Abstract
Despite 50+ years of clinical use as anxiolytics, anti-convulsants, and sedative/hypnotic agents, the mechanisms underlying benzodiazepine (BZD) tolerance are poorly understood. BZDs potentiate the actions of gamma-aminobutyric acid (GABA), the primary inhibitory neurotransmitter in the adult brain, through positive allosteric modulation of γ2 subunit containing GABA type A receptors (GABAARs). Here we define key molecular events impacting γ2 GABAAR and the inhibitory synapse gephyrin scaffold following initial sustained BZD exposure in vitro and in vivo. Using immunofluorescence and biochemical experiments, we found that cultured cortical neurons treated with the classical BZD, diazepam (DZP), presented no substantial change in surface or synaptic levels of γ2-GABAARs. In contrast, both γ2 and the postsynaptic scaffolding protein gephyrin showed diminished total protein levels following a single DZP treatment in vitro and in mouse cortical tissue. We further identified DZP treatment enhanced phosphorylation of gephyrin Ser270 and increased generation of gephyrin cleavage products. Selective immunoprecipitation of γ2 from cultured neurons revealed enhanced ubiquitination of this subunit following DZP exposure. To assess novel trafficking responses induced by DZP, we employed a γ2 subunit containing an N terminal fluorogen-activating peptide (FAP) and pH-sensitive green fluorescent protein (γ2pHFAP). Live-imaging experiments using γ2pHFAP GABAAR expressing neurons identified enhanced lysosomal targeting of surface GABAARs and increased overall accumulation in vesicular compartments in response to DZP. Using fluorescence resonance energy transfer (FRET) measurements between α2 and γ2 subunits within a GABAAR in neurons, we identified reductions in synaptic clusters of this subpopulation of surface BZD sensitive receptor. Moreover, we found DZP simultaneously enhanced synaptic exchange of both γ2-GABAARs and gephyrin using fluorescence recovery after photobleaching (FRAP) techniques. Finally we provide the first proteomic analysis of the BZD sensitive GABAAR interactome in DZP vs. vehicle treated mice. Collectively, our results indicate DZP exposure elicits down-regulation of gephyrin scaffolding and BZD sensitive GABAAR synaptic availability via multiple dynamic trafficking processes.
Introduction
GABAARs are ligand-gated ionotropic chloride (Cl−) channels responsible for the majority of fast inhibitory neurotransmission in the adult CNS. The most prevalent synaptic GABAAR subtype is composed of two α, two β, and a γ2 subunit forming a heteropentamer (1). Benzodiazepines (BZD) are a widely used clinical sedative-hypnotic drug class that selectively binds between the interface of a γ2 subunit and either an α1/2/3/5 subunit (2). Receptors containing these α subunits are considered to be primarily synaptic, with the exception of α5, which is localized both synaptically and extrasynaptically (3). Positive allosteric modulation by BZD enhances GABAAR inhibition by increasing the binding affinity of GABA and increasing channel opening frequency (1). This potentiating effect of BZD is lost after prolonged or high dose acute exposure (4, 5), characterized first by a loss of sedative/hypnotic activity followed by the anti-convulsant properties behaviorally (6–9). The induction of BZD tolerance occurs in part due to the uncoupling of allosteric actions between GABA and BZD (10, 11), a process that appears to rely on GABAAR receptor internalization (12, 13). We have previously shown that 24 h BZD treatment leads to decreased surface and total levels of the α2 subunit in cultured hippocampal neurons that was dependent on lysosomal-mediated degradation (14); however, the process by which the α2 subunit is selectively targeted to lysosomes is still unknown. GABAAR subunit ubiquitination and subsequent degradation at proteasomes or lysosomes modulates cell surface expression of receptors (15–19). Ubiquitination of the γ2 subunit is the only currently known mechanism shown to target internalized surface GABAARs to lysosomes (16).
