Mechanisms of activation and desensitization of full-length glycine receptor in lipid nanodiscs AbstractGlycinergic synapses play a central role in motor control and pain processing in the central nervous system. Glycine receptors (GlyRs) are key players in mediating fast inhibitory neurotransmission at these synapses. While previous high-resolution structures have provided insights into the molecular architecture of GlyR, several mechanistic questions pertaining to channel function are still unanswered. Here, we present Cryo-EM structures of the full-length GlyR protein complex reconstituted into lipid nanodiscs that are captured in the unliganded (closed), glycine-bound (open and desensitized), and allosteric modulator-bound conformations. A comparison of these states reveals global conformational changes underlying GlyR channel gating and modulation. The functional state assignments were validated by molecular dynamics simulations, and the observed permeation events are in agreement with the anion selectivity and conductance of GlyR. These studies provide the structural basis for gating, ion selectivity, and single-channel conductance properties of GlyR in a lipid environment. IntroductionGlycine receptors (GlyRs), along with 纬-aminobutyric acid receptors (GABAARs), are the principal determinants of fast inhibitory synaptic neurotransmission in the central nervous system (CNS). GlyR and GABAAR belong to the superfamily of pentameric ligand-gated ion channels (pLGICs). The vertebrate members of pLGIC also include excitatory cation-selective channels (nicotinic acetylcholine receptors鈥攏AChRs and serotonin receptors鈥?HT3Rs). Glycinergic neurotransmission critically regulates motor coordination, sensory reflex activity, and respiratory rhythms. GlyR dysfunctions are associated with neuromotor deficiencies such as hyperekplexia and epilepsy, and underlie chronic inflammatory pain1,2. Agents that potentiate GlyR function produce analgesia making them attractive candidates for pain therapy3. GlyRs assemble as either homopentamers of 伪 subunits (伪1鈭捨?) or heteropentamers of 伪 and 尾 subunits. GlyR has a conserved pLGIC architecture consisting of an extracellular domain (ECD), a transmembrane domain (TMD), and an intracellular domain (ICD). The ECD houses the neurotransmitter binding pocket and the TMD governs the machinery for selective ion permeation. The ICD is important for receptor trafficking, synaptic clustering, and a central site for posttranslational modifications and regulation by endogenous and exogenous modulators4. Broadly, pLGIC function involves transitions between at least three types of distinct functional states, namely the resting, open, and the desensitized states. Upon agonist binding, the channel switches from the resting (Apo) state to open and desensitized conformations. Allosteric ligands modulate pLGIC activity by shifting the equilibrium between these functional states.Recently, there has been ground-breaking progress in structure determination of both cationic and anionic members of the pLGIC family, providing a detailed view of assembly and conformational changes in homomeric and heteromeric channels5,6,7,8,9,10,11. For GlyRs, X-ray crystallographic and cryo-electron microscopic (cryo-EM) studies have provided the first snapshots of the channel in distinct conformations8,11. These landmark studies revealed a conserved mechanism for channel activation involving an agonist-induced global twist in the ECD and TMD, leading to channel opening. However, there are inconsistencies between these structures and previous functional work, particularly concerning charge selectivity, permeation, and single-channel conductance. One concern is that the pore of the previously reported open GlyR conformation (PDB_ID: 3JAE) is considerably wider than what has been seen in other anionic pLGIC structures, particularly at the selectivity filter8. Molecular dynamics (MD) simulations with this structure predict a channel conductance that is higher than that observed in electrophysiology studies and the structure is only partially blocked by the complete pore-blocking antagonist picrotoxin12, suggesting that the pore opening may have been exaggerated. Moreover, in MD simulations performed in the absence of backbone restraints, the open structure has been shown to rapidly collapse at the Leu9鈥?region of the pore13,14. Potential causes for these discrepancies are that the previous structures were solved with a truncated ICD (consisting of 75鈥?25 residues) to improve sample stability and in a detergent environment. The ICD, besides its role in trafficking, has also been implicated to alter single-channel conductance15, modulation of gating16, and desensitization17. In addition, a recent report shows that ICD truncation in GlyR leads to an increase in maximum single-channel open probability and a substantial increase in agonist efficacy18. Furthermore, lipids are essential components for membrane protein stability, and it therefore seems likely that both the ICD and lipid environment are important for maintaining the structural integrity of the channel during gating conformational cycles. Lipid nanodisc derived from membrane scaffolding protein (MSP) is one of the tools used to mimic the lipid bilayer environment9,19. We therefore determined high-resolution structures of the full-length GlyR in a membrane nanodisc environment by single-particle cryo-electron microscopy (Cryo-EM) under conditions that stabilize various functional states. This allowed us to elucidate ligand-induced conformational changes that give a complete mechanistic description of gating.ResultsCryo-EM structures of GlyR gating conformationsThe zebrafish GlyR伪1 (Supplementary Table聽1) exhibits robust macroscopic currents in response to application of glycine with an EC50 of 100鈥壩糓 with current amplitude saturating beyond 1鈥塵M20. In the continued presence of glycine, the channels desensitize within minutes. Among the blockers of open-channel pore of anionic pLGICs, include picrotoxin (PTX), a plant alkaloid that occurs as an equimolar mixture of picrotin and picrotoxinin. While picrotoxinin is a more potent inhibitor in GABAARs, both components are found to be of similar efficacy in GlyRs21. Upon application of PTX, GlyR currents are strongly inhibited with an IC50 of 5鈥?鈥壩糓, and a near complete inhibition is achieved at concentrations 1鈥塵M22. The block of GlyR currents is reversible, and previous studies have demonstrated that PTX block prevents the channel from desensitizing23. Our strategy to capture the GlyR in the resting, open, and desensitized conformations, was therefore to prepare samples of GlyR in the absence of a ligand, in the presence of both glycine and PTX, and in the presence of glycine, respectively (Supplementary Fig.聽1). Ideally, one would hope to capture the channel in the open conformation in the absence of any perturbation (mutation, blocker, or potentiator). However, given the transient nature of the open conformation and the time requirements of cryo-EM sample preparation, extensive efforts toward this goal did not yield the desired outcome. Since reported mutations do not completely obliterate desensitization, we resorted to the use of PTX.The full-length glycine 伪1 receptor (GlyR) gene from Zebrafish was codon-optimized, cloned into pFastBac1 plasmid, and expressed in Spodoptera frugiperda (Sf9) cells (Supplementary Table聽2, Supplementary Fig.聽2). The sample stability was improved by incorporating a lipid mixture consisting of soybean polar extract (asolectin) and cholesteryl hemisuccinate (CHS) during detergent solubilization and purification. The purified pentameric population of GlyR was then reconstituted into asolectin nanodiscs with the E3D1 MSP and the nanodisc samples were used for single-particle cryo-EM analysis (Supplementary Fig.聽3). The samples were imaged in the absence of ligand (GlyR鈥揂po), in the presence of 5鈥塵M glycine (GlyR鈥揋ly), and in the presence of 5鈥塵M glycine and 3鈥塵M picrotoxin (GlyR鈥揋ly/PTX). Three dimensional reconstructions with imposed C5 symmetry for the three samples led to density maps with nominal resolutions of 3.33聽脜, 3.47聽脜 and 3.51鈥壝? respectively (Supplementary Fig.聽4, Supplementary Table聽3). In each case, the final reconstruction contained density for the entire ECD and TMD as well as a select region of the ICD, which were used for model building and refinement (Supplementary Fig.聽5). The ICD region between Phe341 and Lys394, which is a part of the M3鈥揗4 loop, is unstructured and appears as segmented density that is not conducive to model building. The density surrounding the TMD corresponds to the nanodisc belt comprising of helical segments of the MSP and lipid bilayer within the enclosure (Supplementary Fig.聽3C). There are three sets of additional nonprotein densities in each subunit; one appears as an extension from Asn62 corresponding to N-glycans and the other two are lipid-like densities (in GlyR鈥揂po) in the vicinity of M4 (Supplementary Fig.聽3C). The glycan density at Asn62 was also observed in previous structures8,11 and corresponds to the single glycosylation site shown in GlyR伪1. Mutations at this site have been reported to affect surface expression of functional GlyRs by preventing their exit from the endoplasmic reticulum24. The lipid-like densities at the extracellular end was modeled as phosphatidylcholine and the other at the cytosolic end was partially modeled as a phospholipid (Supplementary Fig.聽6). Interestingly, the density at the intracellular end is in a similar location to PIP2 density observed in a recent GABAAR structure9.The overall architecture, consistent with the previous structures of GlyR and other pLGICs8,11, consists of a symmetrical arrangement of the five subunits with dimensions of 125鈥壝?in height and 80鈥壝?in diameter. Each subunit consists of a twisted 尾-sheet of ten strands forming the ECD, four 伪-helical strands comprising the TMD, and a primarily unstructured region between the third and fourth TM helices forming the ICD (Supplementary Fig.