Benzodiazepine treatment induces subtype-specific changes in GABA_A receptor trafficking and decreases synaptic inhibition

Abstract

Benzodiazepines potentiate γ-aminobutyric acid type A receptor (GABA_AR) activity and are widely prescribed to treat anxiety, insomnia, and seizure disorders. Unfortunately, clinical use of benzodiazepines (BZs) is severely limited by tolerance. The mechanisms leading to BZ tolerance are unknown. BZs bind at the interface between an α and γ subunit of GABA_ARs, preferentially enhancing synaptic receptors largely composed of α(1–3, 5), β3, and γ2 subunits. Using confocal imaging and patch-clamp approaches, we show that treatment with the BZ flurazepam decreases GABA_AR surface levels and the efficacy of neuronal inhibition in hippocampal neurons. A dramatic decrease in surface and total levels of α2 subunit-containing GABA_ARs occurred within 24 h of flurazepam treatment, whereas GABA_ARs incorporating α1 subunits showed little alteration. The GABA_AR surface depletion could be reversed by treatment with the BZ antagonist Ro 15-1788. Coincident with decreased GABA_AR surface levels, flurazepam treatment reduced miniature inhibitory postsynaptic current amplitude, which returned to control levels with acute Ro 15-1788 treatment. GABA_AR endocytosis and insertion rates were unchanged by flurazepam treatment. Treatment with leupeptin restored flurazepam lowered receptor surface levels, strongly suggesting that flurazepam increases lysosomal degradation of GABA_ARs. Together, these data suggest that flurazepam exposure enhances degradation of α2 subunit-containing GABA_ARs after their removal from the plasma membrane, leading to a reduction in inhibitory synapse size and number along with a decrease in the efficacy of synaptic inhibition. These reported subtype-specific changes in GABA_AR trafficking provide significant mechanistic insight into the initial neuroadaptive responses occurring with BZ treatment.

GABA_A receptors (GABA_ARs) are Cl⁻-selective ligand-gated ion channels that mediate fast inhibition in the CNS and are important targets for benzodiazepines (BZs), drugs used to treat anxiety, insomnia, and seizure disorders and as adjunct treatments in both depression and schizophrenia. BZs enhance chloride ion conductance primarily by increasing the frequency of receptor channel opening and slowing the decay of miniature inhibitory postsynaptic currents (mIPSCs) (1–3). Structurally, GABA_ARs are pentameric heterooligomers that can be constructed from eight subunit classes with multiple members: α(1–6), β(1–3), γ(1–3), δ, ε, θ, π, and ρ(1–3) (4, 5). Molecular and genetic studies have revealed that synaptic BZ-sensitive GABA_AR subtypes are composed of α(1–3, 5), β, and γ2 subunits, whereas GABA_AR subtypes composed of α4/6, β, and δ subunits form a specialized population of extrasynaptic receptors that mediate tonic inhibition but are insensitive to modulation by BZ (reviewed in ref. 6).

BZs are highly prescribed and of major clinical significance; however, the development of tolerance restricts their usefulness and has prompted research on the mechanism for several decades. Chronic BZ treatment results in allosteric uncoupling of the GABA and BZ binding sites, suggesting changes in receptor subunit composition and/or receptor function (reviewed in ref. 7). Chronic dosing of animals with BZ leads to a reduction in GABA_AR synaptic inhibition (8–10) and produces diverse changes in GABA_AR transcripts across the brain (7). Direct comparison and interpretation of these and other studies assessing mRNA levels has been challenging due to differences in treatment paradigm (time and dose), brain regions assessed, and the BZ ligand used. Radioligand binding studies have reported mixed results, ranging from decreases to no change in CNS BZ binding sites, likely due to methodological limitations in assessing subtype-specific GABA_AR changes.

The surface level of postsynaptic GABA_ARs is a critical determinant of the strength of synaptic inhibition and is largely dependent on regulated trafficking steps to and from the plasma membrane, including receptor insertion, diffusion within the membrane, receptor removal, and degradation (11). Biochemical studies show that at least 40% of the total cell surface population of GABA_ARs is internalized within 1 h (12). Endocytosed receptors are either rapidly recycled back to the surface for reinsertion or targeted for lysosomal degradation. In addition, GABA_AR surface levels can be regulated by controlling the receptor pool available for insertion, the size of which is primarily determined by endoplasmic reticulum-associated degradation (13). Modulation of GABA_AR turnover has been widely invoked to explain the development of tolerance to BZs. However, it remains to be determined whether exposure of neurons to BZs modifies the surface stability of GABA_ARs or the efficacy of synaptic inhibition.

