GABAA receptors (GABAARs) are the major mediators of fast synaptic inhibition in the brain. In neurons, these receptors undergo significant rates of endocytosis and exocytosis, processes that regulate both their accumulation at synaptic sites and the efficacy of synaptic inhibition. Here we have evaluated the role that neuronal activity plays in regulating the residence time of GABAARs on the plasma membrane and their targeting to synapses. Chronic blockade of neuronal activity dramatically increases the level of the GABAAR ubiquitination, decreasing their cell surface stability via a mechanism dependent on the activity of the proteasome. Coincident with this loss of cell surface expression levels, TTX treatment reduced both the amplitude and frequency of miniature inhibitory synaptic currents. Conversely, increasing the level of neuronal activity decreases GABAAR ubiquitination enhancing their stability on the plasma membrane. Activity-dependent ubiquitination primarily acts to reduce GABAAR stability within the endoplasmic reticulum and, thereby, their insertion into the plasma membrane and subsequent accumulation at synaptic sites. Thus, activity-dependent ubiquitination of GABAARs and their subsequent proteasomal degradation may represent a potent mechanism to regulate the efficacy of synaptic inhibition and may also contribute to homeostatic synaptic plasticity.
GABAA receptors (GABAARs), mediate the majority of fast synaptic inhibition in the brain and also represent the major sites of action for both benzodiazepines and barbiturates. These receptors are Cl−-selective ligand-gated ion channels that can be assembled from seven subunit classes: α1–α6, β1–β3, γ1–γ3, δ, ε, and π, providing the structural basis for extensive heterogeneity of GABAAR structure (Sieghart and Sperk, 2002; Rudolph and Mohler, 2004, 2006). A combination of molecular, biochemical, and genetic approaches suggest that, in the brain, the majority of benzodiazepine receptor subtypes are composed of α, β, and γ2 subunits (Rudolph and Mohler, 2004). γ2-containing receptors are highly enriched at synaptic sites in neurons and are responsible for mediating phasic inhibition (Essrich et al., 1998; Kittler and Moss, 2003; Luscher and Keller, 2004).
The number of GABAARs on the neuronal cell surface is a critical determinant for the efficacy of synaptic inhibition, and, at steady state, this is determined by the rates of receptor insertion and removal from the plasma membrane. It is evident that GABAARs are assembled within the endoplasmic reticulum (ER) and then transported to the plasma membrane for insertion, whereas misfolded or unassembled receptor subunits are rapidly targeted for ER-associated degradation (Gorrie et al., 1997; Kittler and Moss, 2003; Luscher and Keller, 2004). Degradation of ER-retained GABAARs is mediated via the activity of the proteasome and is subject to modulation via their association with the ubiquitin-like protein Plic-1 (Bedford et al., 2001). Cell surface GABAARs are dynamic entities that exhibit rapid rates of constitutive endocytosis, with internalized receptors being subject to rapid recycling or lysosomal degradation (Kittler and Moss, 2003; Luscher and Keller, 2004; Kittler et al., 2005). Intriguingly, a number of studies have provided evidence that chronic changes in neuronal activity can modulate inhibitory neurotransmission by regulating both presynaptic components in addition to modifying postsynaptic GABAAR levels (Rutherford et al., 1997; Kilman et al., 2002; Hartman et al., 2006). The mechanisms underlying this homeostatic modulation of GABAAR functional expression remain ill defined.
Here we have addressed the role that neuronal activity plays in regulating GABAAR turnover and membrane trafficking and the subsequent accumulation of these proteins on the neuronal plasma membrane. Our results demonstrate that the level of neuronal activity can regulate the ubiquitination of GABAARs in the secretory pathway enhancing their degradation primarily within the ER and thereby altering the efficacy of synaptic inhibition. This process decreases the number of receptors inserted into the plasma membrane and their subsequent accumulation at inhibitory synapses. Thus, activity-dependent ubiquitination of GABAARs and their subsequent proteasomal degradation may represent a mechanism to regulate the efficacy of synaptic inhibition.
Materials and Methods
Rabbit anti-myc antibody was purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Rabbit polyclonal antibodies against the GABAAR α2, β3, and γ2 subunits have been described previously (Fritschy and Mohler, 1995; Brandon et al., 2000, 2001; Jovanovic et al., 2004), and monoclonal anti-ubiquitin (FK2) was supplied by Biomol (Plymouth Meeting, PA). Synapsin and anti-green fluorescent protein (GFP) were purchased from Synaptic Systems (Goettingen, Germany). Texas Red and cyanine 5 (Cy5)-conjugated secondary antibodies were obtained from Jackson ImmunoResearch (West Grove, PA).