聽2B). The TM helices splay outward toward the extracellular end of the bilayer. The M4 helix extends out of the putative bilayer boundary (referred to as post-M4), by three turns of the helix and makes contact with the ECD. The ICD is the least conserved region among the pLGIC family and varies both in length as well as amino acid sequence. Based on the ICD sequence, one expects extensive structural variations between anionic and cationic pLGICs, consistent with the distinct roles of this domain in the two channel classes. In GlyR, the M3鈥揗4 linker extends as a continuous stretch of 伪-helix from the C-terminal end of M3 up to four turns (referred to as post-M3). Beyond this, it is predicted to be an unstructured region that eventually terminates with a short 伪-helix (referred to as pre-M4) that extends into the M4 helix. In contrast, the ICD in cationic pLGICs consists of an unstructured loop (post-M3 loop), followed by an amphipathic MX helix that runs parallel to the membrane, an unstructured region that ends in a long 伪-helix MA which appears as a single continuous helix with M425. In the resting conformation of cationic pLGICs, both the post-M3 loop and the MA helix occlude the ion permeation pathway. In 5-HT3AR, ligand-binding elicits large conformational changes in this region to create a lateral portal for ion exit6,7. In the absence of extensive structured motifs, it remains unknown if the ICD in the anionic pLGIC exerts such a barrier to ion permeation.The GlyR鈥揂po, GlyR鈥揋ly/PTX, and GlyR鈥揋ly structures reveal distinct conformational states of the ion permeation pathway, which originates at the ECD, extends through the TMD, and terminates shortly within the ICD (Fig.聽1a). Within the TMD, the pore is lined by the M2 helix from each subunit in an arrangement that appears more cylindrical in GlyR鈥揂po and becomes progressively funnel shaped, tapering toward the intracellular end, in GlyR鈥揋ly/PTX and GlyR鈥揋ly. In the GlyR鈥揂po, the pore is relatively narrow and constricted prominently at Leu9鈥?(pore radius ~1.4鈥壝? and Thr13鈥?(pore radius ~2.2鈥壝?. The pore dimensions at positions Ala20鈥?and Pro-2鈥?are also below the Born radius for the solvated chloride ion, which is 2.26鈥壝? (the Pauling radius for chloride ion is 1.81鈥壝? and are therefore likely barriers to ion permeation (Fig.聽1b). The M2 helices show partial unwinding between Gly17鈥?and Ala20鈥? a feature that is consistent with the dynamic behavior of this region previously noted in nuclear magnetic resonance studies of the isolated GlyR TMD26. In the GlyR鈥揋ly/PTX structure, the M2 helices are oriented outward with the Leu9鈥?rotated away from the central axis (Fig.聽1c). The pore radius profile (created with the PTX ligand removed from the pore) shows an expanded ion permeation pathway at each of the constriction points seen in GlyR鈥揂po. In GlyR鈥揋ly, the pore is similar in radii to that of the GlyR鈥揋ly/PTX but reveals a notable constriction (pore radius ~1.8鈥壝? at the level of Pro-2鈥?due to slight bending of the intracellular end of M2. The Pro-2鈥?position is part of the charge selectivity filter in anionic channels comprising of a conserved stretch (Pro-2鈥? Ala-1鈥? and Arg0鈥?27 and mutations in this stretch are associated with hyperekplexia28. While Pro-2鈥?and Ala-1鈥?face the lumen of the pore, Arg0鈥?faces away. Previous studies have shown that both the charge and the side-chain conformation of Arg0鈥?are determinants of anion selectivity29.Fig. 1: Cryo-EM structures of full-length GlyR in multiple conformational states.a Ion permeation pathway generated with HOLE60 for GlyR鈥揂po (salmon red), GlyR鈥揋ly/PTX (deep teal) and GlyR鈥揋ly (slate blue). For clarity, the cartoon representation of only two non-adjacent subunits are shown. Green and purple spheres define radii of 1.8鈥?.3鈥壝?and 3.3鈥壝? respectively. The residues located at various pore constrictions are shown as sticks. b The pore radius is plotted as a function of distance along the pore axis. The dotted line indicates the approximate radius of a hydrated chloride ion, which is estimated at 2.26鈥壝? c A view of M2 helices from the extracellular end for the three GlyR conformations. Positions Leu9鈥?and Pro-2鈥?are shown in ball-and-stick representation and the corresponding distances between C伪 are given in 脜.Full size imageIn addition to the physical dimensions of the pore, the hydrophobicity of the pore-lining residues plays a key role in dictating the conductance state of the channel. In some cases, a pore that is not narrow enough for steric occlusion may still impede ion permeation due to the hydrophobicity of the pore-lining residues that may disfavor water molecules, causing local dewetting that leads to a higher energetic barrier for water and ion permeation30. We assessed the combined effect of both the pore radii and hydrophobicity on pore dewetting and presence of hydrophobic gates using a previously developed simulation-free heuristic model that was derived using machine learning from MD simulations of nearly 200 ion channel structures31. A pore surface profile based on the radii of the permeation pathway and hydrophobicity of the pore-lining amino acid side chains was estimated by CHAP32 (Supplementary Fig.聽7A). On the plot of the local pore radii at each pore-lining residue against its side-chain hydrophobicity, the dotted line marks the divide of the hydrophobicity-radii landscape into regions of high and low likelihood of pore wetting (Supplementary Fig.聽7B). The sum of shortest distances (危d) from positions falling in the low-likelihood region to the line is used as a heuristic score which allows a prediction for the likelihood of the conformation corresponding to a dewetted and nonconductive state. A cut-off value of 危d鈥?gt;鈥?.55 is used for predicting a nonconductive conformation. In the GlyR鈥揂po, several residues (Pro-2鈥? Ile5鈥? Leu9鈥? and Thr13鈥? fall below the dotted line, leading to an overall heuristic score of 危d鈥?鈥?.11. In the GlyR鈥揋ly/PTX, there are no residues below the line and the score is 0.0. On the other hand, in the GlyR鈥揋ly structure, Pro-2鈥?is well below the line with Val1鈥?lying close and the score is 0.80. These findings suggest that the pore in Gly鈥揋ly/PTX structure is likely to be open, whilst those in the GlyR鈥揂po and GlyR鈥揋ly structures are closed. The electrostatic map shows the positively charged cluster at the intracellular end of M2 and the regions of the ICD close to the TMD interface (Supplementary Fig.聽7C). The Arg0鈥?that is implicated to be the selectivity filter in anionic pLGIC lies on the non-pore facing side of M2 at the interface of the TMD and ICD.The GlyR鈥揋ly and GlyR鈥揋ly/PTX structures were solved in the presence of 5鈥塵M glycine and the 3D reconstructions reveal an additional small density in the canonical neurotransmitter-binding pocket wedged at the subunit interface in the ECD (Fig.聽2a). An unambiguous assignment of a small ligand such as glycine is difficult聽at this resolution, however, since the density was not seen in GlyR鈥揂po, and the location matched with the glycine density in the 伪3-GlyR crystal structure33, we modeled the glycine ligand accordingly. The neurotransmitter binding pocket is lined by several conserved aromatic residues and bulky side-chains, from both the principal (Phe123, Phe183, Tyr226, Thr228, and Phe231) and complementary subunits (Phe68, Arg89, Leu141, Ser145, and Leu151). LigPlot analysis shows interaction of glycine with several of these residues (Fig.聽2b). Mutational studies have shown that a significant impact on agonist affinity ensues upon perturbation of this pocket34. An overlay of GlyR鈥揂po and GlyR鈥揋ly shows substantial conformational changes that lead to a rearrangement of the residues in the binding pocket (Fig.聽2c). The major motion includes an inward movement of Loop C (carrying Tyr226, Thr228, and Phe231) and Loop B (Phe 183). Upon glycine binding, the tip of Loop C (measured at C伪 of Thr228) is displaced by 6.2鈥壝?toward the neurotransmitter binding pocket thereby shrinking the pocket and potentially occluding water molecules. The inward or 鈥渃apping鈥?motion of Loop C is well-described in pLGIC literature and is associated with agonist-binding35 and is also observed in the recent structures of the unliganded and agonist-bound conformations7,8,36.Fig. 2: Conformational changes underlying GlyR gating.a Cryo-EM density segments for neurotransmitter binding site residues and glycine ligand as seen in the GlyR鈥揋ly structure. Residues in the principal subunit and complementary subunit are indicated in black and magenta, respectively. b LigPlot analysis of glycine orientation in the binding pocket and residues within 4鈥壝?distance are displayed in red70. c Comparison of the neurotransmitter binding site for the GlyR鈥揂po and GlyR鈥揋ly conformations. The residues that are involved in neurotransmitter binding are shown in sticks. d Cryo-EM map showing the density for M2 and PTX bound in the pore of GlyR鈥揋ly/PTX. The interacting residues are shown in stick representation. For clarity, M2 for only two diagonal subunits are shown. e LigPlot analysis of PTX and the interacting residues. f A close-up of the M2 conformations is shown upon aligning the three GlyR conformations. Positions Leu9鈥?and Pro-2鈥?are shown as sticks.Full size imageIn the presence of glycine, PTX reversibly blocks GlyR currents (Supplementary Fig.聽1). In the GlyR鈥揋ly/PTX reconstruction, in addition to the glycine density at the neurotransmitter binding pocket, an additional density was seen in the pore at the mid-level of M2 nestled between Gly2鈥?and Leu9鈥?(Fig.聽2d). This density was not observed in the other two conditions and likely corresponds to a PTX molecule. A recently solved structure of GABAAR also contains a PTX molecule at a similar location and was used to guide the orientation of the molecule37. The hydrophobic isoprenyl end of PTX is oriented toward the Leu9鈥?and the hydrophilic end closer to the Thr6鈥?(Fig.聽2e). Mutational studies have shown that perturbations at these positions affect PTX binding in anionic pLGICs38. It is noteworthy that mutations at the Thr6鈥?position can abolish PTX sensitivity in 伪1GlyR and enhance PTX sensitivity in PTX-insensitive 伪1尾 GlyR38. An inspection of the M2 helices upon global alignment of all three conformations show that the binding site for PTX does not physically overlap with the region undergoing constriction in the GlyR鈥揋ly (Pro-2鈥?. An alignment of GlyR鈥揋ly and GlyR鈥揋ly/PTX shows that the two structures differ primarily at the level of Pro-2鈥? the desensitization gate (Fig.