We demonstrate here that 24-h treatment with the BZ flurazepam (Flz) dramatically decreases α2 subunit-containing GABA_AR surface and total levels without comparable changes in levels of the α1 subunit. BZ treatment does not appear to alter receptor insertion and removal rates. However, treatment with leupeptin, an inhibitor of lysosomal proteolysis, restores BZ-lowered GABA_AR levels. The observed decrease in GABA_AR surface levels and synaptic inhibition occurs in a BZ binding site-dependent fashion because these alterations can be rescued by treatment with the BZ antagonist Ro 15-1788. Together, these data suggest that Flz exposure enhances degradation of α2-subtype GABA_ARs after endocytosis, leading to a reduction in inhibitory synapse size and number along with a decrease in the efficacy of synaptic inhibition.

Results

Acute BZ Treatment Decreases GABA_AR Surface Levels.

To directly address the role of BZ in regulating GABA_AR surface levels, we performed live-imaging studies of pH-sensitive GFP-tagged GABA_AR α1 (α1^pHGFP) and α2 (α2^pHGFP) subunits in hippocampal neurons. The addition of this reporter to receptor subunits is functionally silent and allows the specific visualization of surface GABA_AR populations in living neurons (14, 15). Furthermore, BZ potentiation of GABA-induced currents is equivalent in HEK-293 cells expressing these tagged or untagged GABA_AR α subunits cotransfected with β3γ2 subunits (Table S1). We measured the surface levels of BZ-sensitive GABA_ARs containing α1^pHGFP or α2^pHGFP subunits with 24-h 250 nM Flz or control treatment with live confocal microscopy at 37 °C in Hepes-buffered saline (HBS) (pH 7.4). Twenty-four-hour Flz treatment dramatically decreased α2 subunit-containing GABA_AR surface levels on three measured parameters, expressed as percentage control: surface cluster area (65.0 ± 2.8%), average cluster fluorescence intensity (87.1 ± 2.2%), and density (58.7 ± 17%) (Fig. 1 A and B). Minor changes were observed in α1-subtype GABA_AR surface levels, with a significant decrease only in average cluster fluorescence intensity (96.5 ± 1.3%) (Fig. 1 A and B).

Fig. 1.

Acute BZ treatment decreases surface and total GABA_AR levels. (A) Neurons expressing α2^pHGFP or α1^pHGFP GABA_AR subunits were incubated with or without 250 nM flurazepam (Flz) for 24 h followed by live confocal microscopy experiments (37 °C in HBS). (Right) Enlargements of dendrites in boxed areas. (B) Mean ± SEM cluster size, fluorescence intensity, and density of surface GABA_AR clusters normalized to control (*significantly different from control, P < 0.05, t test; n = 15–20 neurons, 3 independent cultures). (C) Neurons were perfused at a rate of 1 mL/min with pH 7.4 HBS followed by HBS with 50 mM NH₄Cl (+NH₄Cl) to collapse the intracellular pH gradients and reveal intracellular GABA_AR. The center panels are enlargement of cell bodies, and the right panels are dendrites in boxed areas. (D) The ratio of α2^pHGFP total/surface levels was normalized to Flz-treated neurons (*P < 0.05, t test; n = 6–8 neurons per culture, 3 independent cultures; error bars represent ±SEM). (Scale bars, 10 μm.)

BZ Treatment Alters Total GABA_AR Levels.

Because these initial studies revealed a major decrease in α2-containing GABA_AR surface levels and a minimal change in α1-containing GABA_AR surface levels with 24-h BZ treatment, subsequent experiments focused on the α2 subunit. We tested whether total BZ-sensitive GABA_AR protein levels were altered in Flz-treated neurons. To do so, we visualized pHGFP-tagged GABA_ARs resident inside the neurons by equalizing the pH of all intracellular compartments to pH 7.4 with a brief perfusion of NH₄Cl in HBS (NH₄Cl in place of equimolar NaCl) (Fig. 1C). This experiment revealed that total α2-subunit levels were decreased dramatically, with the ratio of total/surface α2 being approximately threefold lower following BZ treatment (control total/surface ratio, 277 ± 36%, normalized to Flz total/surface ratio) (Fig. 1 C and D).

