Grp94 Protein Delivers γ-Aminobutyric Acid Type A (GABAA) Receptors to Hrd1 Protein-mediated Endoplasmic Reticulum-associated Degradation*

Proteostasis maintenance of γ-aminobutyric acid type A (GABAA) receptors dictates their function in controlling neuronal inhibition in mammalian central nervous systems. However, as a multisubunit, multispan, integral membrane protein, even wild type subunits of GABAA receptors fold and assemble inefficiently in the endoplasmic reticulum (ER). Unassembled and misfolded subunits undergo ER-associated degradation (ERAD), but this degradation process remains poorly understood for GABAA receptors. Here, using the α1 subunits of GABAA receptors as a model substrate, we demonstrated that Grp94, a metazoan-specific Hsp90 in the ER lumen, uses its middle domain to interact with the α1 subunits and positively regulates their ERAD. OS-9, an ER-resident lectin, acts downstream of Grp94 to further recognize misfolded α1 subunits in a glycan-dependent manner. This delivers misfolded α1 subunits to the Hrd1-mediated ubiquitination and the valosin-containing protein-mediated extraction pathway. Repressing the initial ERAD recognition step by inhibiting Grp94 enhances the functional surface expression of misfolding-prone α1(A322D) subunits, which causes autosomal dominant juvenile myoclonic epilepsy. This study clarifies a Grp94-mediated ERAD pathway for GABAA receptors, which provides a novel way to finely tune their function in physiological and pathophysiological conditions.

Proteostasis maintenance of membrane proteins is essential for their function in normal physiology (1)(2)(3)(4). Because of the complex structure of membrane proteins, their folding and assembly in the endoplasmic reticulum (ER) 2 are often inefficient. Many pathogenic mutations aggravate such inefficient processes. Misfolded and unassembled membrane proteins are rapidly disposed via the ER-associated degradation (ERAD) pathway (5)(6)(7)(8). Excessive ERAD is linked to an increasing number of human diseases (9). The ER proteostasis network regulates the protein folding, assembly, and degradation processes. However, such processes and ER proteostasis network components for integral membrane proteins remain poorly understood.
We use ␥-aminobutyric acid type A (GABA A ) receptors as a model to elucidate the ERAD pathway. GABA A receptors are the primary inhibitory neurotransmitter-gated ion channels in the mammalian central nervous system (10), belonging to the Cys loop superfamily, which also includes nicotinic acetylcholine receptors (11,12). Remarkably, under physiological conditions only 20 -30% of newly synthesized wild type (WT) subunits of nicotinic acetylcholine receptors reach their intended destination at the plasma membrane (13). This is likely the case for GABA A receptors as well (14). Moreover, proteasome inhibition substantially accumulates WT ␣1 subunits of GABA A receptors in cells, supporting that they are an ERAD substrate (15). Such an unproductive surface trafficking and extensive ERAD is also observed for many other WT membrane proteins, including cystic fibrosis transmembrane conductance regulator (16,17) and peripheral myelin protein 22 (18). In humans, GABA A receptors have at least 19 subunits (␣1-6, ␤1-3, ␥1-3, ␦, ⑀, , , 1-3). The most common type in the human brain contains two ␣1 subunits, two ␤2 subunits, and one ␥2 subunit. Individual subunits need to fold into their native structures in the ER (19,20) and assemble with other subunits correctly on the ER membrane to form a heteropentamer (14,21,22) for subsequent trafficking to the plasma membrane. The recent crystal structure of human ␤3 subunits confirms the long-predicted GABA A receptor subunit topology (23). Each subunit has the following: a large extracellular (or ER luminal) N terminus; four transmembrane (TM) helices (TM1-TM4, with the TM2 domain lining the interior of the pore); a large intracellular loop connecting TM3 and TM4; and a short extracellular (or ER luminal) C terminus (Fig. 1A). Loss of function of GABA A receptors causes idiopathic epilepsies (24 -26) and many other psychiatric and neurological disorders (27). Numerous mutations in GABA A receptors predispose them to misfolding or insufficient assembly and thus excessive ERAD (28,29). For example, the missense A322D mutation in the TM3 domain of the ␣1 subunit leads to its misfolding and thus loss of functional channels on the plasma membrane (30), causing autosomal dominant juvenile myoclonic epilepsy (31).
Cells use the ERAD pathway to recognize misfolded proteins in the ER, dislocate them from the ER membrane, ubiquitinate them using various ubiquitin E3 ligases, and target them for degradation by cytosolic 26S proteasome (5,6,(32)(33)(34). This process is accomplished by the synchronized action of a series of chaperones and ERAD factors in the ER and cytosol, which are collectively called the ERAD machinery. Much of our understanding about the basic principles of ERAD comes from genetic and biochemical studies in yeast (8,35), and recently, more knowledge has been gained in mammalian cells (36). The core ERAD machinery is well conserved from yeast to human, although the ERAD system in mammalian cells is much more complex. Here, we focused on elucidating the ERAD pathway of GABA A receptors in HEK293 cells, which is largely unexplored. We demonstrated that glucose-regulated protein 94 (Grp94) and osteosarcoma amplified 9 (OS-9) recognize misfolded WT ␣1 subunits in the ER lumen and deliver them for ubiquitination by Hrd1 (gene name SYVN1).
To generate monoclonal HEK293 cells stably expressing ␣1(A322D)␤2␥2 receptors, the ␣1(A322D) sequence was subcloned into a pIRES2-EGFP bicistronic vector (Clontech) using EcoRI and SacII restriction sites, which would allow the simultaneous expression of ␣1 subunits and enhanced GFP separately but from the same RNA transcript. After transfection and G-418 treatment, cells that are GFP-positive were considered as those successfully transfected with the ␣1(A322D) subunit. GFP-positive cells were further diluted into 96-well plates, allowing a single cell distribution in each well. Cells with robust GFP signals were further selected to grow to population.
Stable shGrp94 and shControl HEK293T cells were a kind gift of Professor Yair Argon (University of Pennsylvania) (39). They were maintained in DMEM with 10% heat-inactivated fetal bovine serum (Sigma) and 1% penicillin/streptomycin (Hyclone) at 37°C in 5% CO 2 . Two g/ml puromycin (Dot Scientific) was added to maintain the colony.