聽2f, Supplementary Fig.聽8, green box). It is interesting to note that PTX does not induce additional conformational changes at the binding site (Supplementary Fig.聽8, gray box). Therefore, it appears that PTX exerts an allosteric effect on the intracellular gate Pro-2鈥?and prevents it from closing in the presence of glycine. In addition, we note positional differences in PTX in GlyR鈥揋ly/PTX and GABAAR (伪1尾3纬2) structures compared with GluCl (Supplementary Fig.聽9) where PTX interacts with Thr2鈥?instead. We think this variation may arise from the nature of the side chain at the 2鈥?position (Gly in homomeric 伪1GlyR, Val/Ala/Ser in 伪1尾3纬2 GABAAR, or a Thr in GluCl) or from slightly different mechanisms of PTX action as is also exemplified by use-dependent nature of PTX block in GABAAR and GluCl but not in the case of GlyR39,40.When compared to GlyR鈥揂po, the glycine-bound structures undergo an anti-clockwise twist of the ECD around the pore axis (when viewed from the extracellular end) (Supplementary Movie聽1). The ECD twist is accompanied by an inward motion of Loop C described above (Supplementary Fig.聽10). In addition, there is major repositioning of the interfacial loops, particularly the 尾1鈥撐? loop and Cys-loop (尾6鈥撐?) from the principal subunit and Loop F (尾8鈥撐?) on the complementary subunit (Fig.聽3a). The TMD helices are rotated clockwise and tilted outward in the GlyR鈥揋ly/PTX and GlyR鈥揋ly structures compared with GlyR鈥揂po, leading to an iris-like opening of the TMD (Supplementary Fig.聽10, Fig.聽3b). In addition to the M2 helices moving outward to open the pore, the M2鈥揗3 linker is also pulled away from the pore axis. This twisting motion is similar to the previously observed changes in GlyR and other pLGICs6,7,8. The described motions have a substantial effect on interactions between subunits at both the ECD and TMD interfaces as well as intra-subunit interactions both in the ECD and TMD (Fig.聽3c). In GlyR鈥揂po, the Cys-Loop, 尾1鈥撐? linker, and Loop F from the ECD are in close contact with the M2鈥揗3 linker, pre-M1, and post-M4. In addition, there are several interactions between M1, M3, and M4 helices between two adjacent subunits. In the glycine-bound conformations, these interactions are reduced due to the ECD rotation and the TMD expansion at the extracellular end (Fig.聽3c). The overall buried surface area (regions inaccessible to water) of the pentameric assembly in GlyR鈥揂po is 36,530鈥壝?sup>2 and reduces to 26,210鈥壝?sup>2 and 26,720鈥壝?sup>2 in GlyR鈥揋ly/PTX and GlyR鈥揋ly, respectively. The interaction network at the domain interfaces are crucial signal transduction elements and, unsurprisingly41, mutational perturbations at several of these positions (marked in green, Fig.聽3c) are reported to significantly alter function, and are associated with disease pathology including hyperekplexia2.Fig. 3: Global conformational changes and altered inter-domain interactions during GlyR gating.a A side-view of the ECD interface in GlyR鈥揂po and GlyR鈥揋ly. The principal subunits are colored (GlyR鈥揂po: salmon red and GlyR鈥揋ly: slate blue) while the complementary subunits are shown in shades of gray. b A side view of the TMD in GlyR鈥揂po and GlyR鈥揋ly conformations. c Interactions at the subunit- and domain-interfaces between ECD鈥揈CD, ECD鈥揟MD, and TMD鈥揟MD in GlyR鈥揂po (left) and GlyR鈥揋ly (right). The residues involved in the interactions are shown in sticks and interactions are highlighted by black dashes. Mutations at positions shown in green are associated with hyperekplexia2.Full size imageGlycine-induced conformational changes within the TMD leads to expansion of the TM helical bundle (Fig.聽4a). In GlyR鈥揂po, M4 helices are oriented closer to the rest of the TM helices within the same subunit. The post-M4 region extends three turns above M4, and extends above the putative membrane and is positioned close to the pre-M1 region and the 尾8鈥撐? strand of the ECD (Supplementary Fig.聽2B). These structural elements are implicated in relaying signals from the ECD to the TMD42. In GlyR鈥揋ly, these regions are placed farther away. At the intracellular end, the M4 and the pre-M4 helices (intracellular extension) are bent in the vicinity of Pro419 and positioned in close proximity to a lipid-like density (Supplementary Fig.聽6). In the GlyR鈥揋ly structure, the M4 and pre-M4 helices are further straightened (Fig.聽4b), and there is no obvious lipid-like density at this position. The movement of post-M3 helices is in the same direction as the pre-M4, and is potentially coupled by the unstructured loop that is currently not modeled in our structures. Among other related movements at the intracellular end is the movement of the M1鈥揗2 linker (Fig.聽4b). In the GlyR鈥揋ly structure, the M1鈥揗2 linker is moved away from the pore axis as a consequence of (or to facilitate) the pinching of M2 helices at Pro-2鈥? In agreement, both the length and the sequence of the M1鈥揗2 linker affect GlyR gating and desensitization, presumably by affecting the positioning of the selectivity filter region (Pro-2鈥? Ala-1鈥? and Arg0鈥?17,43. Another significant consequence of structural changes at the TMD and ICD are the appearance of intra-subunit and inter-subunit cavities at both the extracellular and intracellular ends of the TMD (Fig.聽4c). GlyRs are targets to numerous endogenous and exogenous ligands including lipids, neurosteroids, cannabinoids, alcohols, and anesthetics that bind within the TMD cavities. A change in the volume and polarity of these cavities implicates a role in state-dependent effects of the modulators on channel function44. Several residues lining these pockets (Fig.聽4c) are shown to modulate binding of these allosteric ligands and their accessibility is increased in glycine-bound conformation compared to the resting state45.Fig. 4: Conformational changes in the M4 helix and the effects on internal cavities.a An overlay of GlyR鈥揂po and GlyR鈥揋ly. The TMD viewed from the top reveals repositioning of TM helices. The intra-subunit and inter-subunit cavities are indicated by green and gray arrows, respectively. b Extent of conformational change in the ICD formed by the pre-M4 region, the M1鈥揗2 linker, and the post-M3 region. The direction of movement is highlighted by black arrows. c A comparison of the inter-subunit (gray dots) and intra-subunit (green dots) cavities in GlyR鈥揂po (top) and GlyR鈥揋ly (bottom) predicted using F pocket algorithm61. The cavities predicted on the surface were removed for clarity. The right and left panels are side and top views, respectively. The residues shown in sticks on the left panel are implicated in binding allosteric ligands45.Full size imageAllosteric modulation of GlyR gatingAmong the several types of allosteric modulators of GlyR that bind within the TMD cavities, we studied a well-described potentiator, ivermectin (IVM), which is known to enhance Gly sensitivity and increase open probability (Po) at lower glycine concentrations. There are several IVM-bound structures (both from cryo-EM and X-ray crystallography) for GlyR and GluCl that have provided initial views of the IVM-binding site8,33,46. We solved the structures of full-length GlyR in nanodisc in the presence of varying concentrations of glycine and IVM (1鈥塵M Gly/0.5鈥壩糓 IVM and 0.1鈥塵M Gly/30鈥壩糓 IVM) (Fig.聽5). The first condition led to two states (Gly-IVM-1 and Gly-IVM-2) and the second condition led to one state (Gly-IVM-3). These structures are in the 3.0鈥?.3鈥壝?resolution range with excellent map quality (Supplementary Figs.聽11 and 12; Supplementary Table聽4). Each of these three states reveal clear densities for both glycine and IVM (Supplementary Fig.聽13). The IVM molecule binds at the subunit interface between M3 (from the principal subunit) and M1 (from the complementary subunit) and makes polar contact with Arg295 in M2 (from the principal subunit). The binding pocket is lined by several hydrophobic residues (Ile249, Ile253, Pro254, and Leu257 in M1, Val304, Ala312, Leu315 in M3) with TM helices. The orientation of IVM is as previously observed in other GlyR鈥揑VM structures.Fig. 5: Analysis of GlyR鈥揋ly/IVM structures.a TEVC recording of 1鈥塵M glycine-induced currents in the presence and absence of 0.5鈥壩糓 IVM (left). Representative trace from multiple independent oocyte recordings (n鈥?鈥?). The traces are normalized to the peak current to show the current decay. Ion permeation pathway for GlyR鈥揋ly/IVM-1 GlyR鈥揋ly/IVM-2 (right). Green and purple spheres define radii of 1.8鈥?.3鈥壝?and 3.3鈥壝? respectively. The residues located at various pore constrictions are shown as sticks. b Current recording for 0.1鈥塵M glycine and 0.1鈥塵M glycine and 30鈥壩糓 IVM (left). Representative trace from multiple independent oocyte recordings (n鈥?鈥?). The pore profile for GlyR鈥揋ly/IVM-3 structure. c A view of M2 helices from the extracellular end Positions Leu9鈥?and Pro-2鈥?and the corresponding distances between C伪 in 脜. d Alignment of GlyR structures showing the conformational differences at the level of Pro-2鈥? Dotted arrows indicate the direction of rotation going from the GlyR鈥揂po (closed) to the GlyR鈥揋ly (desensitized) state.Full size imageThe overall conformation of the three structures are very similar to GlyR鈥揋ly. At the level of M2, the channel is open at Leu9鈥?and constricted at Pro-2鈥? broadly reminiscent of GlyR鈥揋ly (Fig.聽5). A closer inspection of the six GlyR structures reveals that the Pro-2鈥?side chains are positioned clockwise going from Apo, Gly/PTX, to Gly conformations. The Pro-2鈥瞫idechain in each of the three IVM structures lies between that of the Gly/PTX and Gly, suggesting that these conformations may represent intermediate states between the open and desensitized states (Fig.聽5d).We compared the full-length GlyR states with the previously solved GlyR cryo-EM conformations by aligning GlyR鈥揂po with StryR (strychnine-bound), GlyR鈥揋ly/PTX with Gly (glycine-bound聽open), and GlyR鈥揋ly and GlyR鈥揋ly/IVM-3 with IVM (glycine/ivermectin-bound)8. While the structures overlap well (Supplementary Fig.