Decreases in GABA_AR Surface Levels Occur via a BZ Binding Site-Dependent Mechanism.

Next, we tested whether the decrease in α2-subtype GABA_AR surface levels was specifically induced by BZ binding. Live-cell confocal microscopy was used to measure α2^pHGFP fluorescence in neurons with 24-h Flz or control treatment followed by an acute 1-h treatment with 5 μM BZ antagonist Ro 15-1788 (Fig. 2 A and B). One-hour treatment with Ro 15-1788 restored the BZ-lowered GABA_AR surface levels to control levels (cluster area in μm²: control, 0.232 ± 0.008; Flz, 0.184 ± 0.006; Flz plus Ro 15-1788, 0.216 ± 0.010; and total cluster fluorescence expressed as % control: control, 101.4 ± 5.4; Flz, 73.4 ± 3.6; Flz plus Ro 15-1788, 96.5 ± 5.3) (Fig. 2B). Control neurons treated with Ro 15-1788 showed no significant change on any measured parameter (control plus Ro 15-1788 cluster area in μm², 0.242 ± 0.006; and total cluster fluorescence expressed as % control, 104.9 ± 5.4) (Fig. 2B). Interestingly, the BZ antagonist did not significantly increase GABA_AR surface cluster density in Flz neurons, suggesting that existing inhibitory synapses are preferentially restored (clusters per 20 μm: control, 6.8 ± 2.2; Flz, 5.4 ± 1.2; control plus Ro 15-1788, 6.9 ± 1.8; Flz plus Ro 15-1788, 5.4 ± 1.6).

Fig. 2.

Decreases in GABA_AR surface levels and GABAergic mIPSC amplitude occur via a BZ binding site-dependent mechanism. (A) Neurons expressing α2^pHGFP were incubated with or without 250 nM Flz for 24 h and then treated for 1 h with 5 μM Ro 15-1788 (BZ antagonist) immediately before live imaging confocal microscopy experiments (37 °C in HBS). (Right) Enlargements of dendrites in boxed areas. (Scale bars, 10 μm.) (B) Surface cluster histograms showing mean area and cumulative α2^pHGFP fluorescence in neurons with 24-h control (Con) or 250 nM Flz conditions with or without 1-h 5 μM Ro 15-1788 treatment (*P < 0.05, one-way ANOVA analysis and Bonferroni’s multiple-comparison test; 22–30 neurons; n = 3 cultures; error bars represent ±SEM). (C) Sample traces of mIPSCs in control (Con) neurons and those treated with 250 nM Flz for 24 h with or without 1-h 5 μM Ro 15-1788 treatment. (D) Bar graph of the mean ± SEM mIPSC amplitudes and frequency (*P < 0.05, Kolmogorov–Smirnov test; n = 7–10 cells for each condition).

Prolonged Exposure of Neurons to BZ Reduces mIPSCs.

Having determined that surface and total levels of α2-subtype GABA_ARs are decreased by 24-h BZ exposure, we next measured the functional effects of prolonged BZ treatment on GABAergic inhibition. It is well established that rapid application of BZ to neurons enhances GABA_AR function due to an increase in the probability of channel opening. Although this results in an increased amplitude and slowed decay of mIPSCs (16, 17), the effects of prolonged initial treatment remain to be addressed. We directly compared the properties of mIPSCs in control neurons and those pretreated with Flz for 24 h before recording. BZ treatment resulted in a significant decrease in mIPSC amplitude from 52.5 ± 2.7 in control neurons to 42.2 ± 2.4 pA in those exposed to Flz for 24 h. In contrast mIPSC frequency (in Hz: control, 1.08 ± 0.34; Flz, 0.69 ± 0.12; Flz plus Ro 15-1788, 0.68 ± 0.10), 10–90% rise time (in ms: control, 1.46 ± 0.08; Flz, 1.68 ± 0.11; Flz plus Ro 15-1788, 1.59 ± 0.11), and decay times (in ms: control, 43.4 ± 1.19; Flz, 47.6 ± 2.34; Flz plus Ro 15-1788, 46.1 ± 1.66) were unaffected by prolonged exposure to BZ (Fig. 2 C and D). One-hour treatment with the BZ antagonist Ro 15-1788 restored mIPSC amplitude to control levels (50.2 ± 5.7 pA). The rapid timescale of Ro 15-1788 action is consistent with BZ-induced changes affecting receptor trafficking pathways rather than protein synthesis. Furthermore, these results are consistent with the efficacy of Ro 15-1788 reversing in vitro and in vivo BZ tolerance (18).