Neurons or HEK-293 cells grown in 60 mm dishes were chilled on ice for 5 min and then washed twice in PBS with 1 mm CaCl2 and 0.5 mm MgCl2 at 4°C. Cells were incubated for 15 min at 4°C in 1 mg/ml NHS-SS-Biotin (Pierce, Rockford, IL) followed by two 5 min washes in 50 mm glycine to quench unreacted biotin. For the endocytosis assay, neurons were returned to a 37°C incubator for 15 min to allow endocytosis (in the presence of 100 μg/ml leupeptin) and then placed on ice and incubated twice with cleavage buffer (50 mm glutathione, 75 mm NaCl, 10 mm EDTA, 1% bovine serum albumin, and 0.075N NaOH) for 15 min to cleave biotin from cell surface receptors. For the cell surface receptor degradation assay, biotinylated neurons were returned to a 37°C incubator for 24 h in the presence or absence of tetrodotoxin (TTX) (2 μm). After these procedures, neurons were lysed in radioimmunoprecipitation assay (RIPA) buffer: 1% NP-40, 0.5% sodium deoxycholate, 50 mm Tris, pH 8, 150 mm NaCl, and 2 mm EDTA. After correction for protein content using the micro BCA protein assay (Pierce), biotinylated proteins were purified on immobilized avidin (Neutravidin; Pierce) and resolved by SDS-PAGE, and GABAAR levels were measured by immunoblotting with anti-β3 antibody followed by detection with ECL. Blots were imaged in the Fujifilm (Stamford, CT) LAS-3000 imaging system, and band intensities were measured using Fujifilm Multi Gauge software.
Cell culture and transfection.
HEK-293 cells were transfected using electroporation with 10 μg of DNA as outlined previously (Taylor et al., 1999, 2000). Cortical and hippocampal neurons were obtained from embryonic day 18 (E18) rats (Kittler et al., 2000; Jovanovic et al., 2004). Dissociated E18 rat neurons were transfected with 3 μg of plasmid DNA per 5 × 106 neurons using the Rat Nucleofector kit (Amaxa, Gaithersburg, MD) (Couve et al., 2004; Kittler et al., 2004; Jacob et al., 2005). For pulse-chase analysis, cortical neurons were incubated in methionine free medium (DMEM) for 20 min and then labeled with 500 mCi/ml [35S]methionine (PerkinElmer, Waltham, MA) for 30–40 min. Neurons were then washed and incubated in complete neurobasal media with an excess of cold methionine (100×) for an additional 0–8 h.
Membrane currents from HEK-293 cells were measured as outlined previously (Taylor et al., 1999, 2000; Bedford et al., 2001; Kittler et al., 2004, 2005; Jacob et al., 2005). Miniature IPSCs (mIPSCs) were recorded from 14–21 d in vitro (DIV) cultured hippocampal neurons in the whole-cell voltage-clamp configuration (−70 mV) as detailed previously (Kittler et al., 2004, 2005). Briefly, all experiments were performed at room temperature (22–23°C). Extracellular bath solution was composed of the following (in mm): 140 NaCl, 4.7 KCl, 10 HEPES, 11 glucose, 1.2 MgCl2, and 2.5 CaCl2, adjusted to pH 7.4 with NaOH. Borosilicate pipettes (3–6 MΩ) were filled with the following (in mm): 150 CsCl, 10 HEPES, 1.1 EGTA, 2 MgCl2, 0.1 CaCl2, and 1 Mg2+-ATP, adjusted to pH 7.2 with CsOH. Recordings were started 5–10 min after a stable whole-cell access was obtained. mIPSCs were recorded in the presence of 200 nm tetrodotoxin (Sigma-Aldrich, St. Louis, MO), 10 μm 6-cyano-7-nitroquinoxaline-2,3-dione (Sigma-Aldrich), and 20 μm 2-d-aminophosphonopentanoic acid (Sigma-Aldrich). An Axopatch 200B amplifier and Digidata 1322A (Molecular Devices, Sunnyvale, CA) software were used for pulse generation, data acquisition (10 kHz), and filtering (5 kHz, four-pole Bessel filter). Synaptic currents were recorded using an Axopatch 200B amplifier (Molecular Devices), filtered at 2 kHz, sampled at 5 kHz, digitized (Digidata 1320A; Molecular Devices), and stored for off-line analysis [using Minianalysis (Synaptosoft, Decatur GA) and IGOR Pro (WaveMetrics, Lake Oswego, OR)]. Access resistance (10–18 MΩ; 80% compensation) was monitored using a −5 mV voltage, applied every 120 s, and data from cells were discarded when >15% change occurred. Miniature events were analyzed using pClamp 9.2 (Clampfit; Molecular Devices) and MiniAnalysis 6.0.3 software (Synaptosoft, Leonia, NJ) using pooled population data from expressed as mean ± SEM. An individual event was defined as being >2.5 SDs from the average background noise. The decay phase was fitted with a monoexponential function (τ decay). Rise time was also analyzed by comparing the mean 10–90% rise time.