聽14), the most notable differences for each alignment pair are in the TMD and the domain interfaces. The buried surface area of StryR is 27,392鈥壝?sup>2 with greater numbers of inter-domain interactions compared with that of 36,530鈥壝?sup>2 in GlyR鈥揂po. The structural elements comprising the ICD adopt distinct conformation, namely the M1鈥揗2 linker and the post-M3 and pre-M4 helices. In addition, in comparison with the GlyR鈥揋ly/PTX, the glycine-bound open state is much wider, particularly at the intracellular end where the charge selectivity filter of the channel resides (Arg0鈥?. We believe that some of these differences may arise from truncation of the construct聽and the presence of a detergent environment. A comparison of the apo, open, and desensitized conformations of GlyRs with the corresponding states from other anionic pLGIC (Supplementary Fig.聽15), shows an overall conservation of the direction of agonist-induced movements in the ECD and TMD. However, the structures differ in the extent of repositioning of the loops and TM helices.Assessing the conductance state of GlyR structuresBy evaluating the hydrophobicity and radius of the transmembrane pore, the constrictions at Leu9鈥?and Pro-2鈥?were predicted to form energetic barriers to varying extents in each of the GlyR conformational states. To assess the behavior of water molecules and ions for the GlyR鈥揂po, GlyR鈥揋ly/PTX, and GlyR鈥揋ly conformations in a membrane environment, MD simulations were carried out upon embedding the GlyR structures within a 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine lipid bilayer (Fig.聽6). The picrotoxin molecule in the pore of GlyR鈥揋ly/PTX structure was removed prior to equilibration in the membrane. Positional restraints were placed on the protein backbone during these simulations, preserving the overall experimentally determined conformational state while permitting rotameric flexibility in amino acid side chains. Water molecules (using the TIP4P/2005 model) and ~150鈥塵M NaCl were included on either side of the bilayer. During simulations, the side-chain movements of the pore-lining residues caused fluctuations in the pore-radius, but there were no major changes to the pore profile (Fig.聽6a). An analysis of simulated water density along the pore axis suggests that in GlyR鈥揂po the pore is dewetted (i.e. devoid of water molecules) at Leu9鈥?(at ~0鈥壝?, conferring closure to water (with an energetic barrier of ~5鈥塳J鈥塵ol鈭?) (Fig.聽6b). This barrier disappears in both the GlyR鈥揋ly/PTX and GlyR鈥揋ly structures. However, while the pore is hydrated in GlyR鈥揋ly, a small energetic peak (~2.5鈥塳J鈥塵ol鈭?) appears at the Pro-2鈥?position (~鈭?0 脜), which corresponds to a maximum free energy value of ca. 1 RT (i.e. comparable to that of thermal fluctuations). While water permeation across the pore is essential for ion permeation, it is not necessarily sufficient. To assess ion conductance for the three states, in a second set of simulations, a transmembrane potential of +500鈥塵V (i.e. positive at the cytoplasmic side) was applied at 500鈥塵M NaCl concentration. The elevated concentration and voltage were used to increase the likelihood of observing permeant ions in microsecond simulation times (in separate control simulations we have checked the linear dependence of simulated conductance on NaCl concentration over the range 0鈥?00鈥塵M). As expected, no ion permeation events were observed for GlyR鈥揂po, even at this elevated potential and concentration. Interestingly, while water permeates through GlyR鈥揋ly, no chloride ions passed through the channel during 1鈥壩約 of total simulation time (Fig.聽6c). In contrast, 173 chloride ion permeation events (defined as ions passing from end to end of the TMD of the intact receptor used in the simulation) were observed over the course of 1鈥壩約 (= 5鈥壝椻€?00鈥塶s) of total simulation trajectory of the GlyR鈥揋ly/PTX structure (Supplementary Movie聽2). This corresponds to an estimated single-channel conductance of ~20鈥塸S (corrected to the value at 0.15鈥塎 NaCl) for chloride. The conductance is smaller than experimental measurements from single-channel experiments, which showed a conductance of 80鈥?8 pS47. We believe that some of the differences in the conductance estimate may be due to the missing residues in the ICD that may exert an effect on the single-channel conductance through electrostatic mechanisms15, but they may also reflect limitations of the computational methodology. It is noteworthy that sodium ions do not pass past the selectivity filter region comprising of Arg0鈥?~鈭?0鈥壝?, confirming the anionic selectivity for GlyR. Our computational assessment of the three GlyR conformations thus indicate that the GlyR鈥揋ly/PTX structure should be assigned to an open state, whereas the GlyR鈥揂po and GlyR鈥揋ly structures are nonconductive.Fig. 6: Molecular dynamics simulations of GlyR conformations.a Mean pore radius profiles and standard deviations averaged across three independent 30鈥塶s equilibrium simulations for GlyR鈥揂po (left), GlyR鈥揋ly/PTX (middle), and GlyR鈥揋ly (right) along the central pore axis. The final 20鈥塶s of each 30鈥塶s simulation trajectory was used to evaluate these profiles. The one-standard-deviation range between calculations (n鈥?鈥?, independent repeats of 30鈥塶s simulations) are shown as a gray band and the thick line is the mean between the triplicate 30鈥塶s simulations. Major constriction sites are indicated and the dotted line denotes the radius of hydrated chloride ion. The black traces are the pore radius profile calculated from the cryo-EM structures. b Corresponding mean water free energy profiles and standard deviations. Peaks in free energy profiles are highlighted. c Trajectories along the pore (z)-axis of water molecules and chloride ion coordinates within 5鈥壝?of the channel axis inside the pore, in the presence of a +500鈥塵V transmembrane potential difference (i.e., with the cytoplasmic side having a positive potential). One of five independent 200鈥塶s replicates is shown for each structure. During these and the preceding simulations, positional restraints were placed on the protein backbone, in order to preserve the experimental conformational state while permitting rotameric flexibility in amino acid side chains. The energetic barriers due to the ring of Leu9鈥?and Pro-2鈥?are at z ~0 and 鈭?0鈥壝? respectively.Full size imageSeveral previous simulation studies have noted the instability of the pLGIC conformations in the absence of protein backbone restraints12,13,14. While the fluctuations in the pore radii are understandably larger, the defining pore features of the three conformations were stably maintained, namely, the constrictions and energetic barriers to water at Leu9鈥?and Pro-2鈥?in GlyR鈥揂po and at Pro-2鈥?in GlyR鈥揋ly. In addition, unrestrained simulations (3鈥壝椻€?00鈥塶s) indicate that the GlyR鈥揋ly/PTX is stable, remaining open and presenting no free energy barrier to water. Overall, based on the pore-profile features of GlyR鈥揋ly/PTX and GlyR鈥揋ly which show that there are no additional effects of PTX binding besides changes at the P-2鈥?position, and the MD simulations, which confirm that the pore conformation of GlyR鈥揋ly/PTX has close to the expected selectivity and single-channel conductance, we believe that this structure may be close to the physiological open conformation.To access the conductance states of the IVM structures, we carried out MD simulations as described above. Under these conditions, no ion permeation event was observed at Pro-2鈥?on a 200鈥塶s timescale) suggesting that the IVM-bound structures are nonconducting (or at most have a reduced single channel conductance compared to the Gly/PTX) (Supplementary Fig.聽16). Interestingly, none of three structures reveals a sizable energetic barrier to water permeation at Pro-2鈥? which is in contrast to that of GlyR鈥揋ly. These findings further point toward the idea that IVM-bound structures may represent intermediate conformations, and subtle changes in Pro-2鈥?side-chain orientation underlie transitions between open and desensitized states.DiscussionHere, we present a set of structures that reveal the complete range of agonist-induced conformational changes in the full-length GlyR reconstituted in a lipid nanodisc environment. At the heart of pLGIC gating mechanisms is the signal transduction machinery that communicates agonist-binding events to the channel pore leading to channel opening. We show that glycine induces a global conformational change that encompasses the ECD, TMD, and the structured regions of the ICD. In the resting conformation, the GlyR has tightly coupled subunit and domain interfaces that weaken in response to glycine binding, leading to a relaxed open and desensitized conformations. MD simulations were used to assess the energy landscape of pore hydration and ion permeation, which allowed us to annotate these structures as resting, open, and desensitized conformations. The single-channel conductance and ion selectivity predicted for the open conformation are in agreement with the wealth of previous functional studies. Overall, these studies set the stage for a detailed characterization of the functional modulation of this clinically important class of channels.MethodsTwo-electrode voltage-clamp (TEVC) electrophysiologyThe gene encoding zebrafish GlyR伪1 (purchased from GenScript) was inserted into pTLN vector for expression in Xenopus laevis (Supplementary Table聽1). To linearize the DNA, the plasmid was incubated with Mlu1 restriction enzyme at 37鈥壜癈 overnight. The mRNA was synthesized from linearized DNA using the mMessage mMachine kit (Ambion) per instructions in the manufacturer鈥檚 manual. The RNA was then purified with RNAeasy kit (Qiagen). About 2鈥?0鈥塶g of mRNA was injected into X. laevis oocytes (stages V鈥揤I) and experiments were performed 2鈥? days after injection. For control experiments to verify that no endogenous currents were present, oocytes were injected with the same volume of water. Dr. W.F. Boron kindly provided oocytes used in this study. Female X. laevis were purchased from Nasco. We