Glycine Receptor β–Targeting Autoantibodies Contribute to the Pathology of Autoimmune Diseases

Fifty-eight patient serum samples were submitted to our laboratory. In 48 patients, SPS was suspected, and they were negative for antiglutamate-decarboxylase and antiamphiphysin aAb. In 10 patients, a focal epilepsy of unknown cause was present, and GlyR aAb were found by routine screening. None of the patients with epilepsy had motor symptoms, hyperekplexia, or startle reaction.¹⁹ Disease pattern of 2 patients with GlyRαβ aAb (Pat31 and Pat36) in comparison with 1 patient with GlyRα aAb only (Pat11) are further described in the Table.

Experiments using patient material have been approved by the Ethics Committee of the Medical Faculty, University of Würzburg, Germany ("Glycine receptor autoantibodies and spinal disinhibition," 20190424 01).

Patients underwent a neurologic examination, and Pat11 and Pat36 were given questionnaires about chronic pain,^20,21 anxiety (Liebowitz Social Anxiety Scale [LSAS]; Anxiety Sensitivity Index [ASI]),^22,23 and heightened sensitivity inducing spasms and falls (see Table for results).²⁴

HEK-293 cells (Human Embryonic Kidney cells; CRL-1573; ATCC—Global Bioresource Center) were used for in vitro experiments. Cells were grown in minimum essential medium (Life Technologies, Carlsbad, US) supplemented with 10% fetal bovine serum, l-glutamine (200 mM), and 100 U/mL penicillin and 100 μg/mL streptomycin at 37°C and 5% CO₂.

Neurons were prepared from embryos of a novel generated hybrid mouse line Glra1^spdot/Glrb^eos from Glrb^eos14 and Glra1^spdot (oscillator, JAX stock #000536, JAX:000536, Jackson Laboratory, Bar Harbor, US) at the embryonic stage 12–13. Experiments were approved by the local veterinary authority (Veterinäramt der Stadt Würzburg, Germany) and the Ethics Committee of Animal Experiments, i.e., Regierung von Unterfranken, Würzburg, Germany (license no.:55.2.2-2532.2-949-31). Mixed spinal cord neuronal cultures were prepared as previously described.²⁵ Genotyping for Glrb^eos and Glra1^spdot was performed according to Maynard et al.¹⁴ Stainings and electrophysiologic measurements were performed after 16–18 days in culture.

HEK-293 cells were transiently transfected by using a modified calcium phosphate precipitation method.¹⁸

Living transfected HEK-293 cells or primary neurons were incubated for 2 hours at 4°C with patient sera, healthy control serum (1:50), or commercial antibody against GlyRα1 (146111, 1:500, Synaptic Systems, Göttingen, Germany). After fixation using 4% paraformaldehyde/4% sucrose in phosphate-buffered saline (PBS) at pH 7.4 for 20 minutes at room temperature (RT), cells were blocked and permeabilized with 5% goat serum/0.2% Triton-X-100 in PBS for 30 minutes. Primary antibodies against myc-tagged GlyRβ (303008, 1:250, Synaptic Systems), gephyrin (147111, 1:500, Synaptic Systems), synapsin (574778, 1:500, Merck), and pan-α-GlyR (146011, 1:250, Synaptic Systems) were incubated for 1 hour, followed by incubation with secondary antibodies (all 1:500, 111-546-003, 109-165-003, 115-175-146, and 111-175-006, all from Dianova, Hamburg, Germany) for 1 hour. Cell nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI) and mounted in Mowiol.

Living HEK-293 cells transfected with the GlyRα1 subunit were incubated with patient sera, healthy control serum (1:50), or GlyRα1 antibody (146111, 1:500, Synaptic Systems) for 1 hour at RT. The supernatant containing unbound antibodies was transferred to another coverslip with 3 repetitions. Finally, the supernatant was transferred to HEK-293 cells transfected with the zebrafish GlyRα1 and human GlyRβ subunit.

GlyRα1 ECD preparation and enzyme-linked immunosorbent assay (ELISA) were performed as described previously.²⁶ Spinal cord neurons were stained afterward with patient serum (1:50).