Transfected neurons were fixed in 4% paraformaldehyde, stained without membrane permeabilization with rabbit anti-GFP, and then permeabilized with 0.1% Triton X-100 for 4 min. They were then labeled with anti-synapsin antibodies, visualized by confocal microscopy (blind to experimental condition), and analyzed using MetaMorph (Molecular Devices) software. Receptor clusters were defined as being ∼0.5–2 μm in length and approximately twofold to threefold more intense than background diffuse fluorescence and were colocalized with or directly apposed to synapsin staining. Clusters farther than 1 μm from presynaptic marker staining were considered extrasynaptic. To quantify the number of synapses, thresholds were set and kept constant for control and test neurons. To measure the role of ubiquitination in regulating GABAAR membrane insertion, we used β3 and β3K12R expression constructs modified at the N terminus with a pHluorin reporter and the minimal α-bungarotoxin (Bgt) binding site peptide (BBS). These are termed BBSβ3 and BBSβ3K12R, respectively (Scherf et al., 2001; Katchalski-Katzir et al., 2003; Sekine-Aizawa and Huganir, 2004; Bogdanov et al., 2006). BBSβ3 synaptic puncta were counted from 1-bit binary masks. Data were analyzed from 15–20 neurons for each condition (50 μm per dendrite per cell). To quantify fluorescence intensity of BBSβ3 synaptic staining, images of neurons were thresholded to a point at which dendrites were outlined. Synapsin staining was thresholded to a set value and kept constant for control and test neurons. Next a 50 μm section along a given proximal dendrite was selected, and a 1-bit binary image (exclusive) was made of the synapsin staining in the outlined dendrite. We then subtracted away all BBSβ3 staining that did not colocalize with the binarized synapsin staining. As a result, only BBSβ3 staining that colocalized with synapsin remained, and the average fluorescence intensity of these BBSβ3puncta was determined. Data were analyzed from 15–20 neurons for each condition from at least two to three different cultures. Finally, puncta selection and analysis were all performed blind to experimental condition.
HEK-293 cells or neurons were lysed in 1% SDS, 25 mm Tris, pH 7.4. Lysates were diluted 10-fold with RIPA buffer lacking SDS: 50 mm Tris, pH 8, 150 mm NaCl, 1% NP-40, 0.5% sodium deoxycholate, 2 mm EDTA plus 1 μm ubiquitin aldehyde (Biomol), and mammalian protease inhibitor cocktail (Sigma-Aldrich). After correction for protein content, lysates were immunoprecipitated with rabbit anti-myc IgGs (Santa Cruz Biotechnology) or anti-β3 antibodies as detailed previously (Brandon et al., 2000, 2001; Jovanovic et al., 2004; Kittler et al., 2004). Precipitated material was then subject to SDS-PAGE and immunoblotting.
Membrane insertion assays.
HEK-293 cells expressing GABAAR α1 and BBSβ3 subunits were incubated with 10 μg/ml unlabeled α-Bgt for 20 min at 4°C (to inhibit endocytosis), washed in PBS, and incubated with 10 μg/ml biotin-conjugated Bgt at 37°C for up to 20 min to label newly inserted receptors. After lysis, biotin–Bgt/BBSβ3 complexes were purified on neutravidin and immunoblotted with anti-β3 antibodies. To measure insertion in neurons nucleofected 15 DIV hippocampal cultures were labeled at 37°C with 10 μg/ml rhodamine (Rd)-conjugated Bgt for 15 min, washed, and incubated for an additional 5 min with 10 μg/ml Alexa-Fluor-647 (Alx)-conjugated Bgt. All incubations were performed in the presence of 5 μm tubocurarine (Sigma-Aldrich) to block Bgt binding to endogenous acetylcholine receptors (Sekine-Aizawa and Huganir, 2004; Bogdanov et al., 2006). All Bgt derivatives were obtained from Invitrogen (Carlsbad, CA). Cells were fixed in 4% paraformaldehyde, and confocal images were collected using a 60× objective lens acquired with Olympus Optical (Tokyo, Japan) Fluoview version 1.5 software, and the same image acquisition settings for BBSβ3 and BBSβ3K12R (Bogdanov et al., 2006) were used. These images were analyzed using MetaMorph imaging software (Universal Imaging, Downingtown, PA). First, a three-dimensional reconstruction of an imaged neuron was made from a series of Z sections, and then the average fluorescence intensity of Alexa-Fluor-647–Bgt staining per 250 μm of dendrite was measured from three dendrites per neuron, after subtraction of background fluorescence.
Site-directed mutagenesis of the β3 subunit was performed using the QuickChange mutagenesis kit (Stratagene, La Jolla, CA).
Statistical significance was assessed using either Student's paired t test or the Kolmogorov–Smirnov two-sample test, in which p < 0.05 was considered significant.
Chronic changes in neuronal activity regulate the numbers of GABAAR expressed on the neuronal cell surface
To commence our studies, we used cultured cortical neurons that had been maintained for at least 14 d DIV and have been demonstrated to exhibit hyperpolarizing GABAAR responses (Woodin et al., 2003; Fiumelli et al., 2005). The effects of the voltage-gated Na+ channel blocker TTX on the number of cell surface GABAARs containing β3 subunits was then examined. After cell lysis and correction for protein content, biotinylated proteins were purified on avidin and immunoblotted with antibodies specific to the β3 subunit (Jovanovic et al., 2004; Kittler et al., 2004). We chose to study primarily GABAAR subtypes containing β3 subunits because gene knock-out has revealed that these proteins are critical for neuronal function and animal survival (Homanics et al., 1997). Treatment of cultures with 2 μm TTX for 24 h resulted in a significant decrease (to 47.8 ± 5.3% of control) (Fig. 1A) in cell surface levels of GABAARs containing β3 subunits. The total expression of receptor β3 subunits was also reduced by this treatment (to 45 ± 6.7% of control) (Fig. 1A). Similar statistically significant decreases in the cell surface expression levels of GABAAR α2 and γ2 subunits were also evident (supplemental Fig. 1, available at www.jneurosci.org as supplemental material), suggesting that chronic neuronal inactivity leads to the loss of functional heteromeric cell surface GABAARs. In addition TTX treatment also significantly decreased the number of cell surface GABAARs containing β3 subunits in 16 DIV hippocampal neurons (supplemental Fig. 2, available at www.jneurosci.org as supplemental material).