complied with all relevant ethical regulations for animal testing and research. Animal experimental procedures were approved by Institutional Animal Care and Use Committee (IACUC) of Case Western Reserve University. Oocytes were maintained at 18鈥壜癈 in OR3 medium (GIBCO-BRL Leibovitz medium containing glutamate, and 500 units each of penicillin and streptomycin, with pH adjusted to 7.5 and osmolarity to 197鈥塵Osm). TEVC experiments were performed on a Warner Instruments Oocyte Clamp OC-725. Currents were sampled and digitized at 500鈥塇z with a Digidata 1332A, and analyzed by Clampfit 10.2 (Molecular Devices). Oocytes were clamped at a holding potential of 鈭?0鈥塵V and solutions were changed using a syringe pump perfusion system flowing at a rate of 6鈥塵l/min. The electrophysiological solutions consisted of (in mM) 96 NaCl, 2 KCl, 1.8 CaCl2, 1 MgCl2, and 5 HEPES (pH 7.4, osmolarity adjusted to 195 mOsM). Chemical reagents (glycine and picrotoxin) were purchased from Sigma-Aldrich. Current traces were plotted using Origin Version b9.5.0.193.Full-length GlyR cloning and transfectionThe zebrafish GlyR伪1 shares 92% amino acid similarity with the human GlyR伪1. Codon-optimized zebrafish GlyR伪1 (NCBI Reference Sequence: NP_571477) was purchased from GenScript (Supplementary Table聽2). The sequence included the full-length GlyR伪1 gene (referred to in the text as GlyR) followed by a thrombin sequence (LVPRGS) and a C-terminal octa-His tag. The gene was subcloned into the pFastBac1 vector for expression in Spodoptera frugiperda (Sf9) cells. The Sf9 cells (purchased from Invitrogen) were cultured in Sf-900鈩?/sup> II SFM medium (Gibco庐) without antibiotics and incubated at 28鈥壜癈 without CO2 exchange. Transfection of sub-confluent cells was carried out with recombinant GlyR bacmid DNA using Cellfectin II transfection reagent (Invitrogen) according to manufacturer鈥檚 instructions. Cell culture supernatants were collected at 5 days post-transfection and centrifuged at 1000脳g for 5鈥塵in to remove cell debris to obtain progeny 1 (P1) recombinant baculovirus. Sf9 cells were infected with P1 virus stock to produce P2 virus, which were then used for subsequent infection to produce P3 viruses and so on till the P6 generation. The P6 virus was used for recombinant protein production based on expression levels and eventually quality of the purified sample.GlyR purification and nanodisc reconstitutionApproximately, 2鈥壝椻€?06 per ml Sf9 cells were infected with P6 recombinant viruses. After 48鈥塰 post infection, the cells were harvested and centrifuged at 8000g for 20鈥塵in at 4鈥壜癈 to separate the supernatant from the cell pellet. The cell pellet was resuspended in dilution buffer (20鈥塵M Tris-HCl, pH 7.5, 36.5鈥塵M sucrose) supplemented with 1% protease inhibitor cocktail (Sigma-Aldrich). Cells were disrupted by sonication on ice and nonlysed cells were removed by centrifugation at 3000脳g for 15鈥塵in. The supernatant was subjected to ultracentrifugation at 167,000鈥塯 for 1鈥塰 to separate the membrane fraction. The membrane pellet was solubilized with 15鈥塵M n-dodecyl-尾-d-maltopyranoside (DDM, Anatrace) in a buffer containing 200鈥塵M NaCl and 20鈥塵M HEPES, pH 8.0 (buffer A), supplemented with 0.05鈥塵g/ml soybean polar extract (asolectin, Avanti Polar Lipids) and 0.05% CHS (Avanti Polar Lipids) for 2鈥塰 at 4鈥壜癈. Nonsolubilized material was removed by ultracentrifugation (167,000脳g for 25鈥塵in). The supernatant containing the solubilized protein was incubated with TALON resin pre-equilibrated with buffer A, 1鈥塵M DDM, 0.05鈥塵g/ml asolectin, and 0.05% CHS for 2鈥塰 at 4鈥壜癈. The beads were then washed with 10 column volumes of buffer A, 1鈥塵M DDM, 0.05鈥塵g/ml asolectin, 0.05% CHS, and 35鈥塵M imidazole. The GlyR protein was eluted with buffer A, 1鈥塵M DDM, 0.05鈥塵g/ml asolectin, 0.05% CHS, and 250鈥塵M imidazole. Eluted protein was concentrated and applied to a Superose 6 column (GE healthcare) equilibrated with buffer A and 1鈥塵M DDM. Fractions containing the GlyR pentamers were collected and concentrated to 0.5鈥塵g/ml using 50-kDa MWCO Millipore filters (Amicon).The nanodisc reconstitution was carried out with the membrane scaffold protein (MSP1E3D1). MSP1E3D1 was expressed and purified with some modifications to the previously described protocols19. The MSP1E3D1 gene in pET 28a (a gift from Stephen Sligar: Addgene plasmid # 20066) was expressed in E. coli BL21(DE3) cells grown in Terrific broth medium supplemented with kanamycin (25鈥壩糶鈥塵L鈭?) and 0.2% glucose. The culture was grown at 37鈥壜癈 with shaking to an OD600 of ~1.0, and induced by with 1鈥塵M IPTG for 4鈥塰 at 37鈥壜癈. The cell pellet was lysed in a buffer containing 100鈥塵M NaCl, 20鈥塵M Tris base, pH 7.4 and supplemented with 1鈥塵M PMSF and complete EDTA-free protease inhibitor cocktail tablet (Roche). The lysate was centrifuged at 30,000脳g for for 30鈥塵in and the supernatant was bound to Ni-NTA, washed with four bed volumes of buffer (buffer B: 300鈥塵M NaCl, 40鈥塵M Tris, pH 8.0) and 1% Triton X-100, four bed volumes of buffer with 50鈥塵M sodium cholate, four bed volumes of buffer B, four bed volumes of buffer B containing 20鈥塵M imidazole, and eluted with buffer B containing 300鈥塵M imidazole. The eluted MSP1E3D1 was passed through a desalting column equilibrated with 100鈥塵M NaCl, 50鈥塵M Tris, 0.5鈥塵M EDTA, and pH 7.5, and the concentration was determined by absorbance at 280鈥塶m (extinction coefficient鈥?鈥?9,910鈥塎鈭?鈥塩m鈭?). The purity was assessed by sodium dodecyl sulfate polyacrylamide gel electrophoresis and size-exclusion chromatography.For reconstitution, the soybean polar lipid extract (Avanti Lipids) was dried in a stream of nitrogen and equilibrated in buffer A with protein, DDM, and MSP1E3D1 such that final molar ratio of Protein: MSP: lipid: DDM was 1:3:120:5. The mixture was incubated for 1鈥塰 with gentle rotation. After 1-h incubation, Bio-bead SM-2 (Bio-Rad Laboratories) were added and mixture was subjected to gentle rotation for 12鈥塰 at 4鈥壜癈. The reconstituted protein was applied to a Superose 6 column (GE healthcare) equilibrated with buffer A. Fractions containing protein nanodiscs were collected and concentrated to 0.1鈥塵g/ml using 50-kDa MWCO Millipore filters (Amicon) for cryo-EM studies.Preparation of sample cryo-EM imaging and data acquisitionGlyR nanodiscs samples from gel filtration (~0.1鈥塵g/ml) was filtered and used for cryo-EM. For the Apo condition, the sample was used as such without any ligand (GlyR鈥揂po). For GlyR鈥揋ly and GlyR鈥揋ly/PTX, the samples were incubated for 1鈥塰 with 5鈥塵M glycine and a combination of 5鈥塵M glycine and 3鈥塵M picrotoxin (PTX), respectively. For GlyR鈥揋ly/IVM the samples were incubated for 1鈥塰 with either 1鈥塵M glycine and 0.5鈥壩糓 Ivermectin (IVM) or 0.1鈥塵M glycine and 30鈥壩糓 IVM. The sample was blotted thrice with 3.5鈥壩糽 sample each time onto glow-discharged Cu 300 mesh Quantifoil 1.2/1.3 grids (Quantifoil Micro Tools) and the grids were plunge frozen into liquid ethane using a Vitrobot (FEI). The grids were imaged using a 300鈥塳V FEI Titan Krios microscope equipped with a Gatan K2-Summit direct electron detector camera (GlyR鈥揂po and GlyR鈥揋ly/PTX datasets) or Gatan K3 direct electron detector camera for GlyR鈥揋ly and GlyR鈥揋ly鈥揑VM datasets. The parameters for data acquisition for the different conditions are as follows. GlyR鈥揂po: Movies containing 40 frames were collected at 130,000脳 magnification (set on microscope) in super-resolution mode with a physical pixel size of 1.064鈥壝?pixel, dose per frame 1.30鈥塭鈭?/sup>/脜2. Defocus values of the images ranged from 鈭?.5 to 鈭?.5鈥壜祄 (input range setting for data collection) as per the automated imaging software Latitude (Gatan). GlyR鈥揋ly/PTX: 40 frames/movie were collected with EPU (ThermoFisher Scientific) at 81,000脳 magnification (set on microscope) in counting mode with a physical pixel size of 1.06鈥壝?pixel, dose per frame 1.45鈥塭鈭?/sup>/脜2. Defocus values ranged from 鈭?.5 to 鈭?.5. GlyR鈥揋ly: 40 frames/movie were collected using Latitude at 81,000脳 magnification (set on microscope) in super-resolution mode with a physical pixel size of 1.08鈥壝?pixel, dose per frame 1.25鈥塭鈭?/sup>/脜2. Defocus values ranged from 鈭?.5 to 鈭?.5鈥壜祄. GlyR鈥揋ly鈥揑VM: For the dataset with 1鈥塵M glycine and 0.5鈥壩糓 Ivermectin, movies containing 40 frames were collected at 81,000脳 magnification (set on microscope) in super-resolution mode with a physical pixel size of 1.08鈥壝?pixel, dose per frame 1.25鈥塭鈭?/sup>/脜2. Defocus values of the images ranged from 鈭?.5 to 鈭?.5鈥壜祄 (input range setting for data collection) as per the automated imaging software EPU. For the 0.1鈥塵M glycine and 30鈥壩糓 Ivermectin, movies containing 40 frames were collected with Latitude at 105,000脳 magnification (set on microscope) in super-resolution mode with a physical pixel size of 0.84鈥壝?pixel, dose per frame 1.25鈥塭鈭?/sup>/脜2. Defocus values of the images ranged from 鈭?.5 to 鈭?.25鈥壜祄 (input range setting for data collection).Image processingMotionCor248 (v 1.2.3) with a B-factor of 300 pixels2 was used to correct beam-induced motion. Super-resolution images (for GlyR鈥揂po and GlyR鈥揋ly datasets) were binned (2鈥壝椻€?) in Fourier space, making a final pixel size of 1.064 and 1.08鈥壝? respectively. Initially, a small subset of micrographs were CTF-corrected using CTFFIND449 and processed in cisTEM50 to generate 2D-templates for autopicking. All subsequent data processing was conducted in RELION 3.051. The defocus values of the motion-corrected micrographs were estimated using Gctf software52.For GlyR鈥揂po dataset: 2, 042 micrographs of a total of 2, 389 were manually sorted. Totally, 444, 921 particles were auto-picked from 2, 042 micrographs using templates generated in cisTEM and were subjected to 2D classification to remove suboptimal particles. Totally, 61,000 good particles were selected and subjected to initial round of 3D auto-refinement using 30鈥壝?low-pass filtered map of related human synaptic 伪1尾3纬2 GABAAR (EMD-4411) from which Megabody38 was manually erased in chimera9. Multiple rounds of 2D classification and 3D classification without image alignment were done to remove low resolution and broken particles. All the particles corresponding to bottom views of the receptor were removed from subsequent processing to prevent misalignment of the particles in 3D auto-refinement. The nanodisc belt was subtracted to improve the angular accuracy of alignment and remove sub-optimal particles. A set of 22, 643 particles resulted in a 3.7鈥壝?map, which was used for training parameters for Bayesian polishing51,53. Optimized parameters (s_vel 1.029 --s_div 9585 --s_acc 2.805) were then used to polish these particles and the polished particles were then subjected to multiple rounds of 3D auto-refinement and 3D classification without image alignment. Multiple rounds of CTF refinement were performed before the final round of 3D auto-refinement. A final subset 19, 653 particles used for auto-refinement, with soft mask accounting for density of nanodisc, resulted in 3.39鈥壝?map which was post-processed resulting in final resolution of 3.33鈥壝?