Western Blot Analysis-Cells were harvested with TrypLE Express and then lysed with lysis buffer (50 mM Tris, pH 7.5, 150 mM NaCl, and 1% Triton X-100) supplemented with the Roche Applied Science complete protease inhibitor mixture. Lysates were cleared by centrifugation (15,000 ϫ g, 10 min, 4°C). Protein concentration was determined by MicroBCA assay (Pierce). Aliquots of cell lysates were separated in an SDS-8% polyacrylamide gel, and Western blot analysis was performed using the appropriate antibodies. Band intensity was quantified using ImageJ software from the National Institutes of Health.
Quantitative RT-PCR-The relative expression levels of target genes were analyzed using quantitative RT-PCR. Total RNA was extracted from cells using the RNeasy mini kit (Qiagen catalogue no. 74104). cDNA was synthesized from 500 ng of total RNA using the QuantiTect reverse transcription kit (Qiagen catalogue no. 205311). Quantitative PCRs (40 cycles of 15 s at 94°C, 30 s at 57°C, and 30 s at 72°C) were performed using cDNA, the QuantiTect SYBR Green PCR kit (Qiagen catalogue no. 204143), and corresponding primers in the StepOnePlus system (Applied Biosystems). The results were analyzed using StepOne version 2.2 software (Applied Biosystems). The forward and reverse primers for CHOP are 5Ј-GGAAACAGAG-TGGTCATTCCC-3Ј and 5Ј-CTGCTTGAGCCGTTCATT-CTC-3Ј; the forward and reverse primers for Grp94 are 5Ј-GGCCAGTTTGGTGTCGGTTT-3Ј and 5Ј-CGTTCCCCGT-CCTAGAGTGTT-3Ј; the forward and reverse primers for GAPDH (housekeeping gene control) are 5Ј-GTCGGAGTCA-ACGGATT-3Ј and 5Ј-AAGCTTCCCGTTCTCAG-3Ј. The threshold cycle (C T ) was extracted from the PCR amplification plot, and the ⌬C T value was defined as follows: ⌬C T ϭ C T (target gene) Ϫ C T (housekeeping gene). The relative mRNA expression level of target genes of treated cells was normalized to that of control cells as follows: relative mRNA expression level ϭ 2 exp(Ϫ(⌬C T (treated cells) Ϫ ⌬C T (control cells))). Each data point was evaluated in triplicate and measured using two biological replicates.
Cycloheximide (CHX) Chase Assay-HEK293 cells were seeded at 2.5 ϫ 10 5 cells per well in 6-well plates and incubated at 37°C overnight. Cells were then transfected with the indicated siRNAs or plasmids for 48 h prior to CHX chase. To stop protein translation, cells were treated with 100 g/ml CHX (Amresco). Cells were then chased for the indicated time, harvested, and lysed for SDS-PAGE and Western blot analysis.
Biotinylation of Cell Surface Proteins-HEK293 cells and SH-SY5Y cells stably overexpressing ␣1␤2␥2 or ␣1(A322D)␤2␥2 receptors were plated in 10-cm dishes for surface biotinylation experiments according to our published procedure (40). Briefly, intact cells were washed twice with ice-cold PBS and incubated with the membrane-impermeable biotinylation reagent Sulfo-NHS SS-Biotin (0.5 mg/ml; Pierce) in PBS containing 0.1 mM CaCl 2 and 1 mM MgCl 2 (PBS ϩ CM) for 30 min at 4°C to label surface membrane proteins. To quench the reaction, cells were incubated with 10 mM glycine in ice-cold PBS ϩ CM twice for 5 min at 4°C. Sulfhydryl groups were blocked by incubating the cells with 5 nM N-ethylmaleimide in PBS for 15 min at room temperature. Cells were solubilized for 1 h at 4°C in lysis buffer (1% Triton X-100, 50 mM Tris-HCl, 150 mM NaCl, and 5 mM EDTA, pH 7.5) supplemented with the complete protease inhibitor mixture (Roche Applied Science) and 5 mM N-ethylmaleimide. The lysates were cleared by centrifugation (16,000 ϫ g, 10 min at 4°C) to pellet cellular debris. The supernatant contained the biotinylated surface proteins. The concentration of the supernatant was measured using MicroBCA assay (Pierce). Biotinylated surface proteins were affinity-purified from the above supernatant by incubating for 1 h at 4°C with 100 l of immobilized neutravidin-conjugated agarose bead slurry (Pierce). The samples were then subjected to centrifugation (16,000 ϫ g, 10 min, at 4°C). The beads were washed six times with buffer (0.5% Triton X-100, 50 mM Tris-HCl, 150 mM NaCl, and 5 mM EDTA, pH 7.5). Surface proteins were eluted from beads by boiling for 5 min with 200 l of LSB/urea buffer (2ϫ Laemmli sample buffer (LSB) with 100 mM DTT and 6 M urea, pH 6.8) for SDS-PAGE and Western blotting analysis. The Na ϩ /K ϩ -ATPase ␣ chain serves as a loading control for biotinylated membrane proteins.
Immunoprecipitation-For immunoprecipitation using cell lysates (500 g), they were pre-cleared with 30 l of protein A/G plus-agarose beads (Santa Cruz Biotechnology) and 1.0 g of normal mouse IgG for 1 h at 4°C to remove nonspecific binding proteins. The pre-cleared cell lysates were incubated with 2.0 g of mouse anti-␣1 antibody (clone BD24) for 1 h at 4°C and then with 30 l of protein A/G plus agarose beads overnight at 4°C. For immunoprecipitation using the mouse brain homogenates (1 mg), they were pre-cleared with 30 l of protein A/G plus-agarose beads (Santa Cruz Biotechnology) and 1.0 g of normal rat IgG for 1 h at 4°C to remove nonspecific binding proteins. The pre-cleared complex was incubated with 2.0 g of rat anti-Grp94 antibody for 1 h at 4°C and then with 30 l of protein A/G plus agarose beads overnight at 4°C. Afterward, the beads were collected by centrifugation at 8000 ϫ g for 30 s and washed three times with lysis buffer. The complex was eluted by incubation with 30 l of SDS loading buffer in the presence of DTT. The immunopurified eluents were separated in SDS-8% polyacrylamide gel, and Western blot analysis was performed using appropriate antibodies.
