The nicotinic acetylcholine receptor and its prokaryotic homologues: Structure, conformational transitions & allosteric modulation

Neuropharmacology 96 (2015) 137e149

Contents lists available at ScienceDirect

Neuropharmacology journal homepage: www.elsevier.com/locate/neuropharm

Invited review

The nicotinic acetylcholine receptor and its prokaryotic homologues: Structure, conformational transitions & allosteric modulation Marco Cecchini a, *, Jean-Pierre Changeux b, c, d, ** ISIS, UMR 7006 CNRS, Universit e de Strasbourg, F-67083 Strasbourg Cedex, France CNRS, URA 2182, F-75015 Paris, France c Coll ege de France, F-75005 Paris, France d Kavli Institute for Brain & Mind University of California, San Diego La Jolla, CA 92093, USA a

b

a r t i c l e i n f o

a b s t r a c t

Article history: Available online 18 December 2014

Pentameric ligand-gated ion channels (pLGICs) play a central role in intercellular communications in the nervous system by converting the binding of a chemical messenger e a neurotransmitter e into an ion flux through the postsynaptic membrane. Here, we present an overview of the most recent advances on the signal transduction mechanism boosted by X-ray crystallography of both prokaryotic and eukaryotic homologues of the nicotinic acetylcholine receptor (nAChR) in conjunction with time-resolved analyses based on single-channel electrophysiology and Molecular Dynamics simulations. The available data consistently point to a global mechanism of gating that involves a large reorganization of the receptor mediated by two distinct quaternary transitions: a global twisting and a radial expansion/contraction of the extracellular domain. These transitions profoundly modify the organization of the interface between subunits, which host several sites for orthosteric and allosteric modulatory ligands. The same mechanism may thus mediate both positive and negative allosteric modulations of pLGICs ligand binding at topographically distinct sites. The emerging picture of signal transduction is expected to pave the way to new pharmacological strategies for the development of allosteric modulators of nAChR and pLGICs in general. This article is part of the Special Issue entitled ‘The Nicotinic Acetylcholine Receptor: From Molecular Biology to Cognition’. © 2015 Elsevier Ltd. All rights reserved.

Keywords: Acethylcholine receptor Protein allostery Ligand-gated ion channels Gating mechanism Agonism Antagonism

1. Introduction A critical event in the history of biological chemistry was, 44 years ago, the chemical identification of the first neurotransmitter receptor, the nicotinic acetylcholine receptor (nAChR) from fish electric organ (Changeux et al., 1970; Miledi et al., 1971; Corringer et al., 2000; Karlin, 2002; Changeux and Edelstein, 2005) (rev (Changeux, 2012)). The success was due to the convergence of disciplines as diverse as electrophysiology, pharmacology, and biochemistry with the common goal of successfully identifying the molecular switch that converts a chemical input into an electrical output at neuronal synapses. Since then, the cationic nicotinic receptor has become the titular head of a broad family of pentameric ligand-gated ion channels (pLGICs), paving the way to the identification of the homologous inhibitory GABAA (rev (Barnard, 1995; * Corresponding author. ** Corresponding author. CNRS, URA 2182, Institut Pasteur, F-75015 Paris, France. E-mail addresses: [email protected] (M. Cecchini), [email protected] (J.-P. Changeux). http://dx.doi.org/10.1016/j.neuropharm.2014.12.006 0028-3908/© 2015 Elsevier Ltd. All rights reserved.

Olsen and Sieghart, 2009; Miller and Aricescu, 2014)) and glycine (rev (Dutertre et al., 2012)) receptors, along with the excitatory 5hydroxytryptamine receptor (rev (Yan et al., 1999)) and, in invertebrates, the glutamate-gated chloride channel (GluCl) (rev (Hibbs and Gouaux, 2011)). The recent discovery of cationic orthologs in prokaryotes (Tasneem et al., 2005; Bocquet et al., 2007) has extended the superfamily, plunging its evolutionary origins back 3 billion years and leading to the first crystallization and full atomistic structure of a pentameric ligand-gated ion channel (Hilf and Dutzler, 2008, 2009; Bocquet et al., 2009). These oligomeric membrane proteins are allosterically regulated by the binding of a neurotransmitter dthe agonistd to an orthosteric site that is topographically distinct from the transmembrane ion channel (rev (Corringer et al., 2012; Taly et al., 2014)). At rest, the ion channel is closed and binding of the agonist to the extracellular domain triggers a rapid conformational change that results in the opening of the transmembrane pore, a process referred to as gating (Changeux and Edelstein, 1998). Recently, the extension of computational approaches based on Molecular Dynamics (rev (Karplus and McCammon, 2002)) to pentameric receptors (Taly

138

M. Cecchini, J.-P. Changeux / Neuropharmacology 96 (2015) 137e149

et al., 2005; Cheng et al., 2006; Nury et al., 2010; Calimet et al., 2013) has introduced a new temporal dimension in the understanding of the signal transduction mechanism. Also, an important outcome of the most recent structural and functional studies is that many of these regulatory proteins carry a variety of allosteric modulatory sites, e in addition to the main categories of orthosteric regulatory and biologically active sites e which have become important new targets for drug design (Mulle et al., 1992; Vernino et al., 1992; Krause et al., 1998; Bertrand and Gopalakrishnan, 2007) (rev (Changeux, 2013a)). In this review, we briefly present the most recent advances in the structure, conformational transitions and allosteric modulation of nAChR and its prokaryotic homologues. We do not intend to exhaustively review electrophysiological studies of nAChR (see e.g. Auerbach and others in this volume) nor early EM analyses (Brisson and Unwin, 1985; Unwin, 2005), but place special emphasis on the recent X-ray structures and Molecular Dynamic simulations. 2. Molecular architecture of nAChRs and homologues nAChRs and homologues are integral allosteric membrane proteins with a molecular mass of ~290 kDa, comprising five identical or homologous subunits symmetrically arranged around a central ionic channel with a fivefold axis perpendicular to the membrane plane. In mammals, there are multiple types of nAChR oligomers, which differ in their subunit compositions throughout the body (Wang et al., 2015; Zoli et al., 2015). Of these, nine b-subunits and three a-subunits are expressed in the brain. They assemble into various homo- and hetero-pentameric combinations, which differ in their pharmacological, physiological and kinetic properties along with their localization in the brain (Zoli et al., 2015). The 3D-structure of nAChR and homologues is wellcharacterized (Karlin, 2002; Taly et al., 2014, 2009; Thompson et al., 2010). The primary structure of each subunit consists of a large hydrophilic amino-terminal extracellular (EC) domain, a transmembrane (TM) domain comprising four hydrophobic segments (M1eM4), and a variable hydrophilic cytoplasmic or intracellular (IC) domain, which is absent in prokaryotic pLGICs; see Fig. 1. There are 2e5 ACh-binding sites within the EC domain, which are distant (ca. 60 Å) but functionally linked to a unique cationic ion channel, located on the axis of symmetry of the TM domain. The atomic structure of the EC domain was first solved for the AChbinding protein (AChBP) d a soluble pentameric homologue of the EC domain of nAChR d which was initially cloned from invertebrate snails (Smit et al., 2001; Brejc et al., 2001). Complete structures of pLGICs at atomic resolution were then obtained with two prokaryotic homologues of nAChR from Gloeobacter violaceus (GLIC) (Hilf and Dutzler, 2009; Bocquet et al., 2009) and Erwinia chrysanthemi (ELIC) (Hilf and Dutzler, 2008) and only recently with three eukaryotic receptors: GluCl from Caenorhabditis elegans (Hibbs and Gouaux, 2011), the 5HT3 receptor from mouse (Hassaine et al., 2014), and the GABAA receptor from human (Miller and Aricescu, 2014). At this stage, no full X-ray structure of any nAChR has ever been obtained. Only low resolution images of the electric organ/muscle nAChR have been published (see (Unwin, 2013), however see also (Taly et al., 2014)). In agreement with structural studies of the AChBP (Brejc et al., 2001), the EC domain of all prokaryotic and eukaryotic pLGICs is folded into a highly conserved immunoglobulin-like b-sandwich stabilized by inner hydrophobic residues. However, the connecting loops, as well as the N-terminal a-helix present in most eukaryotic pLGICs but not in the prokaryotic ones, are variable in length and structure. To date, the role of this N-terminal a-helix, which is duplicated in the GABAA b3 receptor (Miller and Aricescu, 2014), remains unknown (see (Corringer et al., 2012)). Consistent with