(Fourier shell coefficient (FSC)鈥?鈥?.143 criterion). The B-factor estimation and map sharpening were performed in the post-processing step. Local resolutions were estimated using the RESMAP software54.For GlyR鈥揋ly dataset, a total of 5344 movies were used. For autopicking in RELION 3.0, the final 2D class from the GlyR鈥揂po dataset was used as template and a total of 466,174 were auto-picked. Multiple rounds of 2D classification, 3D classification without image alignment was done for datasets. All the particles corresponding to the bottom views of receptor were removed from subsequent processing to prevent misalignment of the particles in 3D auto-refinement. A total of 15,000 particles were subjected to independent 3D auto refinement using GlyR鈥揂po map low-pass filtered to 20鈥壝?as a template. The final ~10,000 particles were used for training parameters for Bayesian polishing. Multiple rounds of CTF refinement and classification were performed before the final round of 3D autorefinement. This resulted in a final map containing 8255 particles at 3.47鈥壝?after post-processing. The B-factor estimation and map sharpening were performed in the post-processing step. Local resolutions were estimated using the RESMAP software54.For GlyR鈥揋ly/PTX: a total of 2, 280 movies, collected on counting mode, were used from two imaging sessions. The grids for these imaging sessions were from the same batch, collected on the same microscope using identical parameters. For auto-picking in RELION3.0, the final 2D classes from the GlyR鈥揂po dataset was used as template and a total of 440,516 particles were picked. Multiple rounds of 2D classification and 3D classification without image alignment was used to get rid of low resolution particles. All the particles corresponding to the bottom views of receptor were removed from subsequent processing to prevent misalignment of the particles in 3D auto-refinement. A total of 8, 288 particles from dataset 1 (1, 470 movies) and 11, 331 particles from dataset 2 (810 movies) were subjected to independent 3D autorefinement using GlyR鈥揂po map low pass filtered to 20鈥壝?as a template. The two datasets were individually used for Bayesian polishing and the final polished datasets were merged in RELION 3.0.6. Few rounds of CTF refinement was performed before the final round of 3D autorefinement. The region corresponding to unstructured ICD was subtracted using RELION 3.0.6 and subtracted particles were further classified. The merged dataset with 12,503 particles was used for 3D autorefinement and further classification and resulted in a final map containing 10, 375 particles at 3.51鈥壝?after post-processing. The B-factor estimation and map sharpening were performed in the post-processing step. Local resolutions were estimated using the RESMAP software54.For GlyR鈥揋ly鈥揑VM dataset with 1鈥塵M glycine and 0.5鈥壩糓 IVM, a total of 9700 movies were used. For autopicking in RELION 3.1, the final 2D class from the GlyR鈥揂po dataset was used as a template and a total of 860, 560 particles were auto-picked. A total of 45,000 particles were subjected to independent 3D auto refinement using GlyR鈥揂po map low pass filtered to 20鈥壝?as a template. Multiple rounds of 2D classification, 3D classification without image alignment was done for datasets. All the particles corresponding to bottom views of the receptor were removed from subsequent processing to prevent misalignment of the particles in 3D autorefinement. The final ~35,000 particles were used for training parameters for Bayesian polishing. Few rounds of CTF refinement and 3D-classification were performed before the final round of 3D auto-refinement. This resulted in two distinct classes with 19,600 and 15,035 particles which resulted in 3.14 and 3.34鈥壝?final maps for GlyR鈥揋ly/IVM-1 and GlyR鈥揋ly/IVM-2, respectively after post-processing. The B-factor estimation and map sharpening were performed in the post-processing step.For GlyR鈥揋ly鈥揑VM dataset with 0.1鈥塵M glycine and 30鈥壩糓 Ivermectin, a total of 2,193 movies were used. For autopicking in RELION 3.1beta, the final 2D classes from the GlyR鈥揂po dataset was used as template and a total of 288,811 were auto-picked. A total of 45,000 particles were subjected to independent 3D auto refinement using GlyR鈥揂po map low pass filtered to 20鈥壝?as a template. All the particles corresponding to the bottom views of receptor were removed from subsequent processing to prevent misalignment of the particles in 3D auto-refinement. The final ~30,000 particles were used for training parameters for Bayesian polishing. Few rounds of CTF refinement and 3D-classification were performed before the final round of 3D autorefinement. A final set of 27,516 particles was used for autorefinement which resulted in a 3.01鈥壝?final map after post-processing. The B-factor estimation and map sharpening were performed in the post-processing step. Local resolutions were estimated using the RESMAP software55.GlyR model buildingThe 3D cryo-EM maps for the GlyR鈥揂po, GlyR鈥揋ly, GlyR鈥揋ly/PTX, GlyR鈥揋ly/IVM datasets used for model-building contained density for the entire ECD, TMD, and a small region of the ICD. The final refined model of GlyR鈥揂po comprised of residues Pro31鈥揚he341 and Lys394鈥揋ln444. The missing region (342鈥?93) is of the unstructured ICD. The previously solved structure of GlyR鈥搒trychnine (PDB ID: 3JAD) was used as a starting model for the GlyR鈥揂po data set. The residues were renumbered as per the sequence available in UniProt database (O93430). For model building of GlyR鈥揋ly, GlyR鈥揋ly/PTX, GlyR鈥揋ly鈥揑VM data sets, the GlyR鈥揂po structure was used as a template. The M3 and M4 helices were truncated based on the quality of density in the respective maps for GlyR鈥揋ly and GlyR鈥揋ly鈥揚TX. The cryo-EM map was first converted to the.mtz format using CCP4i software56 with mapmask and sfall tools and then used for manual model building in COOT57. After initial model building, the three states were refined against their corresponding EM-derived maps using the phenix.real_space_refinement tool from the PHENIX software package58, using rigid body, local grid, NCS, and gradient minimization. The individual models were then subjected to additional rounds of manual model fitting and refinement. The refinement statistics, the final model to map cross-correlation evaluated using phenix module mtriage59, and the stereochemical properties of the models as evaluated by Molprobity55 are detailed in the Supplementary Tables聽3 and 4. Protein surface area and interfaces were analyzed using the PDBePISA server (http://www.ebi.ac.uk/pdbe/pisa/). Buried surface area refers to the solvent-accessible surface area of monomeric units that gets buried upon assembly formation, in this case a pentamer. PDBePISA server is used to calculate buried surface area (or water inaccessible regions 鈥?to a 1.4鈥壝?diameter water probe) as a way to assess changes in interaction interfaces. The pore profile was calculated using the HOLE program60. The cavities were analyzed using Fpocket algorithm61. Figures were prepared using PyMOL v.2.0.4 (Schr枚dinger, LLC), CorelDraw v.20.1.0.708).MD simulationsMolecular structures of the GlyR in various conformational states were separately embedded within phospholipid (POPC, 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine) bilayer membranes that were solvated on either side to form 13鈥壝椻€?3鈥壝椻€?7鈥塶m3 simulation cells. Each protein-membrane system was assembled and equilibrated via a multiscale protocol62. Simulations were performed with GROMACS 201863, using the TIP4P/2005 water model64 and the OPLS all-atom protein force field with united-atom lipids65. The integration time-step was 2鈥塮s. Bonds were constrained through the LINCS algorithm66. A Verlet cut-off scheme was applied, and long-range electrostatic interactions were calculated using the Particle Mesh Ewald method67. Temperature and pressure were maintained at 37鈥壜癈 and 1鈥塨ar during simulations, using the velocity-rescaling thermostat68 in combination with a semi-isotropic Parrinello and Rahman barostat69, with coupling constants of 蟿T鈥?鈥?.1鈥塸s and 蟿P鈥?鈥?鈥塸s, respectively.Pore water free energy profiles were computed for alternative conformations of the protein using the Channel Annotation Package32, in each case based on three replicates of 30鈥塶s equilibrium simulations at physiological salt (150鈥塵M NaCl) concentration. To preserve the conformational state of each cryo-EM structure (while permitting rotameric flexibility in amino acid side chains) during simulations, harmonic restraints at a force constant of 1000鈥塳J鈥塵ol鈭?鈥塶m鈭? were placed on protein backbone atoms. Simulation trajectories were analyzed at 500鈥塸s intervals, with a bandwidth of 0.14鈥塶m applied for water density estimation.Chloride conduction was measured in five replicates of 200鈥塶s simulations for each (backbone-restrained) receptor structure, at 500鈥塵M NaCl concentration and in the presence of a +500鈥塵V transmembrane potential difference, with positve potential on the cytoplasmic side. This was applied by imposing an external, uniform electric field in the membrane normal direction. Conductance values were calculated from the number of permeation events and averaged between replicates.A separate set of equilibrium simulations (at 150鈥塵M NaCl, in the absence of a transmembrane potential difference) were used to monitor the behavior of each protein conformation by sequential relaxation followed by removal of positional restraints. During an initial 10-ns simulation, harmonic restraints (again with force constant 1000鈥塳J鈥塵ol鈭?