HEK293 cells stably expressing (FLAG-␣1)␤2␥2 GABA A receptors were transfected with the indicated siRNA or plasmids for 48 h. Then Triton X-100 cell extracts (500 g) were pre-cleared with 30 l of protein A/G plus-agarose beads (Santa Cruz Biotechnology) and 1.0 g of normal mouse IgG for 1 h at 4°C to remove nonspecific binding proteins. The pre-cleared cell lysates were incubated with anti-FLAG M2 magnetic beads (Sigma) for 1 h at room temperature. Afterward, the beads were collected by a magnetic separator (Promega) and washed three times with lysis buffer. The complex was eluted by incubation with 30 l of FLAG peptides (1 mg/ml) or SDS loading buffer in the presence of DTT. The immunopurified eluents were separated in SDS-8% polyacrylamide gel, and Western blot analysis was performed using appropriate antibodies.
The cross-linking reaction was carried out as before with modification (41). Cells in 10-cm dishes were treated with 10 M proteasome inhibitor MG-132 for 2 h before harvesting. Cells were then washed with DPBS and cross-linked by incubation with 1.5 mM dithiobis(succinimidyl propionate) (DSP) in DPBS for 15 min at room temperature. The DSP reaction was quenched by the addition of 10 mM Tris buffer, pH 7.5. Cells were then scraped into microtubes, pelleted, and washed with DPBS. Then cells were lysed in the Triton X-100 lysis buffer, and the total cell lysates were immunoprecipitated with a mouse anti-␣1 antibody or anti-FLAG M2 magnetic beads. The immunoisolated eluents were blotted with rabbit anti-OS-9 and rabbit anti-␣1 subunit antibodies.
In Vivo Ubiquitination Assay-HEK293 cells stably expressing (FLAG-␣1)␤2␥2 GABA A receptors were transfected with HA-tagged ubiquitin constructs together with the indicated siRNAs or plasmids. Two hours before harvesting cells, 10 M proteasome inhibitor MG-132 was added. Forty eight hours post-transfection, Triton X-100 cell extracts (500 g) were immunoprecipitated with anti-FLAG M2 magnetic beads, and the immunoisolated eluents were subjected to SDS-PAGE and blotted with a mouse anti-ubiquitin antibody.
Mouse Brain Homogenization-C57BL/6J mice (The Jackson Laboratory) at 6 -10 weeks were sacrificed, and the cortex was isolated and homogenized in lysis buffer (25 mM Tris, pH 7.6, 150 mM NaCl, 1 mM EDTA, and 2% Triton X-100) supplemented with the complete protease inhibitor mixture from Roche Applied Science. The sample was centrifuged at 800 ϫ g for 10 min at 4°C. The pellet was re-homogenized in the same lysis buffer and centrifuged at 800 ϫ g for 10 min at 4°C. The combined supernatants were placed on a rotating device for 2 h at 4°C and then centrifuged at 15,000 ϫ g for 30 min at 4°C. The resulting supernatant was collected as mouse brain homogenate, and its protein concentration was determined by a MicroBCA assay (Pierce). This animal study followed the guidelines of the Institutional Animal Care and Use Committees (IACUC) at Case Western Reserve University and was carried out in agreement with the recommendation of the American Veterinary Medical Association Panel on Euthanasia.
Whole-cell Patch Clamp Electrophysiology Recording-Wholecell currents were recorded using monoclonal HEK293 cells stably expressing ␣1(A322D)␤2␥2 receptors. The glass electrodes were pulled from thin-walled borosilicate capillary glass . The whole-cell GABA-induced currents were recorded at a holding potential of Ϫ60 mV in voltage clamp mode using an Axopatch 200B amplifier (Molecular Devices). The signals were filtered at 2 kHz and detected at 10 kHz using pClamp10 acquisition software.
Statistical Analysis-All data are presented as mean Ϯ S.E., and any statistical significance was calculated using two-tailed Student's t test.

Results
Grp94 Positively Regulates the ERAD of the ␣1 Subunit of GABA A Receptors-Grp94, a metazoan-specific member of the Hsp90 family in the ER lumen, plays an important role in the recognition of the following two ERAD substrates: mutant ␣ 1 -antitrypsin (42,43) and mutant myocilin (44). Probably a more prominent role of Grp94 is to assist the folding of a very limited subset of client proteins, including immunoglobulins (45), multiple Toll-like receptors and integrins (46,47), insulinlike growth factors (48), and low density lipoprotein receptorrelated protein 6 (LRP6) (49). Therefore, we began to clarify whether Grp94 promotes the folding or degradation of the WT ␣1 subunit of GABA A receptors. Transient knockdown of Grp94 using small interfering RNA (siRNA) (49% knockdown efficiency) significantly increased the total ␣1 subunit protein level in HEK293 cells stably expressing WT ␣1␤2␥2 GABA A receptors (Fig. 1B). This suggested a possible role of Grp94 in positively regulating the ERAD of ␣1 subunits. Furthermore, stable depletion of Grp94 using shRNA (Ͼ95% knockdown efficiency) in HEK293T cells resulted in a more dramatic increase of the steady state level of the ␣1 subunit (Fig. 1C), indicating that the effect of Grp94 on the ␣1 subunit does not depend on FIGURE 1. Grp94 positively regulates the ERAD of the ␣1 subunit of GABA A receptors. A, subunit topology of GABA A receptors. The schematic is built from the crystal structure of the ␤3 subunit (Protein Data Bank code 4COF). It has a large extracellular (or ER luminal) N terminus, four transmembrane (TM) helices (TM1-TM4), a large intracellular loop connecting TM3 and TM4, and a short extracellular (or ER luminal) C terminus. The two cysteines that form the signature disulfide bond are shown as sphere models. B, knocking down Grp94 increases the total and cell surface protein levels of WT ␣1 subunits. HEK293 cells stably expressing ␣1␤2␥2 GABA A receptors were transfected with non-targeting (NT) control siRNA or siRNA against Grp94. Forty eight hours post-siRNA transfection, cells were lysed, and the total cell lysates were subjected to SDS-PAGE and immunoblotting (IB) with mouse anti-␣1 or rat anti-Grp94 antibodies (n ϭ 3). ␤-Actin serves as a total protein loading control. Alternatively, the surface ␣1 subunits were measured using a cell surface protein biotinylation assay (n ϭ 3). The Na ϩ /K ϩ -ATPase ␣ chain serves as a loading control for biotinylated membrane proteins. Protein band quantifications using ImageJ are shown on the bottom panel. C, stably depleting Grp94 increases the total and cell surface protein levels of WT ␣1 subunits. HEK293T cells stably expressing control (Ctr) or Grp94 shRNA were transiently transfected with WT ␣1 plasmids together with ␤2 and ␥2 subunit plasmids. Forty eight hours post-transfection, cells were lysed and treated as in B (n ϭ 3). D, Grp94 knockdown does not significantly influence the mRNA level of CHOP by using quantitative RT-PCR analysis. HEK293 cells stably expressing ␣1␤2␥2 GABA A receptors were transfected with non-targeting control siRNA or siRNA against Grp94 for 48 h, or cells were treated with thapsigargin (1 M, 6 h) as a positive control, which induces CHOP mRNA expression. Then total RNA was extracted from the cells and reverse-transcribed to cDNA before being subjected to quantitative PCR analysis. The experiments were done using two biological replicates in triplicate each time. NS, not significantly. E, knocking down Grp94 increases the half-life of the ␣1 subunit determined by CHX chase analysis. HEK293 cells stably expressing ␣1␤2␥2 GABA A receptors were transfected with non-targeting control siRNA or siRNA against Grp94 for 48 h and then chased for the indicated time with CHX (100 g/ml), which inhibits protein synthesis. Cells were lysed and subjected to SDS-PAGE and immunoblotting analysis (n ϭ 3). Degradation kinetics was plotted by quantifying ␣1 intensity against time after CHX addition on the bottom panel. F, stably depleting Grp94 increases the half-life of the ␣1 subunit determined by CHX chase analysis. HEK293T cells stably expressing control (Ctr) or Grp94 shRNA were transiently transfected with WT ␣1 plasmids together with ␤2 and ␥2 subunit plasmids for 48 h and then chased for the indicated time with CHX (100 g/ml). Cells were then treated as in E (n ϭ 3). whether the depletion of Grp94 is short term or long term (also see under "Discussion").