BZ Treatment Does Not Alter the Rates of GABA_AR Insertion and Removal.

It is increasingly apparent that GABA_ARs on the surface of neurons are dynamic entities that undergo regulated exocytosis and constitutive clathrin-mediated, dynamin-dependent endocytosis. Changes in the rates of exocytosis or endocytosis could clearly contribute to a mechanism for BZ-dependent alteration of GABA_AR surface levels. To measure the insertion and removal rate of α2-containing GABA_ARs, we generated a construct with a bungarotoxin (Bgt) binding site (BBS) and pHGFP encoded in the N terminus of the GABA_AR subunit (α2^pHBBS). The BBS allows the monitoring of receptor insertion and endocytosis with application of exogenous fluorescent Bgt and does not modify the assembly or functional properties of GABA_ARs (13, 19, 20). Bgt insertion assays were performed on α2^pHBBS-expressing neurons undergoing control and 24-h 250 nM Flz treatments. Neurons were first incubated with unlabeled Bgt at 15 °C to block existing surface GABA_AR receptor populations, and then washed followed by incubation with Alexa 594-conjugated Bgt at 37 °C to label newly inserted receptors. At 5, 10, and 15 min, samples were removed, fixed, and permeabilized, and the total receptor population was labeled with anti-GFP antibody. Quantification and analysis of surface receptor signal in control and Flz-treated neurons produced similar results to live imaging (Fig. 1), with BZ-treated neurons showing a decrease to 51.59 ± 7.3% control (Fig. 3 A and B). The fluorescence intensity of newly inserted α2^pHBBS over a 15-min time course (normalized to the initial fluorescence value of each treatment) was compared in control and Flz conditions (Fig. 3 A and C), revealing that the rate of GABA_AR insertion was not significantly changed by BZ treatment (Bgt fluorescence count at t = 10 min: control, 364.5 ± 112.1%; Flz, 305.3.1 ± 12.2; and at t = 15 min: control, 437 ± 129.5%; Flz, 389 ± 22.6%).

Fig. 3.

BZ treatment does not alter the insertion rate of α2 subunit-containing GABA_AR. (A) Confocal imaging of GABA_AR insertion in control and Flz-treated neurons. Neurons expressing α2^pHBBS were treated with or without 250 nM Flz for 24 h, and then incubated with Alexa594::Bgt (red) at 37 °C for 5, 10, and 15 min to label newly inserted GABA_AR, followed by fixation. Total receptor number was determined by permeabilization and staining with anti-GFP antibody (green). Panels below are enlargements of boxed dendrites. (Scale bars, 10 μm.) (B) Analysis of total receptor signal in control and Flz-treated α2^pHBBS neurons. (*P < 0.01, paired t test; 10–12 neurons per condition for each culture, n = 4 cultures; error bars represent ±SEM). (C) Graph shows Alexa594::Bgt fluorescence intensity of newly inserted α2^pHBBS over time in control (Con) or Flz conditions (normalized to initial fluorescence count at 5 min) (10–12 neurons per condition for each culture; n = 4 cultures; error bars represent ±SEM).