Another major regulator of GABAAR efficacy is postsynaptic scaffolding. Confinement at synaptic sites maintains receptors at GABA axonal release sites for activation. Furthermore, this limits receptor diffusion into the extrasynaptic space where internalization occurs (20, 21). The scaffolding protein gephyrin is the main organizer of GABAAR synaptic localization and density, as gephyrin knock-down and knock-out models show dramatic reductions in γ2- and α2-GABAAR clustering (22, 23). Evidence suggests gephyrin interacts directly with GABAAR α1, α2, α3, α5, β2, and β3 subunits (3, 24, 25). Gephyrin recruitment is involved in inhibitory long term potentiation (26, 27), while its dispersal coincides with GABAAR diffusion away from synapses (22, 28). Extensive post-translational modifications influence gephyrin function (29, 30). Accordingly, expression of gephyrin phosphorylation site mutants revealed complex effects on GABAAR diffusion and gephyrin ultrastructure and scaffolding (30, 31). Phosphorylation at the gephyrin serine 270 (Ser270) site has been particularly characterized to negatively modulate scaffold clustering and density, in part by enhancing calpain-1 protease mediated degradation of gephyrin (32). Given the well-established interdependent relationship between gephyrin and the γ2 subunit in maintaining receptor synaptic integrity (22, 23, 33–36), impaired postsynaptic scaffolding should affect both pre-existing and newly inserted GABAAR clustering and ultimately the efficacy of inhibitory neurotransmission. Thus a central unanswered question is if BZD exposure causes changes in gephyrin phosphorylation or protein levels.
Here we demonstrate that 12-24 h treatment with the BZD, diazepam (DZP), leads to a reduction in total γ2 subunit and gephyrin levels in vitro and in vivo. This reduction occurred coincident with enhanced γ2 subunit ubiquitination, but resulted in no significant change in overall γ2 surface levels. Using our recently published dual fluorescent BZD-sensitive GABAAR reporter (γ2pHFAP), we further show that cell surface γ2-GABAARs are more frequently targeted to lysosomes after DZP exposure. Forester resonance energy transfer (FRET) experiments further confirmed specific loss of synaptic α2/γ2 GABAAR levels following DZP. The scaffolding protein gephyrin also demonstrated augmented phosphorylation at Ser270, increased cleavage and was significantly decreased in membrane and cytosolic compartments. Fluorescence recovery after photobleaching (FRAP) assays identified that DZP treatment increased the simultaneous recovery of γ2-GABAAR and gephyrin at synaptic sites, indicating reduced receptor confinement and accelerated exchange between the synaptic and extrasynaptic GABAAR pool. This process could be reversed by the BZD site antagonist Ro 15-1788. Lastly, coimmunoprecipitation, quantitative mass spectrometry and bioinformatics analysis revealed shifts in the γ2-GABAAR interactome towards trafficking pathways in vivo. Together, these data suggest that DZP exposure causes compensatory decrease in inhibitory neurotransmission by reducing BZD-sensitive GABAAR and gephyrin confinement at synapses, and via ubiquitination and lysosomal targeting of γ2.
Results
DZP Exposure Modifies γ2-GABAAR and Gephyrin Levels
We first examined if DZP exposure reduced surface levels of γ2-GABAARs and altered gephyrin S270 phosphorylation in cortical neurons by immunofluorescence (Fig. 1A). Cortical neurons were treated for 24 h with vehicle or 1 μM DZP, then immunostained for surface γ2, followed by permeabilization and immunostaining with GAD65 (glutamic acid decarboxylase 65, a marker for presynaptic GABAergic terminals) and the phospho-S270 specific gephyrin mAb7a antibody (37, 38). Image analysis identified no sizable change in surface synaptic (91.6 ± 5.3%) or extrasynaptic (93.3 ± 3.8%) γ2 intensity in DZP treated neurons relative to control, but DZP induced a significant 18.9 ± 7.4% increase in synaptic phospho-gephyrin (Fig. 1B). No change in extrasynaptic phosphorylated Ser270 gephyrin was measured. We repeated this DZP treatment and examined total and phospho-gephyrin levels in dendrites (Fig. 1C). Again DZP significantly enhanced phospho-S270 gephyrin compared to vehicle (132 ± 12%), while a decrease in overall gephyrin levels was found (69.7 ± 5.4%) (Fig. 1D). Accordingly, the mean ratio of phospho/total gephyrin was 78.1± 21% higher following DZP (Fig. 1D). Complimentary biochemical studies using membrane fractionation were used to compare cytosolic, membrane, and total protein pools in cortical neurons. In agreement with immunofluorescence data, membrane levels of γ2 (97.1 ± 10.2%) were not reduced after 1 μM DZP, although the total pool of γ2 was diminished (79.3 ± 7.3%) (Fig. 2A,B) compared to vehicle. Comparatively, DZP reduced gephyrin in every compartment measured relative to control (cytosol: 87.3 ± 4.6%; membrane: 70.3 ± 10 %; total: 59.1 ± 3.4%). We confirmed the integrity of our fractions using cytosolic and membrane specific markers (Supplemental Fig. 1).