Quantitative RT-PCR analysis showed that reducing the Grp94 level genetically did not significantly increase the mRNA level of CHOP, a pro-apoptotic marker, indicating that this operation did not cause cell toxicity (Fig. 1D). Also, it has been reported that silencing Grp94 only induces the expression of a limited set of proteostasis network components, which is distinct from activating the unfolded protein response (39). In response to misfolded proteins in the ER, the unfolded protein response adjusts the cellular proteostasis by translational attenuation and transcriptional remodeling of the ER proteostasis network, including some ERAD components (50,51). Therefore, it is unlikely that Grp94 acts as a general ERAD substraterecruiting chaperone. We next manipulated the Grp94 level genetically and evaluated the influence on the degradation rate of ␣1 subunits using the CHX chase assay. The ␣1 subunit has a half-life of 1.6 h when fitted to a single exponential function (Fig. 1E). Remarkably, transiently depleting Grp94 using siRNA substantially increased its half-life to 3.5 h (Fig. 1E), indicating that Grp94 positively regulates the ␣1 subunit degradation. Consistently, permanent inactivation of Grp94 reduced the degradation rate of the ␣1 subunit (Fig. 1F), and overexpression of HA-tagged Grp94 significantly decreased the half-life of the ␣1 subunit to 0.8 h (Fig. 1G).
Moreover, we determined the effect of Grp94 in regulating the ubiquitination of the ␣1 subunit. HEK293 cells expressing FLAG-␣1␤2␥2 GABA A receptors were transfected with HAtagged ubiquitin, immunoprecipitated with anti-FLAG magnetic beads, and blotted for ubiquitin. The ratio of ubiquitin to ␣1 subunit post-immunoprecipitation represents the relative ubiquitination level of the ␣1 subunit. This in vivo ubiquitination assay clearly demonstrated that knockdown of Grp94 (55% knockdown efficiency) decreased the relative ubiquitination level of the ␣1 subunit (Fig. 1H), indicating that Grp94 positively regulates the ubiquitination and ERAD of the ␣1 subunit. As a result, transiently or stably depleting Grp94 significantly increased the surface ␣1 subunit protein level using a cell surface biotinylation assay (Fig. 1, B and C).
Grp94 Uses Its Middle Domain to Bind the ␣1 Subunit of GABA A Receptors-Because operating Grp94 genetically could influence the ER proteostasis network, we asked whether Grp94 directly binds the ␣1 subunit. We carried out an in vitro binding assay using recombinant GST-tagged ␣1 subunits and recombinant His-tagged Grp94 proteins. The His tag antibody pulldown led to the detection of the ␣1 subunit in the GST-␣1 complex using an anti-␣1 antibody or an anti-GST antibody, although we did not detect any ␣1 band in the GST control complex ( Fig. 2A), indicating that Grp94 directly binds the ␣1 subunit. However, because the folding degree of the recombinant ␣1 subunit was not characterized, whether such an interaction is functionally relevant to its ERAD remains to be established. To show an endogenous interaction between Grp94 and the ␣1 subunit in the mammalian central nervous systems, we performed a co-immunoprecipitation assay using the mouse brain cortex homogenates. Pulling down Grp94 led to the detection of the ␣1 subunit (Fig. 2B), confirming an endogenous interaction between Grp94 and the ␣1 subunit.
We next characterized the cellular interaction between Grp94 and the ␣1 subunit. FLAG-tagged ␣1 subunits strongly interacted with endogenous Grp94 in HEK293 cells (Fig. 1H). Grp94 is divided into three domains as follows: an N-terminal domain containing an ATP-binding site (N domain), a middle domain (M domain), and a C-terminal domain containing dimerization sites (C domain) (52). The specific domain that Grp94 uses to interact with its substrates may vary for different substrates. For example, it was reported that Grp94 uses its C-terminal domain to bind Toll-like receptors (53) but uses its M domain to interact with OS-9 (54). To determine which domain of Grp94 interacts with the ␣1 subunit in the ER lumen, we constructed plasmids containing an HA-tagged full-length N domain, M domain, C domain, NM domain, or MC domain of Grp94 (Fig. 2C). HEK293 cells expressing (FLAG-␣1)␤2␥2 GABA A receptors were transiently transfected with these HAtagged Grp94 variant plasmids for 48 h before being immunoprecipitated using an anti-FLAG antibody and blotted for HA. Clearly, the M domain, but not the N domain or the C domain, was pulled down with the ␣1 subunit (Fig. 2D, cf. lanes 13 to 12  and 14), indicating that the M domain alone is sufficient to bind the ␣1 subunit. The M construct migrated as two bands in the SDS-polyacrylamide gel; the lower band corresponds to the molecular weight of a monomer, and the upper band corresponds to that of a dimer, which was resistant to the SDS denaturation (Fig. 2D, lane 6). Only the fast migrating M species interacted with the ␣1 subunit (Fig. 2D, lane 13), possibly because the putative dimerization of the M domain prevented its association with the ␣1 subunit. Intriguingly, the ␣1 subunit interacted with the M domain and the MC domain much stronger than the full-length Grp94 (Fig. 2D, cf. lanes 11 and 13 to 9). However, the NM domain was not pulled down with the ␣1 subunit (Fig. 2D, lane 10), which could be due to the possibilities that the addition of the N domain to the M domain blocked the interaction between the M domain and the ␣1 subunit or altered the conformation of the M domain.