We next measured the rate of endocytosis in control versus Flz-treated neurons expressing α2^pHBBS using an established fluorescent endocytosis assay (19). Surface GABA_AR were live-labeled with Alexa594::Bgt for 5 min at 37 °C, followed by washing to remove unbound Alexa594::Bgt, and then incubated at 37 °C. At t = 0, 7.5, and 15 min, neurons were removed from the incubator and fixed. Total receptor number was visualized by permeabilization and staining with anti-GFP antibody. Quantification of confocal microscopy images showed loss of Alexa594::Bgt fluorescence over time with endocytosis of surface GABA_AR (Fig. 4 A and B). The Alexa594::Bgt total fluorescence count for control and Flz-treated neurons at each time point was compared with the signal at 0 time, set to a value of 100% (Fig. 4B). Analysis of the endocytosis assay time course indicated that the rate of GABA_AR endocytosis was also not changed by BZ treatment (remaining surface Bgt fluorescence: at t = 7.5 min, control, 30.3 ± 8.4%, and Flz, 26.7 ± 8.4%, and at t = 15 min, control, 10.0 ± 5.3%; Flz, 20.7 ± 3.9%). The lack of change in exocytosis and endocytosis rates, combined with the observed dramatic decrease in total GABA_AR levels, suggested that BZ treatment promotes GABA_AR degradation, likely through increased lysosomal degradation of endocytosed receptors or increased endoplasmic reticulum-associated protein degradation via the proteasome.

Fig. 4.

GABA_AR endocytosis rate is not increased by BZ treatment. (A) Confocal imaging of GABA_AR endocytosis in control and Flz-treated neurons. Surface GABA_AR in α2^pHBBS neurons were live-labeled with Alexa594::Bgt (red), and then unbound Alexa594::Bgt was removed by washing, followed by incubation at 37 °C. At t = 0, 7.5, and 15 min, samples were removed and fixed. Total receptor number was assayed by permeabilization and staining with anti-GFP antibody (green). Panels below are enlargements of boxed dendrites. (Scale bars, 10 μm.) (B) Graph represents the surface GABA_AR Alexa594::Bgt fluorescence loss over time with endocytosis in control (Con) and Flz-treated neurons (normalized to the t = 0 Alexa594::Bgt surface signal for each treatment, respectively) (10–12 neurons per condition; n = 3 cultures; error bars represent ±SEM).

GABA_AR Surface Levels Are Restored in BZ-Treated Neurons by Leupeptin Treatment.

Cell surface GABA_ARs undergo constitutive endocytosis and can be recycled or targeted for degradation in the lysosome (11). We next analyzed the role of lysosomal degradation in the decrease in surface and total GABA_AR levels observed with BZ exposure. Neurons were treated with or without 250 nM Flz for 24 h followed by control or 2-h 200-μM treatment with the lysosomal protease inhibitor leupeptin. Live confocal microscopy experiments were used to measure surface α2^pHGFP-subtype GABA_AR levels (Fig. 5 A and B). Quantification of receptor surface fluorescence showed a dramatic increase in Flz-conditioned neurons treated with leupeptin (control, 17,808 ± 1,954; Flz, 8,697 ± 891; Flz plus Leu, 16,359 ± 1,219). Leupeptin caused no significant changes in surface GABA_AR cluster fluorescence for control neurons (control plus Leu, 17,447 ± 1,473). A similar rescue of GABA_AR surface cluster area was observed (in μm²): control, 0.30 ± 0.11; Flz, 0.17 ± 0.05; Con plus Leu, 0.29 ± 0.08; Flz plus Leu, 0.27 ± 0.08 (Fig. 5C). Analysis of α2-containing GABA_AR cluster density showed a trend toward higher values with leupeptin treatment, although these did not reach statistical significance (expressed as % control: control, 101 ± 34; Flz, 60 ± 22; Con plus Leu, 138 ± 67; Flz plus Leu, 100 ± 41; n = 8–15 neurons from two to three independent cultures). Together, these data suggest that exposure of neurons to Flz results in enhanced lysosomal degradation of α2-subtype GABA_ARs after their removal from the plasma membrane, reducing the total available GABA_AR pool, decreasing inhibitory synapse size and number along with the efficacy of synaptic inhibition.

Fig. 5.

GABA_AR surface levels are restored in BZ-treated neurons by leupeptin treatment. (A) Live confocal imaging of neurons treated with or without 250 nM Flz for 24 h followed by control (Con) or 2-h 200 μM treatment with the lysosomal inhibitor leupeptin (Leu). (Right) Enlargements of boxed dendrites. (Scale bars, 10 μm.) (B and C) Mean ± SEM surface receptor α2^pHGFP fluorescence and cluster area [*P < 0.01, one-way ANOVA (Bonferonni post test); 10–20 neurons per condition; n = 3 cultures].