early EM data of Torpedo nAChR at low (maximum 4 Å) resolution (Unwin, 2005), the four transmembrane segments fold into a-helices and are organized as a well conserved bundle strikingly different from the Kþ channel pore (Zhou et al., 2001). The second segment M2 lines the channel walls (Giraudat et al., 1986, 1987; Hucho et al., 1986; Imoto et al., 1988, 1986) and is surrounded by a ring of a helices made of M1 and M3. M4 lies at the side of the bundle. In all X-ray structures investigated to date, the TM domain makes extensive interactions with the lipid bilayer, which is also thought to play a role in modulating ion permeation. Interestingly, the structure of GLIC reveals three lipid molecules per subunit that are bound in the crevices between M4 and either M1 or M3 (Bocquet et al., 2009; Nury et al., 2011). Higher resolution structures (2.4 Å) of the prokaryotic channel GLIC in its open conformation revealed the presence of ordered pentagons of water molecules within the ion pore at the level of two rings of hydroxylated residues (named Ser60 and Thr20 ), which are thought to contribute to the ion selectivity filter (Sauguet et al., 2013a). Finally, the very recent structure of the eukaryotic 5HT3 receptor, which first visualized the conformation of the intracellular domain (IC) in pLGICs, suggests the IC domain would also contribute to ion permeation (Hassaine et al., 2014). The major loop contributing to the EC/TM domains interface, namely the Cys loop carrying the canonical FPFD motif (Rendon et al., 2011), is not ordered in the structure of the isolated EC domain of GLIC (Nury et al., 2010) but adopts a well-defined conformation in the full-length receptor structure through extensive interactions with the four a-helical bundle of the TM domain (Bocquet et al., 2009). The superposition of GLIC and GluCl reveals striking similarities, but also differences that are found at the subunitesubunit interfaces including the b8eb9 loop (Loop F, see below), which is located on the outside of the EC domain, and an insertion loop in strand b5 that faces the inner part of the vestibule (Miller and Aricescu, 2014; Hibbs and Gouaux, 2011). The former region is important for ligand binding (Bourne et al., 2005), the latter is thought to influence assembly specificity in GluCl and the GABAA receptor (Miller and Aricescu, 2014) and be involved in the Zinc modulation of Gly receptors (Miller et al., 2008). Overall, the available X-ray structures of prokaryotic and eukaryotic pLGICs reveal a striking similarity in the 3D structure of the nAChR with its homologues (Corringer et al., 2012; Taly et al., 2014); see Fig. 1. Yet, the absence of high-resolution structures of heteromeric pseudo-symmetrical pLGICs makes it difficult to extend all conclusions drawn on the prokaryotic channels to the eukaryotic members of the superfamily. From a physiological perspective, the particularly well-conserved architecture is suggestive of conservation of function despite significant differences in the primary structure do exist. Indeed, chimeric constructs produced by merging the EC and TM domains from different pentameric receptors were shown to fully preserve function. For instance, functional chimeras of eukaryotic a7, a4, and b2 nAChR subunits with the 5HT3 receptor (Eisele et al., 1993; Cooper et al., 1999) and a7 nAChR with a1 GlyR (Grutter et al., 2005) have been produced. Even more striking, functional chimeras have been constructed using the bacterial GLIC and the eukaryotic a1 GlyR (Duret et al., 2011), demonstrating a remarkable conservation of the structural organization from bacteria to humans. 3. Functional interpretation of structures Signal transduction by nAChR was proposed since the 60's to be mediated by a global isomerization of the receptor coupling the neurotransmitter binding site in the EC domain and the transmembrane ion channel, which was referred to as an allosteric transition (Changeux, 1964, 1966; Changeux et al., 1967; Changeux

M. Cecchini, J.-P. Changeux / Neuropharmacology 96 (2015) 137e149

139

Fig. 1. General topology of pLGICs. On the left, the gross architecture of the heteromeric a2bgd muscle nAChR from Torpedo is shown as visualized by the low (maximum 4 Å) resolution cryo EM reconstruction of Unwin (Unwin, 2005). On the right, the topology of the prokaryotic channel from Gloeobacter violaceus (GLIC) is shown as visualized by its Xray structure at pH4 solved at 2.9 Å resolution. (Bocquet et al., 2009) (Upper) Side view of the channels along the membrane plane (gray). Each subunit is color-coded differently. The extracellular (EC), the transmembrane (TM), and the intracellular (IC) domains are labeled. The topographical position of the neurotransmitter acetylcholine (ACh) binding sites in the EC domain of nAChR is indicated. These sites are located at the interfaces between the a-g and a-d subunits with loop C acting as a binding-site lid shown in red. (Lower) Top view of the TM domain from the extracellular side. The different subunits are labeled. The location of the ion pore in the TM domain is indicated. The neurotransmitter binding site(s) and the ion pore are far apart and must be coupled by an allosteric mechanism. Apart from the IC domain, which is absent in prokaryotes, the overall topology of nAChR and the bacterial channel GLIC are essentially indistinguishable. The greater simplicity of the prokaryotic receptor, which is homopentameric and devoid of the IC domain, is one of the reasons of its successful crystallization, which provided the first high-resolution structure of any full length pLGIC (Bocquet et al., 2009).

et al., 1969). Several structural models have been proposed for ion channel's activation and deactivation. Among them, the MonodWyman-Changeux (Monod et al., 1965) (MWC) model postulates that allosteric ligand-gated ion channels exist in reversible equilibrium between a few e at least two e global conformational states, namely a resting (R) closed-channel state and an active (A) open-channel state, even in the absence of agonist (Changeux and Edelstein, 2005) and that a conformational selection e or shift of conformers population e takes place in the presence of agonist (Changeux and Edelstein, 2005; Cui and Karplus, 2008). In the MWC model the effect of the ligand(s) is to stabilize the conformation of the receptor for which it has the highest affinity. In addition, to account for desensitization, i.e. the time-dependent

decrease of ion conductance that follows receptor's activation, several slowly accessible, high-affinity, closed-channel states were introduced (rev (Changeux and Edelstein, 2005), see also (Lape et al., 2008, 2012)). As pLGICs functions are mediated by a set of conformational transitions between more than two states in interaction with ligands, it is of fundamental importance to provide a functional interpretation of the presently available highresolution structures for a mechanistic description of channel gating and its allosteric modulation. Until recently, the only well characterized structure of pLGICs was arguably the open-channel state visualized by GLIC at pH4 (Hilf and Dutzler, 2009), (Bocquet et al., 2009). The striking similarity with the open form of GluCl (Hibbs and Gouaux, 2011) solved in

140

M. Cecchini, J.-P. Changeux / Neuropharmacology 96 (2015) 137e149

complex with the positive allosteric modulator ivermectin strongly indicates that these structures are adequate models of the active state. Also, close comparison with the recently determined closedchannel forms of GLIC (Sauguet et al., 2014a) and GluCl (Althoff et al., 2014) reveals that a major re-organization of the receptor takes place during activation, consistent overall with a large quaternary transition. Interestingly, an intrinsic structural variability was observed in the closed state of GLIC pH7 (Sauguet et al., 2014a) with significant deviations from the C5 symmetry (root-meansquare deviation of 1.2 ± 0.2 Å in the EC domain compared with 0.2 Å in the TM domain). These fluctuations, which were not detected with the open state of the same receptor molecule, prefigure the transition mediating activation and indicate the coexistence of multiple closed-channel conformations in the closed state. The structure of ELIC, although well resolved and with a closed channel (Hilf and Dutzler, 2008), is not universally accepted as a model of the resting state (Gonzalez-Gutierrez and Grosman, 2010). First, close comparison with the pH7 state of GLIC shows a different configuration of the ion pore along with a profound quaternary reorganization of the EC domain (Sauguet et al., 2014a). Second, the structure of ELIC appears to be virtually unaffected by the cocrystallization with the agonist (Spurny et al., 2012). Finally, the conformation of the C loop in ELIC is surprisingly similar to the one adopted in the active state of GLIC. These observations are consistent with the structure of ELIC being a desensitized state. Along the same line, the conformation of the “locally closed” (LC) state of GLIC (Prevost et al., 2012), which was stabilized by disulfide cross-linking at various positions at the EC/TM domains interface, exhibits globally the conformation of the active state but carries a closed channel due to a concerted inward bending of the upper part of the M2 helices. This structure has been plausibly assigned to a fast-desensitised or I state (or alternatively a gating intermediate) (Prevost et al., 2013a). Finally, the very recent structure of the GABAA b3 receptor (Miller and Aricescu, 2014), which was solved in complex with the orthosteric agonist benzamidine, shows a non-conducting state produced by the inward tapering of the M2 helices (Miller and Aricescu, 2014) at the intracellular end (residues -20 ). Because of the presence of the agonist in the orthosteric site, this structure was also interpreted as a desensitized state. Given the above, X-ray crystallography alone is not sufficient for a non-ambiguous functional interpretation of structures and, as we shall see, it needs to be complemented with time-resolved analyses. 4. The ligand binding sites The recent deciphering of the 3D structure of several pLGICs has led to considerable progress in our understanding of the fine topography of the multiple ligand-binding sites carried by nAChR and its prokaryotic homologues: the orthosteric site, the allosteric site(s) and the ion channel. In the following, we briefly review the recent advances on the structural characterization of these regulatory sites in pLGICs. See Sauguet et al., 2014a,b for a general review. 4.1. The orthosteric binding site In the whole pLGIC family, the neurotransmitter or orthosteric binding site is located in the EC domain at the interface between subunits; see Fig. 2. Three regions from the ‘‘principal’’ subunit (loops A, B, and C) referred to as (þ) and four from the ‘‘complementary’’ subunit (loops D, E, F, and G) referred to as () contribute to the binding pocket (Corringer et al., 2000). X-ray structures of the orthosteric site have been reported for AChBP (Brejc et al.,