OS-9 Acts Downstream of Grp94 to Recognize Misfolded ␣1 Subunits of GABA A Receptors through a Glycan-dependent
Manner-The ␣1 subunit has two N-linked glycosylation sites, Asn-38 and Asn-138. The oligosaccharyltransferase complex catalyzes this glycosylation and transfers 14-monosaccharide residues Glc 3 Man 9 GlcNAc 2 to an Asn residue contained in an Asn-Xaa-(Ser/Thr) sequence motif (Xaa can be any residue except Pro). N-Glycans serve as sensors of glycoprotein folding in the ER (55,56). The recognition of terminally misfolded N-linked glycoproteins is mediated by mannose trimming events, which are facilitated by ER-mannosidase I and EDEM proteins after the release of glycoproteins from the calnexin/ calreticulin cycles (57,58). Afterward, two ER-resident lectins, OS-9 and XTP3-B, recognize the trimmed oligosaccharides (42). OS-9 and XTP3-B seem to have distinct sets of substrates. OS-9, but not XTP3-B, robustly binds Grp94 (42). Therefore, it is possible the Grp94-mediated ERAD substrates depend on OS-9. Because Grp94 positively regulates the ERAD of the ␣1 subunit, we next determined the role of OS-9 in this degradation process and the potential sequential effect of OS-9 with Grp94. Knockdown of OS-9 (71% knockdown efficiency) significantly increased the total ␣1 subunit protein level 3.0-fold (Fig.  3A), supporting that OS-9 recognizes the misfolded ␣1 subunit for ERAD. To access the role of N-glycans in the function of OS-9, we introduced the N38Q/N138Q double mutation at both glycosylation sites (Asn-38 and Asn-138) in the ␣1 subunit. This QQ mutation reduced the mass and protein level of the ␣1 subunit (Fig. 3B, cf. 2nd to 1st lane and 6th to 5th lane), indicating that glycosylation is required for the maturation of this subunit. A co-immunoprecipitation assay revealed that the QQ mutation significantly decreased the interaction between OS-9 and the ␣1 subunit according to the quantification of the ratio of OS-9 to ␣1 after immunoprecipitation using anti-␣1 antibody in HEK293 cells (Fig. 3B, cf. 6th to 5th lane). This indicates that OS-9 binds the N-glycans in the ␣1 subunit. The above data (Fig. 3, A and B) supported the assertion that OS-9 recognizes misfolded ␣1 subunit in a glycan-dependent manner.
We next determined the sequential interactions of calnexin, Grp94, and OS-9 with the ␣1 subunit during its biogenesis pathway. We began to evaluate the binding sequence of Grp94 and OS-9 with the ␣1 subunit. Knocking down Grp94 decreased the interaction between the ␣1 subunit and OS-9 significantly (Fig. 3C, cf. 6th to 5th lane), and consistently depleting OS-9 increased the interaction between the ␣1 subunit and Grp94 significantly in HEK293 cells (Fig. 3D), indicating that Grp94 acts upstream of OS-9 during the recognition of misfolded ␣1 subunits. Previously, we reported that calnexin binds the ␣1 subunits through the glycans and promotes their maturation (40). If the ␣1 subunits that exit the calnexin cycle fail to fold into their native structures, they are likely the substrate of Grp94. To determine whether Grp94 functions downstream of calnexin, we carried out a co-immunoprecipitation assay. Knockdown of Grp94 significantly increased the interaction between the ␣1 subunit and calnexin in HEK293 cells expressing ␣1␤2␥2 receptors (Fig. 3E), indicating that Grp94 acts downstream of calnexin and accepts misfolded ␣1 subunits after they depart from calnexin. Afterward, Grp94 delivers the substrate to OS-9 for processing, and then the substrate will be further tagged by ubiquitin for degradation.

FIGURE 2. Grp94 uses its middle domain to bind the ␣1 subunit of GABA A receptors.
A, recombinant His-tagged Grp94 binds the recombinant GST-tagged ␣1 subunit of GABA A receptors in vitro. One g of GST or GST-␣1 was mixed with 4 g of His-Grp94 in buffers containing 1% Triton X-100. The protein complex was isolated by immunoprecipitation using an anti-His antibody, and the immunopurified eluents were separated by SDS-PAGE and blotted with a rabbit anti-GABA A R ␣1 subunit antibody and a mouse anti-GST antibody (n ϭ 4). B, endogenous Grp94 binds the endogenous ␣1 subunit of GABA A receptors in the mouse brain. Total lysates from the cortex of mouse brains were used for immunoprecipitation with a rat anti-Grp94 antibody or control rat IgG, and the immunoisolated eluents were subjected to SDS-PAGE and Western blot analysis with rat anti-Grp94 and goat anti-␣1 antibodies (n ϭ 2). C, schematic representation of HA-tagged Grp94 constructs. Full-length Grp94 contains a signal peptide (SP, residues , an N-terminal domain (N, residues 22-341), a middle domain (M, residues 342-601), and a C-terminal domain (C, 602-785). All constructs contain a signal peptide (residues 1-21), an HA tag that was inserted between residues 796 and 797, and a sequence containing KDEL for ER localization (786 -803).