Next we assessed if the decrease in gephyrin and γ2 total levels at 24 hours was a result of altered gene expression. qRT-PCR experiments revealed no difference in gephyrin, γ2, or control GABAAR β3 subunit mRNA levels between vehicle and DZP treated neurons (Fig. 2C). To determine if post-translational modification of γ2 also occurs coincident with decreased γ2 protein levels, we examined ubiquitination of γ2 in response to DZP exposure. We reasoned that changes in ubiquitination of γ2 would likely precede the loss of total γ2 seen at 24 h (Fig. 2A,B). GFP-ubiquitin transfected cortical neurons were treated with vehicle or 1 μM DZP for 12 h. Neurons were lysed under denaturing conditions to isolate the γ2 subunit from the receptor complex (Supplemental Fig. 2). Immunoprecipitation of the γ2 subunit revealed a 2.13 fold increase in ubiquitination in DZP treated neurons relative to vehicle (Fig. 2D,E). Furthermore, just as observed with 24 h DZP treatment, a reduced total pool of γ2 was also found at 12 h (Fig. 2D,E). Notably, this is the first demonstration of endogenous γ2 ubiquitination occurring in neurons (previous findings were of recombinant receptors in HEK cells) (15, 16). To investigate mechanisms underlying reduced gephyrin levels, we examined gephyrin cleavage. Gephyrin is degraded post-translationally by the protease calpain-1 (32, 39, 40), and gephyrin Ser270 phosphorylation promotes cleavage by calpain-1 (32). Consistent with the enhanced gephyrin S270 phosphorylation (Fig. 1) and reduced total levels (Fig. 1,2) we found a significant increase in the ratio of cleaved/full length gephyrin after 24 h DZP in vitro (Fig. 2F,G). We confirmed the identity of the gephyrin cleavage product using a well-characterized glutamate stimulation protocol that induces gephyrin cleavage in cultured neurons (39, 40), a process blocked by calpain-1 inhibition (Supplemental Fig. 3).
Finally, we wanted to determine if similar mechanisms occur in vivo following DZP treatment. Prior publications show that BZDs and metabolites are not present 24 h post-injection due to rapid drug metabolism in rodents (41–44). Furthermore, BZD uncoupling does not persist 24 h after a single dose (15 mg/kg) or 2 week daily DZP treatment, whereas uncoupling can be seen 12 h after a single injection, indicating this is the appropriate timepoint for measuring in vivo loss of γ2-GABAAR function (4). Accordingly, mice were given a single intraperitoneal (IP) injection of 10 mg/kg DZP or vehicle control, and cortex tissues were harvested 12 h later. We found DZP significantly reduced the total pool of γ2 (87.3 ± 3.0%) and gephyrin (73.9 ± 9.1%) relative to vehicle treated mice at 12 h post injection (Fig. 2H,I). These findings indicate both BZD-sensitive GABAARs and gephyrin are downregulated by post-translational mechanisms after initial DZP treatment in vitro and in vivo to temper potentiation of GABAAR function.
DZP Enhances Intracellular Accumulation and Lysosomal Targeting of γ2-GABAARs
We then investigated if surface DZP-sensitive GABAARs are more frequently targeted to lysosomes after DZP exposure by live-imaging. For these experiments we used our recently characterized optical sensor for synaptic GABAAR (γ2pHFAP). This dual reporter is composed of a γ2 subunit tagged with an N terminal pH-sensitive GFP, myc, and the fluorogen-activating peptide DL5 (45). The pH-sensitive GFP tag selectively identifies cell surface GABAARs and the DL5 FAP binds malachite green (MG) dye derivatives including MG-BTau (46–48). MG-BTau is cell impermeable and non-fluorescent until bound by DL5. Upon binding, MG-BTau fluoresces in the far red spectral region (~670 nM). This FAP-dye system allows for selective labeling of surface γ2-containing GABAARs which can then be tracked through various phases of trafficking (45). As previously shown, γ2pHFAP GABAARs are expressed on the neuronal surface, form synaptic clusters, do not perturb neuronal development and show equivalent functional responsiveness to GABA and DZP both in the absence and presence of MG dyes (45). We transfected neurons with γ2pHFAP and treated them with DZP for 8-16 h. Neurons were then pulse-labeled with 100 nM MG-BTau dye and returned to conditioned media at 37°C +/− DZP for 1 h. The lysosomal inhibitor leupeptin (200 μM) and the lysosomal specific dye, Lysotracker (50 nM), were added after 30 min. At the end of the incubation, neurons were washed in 4°C saline to inhibit trafficking and immediately used for live-imaging experiments. Representative images demonstrate MG-BTau labeled γ2pHFAP-GABAARs localized on the cell surface (Fig. 3A) and at synaptic clusters on dendrites (Fig. 3B) based on colocalization with surface specific pHGFP signal. MG-BTau further reveals internalized receptors at lysosomes (Fig. 3C). Image quantification showed synaptic γ2-GABAAR intensity remained largely unchanged (Fig. 3D). Importantly, we found a significant 8.0 ± 2.5% enhancement in the mean intensity of GABAARs labeled with MG-BTau at lysosomes following DZP (Fig. 3E).