Discussion

We have begun to dissect the molecular mechanism underlying the neuronal adaptations in GABAergic inhibition induced by BZ treatment. We used live confocal imaging to identify subtype-specific changes in GABA_AR neuronal surface levels and electrophysiology to measure BZ-dependent changes in inhibition. A comparison of α1- and α2-subtype GABA_ARs showed α2-containing GABA_AR surface levels were greatly diminished by 24-h BZ treatment. Further analysis showed that control neurons had nearly threefold higher total α2 levels compared with Flz-treated neurons. Concurrent with the decrease in GABA_AR surface levels, Flz treatment resulted in a significant decrease in mIPSC amplitude. We demonstrated that the decrease in GABA_AR surface levels occurs in a BZ binding site-dependent manner, as acute treatment with the BZ antagonist Ro 15-1788 restored the BZ-lowered GABA_AR surface levels to control levels. Furthermore, acute Ro 15-1788 treatment returned mIPSC amplitude to control levels. Interestingly, BZ treatment did not alter the basal insertion and endocytosis rates of α2-containing GABA_AR. Rather, BZ treatment promoted the degradation of α2-containing GABA_ARs, decreasing the total available α2 GABA_AR pool. Finally, blockade of lysosomal degradation was able to return α2 GABA_AR surface levels in BZ-treated neurons to control levels. We anticipate that the sustained loss in surface and total GABA_ARs during initial 24-h BZ treatment initiates a series of adaptations that develop into long-term changes in inhibition coincident with the development of BZ tolerance.

There is substantial evidence that a number of molecular processes are altered by short- and long-term BZ treatment, but comparatively little is understood regarding the initiation of these events. BZ treatment in vivo, in cultured neurons and in GABA_AR-transfected cell lines, leads to a reduction in the allosteric coupling of BZ and GABA binding sites (7), suggesting that homeostatic down-regulation occurs to diminish BZ-dependent enhancement of GABA_AR activity. Although earlier experiments suggested that uncoupling manifested over several weeks, BZ allosteric uncoupling has been shown to occur after a single dose of diazepam, with peak attenuation occurring between 4 and 12 h and returning to normal after 24 h (21). This rapid uncoupling is consistent with reports in mammalian (22) and chick neuronal culture: Flz treatment of cultured chick neurons showed uncoupling of the BZ and GABA sites with a t_1/2 of ∼18 h (23). Combined with the rapid 24–48 h recovery from uncoupling both in animals chronically dosed or singly treated with BZ (21, 24), the current evidence suggests that BZs trigger an acute neuronal adaption modifying GABAergic inhibition. Studies of recombinant GABA_AR expressed in Sf9 insect cells reported diazepam-induced uncoupling of both α1(α1β2γ2) and α2 (α2β3γ2), with greater uncoupling of α2-containing receptors (25). This observation in a nonneuronal system suggests that individual subunits are uncoupled by distinct mechanisms, likely to be a posttranslational modification such as phosphorylation of GABA_ARs or an interacting protein (26).

Importantly, many groups have shown that uncoupling occurs in vivo and in cultured systems without any contemporaneous changes in transcription or steady-state mRNA levels (21, 27, 28). These observations and reports of various, differing changes in mRNA levels (and often inverse changes in the same subunit) between different brain regions with chronic BZ treatment suggest that observed transcriptional changes may not directly correspond with functional changes in GABA_AR protein levels, activity, and BZ tolerance (7). Although downstream signaling events after BZ administration are largely unknown, a recent microarray-based analysis following a single dose of diazepam identified a decrease in transcripts for CaMKIIα, BDNF, MKP-1, GIF, c-fos, and NGFI-A (29). Interestingly, the transcript levels had returned to baseline by 40 h for all of the genes except for CaMKIIα, further suggesting that key events may be posttranslational in nature.