Hrd1 Ubiquitinates the ␣1 Subunit of GABA A Receptors-
The mammalian ERAD systems are organized primarily by two E3 ubiquitin ligases, Hrd1 and gp78 (36), and additional ubiq-uitin ligases contribute to the degradation of a limited amount of substrates. Hrd1 and gp78 are both homologous to Hrd1p, an essential E3 ligase in yeast (59), but Hrd1 and gp78 have largely FIGURE 3. OS-9 acts downstream of Grp94 to promote ERAD of the ␣1 subunit of GABA A receptors. A, OS-9 knockdown increases the total protein level of WT ␣1 subunits. HEK293 cells stably expressing ␣1␤2␥2 GABA A receptors were transfected with non-targeting (NT) control siRNA or siRNA against OS-9. Forty eight hours post-siRNA transfection, cells were lysed, and the total cell lysates were subjected to SDS-PAGE and immunoblotting with mouse anti-␣1 or rabbit anti-OS-9 antibodies (n ϭ 3). ␤-Actin serves as a protein loading control. Quantification of the total protein level of ␣1 is shown on the bottom. B, OS-9 interacts with the ␣1 subunit in a glycan-dependent way. The ␣1 subunit has two N-linked glycosylation sites at Asn-38 and Asn-138. The N38Q/N138Q double mutations in the ␣1 subunit disrupt both glycosylation sites. HEK293 cells were transiently transfected with WT ␣1 or N38Q/N138Q ␣1 plasmids (together with ␤2 and ␥2 subunit plasmids). Cells were treated with 10 M proteasome inhibitor MG-132 for 2 h before harvesting. Forty eight hours post-transfection, cells were washed with DPBS and cross-linked by incubation with 1.5 mM DSP for 15 min at room temperature. DSP was quenched by the addition of 10 mM Tris buffer, pH 7.5. Then cells were lysed, and the total cell lysates were immunoprecipitated with a mouse anti-␣1 antibody, and the immunoisolated eluents were blotted with rabbit anti-OS-9 and rabbit anti-␣1 subunit antibodies (n ϭ 4). IgG was included as a negative control for immunoprecipitation. Quantification of the intensity of OS-9/␣1 subunit post-IP represents the relative interaction between OS-9 and the ␣1 subunit and is shown on the bottom. The QQ mutation reduces the mass of the ␣1 subunit (cf. 4th to 3rd lane). Consequently, it leads to significantly weaker interactions between the ␣1 subunit and OS-9. C, Grp94 knockdown decreases the interaction between OS-9 and the ␣1 subunit. HEK293 cells stably expressing (FLAG-␣1)␤2␥2 GABA A receptors were transfected with non-targeting control siRNA or siRNA against Grp94. Cells were treated with MG-132 (10 M, 2 h) before harvesting. Forty eight hours post-transfection, cells were cross-linked with DSP and lysed. Then Triton X-100 cell extracts were immunoprecipitated with anti-FLAG M2 magnetic beads, and the immunoisolated eluents were blotted with rabbit anti-OS-9 and rabbit anti-␣1 subunit antibodies (n ϭ 4). Quantification of the relative intensity of OS-9/␣1 post-IP is shown on the bottom. D, OS-9 knockdown, MG-132 treatment, or EER1 treatment enhances the interaction between the ␣1 subunit and Grp94. HEK293 cells stably expressing ␣1␤2␥2 GABA A receptors were transfected with non-targeting siRNA and siRNA against Grp94 for 48 h or treated with MG-132 (2 M, 6 h), a potent proteasome inhibitor, and eeyarestatin I (EER1) (5 M, 24 h), a potent VCP inhibitor. The cell lysates were then subjected to immunoprecipitation before SDS-PAGE and Western blot analysis with rat anti-Grp94 and rabbit anti-␣1 antibodies (n ϭ 3). Quantification of the intensity of Grp94/␣1 subunit post-IP represents their relative interaction and is shown on the bottom. E, Grp94 knockdown enhances the interaction between the ␣1 subunit and calnexin. HEK293 cells stably expressing ␣1␤2␥2 GABA A receptors were transfected with non-targeting control siRNA or siRNA against Grp94 for 48 h. Then Triton X-100 cell extracts were immunoprecipitated with a mouse anti-␣1 antibody, and the immunoisolated eluents were blotted with rabbit anti-calnexin (CNX) and rabbit anti-␣1 subunit antibodies (n ϭ 3). Quantification of the relative intensity of calnexin/␣1 post-IP is shown on the bottom. Each data point is reported as mean Ϯ S.E. *, p Ͻ 0.05. IB, immunoblot. distinct interaction networks (36). Moreover, gp78 acts downstream of Hrd1 to enhance the ERAD of a model luminal substrate, a truncated major histocompatibility complex class I heavy chain molecule (MHC-1-147) (60). We evaluated the role of Hrd1 and gp78 in degrading the ␣1 subunit. Knockdown of Hrd1 using siRNA increased the total protein level of the ␣1 subunit in HEK293 cells stably expressing ␣1␤2␥2 GABA A receptors (Fig. 4, A and B), whereas knockdown of gp78 did not (Fig. 4C), indicating that Hrd1, but not gp78, plays an essential role for this ERAD substrate. Consistently, overexpression of Hrd1 substantially decreased the total protein level of the ␣1 subunit (Fig. 4, D and E). Moreover, we utilized a C291A mutation, which resides in the critical RING domain in Hrd1, exhibiting a dominant negative effect on its ubiquitin ligase activity (61). As a result, overexpression of C291A Hrd1 significantly increased the total ␣1 subunit level (Fig. 4, D and E). We next employed the cycloheximide chase assay to evaluate the degradation rate of the ␣1 subunit. As expected, overexpression of Hrd1 accelerates the degradation of the ␣1 subunit, but repressing Hrd1 activity by overexpressing C291A Hrd1 slows its degradation (Fig. 4F). We then determined the effect of Hrd1 in ubiquitinating the ␣1 subunit. Cell ubiquitination assays showed that knockdown of Hrd1 (Fig. 4G) or overexpression of C291A Hrd1 (Fig. 4H) decreased the ubiquitinated subunit as well as the ratio of ubiquitinated to total ␣1 subunit. Consistently, overexpression of Hrd1 enhanced the ubiquitination of the ␣1 subunit (Fig. 4H). The above data (Fig. 4, A-H) unambiguously demonstrated that Hrd1 plays an essential role in ubiquitinating the ␣1 subunit during its ERAD pathway. As a result, inhibiting Hrd1 by using siRNA or overexpressing C291A Hrd1 increased the surface protein level of the ␣1 subunit, and overexpressing Hrd1 decreased its surface level significantly (Fig. 4, A and D).