To corroborate our experiments with TTX, we assessed the effects of blocking excitatory synaptic transmission on the cell surface levels of GABAARs with glutamate receptor antagonists, which have been established to reduce neuronal activity in culture (Rao and Craig, 1997; O'Brien et al., 1998). Blockade of glutamate receptors significantly decreased the cell surface expression levels of GABAARs containing β3 subunits to 45.8 ± 13.1% of control (Fig. 1B). We also examined the effects of chronically increasing neuronal activity on the cell surface expression levels of GABAARs using picrotoxin (PTX) (40 μm), a GABAAR antagonist and has been shown to increase the activity of neuronal cultures (Rao and Craig, 1997; O'Brien et al., 1998). In contrast to the effects of TTX and glutamate receptor antagonists, treatment of neurons with PTX significantly increased both the cell surface (by 55 ± 2.3% of control) and total (by 47 ± 8.5% of control) expression levels of the GABAARs β3 subunit (Fig. 1C).
To assess whether the reduced levels of cell surface GABAARs evident with TTX alters synaptic inhibition we compared the properties of mIPSCs in hippocampal neurons (>16 DIV) treated with or without TTX. mIPSCs were isolated by adding the excitatory amino acid antagonists (10 μm CNQX, 20 μm d-AP-5) and TTX (2 μm) to the bath solution as described previously (Kittler et al., 2004, 2005; Chen et al., 2006). Under these conditions remaining events were blocked by picrotoxin (20 μm) indicating that these events were dependent upon the activation of GABAARs (data not shown). As illustrated in Figure 2A–D, TTX-treated neurons exhibited a significant shift in peak amplitude and frequency compared with controls. Quantifying these observations, the mean amplitude of mIPSCs in TTX-treated neurons was 41.8 ± 3.3 pA (n = 9), significantly lower (Fig. 2E) (p < 0.05) than the value of 50.7 ± 1.9 pA (n = 8) evident in control neurons. In addition, the average frequency of mIPSCs in TTX-treated neurons was 0.5 ± 0.1 Hz (n = 14), significantly lower (Fig. 2E) (p < 0.05) than values evident in controls (0.9 ± 0.1 Hz; n = 14). However, the 10–90% rise time and decay time constants were unaltered in neurons treated with TTX compared with control (Fig. 2) (supplemental Fig. 3, available at www.jneurosci.org as supplemental material). Although alterations in mIPSC amplitude may reflect modified transmitter release or altered channel kinetics, they are also indicative of altered numbers of synaptic receptors. Consistent with decreased mIPSC amplitude, deficits in the number of cell surface β3 subunits are also evident in cortical hippocampal neurons treated with TTX as measured via biotinylation (Fig. 1) (supplemental Fig. 2, available at www.jneurosci.org as supplemental material), suggesting that the cell surface expression levels of GABAARs are subject to regulation by the level of neuronal activity.
Chronic changes in neuronal activity regulate the levels of GABAAR β3-subunit ubiquitination
To further examine the role that activity plays in regulating GABAAR expression levels, we tested the effects of TTX on subunit ubiquitination. To do so, we immunoprecipitated β3 subunits under denaturing conditions and immunoblotted precipitated material with an antibody that recognizes polyubiquitin (FK2). Typical of protein ubiquitination (Haglund et al., 2003b; Haglund and Dikic, 2005), a smear of immunoreactivity between 66 and 250 kDa was seen with FK2 antibody immunoprecipitating with anti-β3 but not control IgG (Fig. 3A,B). This reaction product represents the addition of multiple 7.6 kDa ubiquitin moieties to various (>20) lysine residues within the GABAAR β3 subunit. By measuring the intensity of the FK2 reaction product between 66 and 250 kDa, it was evident that activity blockade with TTX produced a large (5.0 ± 0.5-fold) increase in the levels of β3 subunit ubiquitination (Fig. 3A,C) and decreased total cellular levels of the β3 subunit. In contrast, treatment of neurons with PTX induced a substantial (3.1 ± 0.3-fold) decrease in the levels of ubiquitinated β3 and increased total cellular levels of this receptor subunit (Fig. 3B,D).
Polyubiquitination of proteins enhances their targeting to and degradation by the proteasome (Glickman and Ciechanover, 2002); thus, we analyzed the role that blocking proteasome activity with epoxomicin (EP) (20 μm) has on activity-dependent turnover of GABAARs. EP blocked the effects of TTX on cell surface expression levels of GABAARs receptors and also significantly increased their cell surface expression levels compared with control (117.5 ± 2.8%) (Fig. 4A). Moreover, EP enhanced the level of β3 subunit ubiquitination to 250.4 ± 20.8% of control (and significantly increased the cell surface levels of β3-containing GABAARs by 159.3 ± 8.1% of control levels (Fig. 4B–D).