Spinal cords were extracted from anesthetized Glra1^+/+/Glrb^eos/eos and Glra1^spdot/spdot/Glrb^eos/eos mice, bedded in Tissue-Tek and immediately frozen on dry ice. A cryostat (CM1950, Leica, Wetzlar, Germany) with a chamber temperature of −20°C was used to cut spinal cord sections of 9 µm thickness. Sections were mounted on SuperFrost Plus slides (03-0060, Langenbrinck, Niederrohrdorf, Switzerland).

Sections were fixed with ice-cold 2% paraformaldehyde in PBS at pH 7.4 for 30 seconds at RT. After washing, sections were shortly dipped in 50 mM NH₄Cl for quenching and incubated in 0.1 mM glycine for 30 minutes. For blocking, 10% goat serum in PBS at pH 7.4 was used followed by primary antibody incubation with patient serum (1:50) and an anti–mEos-Cy3 (N3102-SC3-L, 1:200, Nanotag, Göttingen, Germany) overnight (ON) at 4°C. Secondary antibody goat-anti-human-IgG-Alexa-Fluor-647 (1:500, JIM-109-605-006, Biozol, Eching, Germany) was incubated for 1 hour at RT. Nuclei were stained with DAPI for another 10 minutes. Sections were covered with Fluor Save Reagent (345789, Calbiochem, Darmstadt, Germany).

The cryo-EM structure (7MLY⁹) of the pentameric GlyR with a subunit stoichiometry of 4α:1β was used to generate structural images. Figures were prepared with the help of Pymol (pymol.org, version 2.0.7).

GlyR subunit ECDs (UniProtKB: P23415, P23416, O75311, Q5JXX5, P48167) were displayed in microarray format as 15mer overlapping peptide library. µSPOT²⁷ peptide microarrays were synthesized using a MultiPep RSi robot (CEM, Matthews, US) on cellulose discs containing 9-fluorenylmethyloxycarbonyl-β-alanine (Fmoc-β-Ala) linkers (average loading: 130 nmol/disc—4mm diameter). Synthesis was performed by deprotecting the Fmoc-group using 20% piperidine in dimethylformamide (DMF). Peptide chains were elongated using a coupling solution consisting of amino acids (0.5 M) with oxyma (1 M) and diisopropylmethanediimine (1 M) in DMF (1:1:1). Coupling steps were conducted 3 times (30 minutes), followed by capping (4% acetic anhydride in DMF). Side chains were deprotected using 90% trifluoracetic acid (TFA), 2% dichloromethane (DCM), 5% H₂O, and 3% triisopropylsilane (TIPS, 150 μL/well) for 1 hour at RT. Afterward, the deprotection solution was removed, and the discs were solubilized ON at RT using a solvation mixture containing 88.5% TFA, 4% trifluoromethanesulfonic acid, 5% H₂O, and 2.5% TIPS. The resulting peptide-cellulose conjugates (PCCs) were precipitated with ice-cold ether and spun down at 2,000×g for 10 minutes at 4°C, followed by 2 additional washes with ice-cold ether. Resulting pellets were dissolved in DMSO. PCC solutions were mixed 2:1 with saline–sodium citrate buffer (150 mM NaCl, 15 mM trisodium citrate, pH 7.0) and transferred to a 384-well plate. For transfer of the PCC solutions to white-coated CelluSpot blank slides (76 × 26 mm, Intavis AG Peptide Services, Tübingen, Germany), a SlideSpotter was used.

Microarray slides were blocked for 1 hour in 5% (w/v) milk powder, 0.05% Tween20, and PBS at pH 7.4. The slides were incubated for 30 minutes with positive and negative sera (1:500) or GlyRα1 and GlyR pan-α antibody (1:2500) in blocking buffer. IgG antibodies were detected using goat-anti-human or goat-anti-mouse-IgG-HRP (31410, 1:2500, 31430, 1:5000, Thermo Fisher, Waltham, US). The readout was detected with an Azure imaging system c400 using SuperSignal West Femto substrate (Thermo Scientific). Microarray binding intensities were quantified with FIJI using the “microarray profile” plugin (OptiNav Inc, Bellevue, US).