鈥塶m鈭?) were placed on all non-hydrogen atoms of the protein. Restraints on side chain atoms were then released, and the protein subjected to a further 20鈥塶s of simulation. In the next 20鈥塶s, only the 伪-carbon atom of each residue remained under the positional restraining force. The final state of the system was subsequently simulated for 200鈥塶s, unrestrained.Reporting summaryFurther information on research design is available in the聽Nature Research Reporting Summary linked to this article. Data supporting the findings of this paper are available from the corresponding authors upon reasonable request. A reporting summary for this article is available as a Supplementary Information file. The atomic coordinates and cryo-EM maps for the GlyR鈥揂po, GlyR鈥揋ly, and GlyR鈥揋ly/PTX structures have been deposited in the Protein Data Bank and Electron Microscopy Data Bank with accession codes PDB-6UBS (EMD-20714), PDB-6UBT (EMD-20715), and PDB-6UD3 (EMD-20731), respectively. The atomic coordinates and cryo-EM maps for the GlyR鈥揋ly-IVM-1, GlyR鈥揋ly-IVM-2, and GlyR鈥揋ly-IVM-3 structures have been deposited in the Protein Data Bank and Electron Microscopy Data Bank with accession codes PDB-6VM0 (EMD-21234), PDB-6VM2 (EMD-21236), and PDB-6VM3 (EMD-21237), respectively. PDB 6UBS [https://doi.org/10.2210/pdb6UBS/pdb] PDB 6UBT [https://doi.org/10.2210/pdb6UBT/pdb] PDB 6UD3 [https://doi.org/10.2210/pdb6UD3/pdb] PDB 6VM0 [https://doi.org/10.2210/pdb6VM0/pdb] PDB 6VM2 [https://doi.org/10.2210/pdb6VM2/pdb] PDB 6VM3 [https://doi.org/10.2210/pdb6VM3/pdb] References1.Harvey, R. J. et al. GlyR alpha3: an essential target for spinal PGE2-mediated inflammatory pain sensitization. Science 304, 884鈥?87 (2004).ADS聽 CAS聽 PubMed聽Google Scholar聽 2.Bode, A. Lynch, J. W. The impact of human hyperekplexia mutations on glycine receptor structure and function. Mol. Brain 7, 2 (2014).PubMed聽 PubMed Central聽Google Scholar聽 3.Zeilhofer, H. U., Acuna, M. A., Gingras, J. Yevenes, G. E. Glycine receptors and glycine transporters: targets for novel analgesics? Cell Mol. Life Sci. 75, 447鈥?65 (2018).CAS聽 PubMed聽Google Scholar聽 4.Langlhofer, G. Villmann, C. The intracellular loop of the glycine receptor: it鈥檚 not all about the size. Front. Mol. Neurosci. 9, 41 (2016).PubMed聽 PubMed Central聽Google Scholar聽 5.Morales-Perez, C. L., Noviello, C. M. Hibbs, R. E. X-ray structure of the human alpha4beta2 nicotinic receptor. Nature 538, 411鈥?15 (2016).ADS聽 CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 6.Polovinkin, L. et al. Conformational transitions of the serotonin 5-HT3 receptor. Nature 563, 275鈥?79 (2018).ADS聽 CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 7.Basak, S., Gicheru, Y., Rao, S., Sansom, M. S. P. Chakrapani, S. Cryo-EM reveals two distinct serotonin-bound conformations of full-length 5-HT3A receptor. Nature 563, 270鈥?74 (2018).ADS聽 CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 8.Du, J., Lu, W., Wu, S., Cheng, Y. Gouaux, E. Glycine receptor mechanism elucidated by electron cryo-microscopy. Nature https://doi.org/10.1038/nature14853 (2015).Article聽 PubMed聽 PubMed Central聽Google Scholar聽 9.Laverty, D. et al. Cryo-EM structure of the human alpha1beta3gamma2 GABAA receptor in a lipid bilayer. Nature 565, 516鈥?20 (2019).ADS聽 CAS聽 PubMed聽Google Scholar聽 10.Zhu, S. et al. Structure of a human synaptic GABAA receptor. Nature 559, 67鈥?2 (2018).ADS聽 CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 11.Huang, X., Chen, H., Michelsen, K., Schneider, S. Shaffer, P. L. Crystal structure of human glycine receptor-alpha3 bound to antagonist strychnine. Nature 526, 277鈥?80 (2015).ADS聽 CAS聽 PubMed聽Google Scholar聽 12.Gonzalez-Gutierrez, G., Wang, Y., Cymes, G. D., Tajkhorshid, E. Grosman, C. Chasing the open-state structure of pentameric ligand-gated ion channels. J. Gen. Physiol. 149, 1119鈥?138 (2017).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 13.Cerdan, A. H., Martin, N. E. Cecchini, M. An ion-permeable state of the glycine receptor captured by molecular dynamics. Structure 26, 1555鈥?562 e1554 (2018).CAS聽 PubMed聽Google Scholar聽 14.Damgen, M. A. Biggin, P. C. A refined open state of the glycine receptor obtained via molecular dynamics simulations. Structure 28, 130鈥?39 e132 (2020).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 15.Carland, J. E. et al. Characterization of the effects of charged residues in the intracellular loop on ion permeation in alpha1 glycine receptor channels. J. Biol. Chem. 284, 2023鈥?030 (2009).CAS聽 PubMed聽Google Scholar聽 16.Song, Y. M. Huang, L. Y. Modulation of glycine receptor chloride channels by cAMP-dependent protein kinase in spinal trigeminal neurons. Nature 348, 242鈥?45 (1990).ADS聽 CAS聽 PubMed聽Google Scholar聽 17.Papke, D. Grosman, C. The role of intracellular linkers in gating and desensitization of human pentameric ligand-gated ion channels. J. Neurosci. 34, 7238鈥?252 (2014).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 18.Ivica, J. et al. The intracellular domain of homomeric glycine receptors modulates agonist efficacy. J. Biol. Chem. https://doi.org/10.1074/jbc.RA119.012358 (2020).Article聽 PubMed聽Google Scholar聽 19.Ritchie, T. K. et al. Chapter 11鈥攔econstitution of membrane proteins in phospholipid bilayer nanodiscs. Methods Enzymol. 464, 211鈥?31 (2009).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 20.Low, S. E., Ito, D. Hirata, H. Characterization of the zebrafish glycine receptor family reveals insights into glycine receptor structure function and stoichiometry. Front. Mol. Neurosci. 11, 286 (2018).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 21.Lynch, J. W., Rajendra, S., Barry, P. H. Schofield, P. R. Mutations affecting the glycine receptor agonist transduction mechanism convert the competitive antagonist, picrotoxin, into an allosteric potentiator. J. Biol. Chem. 270, 13799鈥?3806 (1995).CAS聽 PubMed聽Google Scholar聽 22.Pribilla, I., Takagi, T., Langosch, D., Bormann, J. Betz, H. The atypical M2 segment of the beta subunit confers picrotoxinin resistance to inhibitory glycine receptor channels. EMBO J. 11, 4305鈥?311 (1992).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 23.Gielen, M., Thomas, P. Smart, T. G. The desensitization gate of inhibitory Cys-loop receptors. Nat. Commun. 6, 6829 (2015).ADS聽 CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 24.Griffon, N. et al. Molecular determinants of glycine receptor subunit assembly. EMBO J. 18, 4711鈥?721 (1999).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 25.Unwin, N. Refined structure of the nicotinic acetylcholine receptor at 4A resolution. J. Mol. Biol. 346, 967鈥?89 (2005).CAS聽 PubMed聽Google Scholar聽 26.Mowrey, D. D. et al. Open-channel structures of the human glycine receptor alpha1 full-length transmembrane domain. Structure 21, 1897鈥?904 (2013).CAS聽 PubMed聽Google Scholar聽 27.Keramidas, A., Moorhouse, A. J., French, C. R., Schofield, P. R. Barry, P. H. M2 pore mutations convert the glycine receptor channel from being anion- to cation-selective. Biophys. J. 79, 247鈥?59 (2000).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 28.Saul, B. et al. Novel GLRA1 missense mutation (P250T) in dominant hyperekplexia defines an intracellular determinant of glycine receptor channel gating. J. Neurosci. 19, 869鈥?77 (1999).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 29.Cymes, G. D. Grosman, C. Tunable pKa values and the basis of opposite charge selectivities in nicotinic-type receptors. Nature 474, 526鈥?30 (2011).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 30.Aryal, P., Sansom, M. S. Tucker, S. J. Hydrophobic gating in ion channels. J. Mol. Biol. 427, 121鈥?30 (2015).CAS聽 PubMed聽Google Scholar聽 31.Rao, S., Klesse, G., Stansfeld, P. J., Tucker, S. J. Sansom, M. S. P. A heuristic derived from analysis of the ion channel structural proteome permits the rapid identification of hydrophobic gates. Proc. Natl Acad. Sci. USA 116, 13989鈥?3995 (2019).CAS聽 PubMed聽Google Scholar聽 32.Klesse, G., Rao, S., Sansom, M. S. P. Tucker, S. J. CHAP: a versatile tool for the structural and functional annotation of ion channel pores. J. Mol. Biol. 431, 3353鈥?365 (2019).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 33.Huang, X., Chen, H. Shaffer, P. L. Crystal structures of human GlyRalpha3 bound to ivermectin. Structure 25, 945鈥?50 e942 (2017).CAS聽 PubMed聽Google Scholar聽 34.Vandenberg, R. J., Handford, C. A. Schofield, P. R. Distinct agonist- and antagonist-binding sites on the glycine receptor. Neuron 9, 491鈥?96 (1992).CAS聽 PubMed聽Google Scholar聽 35.Hansen, S. B. et al. Structures of aplysia AChBP complexes with nicotinic agonists and antagonists reveal distinctive binding interfaces and conformations. Embo J. 24, 3635鈥?646 (2005).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 36.Althoff, T., Hibbs, R. E., Banerjee, S. Gouaux, E. X-ray structures of GluCl in apo states reveal a gating mechanism of Cys-loop receptors. Nature 512, 333鈥?37 (2014).ADS聽 CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 37.Masiulis, S. et al. GABAA receptor signalling mechanisms revealed by structural pharmacology. Nature https://doi.org/10.1038/s41586-018-0832-5 (2019).Article聽 PubMed聽 PubMed Central聽Google Scholar聽 38.Shan, Q., Haddrill, J. L. Lynch, J. W. A single beta subunit M2 domain residue controls the picrotoxin sensitivity of alphabeta heteromeric glycine receptor chloride channels. J. Neurochem. 76, 1109鈥?120 (2001).CAS聽 PubMed聽Google Scholar聽 39.Xu, M., Covey, D. F. Akabas, M. H. Interaction of picrotoxin with GABAA receptor channel-lining residues probed in cysteine mutants. Biophys. J. 69, 1858鈥?867 (1995).ADS聽 CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 40.Etter, A. et al. Picrotoxin blockade of invertebrate glutamate-gated chloride channels: subunit dependence and evidence for binding within the pore. J. Neurochem. 