Cell surface GABAARs undergo constitutive endocytosis and are eventually targeted for degradation principally in the lysosome (Kittler et al., 2000, 2004). We thus asked whether chronic activity blockade had any influence on the degradation of cell surface GABAARs, which reflects their levels of endocytosis, endocytic sorting, and lysozomal degradation. Using biotinylation, it was evident that TTX did not significantly alter the degradation of surface β3 subunits (Fig. 4E), suggesting that neuronal activity acts principally to regulate GABAAR stability and/or insertion in the secretory pathway.
To further test this, we used pulse-chase analysis with [35S]methionine after a 24 h incubation period with TTX. This revealed that 89.9 ± 3.5% of newly synthesized β3 subunits were still present after the chase period. However, in the presence of TTX, this value was decreased to 35.8 ± 12.4% of that present at time 0 (Fig. 4F). To determine whether the observed increased rate of degradation of the β3 in the presence of TTX results from increased ER-associated degradation, we performed a pulse-chase experiment in the presence of Brefeldin A, an established inhibitor of ER-Golgi transport. Even in the presence of Brefeldin A, turnover of the β3 subunit was increased in TTX-treated neurons (supplemental Fig. 4, available at www.jneurosci.org as supplemental material). Because GABAAR assembly takes place principally in the ER (Connolly et al., 1996; Gorrie et al., 1997), this decreased stability of newly synthesized receptors in the presence of TTX and Brefeldin A suggest that chronic inactivity acts to principally control receptor stability in the ER.
Identification of major sites for ubiquitination within the GABAAR β3 subunit
To further examine the role of ubiquitination in regulating GABAAR trafficking, we used mutagenesis to convert all 12 candidate lysine residues for ubiquitination within the major intracellular loop of the β3 subunit to arginines in addition to adding the myc epitope at the N terminus of this protein (Mβ3K12R) (supplemental Fig. 5, available at www.jneurosci.org as supplemental material) (Taylor et al., 2000). Mβ3K12R exhibited a large decrease (to 55.4 ± 5.4% of control) (Fig. 5A) in ubiquitination compared with Mβ3, demonstrating that major sites for this covalent modification reside within the intracellular domain of this GABAAR subunit. Other sites for ubiquitination presumably reside within the transmembrane or extracellular domains of the β3 subunit that would be subject to ubiquitination preceding proteasomal degradation (Haglund et al., 2003a). Consistent with the reduced level of ubiquitination, we observed a significant increase (to 145 ± 4.5% of control) (Fig. 5B) in the cell surface expression levels of Mβ3K12R compared with Mβ3.
To control for possible deleterious effects of the mutations on GABAAR functional expression, we assessed the ability of the Mβ3K12R subunit to assemble with receptor α1 and γ2 subunits using patch-clamp recording in HEK-293 cells (Taylor et al., 2000; Jacob et al., 2005; Bogdanov et al., 2006). We first assessed the sensitivity of the expressed receptors to benzodiazepine modulation, a property critically dependent on the production of heteromeric α1β3γ2 receptors (Sieghart and Sperk, 2002; Mohler et al., 2005). Flurazepam (100 nm) produced similar robust enhancements of IGABA of ∼200% for receptors composed of α1β3γ2 or α1β3K12Rγ2 subunits (Fig. 5C). Equilibrium GABA concentration–response curves for receptors incorporating Mβ3K12R or Mβ3 subunits yielded similar EC50 values of 9.40 ± 1.80 and 10.17 ± 2.42 μm, respectively, values that were not significantly altered by preincubation with the proteasome inhibitor MG132 [carbobenzoxy-l-leucyl-l-leucyl-l-leucinal] (Fig. 5D). We also examined the effects of MG132 on the maximal currents. For receptors incorporating Mβ3, there was a large and highly significant increase in the magnitude of IGABA using saturating agonist concentrations, an effect that was not replicated for those incorporating Mβ3K12R subunits (Fig. 5E).
Together, our results demonstrate that major sites for ubiquitination reside within the intracellular domain of the GABAAR β3 subunit and that mutation of these residues leads to increased receptor functional expression. They also suggest that these lysine residues within the intracellular domain of the β3 subunit play a role in ubiquitin-dependent proteasomal degradation of this subunit.
Decreasing β3 subunit ubiquitination increases receptor accumulation on the neuronal cell surface
To examine the relevance of our recombinant studies for GABAARs in their native environment, we expressed Mβ3 and Mβ3K12R subunits in cultured cortical neurons using nucleofection. After immunoprecipitation with rabbit anti-myc antibodies and blotting with FK2 antibody, it was evident that ubiquitination of Mβ3K12R was decreased to 50.1 ± 4.5% of control (Fig. 6A). Consistent with this, the cell surface expression levels of GABAARs incorporating Mβ3K12R subunits were higher than those incorporating Mβ3 subunits (by 141.2 ± 8.2% of control) (Fig. 6B). A concern in these experiments is that there is a possible formation of β3 homomeric receptors, but it should be noted that this phenomenon was only observed when β3 was expressed alone in HEK-293 cells or oocytes (Taylor et al., 1999). This unique property of β3 subunit homomerization is totally suppressed on coexpression with GABAAR α and/or γ2 subunits. It should also be noted that β3 homomers are very unstable with a half-life of 2 h (Wooltorton et al., 1997; Taylor et al., 1999). Given that neurons express high levels of endogenous GABAAR α and γ subunits, the presence of β3 homomers is unlikely to be a factor in our experiments.