Cleavable peptides were synthesized with an additional rink amide linker at the C-terminus of the identified epitope. Microarray slides were blocked using 5% (w/v) milk powder, 0.05% Tween20, and PBS at pH 7.4 for 1 hour. Serum samples from patient 36 were preincubated with cleaved peptide in the amount corresponding to 2 and 4 cellulose discs containing Fmoc-β-Ala linkers. Peptides were resuspended in 100 μL of PBS buffer, and 5 μL of serum was added subsequently. The samples were mixed at 1000 rpm (RT) for at least 30 minutes. A control without peptide was treated in the same manner. Solutions were added to 2.5 mL of blocking solution and incubated for 30 minutes on the slides. Microarray binding intensities were quantified with FIJI and normalized against the untreated slide.

Patch-clamp analysis was performed on transfected HEK-293 cells or mixed primary neuronal cultures using whole-cell recordings. Experiments were performed at 21°C. Recording pipettes were pulled from borosilicate capillaries with open resistances of 3.5–5.5 MΩ and filled with internal buffer in mM (120 CsCl, 20 N(Et)₄Cl, 1 CaCl₂, 2 MgCl₂, 11 EGTA, 10 HEPES for HEK-293 cells; 140 CsCl, 1 EGTA, 10 HEPES, and 6 d-Glucose for neurons; pH 7.2, adjusted with CsOH). For determination of maximal current amplitudes (I_max) and dose-response curves (EC₅₀ values), glycine was applied in a concentration series of 10–1000 µM in external buffer in mM (137 NaCl, 5.4 KCl, 1.8 CaCl₂, 1 MgCl₂, 5 HEPES for HEK-293; 130 NaCl, 3 KCl, 1.5 CaCl₂, 2 MgCl₂, 10 HEPES, 6 d-Glucose, and 10 TEA-Cl for neurons; pH 7.35, adjusted with NaOH). Glycine solutions were applied by the Octaflow II system (ALA Scientific Instruments, Farmingdale, US). Following recordings, 50 µM picrotoxinin (Sigma Aldrich, Darmstadt, Germany) + 100 µM glycine were applied. Homomeric GlyRα, but not heteromeric GlyRαβ, are blocked by picrotoxinin.²⁸ This test was used to discriminate between homomeric and heteromeric receptors. Current responses were amplified with an EPC-10 amplifier and measured at a holding potential of −60 mV using Patchmaster Next software (HEKA Elektronik, Reutlingen, Germany).

Spinal cord, brainstem, and cortex samples were extracted from deeply anesthetized male and female mice and directly frozen at −80°C. Lysate were prepared using 1 mL of brain homogenisate buffer (20 mM HEPES, 100 mM potassium acetate, 40 mM KCl, 5 mM EGTA, 5 mM MgCl₂, 5 mM DTT, 1% TritonX-100, 1 mM PMSF, and protease inhibitors (Roche, Basel, Switzerland). After a 15-minute centrifugation at 10.000×g at 4°C, supernatants were transferred and used for Western blots.

Protein samples were separated by SDS-PAGE using 11% (w/v) gels followed by transfer of proteins onto a nitrocellulose membrane (GE Healthcare, München, Germany). After blocking for 1 hour with 5% BSA in TBS-T (TBS with 1% v/v Tween20), membranes were incubated with primary antibodies antigephyrin, anti-GlyR pan-α, anti-GlyRα1, anti-VGAT (131003) all 1:1,000, Synaptic Systems), and anti-GAPDH (CB1001, 1:1,000, Merck) ON at 4°C. Proteins were visualized by horseradish peroxidase–coupled secondary antibodies (111-036-003 and 115-035-146, 1:15000, Dianova) and detected through chemiluminescence using clarity Western ECL substrate (170-5061, BioRad, Feldkirchen, Germany).

Images were captured using an Olympus Fluoview ix1000 microscope (UPLSAPO 60× oil objective) or a Zeiss Axio Imager 2 microscope (20× air objective). For image analysis and processing, the Fiji/ImageJ software was used.²⁹ Synaptic density/100 μm was analyzed through the plug-ins NeuronJ and SynapscountJ.

Data were analyzed using GraphPad Prism or Origin9 software and represented as mean ± standard error of the mean (SEM).

The numbers of experiments (N; all experiments have been performed from 3 biological replicates or as stated otherwise) and cells (n) are listed in eTable 1 (links.lww.com/NXI/A969). Data were tested for outliers by ROUT (Q = 1%). Normality of the data was reviewed by the Shapiro-Wilk normality test (α = 0.05). Statistical significance was calculated using an unpaired 2-tailed Mann-Whitney test or an unpaired t test. p values are given in the result section or eTable 1. The 0-hypothesis was rejected at a level of p < 0.05.

