activation decreases internalization or increases the trafficking of cytosol-located receptors to the plasma membrane. Modulation of GABAAR membrane expression may be due to receptor subunit phosphorylation or by an indirect mechanism in which CaM kinase II phosphorylates receptor-associated proteins, such as gephyrin, GABAAR-associated proteins, or tubulin (Kneussel and Betz 2000). This would also increase the apparent number of activated receptors in the membrane. However, given the stringent washing steps with iced-cold diH2O and Tris-HCl buffer utilized to isolate crude SPM fractions, it is unlikely that internalized vesicles are retained. Even though possible under physiological conditions, recycling of internalized GABAAR to the plasma membrane would not be a significant mechanism in the present study.
Finally, two-dimensional gel electrophoresis was performed to further eliminate the possibility of interference between CaM kinase II autophosphorylation and CaM kinase II-dependent phosphorylation of the GABAAR subunit proteins. Using differences in the isoelectric points of CaM kinase II (6.9–7.0) and GABAAR subunit (4.5–5), crude SPM was phosphorylated, detergent solubilized, and then separated via two-dimensional electrophoresis. The resulting autoradiograph of the second dimension was used to demonstrate GABAAR α1 subunit phosphorylation. The identity of the GABAAR subunit was confirmed by specific, covalent FNZ binding and by the correlation of its location with that of two-dimensional electrophoresis of affinity purified GABAAR subunits. The phosphorylated peptide comigrated with FNZ binding and also appeared on two-dimensional gels at its appropriate Mr and isoelectric point. The data demonstrated that the GABAAR α1 subunit was phosphorylated, and that the analysis was not contaminated by CaM kinase II subunit autophosphorylation.
Due to the heterogeneity of GABAA receptor subunit expression in the CNS (Wisden et al. 1992), regulation of receptor function by phosphorylation is complicated. Typically, protein kinase A (PKA)-dependent phosphorylation results in rapid desensitization and decreased channel opening which has been shown to be due to phosphorylation of Ser409 on the β subunit (Macdonald et al. 1992; Macdonald and Olsen 1994). One significant exception is the granule cells of the dentate gyrus where PKA-dependent phosphorylation has been shown to augment GABAergic function (Kapur and Macdonald 1996). The effect of PKA-dependent phosphorylation may relate to the β isoform expressed in the GABAAR (McDonald et al. 1998). Protein kinase C (PKC)-dependent phosphorylation has also been shown to decrease channel opening resulting in a decreased Cl– current amplitude (Sigel and Baur 1988; Kapur and Macdonald 1996). PKC activation has been shown to result in decreased surface expression of GABAAR channels (Chapell et al. 1998; Connolly et al. 1999a). PKC did not directly result in receptor internalization, but did prevent recycling of GABAAR complexes to the neuronal surface. Thus, it is important to understand the effect of receptor phosphorylation on specific GABAAR subunits.
This study demonstrated that like agonist binding (Churn and DeLorenzo 1998b), activation of endogenous CaM kinase II positively modulated allosteric modulator binding to GABAARs. In addition, the present study demonstrated that CaM kinase II activation resulted in a direct phosphorylation of at least the GABAAR α1 subunit. The data support the conclusion that CaM kinase II positively modulates GABAAR function in part by receptor subunit phosphorylation.
The authors would like to thank Matt Josse and Jonathan Kurz for their aid and helpful discussions and suggestions. The authors would also like to thank Pierre Weber and Dr Eva-Maria Gutknecht (Hoffmann-La Roche Ltd, Basel, Switzerland) for the generous gift of Ro 07−1986/602. This work was supported by NINDS awards RO1-NS39970 (SBC), R01-NS23350 (RJD) and PO1-NS25630 (RJD, SBC), and the Milton L. Markel Alzheimer and Neuroscience Research Fund.