In principle, the elevated accumulation of receptors incorporating β3K12R on the plasma membrane may reflect changes in either endocytic sorting and/or lysosomal degradation, or in higher rates of insertion from the secretory pathway. Using biotinylation, it was evident that receptors containing Mβ3 and Mβ3K12R exhibited very similar levels of endocytosis over a 15 min time course (21.3 ± 4.2 vs 20.9 ± 6.2%, respectively) (Fig. 6C), comparable with levels seen for the endogenous β3 subunit (Kittler et al., 2004). Moreover, cell surface GABAARs containing Mβ3 and Mβ3K12R exhibited very similar rates of degradation over 10 h of 33.5 ± 3.2 and 34.6 ± 3.6%, respectively (Fig. 6D). Together, these results suggest that ubiquitination of the β3 subunit does not primarily act to regulate the endocytic sorting/lysosomal degradation of GABAARs but may act to modulate their insertion, from the secretory pathway, into the plasma membrane.
Ubiquitination of the β3 subunit modulates GABAAR insertion into the plasma membrane
To measure the role of ubiquitination in regulating GABAAR membrane insertion, we used β3 and β3K12R expression constructs modified at the N terminus with a pHluorin reporter and the minimal BBS. These are termed BBSβ3 and BBSβ3K12R, respectively (Scherf et al., 2001; Katchalski-Katzir et al., 2003; Sekine-Aizawa and Huganir, 2004; Bogdanov et al., 2006). We previously established that the addition of these reporters to GABAARs is functionally silent but allows the selective visualization of cell surface receptor populations in both expression systems and cultured neurons (Jacob et al., 2005; Bogdanov et al., 2006). These engineered receptors also bind Bgt with high affinity (7.2 × 10−9 m) with an off-rate in excess of 5 h, facilitating the analysis of GABAAR insertion in living neurons (Bogdanov et al., 2006).
We initially characterized the insertion rate of GABAARs composed of α1BBSβ3 in HEK-293 cells by first labeling existing cell surface populations with 10 μg/ml Bgt. After extensive washing, cells were labeled with 10 μg/ml biotinylated Bgt for varying time periods at 37°C before lysis. Bgt/GABAAR complexes were then isolated with immobilized avidin and immunoblotted with anti-GFP antibodies. A band of ∼89 kDa representing the BBSβ3 subunit was evident after 2.5 min incubation at 37°C but not at the time 0 point (Fig. 7A). This band was not detected in untransfected cells (Fig. 7A). Its appearance could be blocked by coincubation with 10 μg/ml unlabeled Bgt (data not shown). GABAAR insertion appeared to be linear during the first 5 min of incubation before saturating after 20 min (Fig. 7A). We therefore used a time point of 2.5 min to compare membrane insertion for BBSβ3 and BBSβ3K12R, with this time point being given a value of 100%. This revealed that GABAARs containing BBSβ3K12R subunits exhibited a significantly higher level of insertion (to 141.6 ± 15.8% of control) (Fig. 7B) compared with those incorporating BBSβ3 subunits
We next used fluorescence imaging in neurons to examine the insertion of GABAARs containing either BBSβ3 or BBSβ3K12R subunits. Nucleofected hippocampal neurons expressing BBSβ3 or BBSβ3K12R were incubated with Rd–Bgt to label existing populations of receptors containing BBS binding sites. After washing, neurons were incubated with Alx–Bgt for 5 min at 37°C to label newly inserted receptors. Neurons expressing pH–BBS-tagged GABAAR subunits were identified by their endogenous fluorescence attributable to the presence of the pHluorin reporter in these constructs (Fig. 7C). Abundant levels of Rd–Bgt labeling were evident only in neurons that also exhibited green fluorescence, and therefore Rd–Bgt staining represents existing cell surface populations of GABAAR at time 0 (Fig. 7C). Consistent with our observations in HEK-293 cells, newly inserted receptors (Alx–Bgt staining) were evident after 5 min at 37°C (Fig. 7C). The appearance of these new Bgt binding sites could be totally abolished by coincubation with 20 μg/ml unlabeled Bgt (data not shown) (Bogdanov et al., 2006). Proximal dendrites in acquired images were outlined in MetaMorph to determine the average cell surface fluorescence intensity of newly inserted Alx–Bgt-labeled GABAARs along 20 μm of a given dendrite. Three dendrites were analyzed per expressing neuron, and at least 15 neurons were examined per construct. By calculating the average fluorescence intensity of Alx–Bgt labeling, it was evident that GABAARs incorporating BBSβ3K12R subunits exhibited significantly higher levels of insertion compared with those incorporating BBSβ3 subunits (to134.7 ± 8.6% of control) (Fig. 7D).
Collectively, these experiments demonstrate that ubiquitination of the GABAAR β3 subunit acts to regulate receptor insertion primarily from the secretory pathway.