Data that support the findings of this study are available from the corresponding author on reasonable request.

GlyR aAb have been identified to target a common sequence in the ECD of GlyRα subunits (Figure 1A).⁵ The adult receptor composition is 4α:1β heteromeric.^8,9 The GlyRα ECD shares a high homology to the GlyRβ subunit (Figure 1B). In this study, we investigated 58 serum samples from patients with SPS-like symptoms and focal epilepsy and tested them for binding to GlyRβ (Figures 1A and 2, A–D). GlyRβ subunit alone does not form functional channels^8,9,30 and requires coexpressed α for transport to the plasma membrane (Figure 2B). Thirty samples harbored aAb against human GlyRα1 (eFigure 1A, links.lww.com/NXI/A966), whereas 28 displayed no binding. All GlyRα1-negative sera demonstrated also no binding to human GlyRβ subunits (selected examples; eFigure 1B, eTable 2, links.lww.com/NXI/A970).

To analyze whether GlyRα1-positive sera bind GlyRβ, we used zebrafish GlyRα1^dr to ensure transport and integration of GlyRβ into the cellular membrane (Figure 2B). Previously, we have shown that most human serum samples harboring GlyRα aAb do not bind zebrafish α1.⁵ Three patient sera of the 30 GlyRα1-positive sera bound to the zebrafish α1 subunit and were thus excluded from further analysis. Testing of the remaining 27 human GlyRα1-positive sera resulted in GlyRβ binding of Pat 31 and Pat 36 suggesting that indeed GlyRβ represents a target for GlyR aAb (Figure 2D, eTable 2, links.lww.com/NXI/A970). Both patients had a confirmed autoimmune disorder. Pat31 had focal epilepsy without any other cause than GlyR aAb, and Pat36 had unequivocal SPS/PERM with good response to immunotherapy (Table). Detailed information is published elsewhere (Pat11,³¹ Pat31,^19,32,33 and Pat36²⁶).

Single-nucleotide variations in the GlyRβ subunit gene have been associated with an increased susceptibility to anxiety or panic disorders identified in a genome-wide association study with healthy human volunteers.^34,35 However, anxiety scores of patients with GlyRβ aAb (Pat31, Pat36) were not different from Pat11 with exclusively GlyRα aAb (Table).

Peptide microarray-based screenings enable the identification, mapping, and validation of linear aAb epitopes (Figure 3A).³⁶ To identify possible binding regions of GlyRβ aAb, GlyR subunits α1-4 and β were displayed in form of 372 peptides (15 amino acids (AA) length, 10AA overlap, 5AA shift). The recapitulation of the reported epitopes of commonly used GlyR antibodies mAb2b and mAb4a confirms the microarray capacity to report binding epitopes within the structured GlyR ECD (Figure 3, B–D). The same arrays reported putative aAb binding sites for Pat31 and Pat36, which confirmed binding to the GlyRβ subunit. Pat31 showed reactivity toward GlyRα1/α4/β subunits, whereas Pat36 showed pan-GlyR subunit activity (Figure 3, B and C). A focused GlyRβ library of 232 peptides (15AA length, 14AA overlap, 1AA shift) determined GlyRβ sequences that may mediate aAb binding. An autoantibody epitope for Pat36 was mapped toward GlyRβ residues ¹⁴¹PDLFFANEKSANFHDV¹⁵⁶ and for Pat31 toward ⁷⁷GIPVDVVVNIFINSF⁹¹ (Figure 3D). The observation that GlyRβ aAb binding of Pat 36 partially overlaps with the epitope of mAb4a suggests an elevated intrinsic immunogenic potential of this region. Both identified GlyRβ epitopes are localized within surface-accessible GlyRβ ECD β sheets (Figure 3, E and F).

To confirm the microarray mapping and to verify the observed anti-GlyRβ reactivity, we conducted on-chip neutralization experiments. In this study, serum of Pat36 was preincubated with increasing amounts of the soluble GlyR peptide (DSIWKPDLFFANEKG) that overlaps with the mapped binding site. The observed concentration-dependent signal reduction upon preadsorption (Figure 3G) supports the conclusion that the microarray binding signals resulted from sequence-specific GlyRβ recognition of the patient aAb.