We complemented these lysosomal targeting studies with an NH4Cl live-imaging approach that allows us to compare the ratio of cell surface vs. intracellular GABAARs in living neurons. γ2pHFAP expressing neurons were treated with vehicle or DZP for 24 h. Additional control groups included the BZD antagonist Ro 15-1788 (1-2 h) to reverse the effects of DZP. Neurons were actively perfused with HEPES buffered saline (HBS) treatment and an initial image was taken of surface pHGFP receptor signal (Fig. 3F). Neurons were then exposed to pH 7.4 NH4Cl solution to neutralize the pH gradient of all intracellular membrane compartments, revealing internal pools of γ2 containing GABAARs. Analysis revealed no change in surface γ2 levels between treatments (Fig. 3G) consistent with (Fig. 1, 2). However, the number of large intracellular vesicles (circular area ~0.75 μm) containing receptors was significantly enhanced (Fig. 3H), consistent with increased localization in intracellular vesicles. Ro 15-1788 and DZP + Ro 15-1788 treated neurons were not significantly different from vehicle. Overall, these findings suggest γ2-GABAAR ubiquitination, intracellular accumulation, lysosomal targeting and degradation are part of the adaptive response to DZP.
Surface Levels of Synaptic α2/γ2 GABAAR are Decreased Following DZP
Despite the increase in ubiquitination and lysosomal targeting of γ2-GABAARs after DZP, we did not detect decreased overall surface or synaptically localized surface γ2 levels. This suggested two possibilities, one being that as DZP treatment only reduced total γ2 levels to 80% of control in cortical neurons and 85% in vivo, such a slight decrease in surface γ2-GABAARs would be challenging to detect with current methods. Alternatively, there could be an increase in γ2 subunit assembly with BZD-insensitive α subunits (γ2α4β) (49) with a concomitant reduction in surface levels of BZD-sensitive receptors (γ2α1/2/3/5β). Our previous work showed 24 h BZD exposure in hippocampal neurons causes decreased total and surface levels of the α2 GABAAR subunit via lysosomal mediated degradation, without any changes in receptor insertion or removal rate (14). To determine if α2/γ2 GABAARs are specifically decreased by DZP treatment, we developed and employed an intermolecular FRET assay, using pH-sensitive GFP tagged α2pH (50) as a donor fluorophore and a red fluorescent protein (RFP) tagged γ2 subunit (γ2RFP) as an acceptor. FRET is an accurate measurement of molecular proximity at distances of 10-100 Å and is highly efficient if donor and acceptor are within the Förster radius, typically 30-60 Å (3-6 nM), with the efficiency of FRET being dependent on the inverse sixth power of intermolecular separation (51). Synaptic GABAARs exist as five subunits assembled in γ2-α-β-α-β order forming a heteropentamerc ion channel (Fig. 4A). We first expressed α2pH and γ2RFP in neurons and examined their ability to participate in intermolecular FRET. Photobleaching of the acceptor γ2RFP channel enhanced donor α2pH signal (Supplementary Fig. 4), confirming energy transfer from α2pH to γ2RFP. Next, we confirmed measurable FRET only occurs between α2pH/γ2RFP in surface GABAAR at synaptic sites; FRET was blocked with quenching of donor α2pH when the extracellular pH was reduced from 7.4 to 6.0 (Fig. 4A,B). Following FRET assay validation, α2pH/γ2RFP GABAAR expressing neurons were treated for 24 h with vehicle or DZP and examined for total synaptic α2pH and γ2RFP fluorescence as well as the γ2 FRET signal (Fig. 4A). These studies identified a DZP-induced reduction in synaptic α2 (−12.6%), synaptic γ2 (−14.3%) and diminished association of α2 with γ2 in synaptic GABAARs as measured by decreased FRET γ2 signal (−10.6%) (Fig. 4B). In summary, this sensitive FRET method indicates that cortical neurons show a similar susceptibility for α2 subunit downregulation by BZD treatment as seen in hippocampal neurons (14). Furthermore it identifies a loss in a specific pool of surface synaptic BZD-sensitive γ2-GABAAR.