2001), GLIC (Hilf and Dutzler, 2009; Bocquet et al., 2009), ELIC (Hilf and Dutzler, 2008), GluCl (Hibbs and Gouaux, 2011) and GABAA (Miller and Aricescu, 2014) receptors. In AChBP, loops A (Tyr), B (Trp), C (two Tyr), and D (Trp) form an aromatic ‘‘box’’ chelating the ammonium group of ACh with the tryptophan residue from loop B establishing a direct cation-p interaction (Zhong et al., 1998). In GluCl, the endogenous agonist L-glutamate binds through its ammonium group to corresponding aromatic residues from loop A (Phe), B (Tyr), and C (Tyr), whereas the lateral carboxylate groups form salt-bridging interactions with Arg and Lys residues from loops D and F of the () subunit (Hibbs and Gouaux, 2011). In the GABAA receptor b3, the orthosteric site is also formed by four aromatic residues protruding from both the (þ) and () subunits, one from loop B (Tyr), two from loop C (Tyr, Phe) and one from loop D (Tyr), along with Glu155 delimiting the top of the binding pocket; this residue was shown to be implicated in receptor's activation (Newell et al., 2004). Analogous interactions are required for ligand binding and activation in nAChR (Miller and Aricescu, 2014; Purohit et al., 2012). In particular, electrophysiological studies have shown that only three aromatic residues are crucial in adult nAChRs, whereas the four of them are important in the fetal form (Auerbach, 2015). Cocrystallization of ELIC in complex with the mild agonist bromopropylamine at 4.0 Å resolution (Zimmermann and Dutzler, 2011) or the competitive antagonist acetylcholine at 2.9 Å resolution (Pan et al., 2012) showed that both ligands bind in the EC domain at the subunitesubunit interface. In particular, the latter shows that ligand binding to an aromatic cage at the orthosteric site causes significant contraction of loop C along with a slight increase of the ion-pore diameter in the TM domain (Pan et al., 2012). A striking exception in the pLGIC family is the orthosteric site of the proton-gated channel GLIC, which exposes a cluster of charged and non-aromatic residues with the B-loop adopting a unique upward extended conformation. Nonetheless, cinnamic acid derivatives, which antagonize the proton-elicited response in GLIC, were found to bind at the interface between subunits to a pocket located slightly below the classical orthosteric site (Prevost et al., 2013b). Overall, the structure of the orthosteric neurotransmitter site in pLGICs appears to be remarkably conserved from bacteria to humans. 4.2. The ion channel binding site(s) Channel blockers are an important category of nicotinic drugs that exert valuable therapeutic actions in humans. They bind to the TM domain of nAChR and prevent ion flux by sterically occluding the channel pore (rev (Neher, 1983; Changeux, 1990)). The binding sites for channel blockers are structurally and functionally distinct from the orthosteric site(s) and reciprocally linked by allosteric interactions (Changeux, 1981). Early affinity-labeling experiments with channel blockers such as chlorpromazine (Giraudat et al., 1986, 1987) or trimethyl phenyl phosphonium (TPMP) (Hucho et al., 1986) in nAChR identified the key residues within the segment M2 at positions 20 , 60 , 90 , 130 and 20' (rev (Corringer et al., 2000, 2010)), which is consistent with the notions that the same side of the M2 helices ‘face’ the ion pore and that the walls of the channel consist of superimposed pentameric rings of homologous amino acids. The binding sites for channel blockers are distributed throughout the transmembrane channel; in particular they are located within the gate (the mid-section stretch of hydrophobic residues) in the closed conformation and at the entrance of the ion selectivity filter at the cytoplasmic border. The recent highresolution structures of pLGICs (Sauguet et al., 2014b) demonstrate a remarkable conservation of permeation and selectivity structure/function relationships in the TM domain from prokaryotes to eukaryotes, including the homologs of the

M. Cecchini, J.-P. Changeux / Neuropharmacology 96 (2015) 137e149

chlorpromazine/TMPP binding sites (Giraudat et al., 1986, 1987; Hucho et al., 1986) and the outside/inside anionic rings (Imoto et al., 1988, 1986; Leonard et al., 1988).

4.3. The allosteric binding site(s) Several regulatory allosteric sites distinct from the neurotransmitter-binding site have been identified in pLGICs (Sauguet et al., 2014b). Ca2þ ions were the first positive allosteric modulator recognized with a7 and a4b2 neuronal nAChRs (Mulle et al., 1992; Vernino et al., 1992). Site-directed mutagenesis of the Ca2þ‑binding sites in a7 nAChR identified residues in the EC domain grouped on opposite sides of adjacent subunits near the interface with the TM domain (Galzi et al., 1996; Le Novere et al., 2002); see Fig. 2. More recently, homologs of the Ca2þ-binding sites have been structurally characterized with ELIC, where divalent cations including Ba2þ behave as negative modulators (Zimmermann and Dutzler, 2011), and GLIC, where it forms a well-delimited pocket for yet unidentified ligands (Sauguet et al., 2014a). Another important site for the allosteric regulation of pLGICs has been identified in the TM domain. The modulatory effect of the antihelmintic ivermectin, which strongly enhances the ACh-evoked response of a7 nAChR at micromolar concentration (with increased

Fig. 2. Ligand-binding sites in pLGICs. The side view of the pLGIC along the membrane is shown as visualized by the crystal structure of GluCl (Hibbs and Gouaux, 2011). The two front subunits of the homopentamer, which correspond to the principal (blue sky) and the complementary (white) subunits, are shown in cartoon representations. The remaining three subunits are shown as solvent-accessible surfaces color-coded according to the EC (pale cyan) and TM (violet) domains. Ligand binding at the interface between subunits is highlighted in colors. The endogenous agonist L-glutamate, which binds to the orthosteric site, is shown as green spheres. The positive allosteric modulator ivermectin, which binds to the allosteric intersubunit site in the TM domain, is shown in magenta sticks. A cyan sphere shows the position of the allosteric Ca2þbinding site for the negative modulation of some pLGICs e.g. nAChR by divalent cations. The coordinates of the Ca2þ ion were taken from the structure of ELIC in complex with the negative allosteric modulator Ba2þ (Zimmermann et al., 2012) after optimal superimposition of the TM domain.

141

apparent affinity, cooperativity and maximal response) was shown to be altered by mutations in the TM domain (Krause et al., 1998). Several synthetic modulators of a7 nAChR such as PNU-120596 and LY 2087101 were shown to bind to this site (Bertrand and Gopalakrishnan, 2007; Changeux, 2013b; Forman et al., 2015). Also, ivermectin potently activates the ion-channel response of GluCl even in the absence of the endogenous neurotransmitter. The X-ray structure of GluCl in complex with ivermectin (Hibbs and Gouaux, 2011) showed that the ivermectin-binding site is located at the periphery of the TM domain wedged by the helices M3 (þ) and M1 () at the interface between subunits; see Fig. 2. Ethanol binding sites identified in the X-ray structure of an ethanolsensitized variant of GLIC (Sauguet et al., 2013b) were shown to be closely related to the ivermectin-binding site in GluCl. Interestingly, this intersubunit transmembrane cavity is conserved in human ethanol-sensitive glycine and GABAA receptors and involves residues previously recognized as influencing alcohol and anesthetic action (Li et al., 2006; Olsen et al., 2014). Early efforts to reconstitute neurotransmitter receptors in lipid vesicles demonstrated that the activity of nAChR is also modulated by the composition of the membrane in which the channel is embedded (Heidmann et al., 1980; Ochoa et al., 1983). Similar conclusions were drawn with the GABAA receptor in rat hippocampal neurons (Sooksawate and Simmonds, 2001). In the absence of anionic lipids and cholesterol, nAChRs appear to populate an “uncoupled” conformation that binds agonists with a resting-like affinity but does not undergo the gating isomerization (daCosta and Baenziger, 2009). It was then suggested that lipids could modulate the transition with the uncoupled state, thereby regulating channel gating (daCosta et al., 2013). Experimental measurements based on electron-spin resonance difference spectroscopy (Marsh and Barrantes, 1978), fluorescence quenching (Jones and McNamee, 1988), and substitution of cholesterol with phospholipids-cholesterol hybrids (Addona et al., 1998) support the idea that binding of cholesterol to the TM domain affects the structure and/or the dynamics of pLGICs. Consistently, recent computational analyses of the cryo-EM structure of Torpedo nAChR (Brannigan et al., 2008) and a homology model of the GABAA receptor (Henin et al., 2014) showed the existence of several cavities in the TM domain complementary to cholesterol and suggested that bound cholesterol at these sites would stabilize the open-pore conformation of the channel. Altogether, these results suggest that cholesterol binding to the TM domain is homologous to ivermectin binding to GluCl, with cholesterol possibly acting as an endogenous allosteric modulator in eukaryotic pLGICs. General anesthetics such as propofol and desflurane are known to be positive allosteric modulators of GABAA receptors in eukaryotes (Olsen et al., 2014). They also behave as negative allosteric modulators of GLIC (Weng et al., 2010). The X-ray structure of GLIC pH4 in complex with propofol and desfurane demonstrates a common binding site located within the upper part of the transmembrane subunits in a cavity delimited by M1, M2, and M3 (Nury et al., 2011). Also, it reveals that such an intrasubunit transmembrane cavity is accessible from the lipid bilayer. Because its entrance is obstructed by a lipid alkyl chain, which would clash with propofol binding, it was argued that lipids could be endogenous ligands for this allosteric site (Nury et al., 2011). On the other hand, the recent X-ray structure of apo-GluCl, which was obtained by supplementing POPC lipids in the absence of ivermectin, suggests that lipid binding to the intersubunit rather than to the intrasubunit transmembrane site would stabilize the open-channel conformation (Althoff et al., 2014). Further analysis is required to establish the mechanism of modulation by lipids. Homologous intersubunit and intrasubunit binding sites are present in the TM domain of the Gly, GABAA and nACh receptors and are of considerable

