Thermophoretic analysis of ligand-specific conformational states of the inhibitory glycine receptor embedded in copolymer nanodiscs

The glycine receptor (GlyR), a member of the pentameric ligand-gated ion channel family (pLGIC), displays remarkable variations in the affinity and efficacy of the full agonist glycine and the partial agonist taurine depending on the cell system used. Despite detailed insights in the GlyR three-dimensional structure and activation mechanism, little is known about conformational rearrangements induced by these agonists. Here, we characterized the conformational states of the α1 GlyR upon binding of glycine and taurine by microscale thermophoresis expressed in HEK293 cells and Xenopus oocytes after solubilization in amphipathic styrene-maleic acid copolymer nanodiscs. Our results show that glycine and taurine induce different conformational transitions of the GlyR upon ligand binding. In contrast, the variability of agonist affinity is not mediated by an altered conformational change. Thus, our data shed light on specific agonist induced conformational features and mechanisms of pLGIC upon ligand binding determining receptor activation in native environments.

The superfamily of pentameric ligand-gated ion channels (pLGIC) mediates excitatory and inhibitory synaptic neurotransmission in the central nervous system 1 and is the target for many therapeutic agents 2 . The knowledge of the underlying mechanisms determining apparent affinity and efficacy of small ligands acting on these receptors is fundamental for the understanding of their physiological and pharmacological properties under developmental, normal and pathological conditions of the brain. One of the most pressing questions regarding ligand-specific activity concerns the underlying mechanisms by which partial and full agonists possess variable half-maximal effective concentration (EC 50 ) values and amplitude responses under different cellular conditions. Despite detailed insights into their three-dimensional structures and general activation mechanisms, an accurate knowledge of the underlying structural confinements determining the affinity and efficacy of agonists at pLGICs is still lacking 1,3 . The inhibitory glycine receptor (GlyR), a member of the pLGICs, is an outstanding example displaying exceptional variations in the affinity and efficacy of its agonists 4 . GlyRs mediate, together with GABA receptors, the majority of fast inhibitory neurotransmission in the central nervous system and are implicated in neuronal diseases like hypereplexia 4,5 and forms of epilepsy 6 . GlyRs assemble as homopentamers of α-subunits (α1-4) or heteropentamers, containing both α-and β-subunits 7 and are composed of an extracellular orthosteric ligand binding site (ECD), a transmembrane spanning ion-channel (TMD) and an intracellular domain (ICD). GlyR activation upon agonist binding induces a contraction within the ECD that is transmitted to the TMD leading to the opening of the chloride selective intrinsic channel pore 8 . Remarkably, the efficacy to open the ion channel depends on the agonist used, i.e. glycine acts as a full agonist inducing maximal current responses with opening times of 95-98% seen in single-channel recordings 9 , whereas taurine acts as a partial agonist producing a decreased maximal response relative to the effect produced by glycine that is reflected by lower open-channel probabilities 10 . However, both the efficacy and affinity of glycine and taurine also strongly depend Heterologous expression of α1 GlyR in HEK293 cells. HEK293 cells were cultured in minimum essential medium (MEM) supplemented with 10% (v/v) FCS, 2 mM l-glutamine and streptomycin (100 µg/ml) at 37 °C and 5% CO 2 . For transfection, 14-20 × 10 6 cells were diluted in electroporation buffer 1 M (5 mM KCl, 15 mM MgCl 2 , 120 mM Na 2 HPO 4 /NaH 2 PO 4 pH 7.2, 50 mM mannitol) at a final concentration of 1 × 10 6 cells per 100 µl and 500 ng per 100 µl of the respective plasmid was added. Electroporation was performed using the Amaxa Nucleofector II S system (Lonza, Basel, Switzerland). After transfection, cells were reseeded in 75 cm 2 flasks with MEM and incubated for 48 h. ethical approval. All methods involving animals were carried out in accordance with the guidelines and regulations of the local animal care and use committee. Methods were approved by the Technical University of Darmstadt (II25.3-19c20/15, RP Darmstadt, Germany).