Synaptic Exchange of γ2-GABAARs and Gephyrin are Accelerated after Prolonged DZP Treatment
We previously found 24 h BZD exposure reduces the amplitude of miniature inhibitory postsynaptic currents (mIPSC) (14), suggesting changes in synaptic GABAAR function. Having identified both reductions in gephyrin (Fig. 1,2) and BZD sensitive GABAARs (Fig. 2, 4), we next tested if DZP treatment altered the synaptic retention properties of gephyrin and/or GABAARs. Neurons expressing γ2pHFAP and RFP-gephyrin were used for live-imaging fluorescence recovery after photobleaching experiments (FRAP) to measure synaptic and extrasynaptic exchange following exposure to vehicle, 1 μm DZP, 5 μm Ro 15-1788, or DZP + Ro 15-1788. After an initial image was taken, dendrites were photobleached, and signal recovery was measured every 2 min over 30 min at synaptic sites and extrasynaptic regions (Fig. 5A synapses panel; Fig. 5B larger dendritic region with asterisk denoting extrasynaptic region). MG-BTau dye was added directly after the photobleaching step to immediately re-identify the photobleached surface synaptic GABAARs, and improve spatial measurements (Fig. 5B). These experiments revealed synaptic γ2 turnover rates were nearly doubled in DZP treated neurons, a process reversed by Ro 15-1788 co-treatment (Fig. 5C). DZP also accelerated gephyrin synaptic exchange rates compared to vehicle, with Ro 15-1788 co-treatment restoring exchange to control levels. No significant correlation was found between cluster area measured and fluorescence recovery rates of γ2 and gephyrin across all conditions, suggesting synaptic exchange rate is independent of cluster size (Supplementary Fig. 5). Moreover, no statistical difference was found in γ2 or gephyrin extrasynaptic exchange rates (Fig. 5D). These findings suggest concurrent reduction of gephyrin and GABAAR synaptic confinement is a compensatory response to mitigate prolonged DZP potentiation of GABAARs.
Coimmunoprecipitation and Quantitative Proteomics of γ2 GABAAR following DZP injection
We sought to observe DZP-induced changes in receptor trafficking in vivo. As an orthogonal approach, we utilized label-free quantitative proteomics to measure changes in the quantities of proteins associated with γ2-GABAARs in the cortex of mice after DZP. Cortical tissue was collected from DZP- or vehicle-treated mice 12 h post injection, lysed, and immunoprecipitated with anti-γ2 subunit antibody or IgG control. Following label-free mass spectrometry analysis, spectrum counts were used to assess relative abundance of γ2-associated proteins. A total of 395 proteins was identified using our inclusion criteria: minimum of two peptides; identified in at least three samples overall or in two of three samples in a specific treatment group; demonstrated at least 3:1 enrichment over IgG control across at least three samples overall (Dataset 1). The relative abundance of γ2-GABAAR associated proteins in the DZP group compared to vehicle was used to determine which proteins were significantly (P < 0.1) increased (Table 1) or decreased (Table 2). As a result we identified 46 proteins with elevated levels of interaction with γ2-GABAARs, including 10 proteins that were only found in the DZP treated group (Table 1, not found in vehicle samples, NF-V). Increased interactions were shown for 14-3-3 epsilon, In contrast, 23 proteins were found to coimmunoprecipitate with γ2 less in DZP animals relative to control, seven of which were only present in the vehicle treatment group (Table 2, not found in DZP, NF-DZP). Interestingly, the calcium-sensitive kinase CaMKIIα, which can regulate GABAAR membrane insertion, synaptic retention and drug binding properties (26, 70-72), was found to be significantly decreased in interaction with γ2-GABAAR following DZP injection in vivo.