142

M. Cecchini, J.-P. Changeux / Neuropharmacology 96 (2015) 137e149

pharmacological importance as they may accommodate a large variety of drugs including anesthetics, barbiturates or diuretics (rev Olsen et al., 2014; Li et al., 2010; Chiara et al., 2013). In heteropentameric pLGICs such as the neuronal a4b2 nAChR, not all five homologous binding sites in the EC domain bind ACh. Thus, the non-agonist-binding interface may accommodate modulatory ligands different from the neurotransmitter. Using AChBP as a structural model, conditions were found where ligands such as galanthamine, strychnine, cocaine, and morphine bind at micromolar concentrations (Hansen and Taylor, 2007; Nemecz and Taylor, 2011; Hamouda et al., 2013). Based on data collected on nAChR, the binding site of allosteric modulators that do not target the neurotransmitter site was initially suggested to be homologous to the benzodiazepines binding site in GABAA receptors (Galzi and Changeux, 1994). Although crystallographic evidence is still missing, considerable biochemical, pharmacological and modeling data support the notion that benzodiazepines allosterically potentiate GABAA receptors by binding to intersubunit sites in the EC domain that are homologous to the GABA sites but do not bind GABA (Smith and Olsen, 2000; Sawyer et al., 2002). Finally, other allosteric modulatory sites are present in the IC cytoplasmic domain and may play important roles in clustering, stabilization, and modulation of receptor functions (rev (Changeux, 2013b)) in particular through the interaction with GPCRs.

of a7 nAChR (Taly et al., 2005), is best described by the concerted opposite-direction rotation of the EC domain relative to the TM about the five-fold symmetry axis (in a counterclockwise manner as viewed from the synaptic cleft); see Fig. 3 (bottom). Because this change in conformation involves a quaternary reorganization of the receptor with significant reshaping of the subunit interface,

5. The gating mechanism at atomic resolution The conformational transition(s) of nAChR producing an ionconducting (or active) state in the presence of high levels of neurotransmitter is commonly referred to as the gating isomerization. Based on the MWC model (Monod et al., 1965), this functional response of the receptor was interpreted as a selection between a few global and discrete conformational states elicited by the binding of agonist (Changeux and Edelstein, 2005; Cui and Karplus, 2008). This simple model of gating, which already accounts for a number of experimental data (Changeux and Edelstein, 2005; Auerbach, 2013; Jackson, 1986; Changeux and Taly, 2008), may only provide a static (end-point) picture of signal transduction. To break through the dynamic nature of the phenomenon, complementary and time-resolved analyses such as rate-equilibrium free energy relationships (Auerbach, 2007; Lee et al., 2008) and Molecular Dynamics simulations (Taly et al., 2005; Nury et al., 2010; Calimet et al., 2013) were needed. In combination with highresolution structures of pLGICs, these approaches provided a new temporal dimension, which has shed light onto the sequence of structural events that translate neurotransmitter binding into opening of the ion channel 60 Å away. The emerging picture of gating is that a progressive stepwise isomerization (previously referred to as conformational wave) that starts from the orthostericbinding site (loops A, B, and C), propagates to the EC/TM domains interface (b1eb2 loop and Cys loop) via a rigid-body rearrangement of the extracellular b-sandwiches and moves down to the transmembrane helices (first M2, then M4 and M3) to ultimately open the gate (Calimet et al., 2013; Sauguet et al., 2014a; Grosman et al., 2000; Purohit et al., 2007). In this context, the quaternary character of the gating isomerization, which takes place even in the absence of agonist (Jackson, 1986), provides the link between agonist binding and the functional structural changes of the receptor and is key to elucidate the nature of allosteric regulation in pLGICs (Calimet et al., 2013; Cui and Karplus, 2008). The crystal structures of the prokaryotic homologues GLIC pH4 (open channel) and ELIC or GLIC pH7 (closed channel) unambiguously showed the occurrence of a global twist on receptor's activation (Bocquet et al., 2009). This structural rearrangement, which was first identified by normal mode analysis of a homology model

Fig. 3. The blooming and the twisting components of the gating isomerization. On top, the blooming transition is shown. The conformation of the A state is shown in gray as captured by the X-ray structure of GLIC at pH4 (Sauguet et al., 2013a) with loop C closed on top of the orthosteric neurotransmitter site. For illustration, a hypothetical agonist bound to the EC domain is shown as green spheres; its coordinates correspond to those of L-glutamate in the A state of GluCl (Hibbs and Gouaux, 2011) after optimal superimposition of the TM domain. The extracellular b-sandwiches in the R state are shown in pink; coordinates were extracted from the crystal structure of GLIC at pH7 (Sauguet et al., 2014a) and are shown after optimal superimposition of the TM domain. Pink dashed arrows illustrate the direction of the blooming transition from the active (contracted) to the resting (expanded) state. On bottom, the twisting transition is shown. The conformation of the A state is shown in white as captured by the X-ray structure of GluCl in complex with the allosteric agonist ivermectin (Hibbs and Gouaux, 2011). The positive allosteric modulator ivermectine in the TM domain is shown as magenta sticks. The extracellular b-sandwiches as captured by simulation (Calimet et al., 2013) at the end of the twisting transition are shown in cyan. The large blue dashed arrow indicates the direction of the twisting motion from the active (untwisted) to the resting (twisted). The combination of these two quaternary transitions describes the gating isomerization in pLGICs (Taly et al., 2014).

M. Cecchini, J.-P. Changeux / Neuropharmacology 96 (2015) 137e149

quaternary twisting was initially suggested to be directly involved in ion channel activation (Taly et al., 2005). The subsequent determinations of the X-ray structure of the prokaryotic homologues of nAChR (Hilf and Dutzler, 2008, 2009; Bocquet et al., 2009), confirmed the occurrence of the quaternary twist but made clear that important tertiary changes on activation e in particular the significant tilting of the M2 helices e were not accounted for by the minimal twisting model. In addition, the structure of the “locally closed” (LC) state of GLIC, which showed a closed ion pore in a receptor preserving most features of the open form, seriously questioned the direct coupling between global twisting and M2 tilting (Prevost et al., 2012). Subsequent computational analyses based on all-atom Molecular Dynamics shed new light on the molecular mechanism for gating ions (Nury et al., 2010; Calimet et al., 2013). By monitoring the spontaneous relaxation of the open-channel structure upon agonist (or positive allosteric modulator) unbinding (Nury et al., 2010; Calimet et al., 2013), these analyses provided evidence for the existence of an indirect coupling mechanism. In fact, the simulation of the open state of GLIC at neutral pH (Nury et al., 2010) or that of GluCl with ivermectin removed (Calimet et al., 2013) consistently showed that global twisting initiates the closing transition by facilitating the untilting of the M2 helices, which does not occur in the untwisted (active) state of the receptor. The mechanistic scenario emerging from the simulations thus suggests that receptor twisting would contribute to activation by “locking” the ion channel in its open-pore form. In addition, the simulation of GluCl with ivermectin removed (Calimet et al., 2013) predicted that a large radial reorientation or tilting of the extracellular b-sandwiches in the outward direction would be implicated in the allosteric communication between the neurotransmitter-binding site and the ion pore; see Fig. 3 (top). Remarkably, such a radial expansion or blooming of the EC domain in the resting state has been recently demonstrated by the X-ray structure of GLIC pH7 (Sauguet et al., 2014a). Also, a very recent crystallographic analysis of Lymnaea AChBP in complex with a series 4,6-disubstituted 2-aminopyrimidines exhibiting both positive and negative cooperativity in ligand-binding assays, showed that cooperative binding is associated with the global expansion (blooming) of the AChBP molecule, yet with no significant twisting (Kaczanowska et al., 2014). Finally, the most recent structure of GluCl solved in the absence of agonist confirmed that a large isomerization of the EC domain “resembling the closure of a blossom” occurs on receptor's activation (Althoff et al., 2014); note that using the definition of ref. (Calimet et al., 2013) the X-ray structure of apo GluCl captures a conformation of the receptor that is 10 more twisted than the active state, an evidence that was not discussed in the original paper (Althoff et al., 2014). Importantly, these observations confirm the occurrence of both twisting and blooming during the conformational transitions of GluCl and pLGICs more generally. The structural isomerizations presented above occur with significant restructuring of the interfaces between subunits and therefore can be potentially regulated by ligand-binding events. Recent single-channel electrophysiology of the murine nAChR activated by a series of orthosteric agonists with increasing potency (listed in (Jadey et al., 2011)) unambiguously showed that agonist binding at the orthosteric site modulates the opening rate with little or no effect on closing (Jadey et al., 2011). In sharp contrast, computer simulations of the active state of GluCl made it clear that agonist binding to the TM domain controls ion channel's closing (Calimet et al., 2013). These apparently contradictory observations in reality provide evidence that the rate determining step on gating, that is the highest barrier to be crossed, would correspond to structurally different events in the forward (activation) and in the

143

backward (deactivation) directions. The two-step (blooming/ twisting) gating isomerization presented above would naturally account for it and suggest that the reverse of blooming is rate determining on activation, whereas twisting is so on pore closing. The combination of two distinct quaternary transitions thus leads to the development of an asymmetric two-step model of gating, which is in agreement with kinetic models based on electrophysiological analyses (Purohit et al., 2013). As we shall see, the postulation of two-step mechanism also implies that agonist (or positive allosteric modulator) binding at the orthosteric neurotransmitter site(s) would promote activation by primarily regulating the blooming isomerization, i.e. favoring the contracted form of the EC domain, whereas ligand binding at the allosteric transmembrane site would do so by preventing the global twisting of the receptor.