The effects of neuronal activity on GABAAR cell surface expression levels are mediated via β3 subunit ubiquitination
We examined whether TTX mediates its effects on GABAAR cell surface stability via modulation of β3 subunit ubiquitination. To do so, we measured the insertion of GABAARs containing BBSβ3 and BBSβ3K12R subunits into the plasma membranes of hippocampal neurons in the presence or absence of TTX using Bgt binding as outlined above. Interestingly, neurons expressing GABAARs incorporating BBSβ3 subunits appeared to show reduced levels of Alx–Bgt staining after treatment with TTX compared with those containing BBSβ3K12R (Fig. 8A). Quantifying these data, it was evident that TTX decreased the level of insertion for GABAARs containing BBSβ3 by 34 ± 8% relative to control, but this effect was not replicated for receptors containing pHBBSβ3K12R subunits (Fig. 8B).
Together these results provide a molecular mechanism linking the level of neuronal activity and GABAAR cell surface stability, a mechanism dependent on direct modulation of β3 subunit ubiquitination, modified proteasomal degradation, and altered rates of plasma membrane insertion.
Chronic blockade of neuronal activity regulates the accumulation of GABAARs containing β3 subunits at synaptic sites
As a final series of experiments, we measured the effects of modulating the level of GABAAR ubiquitination on their accumulation at synapses. Neurons were transfected with BBSβ3 or BBSβK12R using nucleofection. At 15 d in vitro, neurons were fixed and immunostained with anti-GFP antibodies and with antibodies against synapsin, an accepted marker for presynaptic terminals, followed by confocal microscopy.
The accumulation of BBSβ3 and BBSβK12R at synaptic sites was compared using immunofluorescence with an anti-GFP antibody under nonpermeabilized conditions to measure only cell surface receptor populations, followed by anti-synapsin antibodies after permeabilization (Fig. 9A). The number of synapses containing recombinant β3 subunits per unit length of dendrite and their relative intensity were then determined using MetaMorph. From this analysis, it was evident that similar numbers of synapses containing BBSβ3 and BBSβ3K12R subunits were present on neuronal processes (Fig. 9B). However, the intensity of the fluorescence signal for synapses containing BBSβ3K12R was significantly higher (126.9 ± 8.4%) compared with synaptic BBSβ3 subunits (Fig. 9B). Thus, these results suggest that decreasing GABAAR ubiquitination increases the receptor number at synaptic sites, consistent with their increased cell surface expression levels.
The effects of TTX treatment on the accumulation of GABAARs containing recombinant β3 subunits at synaptic sites was also ascertained using immunofluorescence (Fig. 9A). A 24 h, treatment with 2 μm TTX significantly decreased the number of synaptic sites containing BBSβ3 from 17.6 ± 1.9 per 50 μm of dendrite in control compared with 8.6 ± 1.1 in the presence of this agent (Fig. 9C). In contrast, TTX did not significantly alter the number of synapses containing BBSβ3K12R subunits (Fig. 9C). In addition, TTX also reduced the intensity of BBSβ3 staining at remaining synapses by 31.8 ± 3.5% relative to control, an effect not replicated for synapses containing BBSβ3K12R subunits (Fig. 9D).
We have begun to assess the role that chronic changes in neuronal activity play in determining the membrane trafficking and synaptic accumulation of GABAARs. We demonstrated that blockade of neuronal depolarization with TTX or the activity of ionotropic glutamate receptors for 24 h decreased the cell surface stability of GABAARs receptors containing β3, α2, or γ2 subunits; in contrast, enhancing neuronal activity increases the accumulation of these proteins on the neuronal plasma membrane. Therefore, these combined experiments suggest that the level of neuron activity is a powerful determinant of the number of GABAARs subtypes containing β3, α2, or γ2 on the plasma membrane of cultured cortical and hippocampal neurons. Previous studies on the role that neuronal activity plays in regulating GABAAR cell surface stability have proven controversial. Electrophysiological approaches in mature cultured cortical neurons have shown that chronic inactivity decreases GABAAR number and reduces both frequency and amplitude of miniature inhibitory synaptic currents (Turrigiano et al., 1998; Kilman et al., 2002). In contrast, blockade of synaptic activity in developing hippocampal neurons does not appear to modify the number of cell surface GABAARs or their accumulation at synaptic sites (Harms and Craig, 2005). This discrepancy may reflect the varying ages of the culture preparations used (Huupponen et al., 2007) or the length of activity blockade, 24 h compared with 18 d. Intriguingly, it has been demonstrated recently that exposure of cultured neurons to TTX over a period of 7–14 d results in retraction of dendrites and a loss of spines (Fishbein and Segal, 2007). Therefore, this phenomenon may complicate the analysis of the long-term effects of activity blockade on the formation of inhibitory synapses.