72, 318鈥?26 (1999).CAS聽 PubMed聽Google Scholar聽 41.Grutter, T. et al. Molecular tuning of fast gating in pentameric ligand-gated ion channels. Proc. Natl Acad. Sci. USA 102, 18207鈥?8212 (2005).ADS聽 CAS聽 PubMed聽Google Scholar聽 42.daCosta, C. J., Dey, L., Therien, J. P. Baenziger, J. E. A distinct mechanism for activating uncoupled nicotinic acetylcholine receptors. Nat. Chem. Biol. 9, 701鈥?07 (2013).CAS聽 PubMed聽Google Scholar聽 43.Lynch, J. W. et al. Identification of intracellular and extracellular domains mediating signal transduction in the inhibitory glycine receptor chloride channel. EMBO J. 16, 110鈥?20 (1997).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 44.Mihic, S. J. et al. Sites of alcohol and volatile anaesthetic action on GABA(A) and glycine receptors. Nature 389, 385鈥?89 (1997).ADS聽 CAS聽 PubMed聽Google Scholar聽 45.Lobo, I. A., Mascia, M. P., Trudell, J. R. Harris, R. A. Channel gating of the glycine receptor changes accessibility to residues implicated in receptor potentiation by alcohols and anesthetics. J. Biol. Chem. 279, 33919鈥?3927 (2004).CAS聽 PubMed聽Google Scholar聽 46.Hibbs, R. E. Gouaux, E. Principles of activation and permeation in an anion-selective Cys-loop receptor. Nature 474, 54鈥?0 (2011).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 47.Legendre, P. Pharmacological evidence for two types of postsynaptic glycinergic receptors on the Mauthner cell of 52-h-old zebrafish larvae. J. Neurophysiol. 77, 2400鈥?415 (1997).CAS聽 PubMed聽Google Scholar聽 48.Zheng, S. Q. et al. MotionCor2: anisotropic correction of beam-induced motion for improved cryo-electron microscopy. Nat. Methods 14, 331鈥?32 (2017).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 49.Rohou, A. Grigorieff, N. CTFFIND4: fast and accurate defocus estimation from electron micrographs. J. Struct. Biol. 192, 216鈥?21 (2015).PubMed聽 PubMed Central聽Google Scholar聽 50.Grant, T., Rohou, A. Grigorieff, N. cisTEM, user-friendly software for single-particle image processing. Elife https://doi.org/10.7554/eLife.35383 (2018).51.Fernandez-Leiro, R. Scheres, S. H. W. A pipeline approach to single-particle processing in RELION. Acta Crystallogr D Struct. Biol. 73, 496鈥?02 (2017).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 52.Mindell, J. A. Grigorieff, N. Accurate determination of local defocus and specimen tilt in electron microscopy. J. Struct. Biol. 142, 334鈥?47 (2003).PubMed聽Google Scholar聽 53.Zivanov, J., Nakane, T. Scheres, S. H. W. A Bayesian approach to beam-induced motion correction in cryo-EM single-particle analysis. IUCrJ 6, 5鈥?7 (2019).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 54.Kucukelbir, A., Sigworth, F. J. Tagare, H. D. Quantifying the local resolution of cryo-EM density maps. Nat. Methods 11, 63鈥?5 (2014).CAS聽 PubMed聽Google Scholar聽 55.Chen, V. B. MolProbity: all-atom structure validation for macromolecular crystallography. Acta Crystallogr. D 66, 12鈥?1 (2010).CAS聽Google Scholar聽 56.Collaborative Computational Project, N. The CCP4 suite: programs for protein crystallography. Acta Crystallogr. Sect. D Biol. Crystallogr. 50, 760鈥?63 (1994). Google Scholar聽 57.Emsley, P. Cowtan, K. Coot: model-building tools for molecular graphics. Acta Crystallogr. Sect. D Biol. Crystallogr. 60, 2126鈥?132 (2004). Google Scholar聽 58.Adams, P. D. et al. PHENIX: building new software for automated crystallographic structure determination. Acta Crystallogr. Sect. D Biol. Crystallogr. 58, 1948鈥?954 (2002).MathSciNet聽Google Scholar聽 59.Afonine, P. V. et al. New tools for the analysis and validation of cryo-EM maps and atomic models. Acta Crystallogr D Struct. Biol. 74, 814鈥?40 (2018).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 60.Smart, O. S., Neduvelil, J. G., Wang, X., Wallace, B. A. Sansom, M. S. HOLE: a program for the analysis of the pore dimensions of ion channel structural models. J. Mol. Graph 14, 354鈥?60, 376 (1996).CAS聽 PubMed聽Google Scholar聽 61.Le Guilloux, V., Schmidtke, P. Tuffery, P. Fpocket: an open source platform for ligand pocket detection. BMC Bioinform. 10, 168 (2009). Google Scholar聽 62.Stansfeld, P. J. et al. MemProtMD: automated Insertion of membrane protein structures into explicit lipid membranes. Structure 23, 1350鈥?361 (2015).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 63.Abraham, M. J., Murtola, T., Schulz, R., Pall, S. Jeremy, C. GROMACS: high performance molecular simulations through multi-level parallelism from laptops to supercomputers. SoftwareX 1, 19鈥?5 (2015).ADS聽Google Scholar聽 64.Abascal, J. L. Vega, C. A general purpose model for the condensed phases of water: TIP4P/2005. J. Chem. Phys. 123, 234505 (2005).ADS聽 CAS聽 PubMed聽Google Scholar聽 65.Jorgensen, W. L., Maxwell, D. S. Tirado-Rives, J. Development and testing of the OPLS all-atom force field on conformational energetics and properties of organic liquids. J. Am. Chem. Soc. 118, 11225鈥?1236 (1996).CAS聽Google Scholar聽 66.Hess, B., Bekker, H., Berendsen, H. J. C. Fraaije, J. G. E. M. LINCS: a linear constraint solver for molecular simulations. J. Comput. Chem. 18, 1463鈥?472 (1997).CAS聽Google Scholar聽 67.Darden, T., York, D. Pedersen, L. Particle mesh Ewald: an N路log(N) method for Ewald sums in large systems. J. Chem. Phys. 98, 10089鈥?0092 (1993).ADS聽 CAS聽Google Scholar聽 68.Bussi, G., Donadio, D. Parrinello, M. Canonical sampling through velocity rescaling. J. Chem. Phys. 126, 14101 (2007). Google Scholar聽 69.Parrinello, M. Rahman, A. Polymorphic transitions in single crystals: a new molecular dynamics method. J. Appl. Phys. 52, 7182鈥?190 (1981).ADS聽 CAS聽Google Scholar聽 70.Wallace, A. C., Laskowski, R. A. Thornton, J. M. LIGPLOT: a program to generate schematic diagrams of protein-ligand interactions. Protein Eng. 8, 127鈥?34 (1995).CAS聽 PubMed聽Google Scholar聽 Download referencesAcknowledgementsWe acknowledge the use of instruments at the National Cryo-Electron Microscopy Facility at the NCI and Stanford-SLAC Cryo-Electron Microscopy Facility. A special thanks to Prof. Wah Chiu for providing us imaging time and supervising the data collection (supported by National Institutes of Health grants: P41GM103832 and S10OD021600). We are grateful to the Cryo-Electron Microscopy Core at the CWRU School of Medicine and Dr. Kunpeng Li for the access to the sample preparation and Cryo-EM instrumentation. We thank Dr. Walter F. Boron for kindly providing us Xenopus oocytes and for unrestricted access of the oocyte rig. We are deeply appreciative of the support provided by Dr. Fraser Moss and Mr. Brian Zeise with the oocyte rig. We are very grateful to the members of the Chakrapani lab for critical reading and comments on the manuscript. This research was, in part, supported by the National Cancer Institute鈥檚 National Cryo-EM Facility at the Frederick National Laboratory for Cancer Research. This work was supported by the National Institutes of Health grants R01GM108921, R01GM131216, R35GM134896, and Cryo-EM supplements: 3R01GM108921-03S1, R01GM108921-5S1, 3R01GM131216-1S1 to S.C. and the AHA postdoctoral Fellowship to A.K. (20POST35210394) and S.B. (17POST33671152).Author informationAffiliationsDepartment of Physiology and Biophysics, Case Western Reserve University, Cleveland, OH, 44106-4970, USAArvind Kumar,聽Sandip Basak,聽Yvonne Gicheru聽 聽Sudha ChakrapaniDepartment of Biochemistry, University of Oxford, Oxford, OX1 3QU, UKShanlin Rao聽 聽Mark S. P. SansomDivision of CryoEM and Bioimaging, SSRL, SLAC National Accelerator Laboratory, Stanford University, Menlo Park, CA, 94025, USAMegan L. MayerDepartment of Neuroscience, School of Medicine, Case Western Reserve University, Cleveland, OH, 44106-4970, USASudha ChakrapaniAuthorsArvind KumarView author publicationsYou can also search for this author in PubMed聽Google ScholarSandip BasakView author publicationsYou can also search for this author in PubMed聽Google ScholarShanlin RaoView author publicationsYou can also search for this author in PubMed聽Google ScholarYvonne GicheruView author publicationsYou can also search for this author in PubMed聽Google ScholarMegan L. MayerView author publicationsYou can also search for this author in PubMed聽Google ScholarMark S. P. SansomView author publicationsYou can also search for this author in PubMed聽Google ScholarSudha ChakrapaniView author publicationsYou can also search for this author in PubMed聽Google ScholarContributionsA.K. and S.C. conceived the project and designed the experimental procedures. A.K. purified the protein, and with assistance from S.B., optimized the cryo-EM sample preparation and performed grid screening. A.K carried out cryo-EM data analysis, model building, and refinement with inputs from S.B. M.L.M. collected the cryo-EM data for GlyR鈥揋ly/PTX. A.K. and Y.G. performed two-electrode voltage-clamp recordings. S.R. performed the MD simulations, under the supervision of M.S. S.C. supervised the execution of the experiments, data analysis, and interpretation. A.K. and S.C. drafted the paper with contributions from S.B., S.R., and M.S. All authors reviewed the final paper.Corresponding authorCorrespondence to Sudha Chakrapani.Ethics declarations Competing interests The authors declare no competing interests. Additional informationPeer review information Nature Communications thanks Anders Jensen, Yuji Sugita and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.Publisher鈥檚 note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.Supplementary informationSupplementary InformationPeer Review FileDescription of Additional Supplementary FilesSupplementary Movie 1Supplementary Movie 2Reporting SummaryRights and permissions Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article鈥檚 Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article鈥檚 Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. Reprints and PermissionsAbout this articleCite this articleKumar, A., Basak, S., Rao, S. et al. Mechanisms of activation and desensitization of full-length glycine receptor in lipid nanodiscs. Nat Commun 11, 3752 (2020). https://doi.org/10.1038/s41467-020-17364-5Download citationReceived: 07 February 2020Accepted: 25 June 2020Published: 27 July 2020DOI: https://doi.org/10.1038/s41467-020-17364-5 CommentsBy submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate. Sign up for the Nature Briefing newsletter 鈥?what matters in science, free to your inbox daily.