Fig. 4. Dynamics of the orthosteric neurotransmitter-binding site. On top, the structural evolution of the orthosteric site during the twisting transition is shown. The principal (dark green) and the complementary (red) subunits are shown as visualized by the crystal structure of GluCl (Hibbs and Gouaux, 2011). The endogenous agonist, Lglutamate, is shown in light-blue sticks. The residues forming the agonist-binding pocket are shown in yellow with a dashed line joining their Ca atoms. The yellow and cyan dashed lines indicate the volume of the binding pocket before and after the twisting isomerization, respectively. The cyan dashed line was drawn using the coordinates of the protein extracted from the simulation of GluCl with ivermectin removed at the end of twisting transition (Calimet et al., 2013). On bottom, the dynamics of the orthosteric site during the blooming of the EC domain is shown. The principal (dark green) and the complementary (red) subunits are shown as visualized by the crystal structure of GLIC pH4 (Sauguet et al., 2013a). The coordinates of the (þ) subunits after blooming (blue) were extracted from the structure of GLIC pH7 (Sauguet et al., 2014a) and are shown here after optimal superimposition of the TM domain excluding M2. The cyan dashed lines indicate the expansion of the orthosteric binding pocket during the blooming isomerization.

144

M. Cecchini, J.-P. Changeux / Neuropharmacology 96 (2015) 137e149

The most recent structural determinations of eukaryotic pLGICs, particularly those of apo GluCl and the mouse serotonin 5HT3 receptor raise additional questions. The structure of apo GluCl, which was solved in complex with Fab, shows that a radial contraction of the EC domain occurs during activation in agreement with the present model of gating. However, this movement was not described as an authentic blooming/unblooming of the EC domain but rather as an inward contraction taking place on activation (Althoff et al., 2014). Whether the absence of enhanced structural variability in the resting state is due to the presence of antibodies, which bind at the EC subunits interface in the crystal, has yet to be established. In addition, the structure of the 5HT3 receptor, which was solved in complex with nanobodies and in the absence of agonist, raises the question of the role of the IC cytoplasmic domain in gating. In fact, this very recent structure visualizes an active-like conformation most similar to GluCl with ivermectin bound, i.e. globally untwisted and with an open transmembrane pore, but shows an apparently closed or non-permeable configuration of the IC domain (Hassaine et al., 2014). Although the functional significance of this structure is yet to be assessed, Hassaine et al. (Hassaine et al., 2014) stated that conformational changes in the IC domain are likely to mediate ion conductance as well. Under these conditions, a complete model of gating of eukaryotic pLGICs should include the contribution of the IC domain.

nAChR starting from bacterial and invertebrate eukaryotic homologues (Taly et al., 2014; Calimet et al., 2013). As discussed previously (Changeux, 2013a, 1990; Galzi et al., 1996), the pharmacological action of an allosteric modulator is mediated by the same global conformational transition that promotes channel opening via orthosteric ligands binding, yet with features of their own. The simulation analysis of the active state of GluCl with and without ivermectin demonstrated that global twisting can be regulated by the binding of positive allosteric modulators (like ivermectin) at the intersubunit transmembrane site (Calimet et al., 2013). At the same time, the structural determination of the resting state of GLIC (Sauguet et al., 2014a) unambiguously showed that a large contraction of the EC orthosteric site occurs on receptor activation. These two evidences made it clear that, in agreement with the MWC model, allosteric coupling in pLGICs is mediated by the dynamic reorganization of the interface between subunits. The recent characterization of the resting state of GLIC by X-ray crystallography (Sauguet et al., 2014a) and that of an intermediate state on activation (at the end of the twisting transition) by Molecular Dynamics (Calimet et al., 2013) provide the first opportunity to disentangle the conformational dynamics of the agonist-binding sites during signal transduction. 6.1. The orthosteric neurotransmitter site

6. Molecular dynamics of allosteric modulation Although structures of the nAChR homologues were first obtained with the prokaryotic GLIC (Hilf and Dutzler, 2009; Bocquet et al., 2009) and ELIC (Hilf and Dutzler, 2008), the more recent structures of the eukaryotic receptors GluCl (Hibbs and Gouaux, 2011), 5HT3 (Hassaine et al., 2014) and GABAA (Miller and Aricescu, 2014) have shown striking homologies that legitimate the attempt to explore gating and its allosteric modulation in

The crystal structure of GluCl solved in complex with the endogenous neurotransmitter (L-glutamate) and the positive allosteric modulator ivermectin revealed a configuration of the orthosteric site in its native agonist-bound state (Hibbs and Gouaux, 2011). As shown by Fig. 4, L-glutamate binding at the orthosteric site bridges residues from adjacent subunits and, opposite to the resting state structure of GLIC (Sauguet et al., 2014a), stabilizes a radially contracted configuration of the EC domain. Perhaps surprisingly, the simulation of GluCl with ivermectin removed

Fig. 5. Dynamics of the allosteric Ca2þ-binding site. The regulatory Ca2þ-binding site is shown as visualized by the structure of ELIC in complex with the negative allosteric modulator Ba2þ (ref. (Zimmermann et al., 2012)). The structural core of the b-sandwich is colored in green, loop F is in red. The coordinates of the Ca2þ ion (cyan) were taken from the structure of ELIC (not shown) upon optimal superposition of the inner b-sheet of one EC subunit to the structure of GLIC pH7, shown on the right. The structure of GLIC pH4 superimposed on the same b-sheet is shown on the left. The comparison of GLIC pH4 (active) with GLIC pH7 (resting) reveal that there is an important tertiary change in the EC subunits during activation, i.e. a rotation of the outer b-sheet relative to the inner sheet that produces a small cavity (orange surface) in the resting state structure. This cavity, which involves residues from both strand b1 and loop F and gives access to the hydrophobic core of the b-sandwich, is not formed in the active state structure. Given its dynamic character during receptor's activation/deactivation, this could be targeted as a ligand-binding site for negative allosteric modulators of pLGICs. Interestingly, this pocket is delimited by the conformation of loop F, which is implicated in binding of regulatory Calcium. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

M. Cecchini, J.-P. Changeux / Neuropharmacology 96 (2015) 137e149

145

Fig. 6. Dynamics of the allosteric intersubunit TM binding site. (A) The intersubunit TM binding site is shown as visualized by the crystal structure of GluCl with ivermectin bound (Hibbs and Gouaux, 2011). Several hydrophobic residues (yellow sticks) protruding from the transmembrane helices M1 and M3 form the binding pocket. The positive allosteric modulator ivermectin is shown as light-blue sticks. (B) Contraction of the intersubunit TM binding site by receptor's twisting. The structure of the receptor at the end of the twisting transition was taken from simulations (Calimet et al., 2013) and is shown here in cyan after optimal superimposition of the M3 helix of the (þ) subunit. The configuration of the binding site after twisting shows that the local re-orientation of the upper part of M1 in the direction of the (þ) subunit results in a significant contraction of the intersubunit transmembrane binding pocket. The three residues, which are primarily involved in the (partial) occlusion of the allosteric site, are labeled. (C) Quaternary nature of twisting. The twisted configuration of the receptor (cyan ribbons) shows that the contraction of the allosteric binding site in the TM domain is coupled to the global movement of the EC domain. Overall, this analysis shows that the binding of ivermectin sterically hinders the local reorientation of the upper part of M1 (up to the totally conserved Proline), which stabilizes the active (untwisted) state of the receptor. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

(Calimet et al., 2013) showed that the orthosteric site does not change its conformation during receptor's twisting and keeps a closed shape; see Fig. 4 (top). In sharp contrast, blooming of the EC domain as visualized by GLIC pH7 (Sauguet et al., 2014a) produces a significant expansion of the orthosteric pocket as a result of two distinct movements: (i) a global displacement of the (þ) subunit that “opens” loop B by 1.7 Å; and (ii) a local rearrangement of loop C, which destructures and opens the orthosteric pocket by 3.2 Å; see Fig. 4 (bottom). Although the opening of loop C likely facilitates agonist binding at the orthosteric site, opening of loop B is the result of a quaternary reorganization of the receptor and may thus transfer information in space (Auerbach, 2015). Correspondingly, closure of loop B in the resting state is expected to be the structural event initiating ion-channel activation, as previously suggested (Sauguet et al., 2014a). This conclusion is also consistent with recent electrophysiological analyses showing that synthetic variants of GLIC with mutated or deleted C-loops preserve native-like responses to agonist (Gonzalez-Gutierrez et al., 2013). Also, similar manipulations of loop C in nAChR were shown to preserve diliganded channel opening though impairing orthosteric agonist binding (Purohit and Auerbach, 2013).

6.2. The allosteric Ca2þ site The precise structural location of the modulatory site for Ca2þ in nAChR is presently unknown but it can be tentatively inferred from the X-ray structure of ELIC in complex with Ba2þ (see (Zimmermann et al., 2012)). In agreement with site-directed mutagenesis and modeling data on nAChR (see above), this structure suggests that the divalent ion-binding site is located at the interface between subunits in the EC domain below the orthosteric site and involves residues of the Cys loop from the (þ) subunit and loop F from the () subunit. Close comparison of the structures of GLIC pH4 and GLIC pH7 (Sauguet et al., 2014a) shows that during activation a rotation of the outer b-sheet relative to the inner bsheet of the EC domain results in a 2 Å displacement of loop F in the

direction of the Cys loop; see Fig. 5. This tertiary change significantly reshapes the subunit interface and is likely to modulate the binding of divalent ions. Nonetheless, the peculiar conformation of loop F in ELIC along with a low conservation of nearby residues within the superfamily makes it difficult to extend this finding to other pLGICs. A closer look at the structures of GLIC, however, shows that the displacement of loop F discussed above actually produces a cavity in the resting state structure that gives access to the hydrophobic core of the b-sandwiches; see Fig. 5. Because this pocket is not formed in the active state, this region could be targeted as a ligand-binding site for the negative allosteric modulation of pLGICs. Strikingly, this region has been recently identified as the binding site for the general anesthetic Bromoform in ELIC (Spurny et al., 2013).