To confirm specific GlyRβ binding of patient sera, GlyRα1 expressed in transfected HEK-293 cells were used for preadsorption of GlyRα1-specific aAbs from patient serum. As a second approach, incubation of patient serum with the purified and refolded GlyRα1 ECD coated to ELISA plates was used.²⁶ We transferred patient serum 3 times to transfected HEK-293 cells expressing only GlyRα1 for 1 hour followed by live staining of the remaining supernatant on HEK-293 cells expressing GlyRα1 and GlyRβ. As control, we used an exclusively GlyRα1-positive serum with the same titer as Pat31 and Pat36 for better comparison (titer 1:500).

Serum signals of Pat31 and Pat36 were already strongly reduced after the first transfer and completely abolished after the second one, whereas binding of Pat12 serum was no longer visible after 3 transfers (Figure 4A). Final transfer of the samples to GlyRα1 and GlyRβ expressing cells revealed again binding of Pat31 and Pat36 sera but not binding of Pat12 serum arguing that GlyRβ-specific aAbs still remained in the serum.

Then patient samples were incubated with the ECD of GlyRα1 bound to ELISA plates, and the remaining supernatant was used for live staining of spinal cord neurons (Figure 4B). Although Pat36 and Pat12 sera both bound very strongly to spinal cord neurons before the adsorption by GlyRα1 ECD, remaining signal was only detectable for Pat36 serum afterward. Together, these findings further confirm that both, Pat31 and Pat36, sera contain aAb targeting GlyRβ and cannot be neutralized by GlyRα1. Due to insufficient material from Pat31, this and some following experiments were performed only with Pat36 serum.

The binding of GlyRβ-positive patient sera to endogenous GlyRs was evaluated using murine mixed spinal cord cultures isolated from a mouse model that allows specifically the detection of GlyRβ. Glra1^spdot/spdot/Glrb^eos/eos animals result from crossing mEos4-tagged Glrb¹⁴ with Glra1 mutant oscillator mice.³⁷ Oscillator mice carry a frameshift mutation resulting in lack of GlyRα1 in homozygous animals. Binding of Pat31 and Pat36 sera were detected in both Glra1^+/+/Glrb^eos/eos (wild-type controls) and Glra1^spdot/spdot/Glrb^eos/eos neurons (absence of GlyRα1). Pat12 serum bound to Glra1^+/+/Glrb^eos/eos but not Glra1^spdot/spdot/Glrb^eos/eos neurons lacking GlyRα1 (Figure 5A). HC serum showed no binding (eFigure 2, links.lww.com/NXI/A967).

The gephyrin signal serves as a postsynaptic marker and allows the quantification of GlyRβ (mEos) in gephyrin-positive clusters (Figure 5A). A comparison of gephyrin and GlyRβ between wild-type controls (Glra1^+/+/Glrb^eos/eos) and homozygous oscillator neurons (Glra1^spdot/spdot/Glrb^eos/eos) revealed significantly less GlyRβ in oscillator neurons (**p = 0.004), while gephyrin levels were indistinguishable between oscillator and wild-type controls (p = 0.35, Figure 5, B and C, eTable 1A (links.lww.com/NXI/A969). The reduced GlyRβ expression in oscillator neurons was not surprising because GlyRα1, the main partner of GlyRβ, is absent. Of interest, no significant differences were identified in the synaptic localization of GlyRβ between Glra1^spdot/spdot/Glrb^eos/eos and Glra1^+/+/Glrb^eos/eos neurons (p = 0.86, Figure 5D). The number of synapses targeted by Pat31 and Pat36 sera showed no differences between Glra1^+/+/Glrb^eos/eos and Glra1^spdot/spdot/Glrb^eos/eosmice (p = 0.11 and p = 0.43, respectively, Figure 5E, eTable 1A).

Lack of GlyRα1 in homozygous oscillator mice was further validated using an α1-specific antibody (mAb2b) in Western blots of spinal cord, brainstem, and cortex tissue lysates (eFigure 3, A and B, links.lww.com/NXI/A968). Nevertheless, other GlyRα subunits (α2 and α3) were present in spinal cord (GlyR pan-α–positive protein samples, GlyRα2-positive staining of dissociated spinal cord neurons; eFigure 3A-B), and hence, binding of patient serum to other GlyRα subunits (α2 and α3) could be possible. However, cell-based analysis revealed no binding of Pat31 and Pat36 sera to GlyRα2 or α3 (eFigure 3C) confirming that binding of Pat31 and Pat36 aAbs is specific to GlyRβ.