Heterologous expression of α1-GFP GlyR in X. laevis oocytes. cRNA was synthesized using the AmpliCap-Max T7 High Yield Message Maker Kit (Cellscript, Madison, WI, USA). Therefore GFP-GlyR α1 in pCDNA3.1(+) was linearized with NotI. Oozytes were surgically taken from female X. laevis after anesthesia with 0.1% Tricaine in water. For SMA-copolymer solubilization, 300-400 oocytes were injected with 50 ng in a volume of 50.6 nl of cRNA. After injection the oocytes were incubated in ND-96 solution (96 mM NaCl, 2 mM KCl, 1 mM CaCl 2 , 1 mM MgCl 2 , 5 mM HEPES, pH 7.4) at 18 °C for 1-2 days. SMA-copolymer solubilization of α1 GlyR. For SMA copolymer solubilization of the heterogeneous expressed α1 GlyR in HEK293 cells, cells were washed with PBS, subsequently scraped off and resuspended in 2 ml PBS. Cells were washed and resuspended in 150 mM NaCl, 50 mM Tris/HCl pH 8.0 and lysed by sonification. Membranes are separated by ultracentrifugation at 100,000g for 1.5 h and 4 °C. For α1 GlyR expressed in oocytes, cells were resuspended in 20 mM Tris-HCl, pH 8.0 and mechanically homogenized by pipetting. To remove cell debris, cells were centrifuged at 1000×g for 15 min at 4 °C. Membranes were separated by ultracentrifugation at 100,000×g for 1 h at 4 °C. To remove additional yolk proteins 22 , pellet was washed with 1 M NaCl, 20 mM Tris-HCl, pH 8.0, followed by an additional ultracentrifugation step as written above. The membrane pellet was resuspended in SMA-solubilization buffer (150 mM NaCl, 10% glycerol, 50 mM Tris/HCl pH 8.0, SIGMAFAST Protease Inhibitor Cocktail (Sigma-Aldrich, St. Louis, MO, USA) to a concentration of 80 mg/ml. A freshly prepared 4% (w/v) SMA copolymer solution (Lipodisq Styrene:Maleic Anhydride Copolymer 3:1, Prehydrolyzed, Sigma-Aldrich) in SMA-solubilization buffer was slowly dropped under stirring to the membrane suspension in a 1:1 ratio and was solubilized for 1 h at room temperature. To remove all non-solubilized cell fragments, the suspension was centrifuged at 100,000g for 45 min and 4 °C.
Ni-NTA purification and size exclusion chromatography. The α1-His GlyR SMA copolymer nanodiscs containing supernatant was incubated with pre-equilibrated 0.2 ml HisPur Ni-NTA spin columns (Thermo Scientific RepoRtS | (2020) 10:16569 | https://doi.org/10.1038/s41598-020-73157-2 www.nature.com/scientificreports/ Fisher Scientific) overnight at 4 °C under gentle rotation. Spin columns were washed three times with SMA-solubilization buffer supplemented with 25 mM imidazole and α1-His GlyR SMALPs were eluted in SMA-solubilization buffer supplemented with 400 mM imidazole. For separation of the previously purified α1 GlyR nanodiscs from other soluble proteins and for further analysis, pooled elution fractions were loaded on a Superdex 200 increase 10/300 GL (GE Healthcare, Chicago, IL, USA) connected to an Äkta pure (GE Healthcare) and buffer was exchanged to 50 mM Tris/HCl pH 7.4 and 150 mM NaCl. Estimation of the molecular weight (MW) was based on a calibration curve by a linear fit of proteins of known MW (Aldolase, Ovalbumin, Conalbumin, Cyanocobalbumin, Thyroglobulin) versus the partition coefficient k av (k av = V el − V 0 /V t − V o , where V el is the elution volume of the protein, V t is the total column volume and V 0 is the void volume). The hydrodynamic radii (R s , Stokes radius) where calculated, according to the Stokes-Einstein relation 23 , by a linear fit of the Stokes radii of known proteins versus the square root of the negative decadic logarithm of the k av .
SDS-PAGE and western blot. SDS-PAGE and western blot analysis was performed as described elsewhere 24 . In brief, proteins were separated using a 10% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE). For SDS-PAGE analysis, the gel was stained with Pierce Silver Stain Kit (Thermo Fisher Scientific). For western blot analysis, separated proteins were transferred to a PVDF membrane (Bio-Rad, Feldkirchen, Germany) and the membrane was blocked for 1 h in TBS-T supplemented with 5% skim milk. Afterwards, the membrane was incubated with 1:500 primary eGFP Polyclonal antibody (CAB4211, Thermo Fisher Scientific) in TBS-T containing 1% skim milk over night at 4 °C. The membrane was washed 3 times for 10 min with TBS-T and incubated with the goat anti-rabbit IgG-HRP (sc-2054, Santa Cruz Biotechnology, Dallas, TX, USA) in TBS-T containing 1% skim milk for 1 h at room temperature. The membrane was washed 3 times for 10 min in TBS-T and protein bands were visualized by adding Pierce Western Blotting Substrate (Thermo Fisher Scientific) and detected with a CCD camera.