Box 1

The most recent structural and simulation results consistently point to a model of gating that involves two distinct quaternary transitions: a radial contraction or (un)blooming of the EC domain, which promotes the opening of the ion pore, followed by a global untwisting of the receptor to lock the channel in its active state (Fig. 3). These transitions mediate the dynamic reorganization of the interfaces between subunits, which host several orthosteric and allosteric binding sites, and are thus regulated by ligandbinding events. Elucidating the link between the functional isomerization of the receptor and the structural changes at the ligand-binding sites would offer new strategies for the design of novel modulators of nAChR and pLGICs more generally.

146

M. Cecchini, J.-P. Changeux / Neuropharmacology 96 (2015) 137e149

Fig. 7. Hypothetical mechanism of signal transduction in pLGICs. The receptor is viewed from the extracellular end of the membrane and is sketched as two concentric layers of gray (large) and white (small) circles indicating the EC and the TM subunits, respectively. Agonist molecules are shown as green bullets. The extent of global twisting, the opening of the transmembrane ion channel, and the degree of contraction/extension of the EC domain are used as signatures to assign the functional states of the receptor to known X-ray (green labels) or simulation (blue labels) structures. Black double-headed arrows indicate reversible equilibrium. Red dashed arrows sketch the molecular dynamics involved in the functional conformational transitions of the receptor. Pink solid arrows highlight the privileged dynamic pathway(s) for activation (R to A) and desensitization (A to D). Shadowing is used to illustrate the conformational flexibility of the EC domain in the R state. The washed-out pictures related to fast and slow desensitization are still speculative due to insufficiently documented structural data (see Main Text). The mechanistic interpretation is as follows. In the resting state (R), the receptor is globally twisted, the channel is closed, and the agonist does not bind the EC domain. Although the structural variability of the resting state is still under debate (i.e. it was observed in GLIC pH7 (Sauguet et al., 2014a) but not in apo GluCl (Althoff et al., 2014)), molecular dynamics studies with GluCl (Cecchini et al. unpublished) show that the enhanced flexibility of the EC domain is a signature of the R state. Neurotransmitter binding stabilizes the opening of the ion channel and is associated with a significant contraction (reverse of blooming) of the EC domain (Sauguet et al., 2014a). Counter-clockwise rotation of the EC domain relative to the TM domain, which corresponds approximately to a 10 un-twisting of the receptor, follows and locks the ion channel in an open ion-conducting or active (A) state. The latter corresponds to a globally un-twisted, agonist binding, open-channel conformation of the receptor that is most consistent with the X-ray structures of GLIC pH4 (Bocquet et al., 2009) or GluCl with ivermectin (Hibbs and Gouaux, 2011). The conformation of the “intermediate state on activation” (MD state), i.e. a short lived conformation that follows the contraction of the EC domain but precedes global un-twisting of the receptor, was captured by the simulation of GluCl with ivermectin removed (Calimet et al., 2013). Upon exposure to steady levels of agonist, conformational changes mediating both fast and slow desensitization occur promoting the transition to closed-channel forms that are structurally distinct from the R state. The un-coupling of the M2 helix from the rest of the TM domain is the structural event initiating the transition to the fast-desensitized state (I) with a shut ion pore in the presence of agonist. This structure is best visualized by the LC state of GLIC (Prevost et al., 2012). Clock-wise twisting of the receptor along with recoupling of the M2eM3 helices in the TM domain is proposed to mediate slow desensitization (D) and is tentatively assigned to the conformation of ELIC (Hilf and Dutzler, 2008). Finally, blooming of the EC domain coupled to agonist unbinding drives the receptor back to the R state.1 (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

M. Cecchini, J.-P. Changeux / Neuropharmacology 96 (2015) 137e149

6.3. The allosteric inter-subunit TM site The crystal structure of GluCl in complex with ivermectin (Hibbs and Gouaux, 2011) provided an atomistic description of the intersubunit transmembrane site in pLGICs for a positive allosteric modulator; see Fig. 6A. Ivermectin binding to this site is stabilized by two hydrogen-bonding interactions, which involve the side chain of S260 on M2 (þ) and the backbone of L218 on M1 (), in addition to extensive hydrophobic contacts with residues protruding from both subunits. Molecular Dynamics of GluCl with ivermectin removed showed that this transmembrane site shrinks substantially by receptor's twisting (Calimet et al., 2013). In fact, the comparison of this binding pocket before and after the twisting isomerization reveals that the large-amplitude rotation of the EC domain is allosterically coupled to a more local reorientation of helix M1 () in the TM domain, which moves in the direction of the (þ) subunit; see Fig. 6B. Because no twisting was observed with ivermectin bound in simulation (Calimet et al., 2013), we conclude that ivermectin binding blocks the twisting isomerization by preventing the local reorientation of the M1 helix at the intersubunit transmembrane site. Because receptor twisting controls ion channel's deactivation (see Box 1), binding of bulky ligands at this transmembrane site is expected to allosterically stabilize the active state of the receptor; see Fig. 6C. This conclusion is fully supported by the X-ray structure of GluCl in complex with POPC molecules, where lipid binding to this site elicits the transition to the active state (Althoff et al., 2014). The above observations consistently point to a model of gating composed of two distinct quaternary transitions with profoundly different (structural) consequences at the interface between subunits (Taly et al., 2014; Calimet et al., 2013). As illustrated in Fig. 7, the radial contraction of the EC domain in the R state would be the first step on activation, which results into a structural modulation of both the orthosteric neurotransmitter site and the allosteric Ca2þ regulatory site with little or no change in the TM domain. Global untwisting would then follow to lock the ion pore in the open form and stabilize an ion conducting or A state of the receptor. This second transition occurs with important restructuring of the interface between subunits in the TM domain and thus of the allosteric modulatory site present there, while leaving the orthosteric site essentially unaltered. In this framework, orthosteric agonist binding is expected to regulate the contraction/expansion (blooming) of the EC domain, which is rate determining on activation (kon), whereas allosteric modulators binding to the TM domain would control the twisting isomerization, which is rate determining for closing (koff). Also, the present interpretation of gating postulates two principal modes of regulation by allosteric modulators binding: one at the level of the EC domain by e.g. Ca2þ and benzodiazepines, the other at the level of the TM domain by e.g. ivermectin and lipids. We note that alternative models to the hypothetical gating mechanism presented here exist in the literature. They include among others the

1 Although the present model is at this point rather speculative, a few distinctive features emerge that could be tested experimentally or computationally. (1) The uncoupling of the M2 and M3 helices is predicted to be an early event on desensitization. Engineering covalent links between these helices (i.e. disulphide bridges, metal ion binding sites, etc.) should prevent or significantly slow down receptor desensitization. (2) Receptor twisting after activation is proposed to mediate slow desensitization. FRET experiments where the donor and acceptor fluorophores are positioned in such a way that the energy transfer between them is efficient in the un-twisted states (A, I) and negligible upon twisting (R, D) should allow to distinguish between fast (I) and slow (D) desensitized states. (3) In the structure of apo GluCl and microsecond simulations of GluCl active state with ivermectin removed (Cecchini et al., unpublished), the pore-lining helices M2 shut the ion pore with no tangential reorientation. Simulations analyses of apo GluCl, GLIC pH7 and ELIC could be used to shed light on the role of M2 tangential tilting in the recovery of the resting (R) state after slow desensitization (D).