At last, binding of Pat36 serum to GlyRβ was demonstrated by immunostaining of spinal cord sections of Glra1^spdot/spdot/Glrb^eos/eos mice (Figure 5F). GlyRβ signals in Glra1^+/+/Glrb^eos/eos spinal cord were stronger than in Glra1^spdot/spdot/Glrb^eos/eos. Similarly, serum staining was more intense to Glra1^+/+/Glrb^eos/eos spinal cord while still present at Glra1^spdot/spdot/Glrb^eos/eos dorsal and ventral horn spinal cord (Figure 5F, right images a₁, a₂, b₁, b₂). Our data clearly evaluated GlyRβ targeting of aAb from some patients suggesting GlyRβ as a new target for GlyR aAb in rare cases.

To investigate physiologic consequences of GlyRβ aAb targeting, ion channel function of GlyRs after preincubation with GlyRβ-positive patient serum was tested to assess whether similar molecular alterations exist as demonstrated for aAb against GlyRα.⁵

Whole-cell patch-clamp measurements of mixed spinal cord neurons from Glra1^spdot/spdot/Glrb^eos/eos and Glra1^+/+/Glrb^eos/eos mice demonstrated almost absent glycine-induced I_max for Glra1^spdot/spdot/Glrb^eos/eos neurons lacking GlyRα1, while wild-type neurons showed large glycine-induced chloride currents (Glra1^+/+/Glrb^eos/eos: 2.1 ± 0.3 nA; Glra1^spdot/spdot/Glrb^eos/eos: 0.04 ± 0.01 nA, ****p < 0.0001) (Figure 6, A and B, eTable 1B, links.lww.com/NXI/A969). This excluded primary neurons from Glra1^spdot/spdot/Glrb^eos/eos for investigation of the functional consequences of GlyRβ aAb. To better discriminate between the effects of aAbs against GlyRα1 and GlyRβ, we turned back and transfected HEK-293 cells with either human GlyRα1^hs subunit alone, or coexpressed zebrafish GlyRα1^dr with human GlyRβ. Similar to findings by Rauschenberger et al.,⁵ patient serum binding to human GlyRα1 subunit led to a rightward shift in dose-response curves (untreated: 74 ± 4 µM; HC: 77 ± 6 µM; Pat31: 108 ± 12 µM, p = 0.003 compared with untreated and p = 0.01 compared with HC; Pat36: 100 ± 9 µM, p = 0.01 compared with untreated and p = 0.04 compared with HC; Pat12: 106 ± 11 µM, p = 0.009 compared with untreated and p = 0.03 compared with HC) and thus a reduced receptor potency (Figure 6, C and D). Binding of Pat36 serum but not Pat12 and Pat31 sera resulted in a slightly decreased maximal current (relative I_max to untreated: 100 ± 12%; HC: 95 ± 14%; Pat31: 84 ± 18%; Pat36: 67 ± 7%, p = 0.02 compared with untreated Pat12: 73 ± 7%; Figure 6E, see also eTable 1C, links.lww.com/NXI/A969). By contrast, binding of patient serum to GlyRβ (zebrafish GlyRα1^dr with human GlyRβ were transfected) did significantly decrease maximal currents after preincubation with both Pat31 (56% ± 5.8%) and Pat36 (52% ± 14.6%) sera compared with untreated (100% ± 8.6%) and HC treated (105% ± 13.8%) cells (Pat 31: **p = 0.001 compared with untreated and **p = 0.002 compared with HC; Pat 36: *p = 0.02 compared with untreated and *p = 0.04 compared with HC; Figure 6, F–H, see also eTable 1C) arguing for less glycine efficacy.

In sum, while binding of patient aAb to GlyRα subunits leads to reduced glycine potency, binding to GlyRβ results in reduced maximal glycine-gated currents and hence reduced glycine efficacy. Therefore, GlyR aAbs targeting distinct GlyR subunits affect glycinergic function differently, thus arguing for a significant contribution of both GlyRα1-specific and GlyRβ-specific aAb to the disease pathology.