Bioinformatics Analysis of the γ2 GABAAR interactome
To better understand the consequences of the DZP-induced shift in the γ2-GABAAR protein interaction network, protein fold change data was subjected to core Ingenuity Pathway Analysis (IPA). Top enriched canonical pathways with −log(p-value) > 6.2 are shown in Fig. 6A. Notably, GABA receptor signaling pathways were highly enriched, as expected, although IPA was unable to determine pathway activation status by z-score analysis. γ2-GABAAR association with proteins involved in 14-3-3 mediated signaling and RhoA signaling pathways were greatly increased after DZP (Fig. 6A, orange), while interaction with proteins involved in EIF2 signaling and sirtuin signaling pathways were reduced (Fig. 6A, blue) relative to vehicle.
We further examined alterations in functional network association, identifying changes in key trafficking, localization, and cell adhesion pathways in DZP treated mice. Fig. 6B lists γ2-GABAAR-associated proteins found to be elevated with DZP, contributing to processes such as endocytosis, developmental process of the synapse, cell-cell contact, and quantity of cell-cell contacts (activation z-scores > 2.5). As an additional measurement, we performed gene ontology (GO) database analysis of proteins which were found to be significantly increased (P < 0.1) in DZP treated mice relative to vehicle control (Table 3). GO analysis identified DZP treatment enriched for a number of intracellular trafficking and cellular localization biological pathways that were consistent with the functional network analysis in IPA. Taken together, these results suggest DZP modifies intracellular and surface trafficking of γ2-GABAARs both in vitro and in vivo.
Discussion
This work identifies key trafficking pathways involved in GABAAR neuroplasticity in response to initial DZP exposure. Using a combination of biochemical and imaging techniques, we identified total γ2 subunit levels are diminished in response to 12-24 h of DZP exposure in vitro and in vivo. Concurrent with the decrease in the overall γ2 pool, we found DZP treatment enhanced ubiquitination of this subunit. Use of an innovative optical sensor for BZD sensitive GABAAR (γ2pHFAP) in combination with MG dye pulse-labeling approaches revealed DZP exposure moderately enhanced targeting of surface γ2-GABAARs to lysosomes. Live-imaging experiments with pH 7.4 NH4Cl revealed increased intracellular receptor pools, providing further evidence that DZP enhances GABAAR lysosomal accumulation, a response reversed by BZD antagonist Ro 15-1788 treatment. We used novel intersubunit FRET based live-imaging to identify that surface synaptic α2/γ2 GABAARs were specifically decreased after DZP, suggesting these receptor complexes were subjected to ubiquitination, lysosomal targeting, and degradation. In addition to DZP modulation of receptor trafficking, the postsynaptic scaffolding protein gephyrin demonstrated significant plasticity including increased Ser270 phosphorylation and production of gephyrin proteolytic fragments, concurrent with a decrease in total and membrane associated gephyrin levels. Given the fundamental role of gephyrin in scaffolding GABAARs and regulating synaptic confinement, we used simultaneous FRAP live-imaging of receptors and scaffold in neurons to monitor inhibitory synaptic dynamics. We found ~24 h DZP exposure accelerates both the rate of gephyrin and GABAAR exchange at synapses as shown by enhanced fluorescence recovery rates. Control experiments using the BZD antagonist Ro 15-1788 were able to reverse the DZP induced loss of synaptic confinement, reducing gephyrin and GABAAR mobility back to vehicle levels. Finally, we used label-free quantitative mass spectrometry and bioinformatics to identify key changes in γ2-GABAAR protein association in vivo suggesting enhanced accumulation in cell surface and intracellular trafficking networks. Collectively, this work defines a DZP-induced reduction of gephyrin scaffolding coupled with increased synaptic exchange of gephyrin and GABAARs. This dynamic flux of GABAARs between synapses and the extrasynaptic space was associated with enhanced γ2-GABAAR accumulation in intracellular vesicles and γ2-GABAAR subtype specific lysosomal degradation. We propose DZP treatment alters these key intracellular and surface trafficking pathways ultimately diminishing responsiveness to DZP.