147

hydrophobic (dewetting-driven) mechanism of gating, which was observed in ion channels with highly hydrophobic transmembrane pores (rev (Aryal et al., 2014)). 7. Desensitization and other conformational states The prolonged exposure (fraction of a second to minutes) of nAChR to high (approximately micromolar) levels of the agonist nicotine was shown already by Langley in 1905 to elicit a timedependent decrease of ion conductance following activation; a phenomenon that was referred to as desensitization (Katz and Thesleff, 1957). Desensitization in nAChR, and more generally in LGICs, modulate the frequency of the ion-channel conducting states and has been suggested to play an important role in shaping neural networks associated with memory and learning (Heidmann and Changeux, 1982; Dehaene and Changeux, 1991). Not surprisingly altered desensitization mechanisms correlate with a number of disease conditions including autosomal dominant nocturnal frontal lobe epilepsy or some congenital myasthenic syndrome, and are implicated in nicotine addiction and lung cancer (Ochoa et al., 1989). In vivo, single-channel electrophysiology of nAChR at micromolar concentrations of acetylcholine demonstrated that receptor desensitization has two distinct kinetic components that correspond to: (i) the transition to a fast desensitized state, which interconverts with the active form on the hundreds of milliseconds; followed by (ii) the transformation to a slow desensitized state, from which recovery may take up to seconds (Sakmann et al., 1980). In vitro, kinetic analyses of cholinergic agonist binding by rapid mixing techniques unambiguously showed that the biphasic nature of the phenomenon originates from distinct conformational transitions of the receptor, which exists under multiple desensitized states all sharing a closed ion pore (Heidmann et al., 1983). These observations have been interpreted in terms of a minimum four-state “allosteric” model (Heidmann and Changeux,1980), which is re-used in Fig. 7. Because the fast (I) and the slow (D) desensitized states have increasing affinity for the neurotransmitter, desensitization proceeds deterministically at high levels of agonist, slowly shifting the equilibrium from the active to the slow desensitized state. This interpretation is fully consistent with the observed modulation of the desensitization rate by mutagenesis (Bouzat et al., 2008) or binding of positive allosteric modulators (Donnelly-Roberts et al., 2011). Given the functional role of desensitization, regulation of these conformational transitions by ligand binding, in particular by allosteric modulators, may open up to unprecedented pharmacological strategies (Arias, 1998). Yet, structural studies based on Xray crystallography and electron microscopy have so far failed in capturing the desensitized state(s) of the receptor, for reasons that are not understood. In fact, although the time required for crystallization in the presence of agonist would be more compatible with an ion channel populating a D state, several crystal structures of the A state were determined for both prokaryotic (Hilf and Dutzler, 2009; Bocquet et al., 2009; Nury et al., 2011) and invertebrate eukaryotic (Hibbs and Gouaux, 2011) receptors, whereas only one structure was so far assigned to the D state (Miller and Aricescu, 2014). This structure captures a non-conducting state of the GABAA b3 receptor with a closed gate (at residues -20 ) produced by the inward tapering of the M2 helices down to the intracellular border. Because the ion channel was crystallized in complex with the orthosteric agonist benzamidine, this structure was interpreted as a desensitized state as much as was done for the structures of ELIC and the LC state of GLIC (see above). Nonetheless, these structures differ substantially the one from the other both at the tertiary and the quaternary level. Because the 3D conformation of the GABAA receptor and the LC state of GLIC are close to the active state, one may speculate that they capture the fast-desensitized state (Prevost

148

M. Cecchini, J.-P. Changeux / Neuropharmacology 96 (2015) 137e149

et al., 2013), which would eventually evolve to a configuration more similar to ELIC during slow desensitization. A speculative interpretation of desensitization in pLGICs based on the recent X-ray structural determinations is given in Fig. 7. Besides X-ray crystallography, biochemical studies based on time-resolved photo labeling experiments have provided complementary information on the conformational transition(s) of the TM domain during desensitization. In particular, using the Torpedo nAChR it was found that two regions of the TM domain, one involving a cluster of four residues at the interface between dM1 and dM2 and one in the channel lumen at the conserved 9’-leucines on M2, undergo significant changes during activation and desensitization (Yamodo et al., 2010). The spatio-temporal pattern of the structural changes elicited by agonist binding shows that the hydrophobic pocket located at the extracellular end of d-subunit, which first appears during activation, is preserved until slow desensitization. By contrast, significant changes in the channel lumen were detected during fast desensitization, with only modest rearrangement on moving to the slow desensitized state (Yamodo et al., 2010). Provided that the structural interpretation of these experiments is not affected by misalignments in the cryo EM reconstruction of the Torpedo nAChR (Taly et al., 2014), these results would be consistent with the two-gate model by Auerbach and Akk (Auerbach and Akk, 1998). In this model, the “desensitization gate”, which sits in the middle of the ion pore, is “locally closed” during fast desensitization, whereas the “activation gate”, which is located at the extracellular end, would be shut only when slow desensitization occurs. Similar conclusions were drawn from physiological analyses of the GLIC channel using site-directed spin labeling and EPR resonance measurements (Velisetty et al., 2012). Based on these observations, we speculate that the tangential reorientation of the porelining M2 helices, solely captured by the structure of ELIC, would shut the activated gate and thus become a signature of the slowly desensitized state. Recent double electroneelectron resonance (DEER) experiments with functional GLIC channels reconstituted into liposomes have provided information on the rearrangement of the interfacial loops (i.e. the b1eb2 loop, the M2eM3 loop and loop F) in response to proton-elicited desensitization. Although it is not clear which desensitized state was probed in these experiments e fast or slow e the distance measurements indicated that the X-ray structure of ELIC is not an appropriate model for the resting state and that the X-ray structure of GLIC pH4 does not correspond to a desensitized state (Dellisanti et al., 2013). Based on the presently available structural and functional results, providing a detailed structural interpretation for desensitization in pLGICs at the atomic level appears to be premature. Crystallization of both fast (I) and slow (D) states for the same receptor will be useful to resolve the issue. 8. Conclusion We have reviewed the most recent advances in the structure, conformational transitions and allosteric modulation of nAChR and its pentameric homologues. The introduction of a new temporal dimension in the description of gating by time-resolved techniques such as single-channel electrophysiology, site-directed photo labeling and Molecular Dynamics has recently opened to an unprecedented view of signal transduction at the atomic level (Fig. 7). Most recent structural, functional, and computational analyses consistently point to a model of gating that involves two distinct quaternary transitions of the neurotransmitter receptor e the blooming of the EC domain and its global twisting e with profoundly different consequences at the interface between subunits. Analysis of the dynamics of the allosteric modulatory sites during ion-

channel opening/closing together with electrophysiological data on the influence of agonist binding on gating consistently point to a two-step asymmetric mechanism in which the rate-determining events on activation (channel opening) and deactivation (channel closing) are structurally distinct. The model of gating presented here offers a molecular interpretation for the allosteric modulation of nAChR and homologues by known orthosteric ligands and allosteric modulators and predicts that lipids like POPC and cholesterol may work as endogenous allosteric modulators in eukaryotic pLGICs by regulating the twisting isomerization. The detailed characterization of the two quaternary components of the gating conformational transition along with the increasing availability of high-resolution structures of pLGICs in complex with ligands will soon offer the opportunity to explore agonistic activity at various modulatory sites by computer simulations. The present model of gating was inferred from the high-resolution structures and dynamics of homo-pentameric homologues of nAChR. As the hetero-pentameric organization is likely to introduce asymmetries in the signal transduction mechanism and may modify (or not) the nature of the allosteric transition (Unwin, 2013), the development of a general model for nAChR's function requires highresolution structures of heteropentameric receptor channels, which are presently not available. The same is true for the physiological role of the intracellular domain, whose implication in gating has just been suggested (Hassaine et al., 2014). These advances demonstrate that the synergy between structural, functional and computational studies is a promising approach to elucidate the chemical basis of allosteric modulation in pLGICs and ultimately guide the design of new active compounds. Acknowledgments This work was supported by the Agence Nationale de la Recherche (ANR) through a LabEx CSC grant to M.C. (Project ID: CSC-MCE-13) and the International Center for Frontier Research in Chemistry (icFRC). J.P.C. is grateful to Dr Ralph Greenspan and the Kavli Institute for Brain & Mind, University of California San Diego. References Addona, G.H., Sandermann Jr., H., Kloczewiak, M.A., Husain, S.S., Miller, K.W., 1998. Biochim. biophys. acta 1370, 299. Althoff, T., Hibbs, R.E., Banerjee, S., Gouaux, E., 2014. Nature 512, 333. Arias, H.R., 1998. Biochim. biophys. acta 1376, 173. Aryal, P., Sansom, M.S., Tucker, S.J., 2014. J. Mol. Biol. Auerbach, A., 2007. J. Gen. Physiol. 130, 543. Auerbach, A., 2013. J. Mol. Biol. 425, 1461. Auerbach, A., Akk, G., 1998. J. Gen. Physiol. 112, 181. Auerbach, A., 2015. Agonist activation of a nicotinic acetylcholine receptor. Neuropharmacology 96, 150e156. Barnard, E.A., 1995. Adv. Biochem. Psychopharmacol. 48, 1. Bertrand, D., Gopalakrishnan, M., 2007. Biochem. Pharmacol. 74, 1155. Bocquet, N., Prado de Carvalho, L., Cartaud, J., Neyton, J., Le Poupon, C., Taly, A., Grutter, T., Changeux, J.P., Corringer, P.J., 2007. Nature 445, 116. Bocquet, N., Nury, H., Baaden, M., Le Poupon, C., Changeux, J.P., Delarue, M., Corringer, P.J., 2009. Nature 457, 111. Bourne, Y., Talley, T.T., Hansen, S.B., Taylor, P., Marchot, P., 2005. EMBO J. 24, 1512. Bouzat, C., Bartos, M., Corradi, J., Sine, S.M., 2008. J. Neurosci. Off. J. Soc. Neurosci. 28, 7808. Brannigan, G., Henin, J., Law, R., Eckenhoff, R., Klein, M.L., 2008. Proc. Natl. Acad. Sci. U. S. A. 105, 14418. Brejc, K., van Dijk, W.J., Klaassen, R.V., Schuurmans, M., van Der Oost, J., Smit, A.B., Sixma, T.K., 2001. Nature 411, 269. Brisson, A., Unwin, P.N., 1985. Nature 315, 474. Calimet, N., Simoes, M., Changeux, J.P., Karplus, M., Taly, A., Cecchini, M., 2013. Proc. Natl. Acad. Sci. U. S. A. 110, E3987.  de Paris, 1964. Changeux, J. P., Universite Changeux, J.P., 1966. Mol. Pharmacol. 2, 369. Changeux, J.P., 1981. Harvey Lect. 75, 85. Changeux, J.P., 1990. Trends Pharmacol. Sci. 11, 485. Changeux, J.P., 2012. J. Biol. Chem. 287, 40207. Changeux, J.P., 2013a. Nat. Rev. Mol. Cell. Biol. 14, 819.