Alterations in the regulated assembly, delivery, and removal of GABA_ARs have been proposed as mechanisms for BZ treatment-induced changes of GABA_AR activity, although they have not been thoroughly explored. Mice chronically treated with lorazepam show an increase in flunitrazepam binding in clathrin-coated vesicles and a decrease in synaptic membranes, suggesting sequestration of GABA_ARs with BZ treatment (30). Studies of uncoupling in recombinant expression systems have also suggested BZ-induced internalization of GABA_ARs into an acidic compartment that could retain BZ binding capability (26). Here, we show a mechanism for rapid neuronal adaptation to BZ treatment that occurs via enhanced degradation of α2-containing GABA_ARs, rather than alterations in receptor insertion or removal. The quick recovery of synaptic inhibition and GABA_AR surface levels with 1 h of BZ antagonist Ro 15-1788 treatment is consistent with BZ-induced changes affecting receptor stability rather than protein synthesis. We surmise that the BZ antagonist Ro 15-1788 acts to restore GABA_AR surface levels by blockade of BZ potentiation of GABA, thereby pushing GABAergic inhibition to low levels that are subsequently compensated for by increased GABA_AR trafficking and reduced degradation.

Acute BZ tolerance has been reported with a single BZ dose (31–33), suggesting a link between rapid allosteric uncoupling and tolerance development. The underlying neuronal adaptations that modulate GABAergic signaling may be a highly evolutionarily conserved mechanism because treatment with the GABA_A agonist muscimol in the nematode Caenorhabditis elegans results in selective removal of GABA_AR from synapses during adaptation to muscimol, a process that occurs on a 24-h timescale and can largely be blocked in a mutant with defective lysosomal degradation (34). By providing a mechanism linking BZ treatment of neurons to changes in surface and total levels of GABA_AR subtypes and the efficacy of neuronal inhibition, our results provide a mechanism for the initial adaptations after BZ treatment. A greater understanding of these events and the subsequent long-lasting changes with chronic BZ treatment at an individual GABA_AR subtype level will advance current treatment paradigms, potentially producing less tolerance and potential for abuse of one of the most widely prescribed medications in the Western world.

Materials and Methods

Cell Culture, Transfection, and Expression Constructs.

Hippocampal neurons were prepared from embryonic day 18 rats, and constructs (14) were nucleofected at plating (Amaxa). The α2^pHBBS construct was generated by standard molecular biology techniques and sequenced. All studies were conducted in accordance with the institutional review boards of Tufts University School of Medicine and University of Pittsburgh School of Medicine.

Chemicals and Antibodies.

Flz and Ro 15-1788 (Sigma) treatments were performed with 250 nM Flz and 5 μM Ro 15-1788. Anti-GFP antibodies were used as previously reported (13).

Live Imaging.

Measurements were made on 14–17 d in vitro (DIV) hippocampal neurons at 37 °C in a closed heated chamber continuously perfused with extracellular Hepes-buffered saline, containing the following (in mM): 135 NaCl, 4.7 KCl, 10 Hepes, 11 glucose, 1.2 MgCl₂, and 2.5 CaCl₂ (adjusted to pH 7.4 with NaOH). Images were acquired using a Nikon A1 confocal microscope with a 60× oil objective (N.A., 1.4) at 2× zoom. Data were analyzed in Metamorph (Molecular Devices) as previously described (20), blind to experimental condition.

Insertion and Endocytosis Assays.

Hippocampal neurons (14-17 DIV) expressing α2^pHBBS were assayed for GABA_AR insertion (35) and endocytosis (20) with Bgt (Invitrogen) as previously and described in Results. All incubations were performed in the presence of 150 μM tubocurarine (Sigma) to block Bgt binding to endogenous acetylcholine receptors (36). See SI Materials and Methods for additional information.

Electrophysiology.

Neurons were plated on 12-mm glass coverslips (German glass; VWR) coated with poly-l-lysine (0.5 mg/mL; Sigma) and cultured for 2–3 wk before the recordings. To measure mIPSCs, coverslips were placed in a recording chamber mounted on the stage of an inverted microscope and continually perfused with the following (in mM): 140 NaCl, 4.7 KCl, 10 Hepes, 11 glucose, 2 MgCl₂, and 2.5 CaCl₂ (adjusted to pH 7.4 with NaOH, 295–315 mOsm). For mIPSC recording, the extracellular solution was supplemented with 200 nM TTX, 10 μM CNQX, and 20 μM d-AP5 (Sigma). Borosilicate pipettes (3-6 MΩ) were filled with the following (in mM): 150 CsCl, 10 Hepes, 1.1 EGTA, 2 MgCl₂, 0.1 CaCl₂, 2 Mg²⁺-ATP (adjusted to pH 7.2 with CsOH). Recordings were started 5–10 min after a stable whole-cell access was obtained. See SI Materials and Methods for additional information.