VCP, an AAA ϩ ATPase, plays an essential role in extracting polyubiquitinated substrates from the ER to the cytosol and driving this retrotranslocation process (62,63). Recently, we showed that VCP binds the WT ␣1 subunit and that inhibiting VCP increased its protein level, supporting VCP's critical role in degrading the ␣1 subunit (15). Because Grp94 acts early in the ERAD recognition step for the ␣1 subunit, we expected that Grp94 functions upstream of Hrd1 and VCP. Indeed, a co-immunoprecipitation assay demonstrated that knockdown of Grp94 significantly decreased the interaction between the ␣1 subunit and VCP/Hrd1 (Fig. 4I), confirming such a sequential binding. Moreover, application of eeyarestatin I, a potent VCP inhibitor (64), increased the interaction between the ␣1 subunit and Grp94 (Fig. 3D, 5th lane), further supporting the idea that Grp94 operates upstream of VCP. In addition, as expected, treatment with MG-132, a potent proteasome inhibitor, accumulated more ␣1 subunits in the ER and increased their interaction with Grp94 (Fig. 3D, 4th lane).
To determine whether the increased surface ␣1(A322D) subunits form functional receptors on the plasma membrane, we performed a whole-cell patch clamping experiment to record GABA-induced chloride current. To ensure that recording was done for successfully transfected cells, Grp94 siRNA was cotransfected with an mCherry plasmid in monoclonal HEK293 cells expressing the ␣1(A322D) ␤2␥2 GABA A receptors, and only mCherry-positive cells were selected for recording. Grp94 knockdown significantly increased the peak GABA-induced chloride current to 46 pA (Fig. 5D), demonstrating a partial restoration of function for epilepsy-associated ␣1(A322D) GABA A receptors. Fig. 6 illustrates the first ERAD pathway model of the ␣1 subunit of GABA A receptors. Upon entering the ER, newly synthesized ␣1 subunits tend to fold with the assistance of BiP (also FIGURE 4. Hrd1 ubiquitinates the ␣1 subunit of GABA A receptors. A and B, Hrd1 knockdown increases the total and cell surface protein level of ␣1 subunit. HEK293 cells stably expressing ␣1␤2␥2 GABA A receptors were transfected with non-targeting (NT) control siRNA or siRNA against Hrd1. Forty eight hours post-siRNA transfection, cells were lysed, and the total cell lysates were subjected to SDS-PAGE and immunoblotting (IB) with mouse anti-␣1 or rabbit anti-Hrd1 antibodies (n ϭ 3). ␤-Actin serves as a total protein loading control. Alternatively, the surface ␣1 subunits were measured using cell surface protein biotinylation assay (n ϭ 3). The Na ϩ /K ϩ -ATPase ␣ chain serves as a loading control for biotinylated membrane proteins. Protein band quantifications using ImageJ are shown in B. C, knocking down gp78 does not change the total protein level of the ␣1 subunit. HEK293 cells stably expressing ␣1␤2␥2 receptors were transfected with non-targeting control siRNA or siRNA against gp78 for 48 h. Then the cell lysates were subjected to SDS-PAGE and Western blotting assay (n ϭ 2). D and E, effect of overexpressing Hrd1 (WT and C291A) on the total and cell surface protein levels of ␣1 subunits. HEK293 cells stably expressing ␣1␤2␥2 GABA A receptors were transfected with a GFP control plasmid, a WT Hrd1 plasmid, or a C291A Hrd1 plasmid. Forty eight hours post-transfection, cells were treated as in A to detect the total and cell surface ␣1 protein levels (n ϭ 3). GFP serves as a negative control. The C291A mutation in Hrd1 has a dominant negative effect on its E3 ligase activity. Protein band quantifications are shown in E. F, influence of overexpressing Hrd1 (WT and C291A) on the degradation rate of the ␣1 subunit determined by CHX chase analysis. HEK293 cells stably expressing ␣1␤2␥2 GABA A receptors were transfected with a GFP control plasmid, a WT Hrd1 plasmid, or a C291A Hrd1 plasmid for 48 h and then chased for the indicated time with CHX (100 g/ml). Cells were lysed and subjected to SDS-PAGE and immunoblotting analysis (n ϭ 3). The degradation kinetics were plotted by quantifying the ␣1 protein intensity against time after CHX addition on the bottom panel. G and H, influence of knocking down or overexpressing Hrd1 on the relative ubiquitination level of ␣1 subunits. HEK293 cells stably expressing (FLAG-␣1)␤2␥2 GABA A receptors were transfected with an HA-tagged ubiquitin plasmid together with non-targeting control siRNA or siRNA against Hrd1 for 48 h (G) or transfected with a HA-tagged ubiquitin plasmid together with a GFP control plasmid, a WT Hrd1 plasmid, or a C291A Hrd1 plasmid for 48 h (H). Then the total cell lysates were immunoprecipitated with anti-FLAG M2 magnetic beads. The immunoisolated complexes were subjected to SDS-PAGE and Western blot analysis. Quantification of the ratio of ubiquitin/␣1 band intensity post-IP, representing relative ubiquitin level of the ␣1 subunit, is shown on the bottom (n ϭ 3). I, Grp94 knockdown decreases the interaction between VCP or HRD1 and the ␣1 subunit. HEK293 cells expressing (FLAG-␣1)␤2␥2 GABA A receptors were transfected with non-targeting control siRNA or siRNA against Grp94 for 48 h. Then the total cell lysates were immunoprecipitated for SDS-PAGE and Western blot analysis. Quantification of the ratio of VCP or HRD1/␣1 band intensity post-IP is shown on the bottom (n ϭ 3). Each data point in B and E-I is reported as mean Ϯ S.E. *, p Ͻ 0.05; **, p Ͻ 0.01.

Discussion
termed Grp78) in a glycan-independent manner and with the support from calnexin in a glycan-dependent way. BiP is known to play a role very early in protein biogenesis by binding to the hydrophobic patches of unfolded proteins in the ER, facilitating their folding while preventing aggregation (45,65). Calnexin, a membrane-bound lectin chaperone, binds monoglucosylated N-glycans installed on a glycoprotein and facilitates its folding with the assistance of protein-disulfide isomerases, such as ERp57 (66). Recently, we demonstrated that both BiP and calnexin facilitate the maturation of WT ␣1 subunits (40). After the collaborative operations of BiP and calnexin, if native structures are obtained, ␣1 subunits assemble with other subunits, such as ␤2 subunits and ␥2 subunits, to form a pentamer on the ER membrane. This pentamer traffics efficiently to the plasma membrane in a fully functional state. However, if ␣1 subunits misfold, Grp94 recognizes their non-native states in the ER lumen. OS-9 acts downstream of Grp94 to further recognize the misfolded subunits in a glycan-dependent manner. Grp94 and OS-9 deliver misfolded ␣1 subunits to Hrd1, which ubiquitinates them. VCP facilities the dislocation of the ␣1 subunits from the ER member to the cytosol, where they will be targeted to the 26S proteasome for degradation.