Numerous classical studies have examined gene and protein expression adaptations in GABAAR subunits after BZD exposure with minimal agreement that a specific change occurs (1, 2, 6). Here molecular mechanistic insight is provided, through direct measurements of enhanced ubiquitination of the γ2 subunit (Fig. 2), lysosomal targeting (Fig. 3), reduced surface synaptic α2/γ2 GABAAR levels (Fig. 4), and reduced synaptic confinement (Fig. 5) of DZP-sensitive GABAARs. Together this suggests BZD exposure primarily decreases synaptic retention of γ2 containing GABAAR while downregulating surface levels of specific α subunit levels. Ubiquitination of the γ2 subunit by the E3 ligase Ring Finger Protein 34 (RNF 34) (36) is the only currently known mechanism targeting internalized synaptic GABAARs to lysosomes (37). Due to the requirement of the γ2 subunit in all BZD-sensitive GABAARs, it is likely that ubiquitination of the γ2 subunit is a contributing factor for increased lysosomal-mediated degradation in response to DZP. Despite a small decrease in the γ2 total protein, changes in surface levels were not significant by biochemical approaches, consistent with evidence that γ2-GABAAR surface levels are tightly regulated to maintain baseline inhibition and prevent excitotoxicity. For example, in heterozygous γ2 knockout mice a 50% reduction in γ2 levels appears to be compensated by increased cell surface trafficking, resulting in only approximately a 20% reduction in BZD binding sites in the cortex and a limited reduction in synaptic GABAAR clusters. In contrast, homozygous γ2 knockout mice show a complete loss of behavioral drug response to BZD and over 94% of the BZD sites in the brain (GABA binding sites unchanged) and early lethality (52). Similarly, studies have shown that prolonged GABAAR agonist or BZD application increases γ2 GABAAR internalization in cultured neurons, while surface GABAAR levels remain unchanged (53, 54). Importantly, by using high sensitivity surface GABAAR intersubunit FRET measurements we were able to detect a decrease in BZD sensitive α2/ γ2 GABAARs (Fig 4).
The role of inhibitory scaffolding changes in responsiveness to BZD has been largely under investigated. Phosphorylation of gephyrin at Ser270 is mediated by CDK5 and GSK3β, while a partnering and functionally relevant Ser268 site is regulated by ERK1/2 (31). While the exact signaling mechanism responsible for gephyrin remodeling and phosphorylation in our study is unclear, we have previously shown 30 min treatment with the GABAAR agonist muscimol leads to ERK1/2/BDNF signaling, decreased gephyrin synaptic and total levels, and decreased γ2-GABAAR at synapses and potentiation by BZDs (55). Thus, changes in receptor and scaffold synaptic level and function can occur on the timescale of minutes. Similarly, calpain mediated gephyrin cleavage can occur within 1 minute in hippocampal membranes (56), and cleavage products are increased following in vitro ischemia at 30 min and up to 48 hours following ischemic events in vivo (39). Additionally, chemically-induced inhibitory long-term potentiation (iLTP) protocols demonstrate gephyrin accumulation occurs concurrent with the synaptic recruitment of GABAARs within 20 min (26). Collectively, these proteins display a high degree of interdependence across different experimental paradigms of inhibitory synapse plasticity occurring over minutes to days.
A recent work has demonstrated 12 h DZP treatment of organotypic hippocampal slices expressing eGFP-gephyrin caused enhanced gephyrin mobility at synapses and reduced gephyrin cluster size (57). Here we found the synaptic exchange rate of γ2 GABAARs and gephyrin to be nearly doubled at synapses in cortical neurons after ~24 h DZP exposure (Fig. 4). This effect occurred coincident with the formation of truncated gephyrin cleavage products (Fig. 2), which has previously been shown to decrease γ2 synaptic levels (39). These findings are also consistent with our previous work showing RNAi gephyrin knockdown doubles the rate of γ2-GABAAR turnover at synaptic sites (22). Later quantum dot single particle tracking studies confirmed γ2 synaptic residency time is linked to gephyrin scaffolding levels (58). Importantly, GABAAR diffusion dynamics also reciprocally regulate gephyrin scaffolding levels (59), suggesting gephyrin and GABAARs synaptic residency are often functionally coupled. Accordingly, γ2 subunit and gephyrin levels both decrease in responses to other stimuli including status epilepticus (60) or prolonged inhibition of IP3 receptor-dependent signaling (61).
Increasing receptor synaptic retention enhances synaptic currents, while enhanced receptor diffusion via decreased scaffold interactions reduces synaptic currents. For example, reduction of gephyrin binding by replacement of the α1 GABAAR subunit gephyrin binding domain with non gephyrin binding homologous region of the α6 subunit results in faster receptor diffusion rates and a direct reduction in mIPSC amplitude (62). Similarly, enhanced diffusion of GABAARs following estradiol treatment also reduces mIPSCs in cultured neurons and in hippocampal slices (63). In contrast, brief DZP exposure (< 1h) reduces GABAAR synaptic mobility (64) without a change in surface levels (65), consistent with initial synaptic potentiation of GABAAR neurotransmission by DZP. Together with our current findings, this suggests post-translational modifications on GABAAR subunits or gephyrin that enhance receptor diffusion are a likely key step leading to functional tolerance to BZD drugs.