Microscale thermophoresis.
Microscale thermophoresis (MST) analysis was performed using a NanoTemper Monolith NT.115 instrument (NanoTemper Technologies, Munich, Germany). Therefore, purified α1-His GlyR nanodiscs or total solubilized cell membranes with α1-GFP GlyR were diluted to a concentration of 400 nM in PBS. α1-His GlyR was fluorescence-labeled using the Monolith NT His-Tag Labeling Kit RED-tris-NTA (NanoTemper Technologies). Labeled GlyR α1 SMALPs were added in a 1:1 ratio to a 1:2 dilution series with a final concentration of 3 mM down to 0.73 µM for glycine or 12.5 mM down to 6 µM for taurine, as well as 0 µM for each ligand as an internal control and loaded into standard capillaries (Monolith NT.115 Capillaries, NanoTemper Technologies). Thermophoresis was measured at 21 °C for 15 or 20 s with 40% LED power and 60% infrared laser power. For MST experiments n = 3-4 independent technical measurements were collected from N = 2-3 independent oocyte or HEK293 cell batches. Data and statistical analysis. For electrophysiological dose-response analysis, normalized current responses were plotted against the agonist concentration and fitted with a sigmoidal Hill equation I/I max = 100 × c n /(c n + aEC n 50 ) in GraphPad Prism 8 (GraphPad Software Inc., La Jolla, USA), where I/I max is the normalized current, c the concentration, n the Hill coefficient and aEC 50 the agonist concentration resulting in a half-maximal response. For thermophoretic binding experiments, the relative thermophoretic fluorescence signal F norm was calculated as ratio of the initial fluorescence (F cold = 1 s) and fluorescence after thermodiffusion (F hot = 15 s) using the following equation in MO.Affinity Analysis software (NanoTemper Technologies): where ∂F/∂T is the fluorescence change due to the fluorophore's temperature dependence and S T the Soret coefficient. For thermophoretic dose-response analysis, the normalized relative fluorescences were plotted against the agonist concentration and fitted with the sigmoidal Hill equation F norm /F norm,max = 100/(1 + (cEC 50 /c) n ) , where F norm /F norm,max is the normalized relative fluorescence, cEC 50 the agonist concentration resulting in a half-maximal response and n the Hill coefficient. All values are given in mean ± SEM, unless indicated otherwise. Statistical significance was determined using a Student's twotailed, unpaired two-side t test with *p < 0.05, **p < 0.01, and ***p < 0.001 levels. Equality of variances was confirmed using a F test.

Results
Detergent-free purification of homomeric α1 GlyR in SMA-copolymer nanodiscs. To probe the general potential of SMA-solubilized GlyRs for pharmacological analysis we solubilized a His-tag fused α1 construct (α1-His) by incubating SMA copolymers with the isolated membrane fraction of α1-His overexpressing Scientific RepoRtS | (2020) 10:16569 | https://doi.org/10.1038/s41598-020-73157-2 www.nature.com/scientificreports/ HEK293 cells (Fig. 1a). The α1-His containing nanodiscs (Fig. 1a) were further isolated with Ni-affinity chromatography, followed by size-exclusion chromatography (SEC). SDS-PAGE analysis of peak fraction one (Fig. 1b) shows a clear band, migrating at ~ 48 kDa, corresponding to the molecular weight of the α1 GlyR monomer (Fig. 1, Supplementary Fig. S1). The SMA copolymer runs at an expected lower molecular weight 26 of approximately 10 kDa. The second SEC peak (Fig. 1b) reveals only SMA copolymer without any additional protein band, which represent probably free polymer that was interacting with the Ni-affinity columns. The molecular weight and size of the purified α1 GlyR nanodiscs were further investigated by the generated SEC data. Thus, the nanodiscs having an average molecular weight of 495 kDa with a calculated stoke radius of 13.2 nm, which is in good agreement with previous studies [26][27][28] . Since the pentameric receptor has a molecular weight of 240 kDa and no accessory proteins in the purified fraction can be detected, we conclude that the residual molecular weight of the nanodiscs must result from its lipid bilayer. We therefore showed that the SMA copolymer solubilization of α1 GlyR in nanodiscs was successful and suitable for further investigation.

Functional characterization of α1 GlyR nanodiscs by microscale thermophoresis.