Many important questions remain to be answered for the ERAD pathway of the ␣1 subunit. E3 ligases play a central role in organizing the ERAD machinery. There are over 500 ubiquitin E3 ligases in mammals. It is believed that one protein uses only a subset of those E3 ligases. What other E3 ligases play an important role in the ubiquitination of the ␣1 subunits? Recently, it was reported that ring finger protein 34 (RNF34) promotes the degradation of the ␥2 subunit by ubiquitinating them (67). Furthermore, our recent proteomics study identified several E3 ligases that interact with the 1 subunit, such as HUWE1 and UBR5 (68). It will be of great interest to determine whether these E3 ligases complement the function of Hrd1 in ubiquitinating the ␣1 subunits. In addition, presumably Grp94 and OS-9 act on the misfolded domains of the ␣1 subunit in the ER lumen. Whether and how the transmembrane domains of the ␣1 subunit contribute to the ERAD recognition steps remain to be determined. Beyond the ERAD factors that were identified here, we still need to decipher the role of many known ERAD factors in the context of degrading the ␣1 subunit, such as EDEM and Derlin family proteins. Therefore, substantial further studies need to be done to fully understand ERAD of the ␣1 subunit. Nonetheless, our report represents a significant progress toward that end.
We clarified that Grp94 promotes the ERAD of the ␣1 subunits of GABA A receptors. Unlike its cytosolic Hsp90 family members, Grp94 only has a few known client proteins or substrates in the ER, and the exact function and mechanism of Grp94 on the folding or ERAD of its substrates remains to be established (52). The effect of Grp94 on the GABA A receptor ERAD does not depend on whether the depletion of Grp94 is short term or long term, which is different from that on the null Hong Kong variant of ␣ 1 -antitrypsin (NHK). Previously, it was reported that transient knockdown of Grp94 delayed the degradation and increased the steady state level of NHK (42,43), whereas stably depleting Grp94 had no apparent effect (54). Up-regulation of BiP by transient knockdown of Grp94 might contribute to such a discrepancy on NHK; alternatively, long term depletion of Grp94 might result in cell adaption for the altered proteostasis network. A similar discrepancy was reported between short term or long term depletion of gp78; transient knockdown of gp78 using siRNA attenuated the degradation of a model ERAD substrate, MHC(1-147), whereas permanent depletion of gp78 using CRISPR cells had no effect (60). It was proposed that cells can adapt to compensate for the function of gp78 during the permanent gp78 inactivation process (60). Therefore, one possibility is that if an ERAD factor is essential for the disposal of one substrate, cells cannot adapt to the loss of that ERAD factor; if not essential, cells can adjust by adapting the proteostasis network. Our results supported that the ERAD of GABA A receptors critically depends on Grp94, adding GABA A receptor subunits to the very limited pool of Grp94 substrates. Because the ␤-sheet-rich ␣1 subunit has distinct structures compared with other known Grp94 substrates, how this subunit is recognized by Grp94 remains to be investigated. Grp94 inhibition stabilizes ␣1 subunits and enhances their functional surface expression. Our results strongly support that Grp94 regulates GABA A receptor function directly because Grp94 binds the ␣1 subunit in vitro and in vivo. Grp94 inhibition has a much greater effect on ␣1(A322D) subunits than WT ␣1 subunits. Therefore, specific Grp94 inhibition provides a promising way to enhance the function of misfoldingprone ␣1 subunits, as a therapeutic strategy to ameliorate idiopathic epilepsy. To achieve further specificity, it is also of interest to disrupt the direct interaction between Grp94 and ␣1 subunits.
For fast-degraded ERAD substrates, it is desirable to enhance folding as well as reducing ERAD to achieve substantial functional rescue. Recently, we identified different classes of small molecule proteostasis regulators that enhance the folding and trafficking and thus the function of epilepsy-associated mutant GABA A receptors (40,69). Suberanilohydroxamic acid increases BiP protein level and the interaction between calnexin and the ␣1(A322D) subunit, which promotes the functional cell surface expression of this mutant subunit (40). Verapamil, an L-type Ca 2ϩ channel blocker, enhances the plasma membrane trafficking of the ␣1(D219N) subunit by promoting calnexin-assisted folding and inter-subunit assembly between ␣1(D219N) subunit and ␤2 subunit (69). In addition, pharmacological chaperones, which are small molecules that directly bind their target protein, stabilize it and enhance its forward trafficking (70). Both proteostasis regulators and pharmacolog- The ␣1 subunits are co-translationally translocated to the ER membrane. They tend to fold with the assistance of BiP in a glycan-independent manner and calnexin (CNX) in a glycan-dependent manner. Properly folded ␣1 subunits assemble with other subunits to form a heteropentamer on the ER membrane for trafficking to the plasma membrane. However, if ␣1 subunits misfold, Grp94 recognizes the non-native states in the ER lumen. In the ER lumen, OS-9 further recognizes the misfolded subunits in a glycan-dependent manner and delivers them to Hrd1. Hrd1 ubiquitinates ␣1 subunits and translocates them from the ER member to the cytosol with the assistance of VCP. Then misfolded ␣1 subunits are targeted to the 26S proteasome for degradation.
ical chaperones are expected to yield an additive/synergistic effect with ERAD inhibitors to enhance the function of GABA A receptors. Indeed, recently, we revealed that combining suberanilohydroxamic acid treatment with eeyarestatin I, a VCP inhibitor, additively restored the functional cell surface expression of the ␣1(A322D) subunit, providing a proof-of-principle case (15). The elucidation of the ERAD pathway of GABA A receptors allows the development of a more effective combination therapy to restore the function of disease-causing GABA A receptors.
GABA A receptors are within the superfamily of Cys-loop receptors, which contain important neuroreceptors, including nicotinic acetylcholine receptors, serotonin type 3 receptors, and glycine receptors (2,11). They are responsible for fast excitatory and inhibitory transmission in nervous systems. Because they share common structural characteristics, much of our knowledge gained on the folding, assembly, and degradation of GABA A receptors could be extrapolated to the whole class of Cys loop receptors.