It is a significant technical challenge to examine dynamic alterations in receptor trafficking occurring in vivo. To overcome this we examined changes in γ2-GABAAR protein association following DZP injection in mice using quantitative proteomics and bioinformatics analysis. This work revealed shifts toward γ2-GABAAR association with new protein pathway networks associated with cell surface adhesion, intracellular junctions, synaptic plasticity, endocytosis & recycling and ubiquitination (Fig. 6, Table 3), confirming similar fluctuations in membrane and intracellular trafficking occur in vivo and in vitro after DZP. Recent inhibitory synapse proteomics studies have identified a number of new protein synaptic constituents or modulators of GABAAR function (66–70). We show here that proteins known to have roles in synaptic function and trafficking of membrane receptors show changes in their association with γ2-receptors. For example, the calcium-sensitive kinase CaMKIIα was found to be significantly decreased in interaction with γ2-GABAAR following DZP, which can regulate GABAAR membrane insertion, synaptic retention and drug binding properties (26, 71–73) (Table 2). DZP was also found to significantly increase γ2 association with 14-3-3 protein family members (Table 1), which are known mediators of GABAAR surface and intracellular trafficking (74, 75). γ2 coassembly with the GABAAR α5 subunit was also enhanced after DZP (Table 1). Interestingly, the α5 subunit is required for the development of BZD sedative tolerance in mice (76). Future follow up studies are needed to dissect the individual roles of proteins found to be significantly altered in their association with GABAAR, and their physiological and pharmacological importance to BZD tolerance and inhibitory neurotransmission.
Through application of novel and highly sensitive fluorescence imaging approaches combined with in vivo proteomics, we provide unprecedented resolution at both the level of the single neuron and cortex of GABAAR synapse plasticity induced by BZDs. Our study reveals that sustained initial DZP treatment diminishes synaptic BZD sensitive GABAAR availability through multiple fundamental cellular mechanisms: through reduction of the post-synaptic scaffolding protein gephyrin; shifts toward intracellular trafficking pathways and targeting of receptors for lysosomal degradation; and enhanced synaptic exchange of both gephyrin and GABAARs. Proteomic and bioinformatics studies using DZP-treated mouse brain tissue provide further evidence that altered γ2-GABAAR surface and intracellular trafficking mechanisms play a critical role to the response to DZP in vivo. These results define key events leading to BZD irresponsiveness in initial sustained drug exposure. Future studies utilizing this dual approach will address the neuroadaptations produced by long term BZD use to systematically identify the effects of a critical drug class that has seen a tripling in prescription numbers over the last two decades (77).
Materials and Methods
Biochemical experiments, confocal imaging, image analysis
Detailed methods of cell culture, reagents, animals, imaging, biochemical protocols, mass spectrometry, bioinformatics and analysis are described in SI Materials and Methods.
Supplementary Data
IN EXCEL FILE
Dataset 1. Proteins Identified by Label-Free Mass Spectrometry and Spectral Counts. Identification criteria were: minimum of two peptides; 96% peptide threshold; 1% FDR; 99% protein threshold; identified in at least 3 samples overall or identified 2 of 3 times in a specific treatment group; demonstrated at least 3:1 enrichment over IgG control across at least 3 samples overall.
ACKNOWLEDGEMENTS
This work was supported by funding from a National Institute of Health Training Grant (T32GM008424), National Institutes of Health Predoctoral Individual National Research Service Award (F31 MH117839), University of Pittsburgh Pharmacology and Chemical Biology Fellowship (William C. deGroat Neuropharmacology Departmental Fellowship), NARSAD young investigator grant and Pharmacology and Chemical Biology Startup Funds. We thank Jonathan Beckel for technical advice on qRT-PCR and Katarina Vajn for assistance with neuronal cultures. Mass spectrometry analyses were conducted at the UTHSCSA Institutional Mass Spectrometry Laboratory, supported in part by UTHSCSA and the University of Texas System for purchase of the Orbitrap Fusion Lumos mass spectrometer. The expert technical assistance of Sammy Pardo and Dana Molleur is gratefully acknowledged.