To examine the functionality of the purified α1 GlyR nanodiscs, we intended to analyze concentration-dependent conformational changes upon binding of the agonist glycine by microscale thermophoresis (MST). Therefore, we diluted the fluorescence labeled α1-His GlyR containing nanodiscs (see "Materials and Methods" section) to a final concentration of 400 nM and added a glycine dilution series of 3 mM down to 0.73 µM. Samples were loaded into glass capillaries and get focused by an infrared (IR) laser, creating a spatial temperature gradient. Thermophoretic movement of the fluorescence labeled α1-His GlyR nanodiscs was measured by fluorescence emission coupled into the IR laser path 29 . We found that addition of glycine to the receptor complex was sufficient to induce a change in thermophoretic mobility of the complex (Fig. 2a). Increasing glycine concentrations resulted in a concentration-dependent shift of the thermophoretic signal saturating at 500 µM glycine (Fig. 2b). Because no concentration-dependent fluorophore quenching could be observed (Fig. 2c), we concluded that the purified SMA copolymer nanodiscs must contain functional pentameric α1 GlyR receptors and that the alterations in particle movement obtained must be due to a glycine-induced alteration of the conformation, size, charge and/or hydration shell 30 of the GlyR. This is consistent with the findings at native GlyRs where (i) pentameric assembly of the GlyR is required for the proper formation of the specific glycine-binding site at the interface of adjacent subunits and (ii) binding of glycine induces a significant conformational rearrangement of the ECD. Analyzing the glycine depended shift of the thermophoretic movements revealed an EC 50 value of 65 ± 22.8 µM (mean ± SEM; n = 4; Fig. 2d), which is similar to the apparent EC 50 value obtained by electrophysiology in α1 GlyR expressing HEK293 cells 12,13 . To distinguish electrophysiological and MST-determined EC 50 values, we termed them in the following as aEC 50 (a for apparent; e-phys) and cEC 50 (c for conformation, MST), respectively. In conclusion, our results obtained upon analyzing glycine concentration-dependent conformational changes of the GlyR by MST embedded in SMA copolymer nanodiscs, indicate that the glycine-induced contraction within the ECD leading to a highly efficient gating of the channel pore, resulting in a 1:1 ratio of cEC 50 and aEC 50 values.
Characterization of glycine binding to α1 GlyR obtained from HEK293 cells and X. laevis oocytes.
One key advantage of solubilizing membrane proteins in SMA copolymer nanodiscs is the retention of their native lipid surrounding, enabling the analysis of ligand-induced conformational changes while preserving the influence of the native membrane environment in in vivo-like conditions. Homomeric α1 GlyR exhibit a major discrepancy concerning their glycine affinity with aEC 50 values of 212.9 ± 21 µM and 68.8 ± 7.4 µM when www.nature.com/scientificreports/ heterologously expressed in X. laevis oocytes or HEK293 cells, respectively (Fig. 3a). The root cause and underlying mechanism of these variations in the efficiency of receptor activation across different expression systems is still unknown. To further explore this concept we analyzed the binding affinity for α1 GlyR extracted from HEK293 cells and oocytes. To maximally mimic the complex environmental conditions found in the cells and to simplify the analytical process, we used an N-terminal GFP fused receptor (α1-GFP GlyR) construct, that was directly measured after SMA-solubilization within the total nanodiscs fraction, that has previously reported for soluble GFP fused proteins 29 . Western blot analysis (Fig. 3b, Supplementary Fig. S2) of the solubilized membrane fractions indicates an adequate incorporation of GlyRs into SMA copolymer nanodiscs. For comparable results, we first probed the binding affinity for glycine from HEK293 cell lysate. α1-GFP GlyRs also showed a concentration-dependent shift in their thermophoretic mobility, similar to purified and fluorescence labeled His-tagged α1 GlyR, with a cEC 50 value of 40.9 ± 13.4 µM (p = 0.22, n = 4). A concentration-dependent GFP quenching was not detected as well (not shown). Thus our GFP tagged α1 GlyR yield similar cEC 50 values compared to pure α1 GlyR nanodiscs and was suitable for the comparative analysis of α1 GlyRs expressed in oocytes. Next we analyzed the apparent glycine binding affinities of α1 GlyR extracted from oocyte membranes (Fig. 3a,c). The determined cEC 50 of 52.6 ± 40.8 µM (n = 4) for glycine reveals no significant difference (p = 0.58; Fig. 3c) compared with the cEC 50 value obtained from HEK293 cells. Thus, the cEC 50 value obtained from HEK293 cells is identical with the aEC 50 (p = 0.23), whereas the cEC 50 value obtained from oocytes is about 4 times lower than the aEC 50 (p < 0.01). Furthermore, the similar thermophoretic signal amplitudes (p = 0.41; Fig. 3d) obtained from both expression systems indicate that the GlyR adopts the same ECD configuration with a similar degree of domain-closure upon glycine binding. These findings implicate, that the efficiency of glycine to induce the conformational change within the ECD is identical in both expression systems and the increased aEC 50 observed in oocytes must be linked to an impaired channel opening. www.nature.com/scientificreports/ conformational changes induced by binding of the partial agonist taurine. Taurine acts as a partial agonist on α1 GlyR in oocytes with an aEC 50 value of 843 ± 16 µM (n = 3) and a maximal current of 61% compared to glycine (Fig. 4a). To investigate how taurine influences the ECD conformation and if glycine may act as a partial agonist in oocytes, we analyzed the conformational transformation of α1-GFP GlyR containing SMA nanodiscs in response to taurine by MST, as well. As expected, the cEC 50 value of 473.8 ± 66.1 µM (n = 3; Fig. 4b) is much higher than for glycine and is approximately half as much as the obtained aEC 50 value (n = 3; p < 0.05). Also the Hill coefficient of 0.8 decreased for taurine, which is in good agreement with previous studies, reporting a decrease for less efficient agonists 15,31 . Since the binding of the partial agonist taurine might stabilize a conformational state with an altered ECD closure, we analyzed the discrepancies in thermophoretic movement between the unbound and agonist bound states. We found, that the signal amplitude between glycine and taurine significantly decreases (p < 0.05, Fig. 4c,d) from 3.38 ± 1.09 to 1.32 ± 0.17, respectively. This finding implicates, that taurine stabilizes a distinct conformational state under saturating conditions that can be clearly distinguished from the conformational state adopted after glycine binding. In contrast, binding of glycine to the GlyR induces in both expression systems the same overall conformational configuration, leading to the conclusion that glycine acts not as a partial agonist in oocytes like taurine.

Discussion
In this study, we characterized the conformational states of the homomeric α1 GlyR in SMA copolymer nanodiscs by MST during receptor activation by glycine and the partial agonist taurine upon heterologous expression in HEK293 cells and Xenopus oocytes. Our results indicate that in the GlyR (i) partial and full agonists induce different conformational transitions and (ii) conformational transitions after agonist-binding affect apparent affinity. Thus our data shed light on the conformational features and mechanisms determining receptor activation by different agonists in different cell systems in the pLGIC superfamily. In principle, two different models for the conformational states induced by a full and partial agonist at pLGICs can be envisaged. First, binding of the agonist simply shifts the equilibrium between the closed state and the open state of the channel. This model is based on the initial work by del Castillo and Katz 32 , where it has been supposed that full and partial agonists differ in their efficiency to activate a receptor, implying that both partial and full agonists stabilize the same open conformational state of the receptor. Consequently, a full agonist would promote efficient channel opening, while a partial agonist would be less effective in stabilizing the open state. Second, it can be assumed that a full and a partial agonist can adopt different conformational states in the agonist binding domain which are correlated to the extent of channel opening, mainly depending on the specific stereochemical properties of the ligand. Recent studies at the GlyR indicate that receptor activation includes one or more intermediate states, so called flipped or primed states, where the ECD is closed but the channel is still shut 9,10,20 . In this model, the origin of partial agonism would be related to a reduced ability of the receptor to reach an intermediated pre-open state, rather than a reduced ability to fully activate the receptor, when the intermediated is reached 10,33 . Nevertheless, little is known if the altered receptor kinetic for partial agonists is also reflected by distinct conformational states, as seen in ionotropic glutamate receptors 34,35 .