In addition to decreasing cell surface expression levels, TTX treatment dramatically enhanced the ubiquitination of the GABAAR β3 subunit over the same time course. Accordingly, the ability of TTX to decrease GABAAR cell surface expression levels was dependent on the activity of the proteasome. Interestingly, enhancing neuronal activity by pharmacological blockade of GABAARs resulted in decreased receptor ubiquitination and enhanced cell surface expression levels. TTX treatment did not alter receptor endocytosis or cell surface half-life, suggesting that neuronal activity primarily acts to regulate receptor insertion into the plasma membrane from the secretory pathway. Consistent with this, pulse-chase analysis revealed that TTX reduced the stability of newly translated receptor subunits primarily in the ER, the principal subcellular site of GABAAR assembly (Gorrie et al., 1997; Bedford et al., 2001). Thus, these results suggest that the level of neuronal activity is a key determinant of GABAAR cell surface stability and strongly suggest that these effects are mediated by altered receptor ubiquitination and proteasomal degradation. Similarly, activity-dependent ubiquitination and degradation of the NMDA receptor subunit NR1 has been observed in hippocampal neurons (Kato et al., 2005). However, activity-dependent ubiquitination of mammalian AMPA receptor subunits has not been demonstrated, although chronic changes in activity regulate the ubiquitination and turnover of associated postsynaptic density proteins (Colledge et al., 2003; Patrick et al., 2003). In contrast, ubiquitination has been illustrated to regulate the endocytosis of AMPA receptors in Caenorhabditis elegans (Burbea et al., 2002).
To further explore the significance of activity-dependent ubiquitination, we used mutagenesis to convert all 12 lysine residues (β3K12R) within the major intracellular domain of the GABAAR β3 subunit to arginines. These mutations significantly reduced β3 subunit ubiquitination but did not compromise assembly with the α1 and γ2 subunits into functional benzodiazepine-sensitive heteromeric GABAARs. Significantly higher levels of cell surface expression of the β3K12R mutant were evident compared with wild-type subunits in both expression systems and neurons. These mutations did not alter GABAARs endocytosis or cell surface stability but significantly enhanced receptor insertion into the plasma membrane. Collectively, these experiments together with our pulse-chase analysis suggest that ubiquitination of the β3 subunit primarily acts to regulate subunit stability within the ER. Consistent with this, it is well established that polyubiquitination is required for retrotranslocation of proteins from the endoplasmic reticulum back into the cytosol, in which they are degraded by the proteasome (Ye et al., 2003). Given the critical role that receptor β subunits play in regulating the ER exit of assembled GABAARs (Luscher and Keller, 2004), this reduced level of β3 subunits would be predicted to reduce the pool of fully assembled heteromeric receptors for insertion into the plasma membrane. In support of this idea, previous studies have shown that blocking ER-associated degradation of nicotinic acetylcholine receptors with proteasome inhibitors leads to increased receptor ubiquitination and also subunit oligomerization, which resulted in increased receptor insertion into the plasma membrane (Christianson and Green, 2004).
To directly test role that activity-dependent ubiquitination plays in regulating the accumulation of GABAAR at synaptic sites, we used β3 subunit expression constructs modified with N-terminal extracellular pHluorin and BBS reporters (Bogdanov et al., 2006). Using imaging with fluorescent Bgt, this approach demonstrated that GABAARs incorporating β3K12R subunits showed significantly higher rates of insertion into the plasma membrane compared with their wild-type equivalents. In agreement with this, higher levels of accumulation of GABAARs containing BBSβ3K12R subunits were evident at synapses compared with those containing wild-type BBSβ3. Moreover, TTX treatment significantly reduced the number of synapses containing BBSβ3 subunits together with receptor number at remaining synapses. However, the synaptic accumulation of receptors containing BBSβ3K12R was unaffected by TTX treatment. Therefore, these studies strongly suggest that neuronal activity can modulate the accumulation of GABAARs at synaptic sites via regulating the ubiquitination of these proteins in the secretory pathway and their subsequent insertion into the plasma membrane.
The molecular details of how altered levels of neuronal activity regulate GABAAR ubiquitination remain to be established but are likely to be highly dynamic and determined by the activity of individual ubiquitin-ligase/hydrolyases and other modulators of proteasomal degradation. It should be noted that GABAARs bind directly to the ubiquitin-like protein Plic-1 (Bedford et al., 2001), an established regulator of ubiquitin-dependent proteasomal degradation, further highlighting a possible mechanism neurons may use to control GABAAR ubiquitination. In addition to controlling the stability of GABAARs, neuronal activity may also regulate the formation of inhibitory synapses at many additional loci, ranging from the stability of cytoskeletal anchors such as gephyrin to the number of innervating presynaptic terminals (Craig et al., 2006).
In summary, our results provide evidence that neuronal activity can regulate the number of cell surface GABAARs by modulating their ubiquitination and subsequent proteasomal degradation in the secretory pathway. This process can directly regulate the level of GABAAR insertion into the plasma membrane and their subsequent accumulation at postsynaptic sites. This putative mechanism may therefore play a critical role in coordinating the level of local synaptic activity in the brain and thus contribute to homeostatic synaptic plasticity.
R.S.S. is supported by a postdoctoral fellowship from the Epilepsy Foundation and G.M. by Deutsche Forschungsgemeinschaft Grant MI 960/1-1 and the Koeln Fortune Program/Faculty of Medicine (University of Cologne). S.J.M. is supported by National Institutes of Health Grants NS046478, NS048045, NS051195, and NS056359, the Medical Research Council (United Kingdom), and the Wellcome Trust. We thank Margie Maronski from the Dichter laboratory for preparation of cultured neurons and Yolande Haydon for manuscript preparation.
- Correspondence should be addressed to Dr. Stephen J. Moss at the above address.