Our thermophoretic analysis during receptor activation induced by agonists reveals large conformational changes of the GlyR. We attribute the detected agonist-induced rearrangements to a contraction within the ECD entrapping the ligand between adjacent subunits 8,36,37 , rather than structural changes within the channel pore itself, which are masked by the surrounding lipid-bilayer of the nanodisc. This is consistent with a recent finding that, although full and partial agonists of the GlyR have the same orientation within the ECD upon binding, www.nature.com/scientificreports/ interactions between the amino groups of the agonists and receptor residues differ remarkably resulting in a different extent of ECD contraction with the volume of the agonist binding pocket smallest in the glycine-bound structure 38 . Since our measurements were performed under equilibrium conditions, our thermophoretic data therefore most likely reflect structural changes between the resting conformation, with an open ECD, and one or multiple intermediate states, as well as the open state, were the ECD is contracted (Fig. 5a). Thus, our MST measurements would reflect exclusively the agonist-induced conformational change within the ECD, whereas the electrophysiological measurements determine the general ability of the agonist to open the ion channel (aEC 50 value). Most strikingly, our MST measurements revealed different conformational changes of the ECD induced by taurine compared to the full agonist glycine. Therefore, we assume that the interactions between taurine and the ECD loops result in a lesser extend of ECD contraction, and therefore to the observed decreased thermophoretic mobility. This indicates that glycine-and taurine-binding induces a different degree of conformational change in the ECD and that the extent of agonist-induced ECD closure is correlated with the efficacy of agonists.
Since we cannot discriminate in our MST measurements between distinct closed ECD conformations, two general activation models for partial agonists can be envisaged (Fig. 5b). First, the taurine-induced conformation reflects a closed intermediate state that further adopts the same closed ECD conformation as seen for glycine during full activation (Fig. 5b, upper row). Second, the taurine-induced conformation reflects the final open conformation, without reaching a fully contracted ECD conformation (Fig. 5b, lower row). However, recent findings support the existence of an intermediated taurine-bound state, that is characterized by a lesser degree of ECD rotation 38 .
Depending on the cell system analyzed, the GlyR displays in electrophysiological measurements a remarkable difference in the apparent affinity (aEC 50 ) of its agonists 4 . Thus, the apparent affinity of an agonist is not exclusively determined by its specific stereochemical properties. Previous studies supposed that intermolecular cooperativity at higher receptor expression levels may influence agonist affinity in different cell systems 39 . Our experiments show that the large variability of the aEC 50 for glycine observed in different cell systems (HEK293 vs. oocyte) is not correlated with the efficiency to induce a contraction within the ECD (cEC 50 ) after glycine binding. In addition, the cEC 50 values in HEK293 and oocytes are similar to the aEC 50 value obtained from electrophysiological measurements in HEK293 cells, indicating an efficient receptor activation in HEK293 cells once the ECD has closed, consistent with previous single-channel measurements 9,10,20 . As a direct consequence, the receptor equilibrium lies strongly on the ligand-bound, open conformation. In contrast, while the cEC 50 is unchanged in oocytes, the aEC 50 apparent affinity is decreased. We therefore conclude, that the decreased apparent affinity of GlyR in oocytes probably arise from a limited ability to open the channel pore once the ECD is closed, rather than the ability to induce a conformational change in the ECD layer upon glycine binding (Fig. 5c). We speculate that the impaired coupling of the ECD and the TMD is driven (i) by lipid or sterol modulations, as described for other members of pLGICs [40][41][42] and supported by recent findings supposing interactions between glycine receptors and cholesterol 43 or (ii) by the impact of the ICD 44 . This would explain the high variability of aEC 50 values seen upon electrophysiological measurements of GlyRs in the brain and in different heterologous expression systems. Thus, the apparent agonist-affinities of the GlyR obtained in electrophysiological measurements are likely determined by conformational transitions after the agonist-induced ECD closure which is in agreement with several postulated intermediate shut states between the resting and open conformation, as indicated by different primed or flipped schemas. Our data may also have some implications for the understanding of the differential impact of (i) mutations 45 and modulators 25 in the ECD and (ii) the TMD and ICD 44 at the GlyR in affecting partial agonism. Therefore mutations and modulators acting in the ECD might affect the conformational change in the ECD layer upon agonist binding whereas modulators acting in the TMD and the ICD may alter intermediate shut states between the resting and open conformation. Future work will need to explore these possibilities by testing the effects of mutations, modulators and the ICD on conformational states by thermophoretic analysis of the GlyR extracted in SMA-copolymer nanodiscs enabling the receptor to occupy physiologically relevant states.
In summary, by analyzing conformational states of the GlyR by MST, our results indicate that first partial agonism in the pLGIC family is reflected by the adaption of distinct receptor conformations and second that modulation within the TMD region and/or the ICD causes the variable apparent affinities seen in different cell systems. Therefore, our approach provides an easy access to correlate structural and functional impacts of ligand binding and may also help in the mechanistic understanding of positive allosteric modulators as well as in rationalized drug design.

Data availability
All data supporting the findings of this study are available within the article or are available from the corresponding author upon reasonable request. All cDNA constructs are available from the corresponding author based on reasonable request.