M. Cecchini, J.-P. Changeux / Neuropharmacology 96 (2015) 137e149 Changeux, J.P., 2013b. Drug Discov. Today Technol. 10, e223. Changeux, J.P., Edelstein, S.J., 1998. Neuron 21, 959. Changeux, J.P., Edelstein, S.J., 2005. Science 308, 1424. Changeux, J.P., Taly, A., 2008. Trends Mol. Med. 14, 93. Changeux, J.P., Thiery, J., Tung, Y., Kittel, C., 1967. Proc. Natl. Acad. Sci. U. S. A. 57, 335. Changeux, J.P., Gautron, J., Israel, M., Podleski, T., 1969. C. R. Acad. Sci. Hebd. Seances Acad. Sci. D. 269, 1788. Changeux, J.P., Kasai, M., Lee, C.Y., 1970. Proc. Natl. Acad. Sci. U. S. A. 67, 1241. Cheng, X., Lu, B., Grant, B., Law, R.J., McCammon, J.A., 2006. J. Mol. Biol. 355, 310. Chiara, D.C., Jayakar, S.S., Zhou, X., Zhang, X., Savechenkov, P.Y., Bruzik, K.S., Miller, K.W., Cohen, J.B., 2013. J. Biol. Chem. 288, 19343. Cooper, S.T., Harkness, P.C., Baker, E.R., Millar, N.S., 1999. J. Biol. Chem. 274, 27145. Corringer, P.J., Le Novere, N., Changeux, J.P., 2000. Annu. Rev. Pharmacol. Toxicol. 40, 431. Corringer, P.J., Baaden, M., Bocquet, N., Delarue, M., Dufresne, V., Nury, H., Prevost, M., Van Renterghem, C., 2010. J. Physiol. 588, 565. Corringer, P.J., Poitevin, F., Prevost, M.S., Sauguet, L., Delarue, M., Changeux, J.P., 2012. Structure 20, 941. Cui, Q., Karplus, M., 2008. Protein Sci. Publ. Protein Soc. 17, 1295. daCosta, C.J., Baenziger, J.E., 2009. J. Biol. Chem. 284, 17819. daCosta, C.J., Dey, L., Therien, J.P., Baenziger, J.E., 2013. Nat. Chem. Biol. 9, 701. Dehaene, S., Changeux, J.P., 1991. Cereb. Cortex 1, 62. Dellisanti, C.D., Ghosh, B., Hanson, S.M., Raspanti, J.M., Grant, V.A., Diarra, G.M., Schuh, A.M., Satyshur, K., Klug, C.S., Czajkowski, C., 2013. PLoS Biol. 11, e1001714. Donnelly-Roberts, D., Bertrand, D., Gopalakrishnan, M., 2011. Biochem. Pharmacol. 82, 797. Duret, G., Van Renterghem, C., Weng, Y., Prevost, M., Moraga-Cid, G., Huon, C., Sonner, J.M., Corringer, P.J., 2011. Proc. Natl. Acad. Sci. U. S. A. 108, 12143. Dutertre, S., Becker, C.M., Betz, H., 2012. J. Biol. Chem. 287, 40216. Eisele, J.L., Bertrand, S., Galzi, J.L., Devillers-Thiery, A., Changeux, J.P., Bertrand, D., 1993. Nature 366, 479. Forman, S.A., Chiara, D.C., Miller, K.W., 2015. Anesthetics target interfacial transmembrane sites in nicotinic acetylcholine receptors. Neuropharmacology 96, 169e177. Galzi, J.-L., Changeux, J.-P., 1994. Curr. Opin. Cell. Biol. 4, 554. Galzi, J.L., Bertrand, S., Corringer, P.J., Changeux, J.P., Bertrand, D., 1996. EMBO J. 15, 5824. Giraudat, J., Dennis, M., Heidmann, T., Chang, J.Y., Changeux, J.P., 1986. Proc. Natl. Acad. Sci. U. S. A. 83, 2719. Giraudat, J., Dennis, M., Heidmann, T., Haumont, P.Y., Lederer, F., Changeux, J.P., 1987. Biochemistry 26, 2410. Gonzalez-Gutierrez, G., Grosman, C., 2010. J. Mol. Biol. 403, 693. Gonzalez-Gutierrez, G., Cuello, L.G., Nair, S.K., Grosman, C., 2013. Proc. Natl. Acad. Sci. U. S. A. 110, 18716. Grosman, C., Zhou, M., Auerbach, A., 2000. Nature 403, 773. Grutter, T., de Carvalho, L.P., Dufresne, V., Taly, A., Edelstein, S.J., Changeux, J.P., 2005. Proc. Natl. Acad. Sci. U. S. A. 102, 18207. Hamouda, A.K., Kimm, T., Cohen, J.B., 2013. J. Neurosci. : Off. J. Soc. Neurosci. 33, 485. Hansen, S.B., Taylor, P., 2007. J. Mol. Biol. 369, 895. Hassaine, G., Deluz, C., Grasso, L., Wyss, R., Tol, M.B., Hovius, R., Graff, A., Stahlberg, H., Tomizaki, T., Desmyter, A., Moreau, C., Li, X.D., Poitevin, F., Vogel, H., Nury, H., 2014. Nature 512, 276. Heidmann, T., Changeux, J.P., 1980. Biochem. Biophys. Res. Commun. 97, 889. Heidmann, T., Sobel, A., Popot, J.L., Changeux, J.P., 1980 Sep. Reconstitution of a functional acetylcholine receptor. Conservation of the conformational and allosteric transitions and recovery of the permeability response; role of lipids. Eur. J. Biochem 110 (1), 35e55. Heidmann, T., Changeux, J.P., 1982. Comptes rendus seances l'Acad. sci. Ser. III, Sci. la vie 295, 665. Heidmann, T., Bernhardt, J., Neumann, E., Changeux, J.P., 1983. Biochemistry 22, 5452. Henin, J., Salari, R., Murlidaran, S., Brannigan, G., 2014. Biophys. J. 106, 1938. Hibbs, R.E., Gouaux, E., 2011. Nature 474, 54. Hilf, R.J., Dutzler, R., 2008. Nature 452, 375. Hilf, R.J., Dutzler, R., 2009. Nature 457, 115. Hucho, F., Oberthur, W., Lottspeich, F., 1986. FEBS Lett. 205, 137. Imoto, K., Methfessel, C., Sakmann, B., Mishina, M., Mori, Y., Konno, T., Fukuda, K., Kurasaki, M., Bujo, H., Fujita, Y., et al., 1986. Nature 324, 670. Imoto, K., Busch, C., Sakmann, B., Mishina, M., Konno, T., Nakai, J., Bujo, H., Mori, Y., Fukuda, K., Numa, S., 1988. Nature 335, 645. Jackson, M.B., 1986. Biophys. J. 49, 663. Jadey, S.V., Purohit, P., Bruhova, I., Gregg, T.M., Auerbach, A., 2011. Proc. Natl. Acad. Sci. U. S. A. 108, 4328. Jones, O.T., McNamee, M.G., 1988. Biochemistry 27, 2364. Kaczanowska, K., Harel, M., Radic, Z., Changeux, J.P., Finn, M.G., Taylor, P., 2014. Structural basis for cooperative interactions of substituted aminopyrimidines with the acetylcholine binding protein. Proc. Natl. Acad. Sci. 111 (29), 10749e10754. http://dx.doi.org/10.1073/pnas.14109921. Karlin, A., 2002. Nat. Rev. Neurosci. 3, 102. Karplus, M., McCammon, J.A., 2002. Nat. Struct. Biol. 9, 646. Katz, B., Thesleff, S., 1957. J. Physiol. 138, 63. Krause, R.M., Buisson, B., Bertrand, S., Corringer, P.J., Galzi, J.L., Changeux, J.P., Bertrand, D., 1998. Mol. Pharmacol. 53, 283. Lape, R., Colquhoun, D., Sivilotti, L.G., 2008. Nature 454, 722. Lape, R., Plested, A.J., Moroni, M., Colquhoun, D., Sivilotti, L.G., 2012. J. Neurosci. : Off. J. Soc. Neurosci. 32, 1336. Lee, W.Y., Free, C.R., Sine, S.M., 2008. J. Gen. Physiol. 132, 265. Leonard, R.J., Labarca, C.G., Charnet, P., Davidson, N., Lester, H.A.,1988. Science 242, 1578.

149

Le Novere, N., Grutter, T., Changeux, J.P., 2002. Proc. Natl. Acad. Sci. U. S. A. 99, 3210. Li, G.D., Chiara, D.C., Sawyer, G.W., Husain, S.S., Olsen, R.W., Cohen, J.B., 2006. J. Neurosci. Off. J. Soc. Neurosci. 26, 11599. Li, G.D., Chiara, D.C., Cohen, J.B., Olsen, R.W., 2010. J. Biol. Chem. 285, 8615. Marsh, D., Barrantes, F.J., 1978. Proc. Natl. Acad. Sci. U. S. A. 75, 4329. Miledi, R., Molinoff, P., Potter, L.T., 1971. Nature 229, 554. Miller, P.S., Aricescu, A.R., 2014. Nature 512, 270e275. http://dx.doi.org/10.1038/ nature13293. Miller, P.S., Topf, M., Smart, T.G., 2008. Nat. Struct. Mol. Biol. 15, 1084. Monod, J., Wyman, J., Changeux, J.P., 1965. J. Mol. Biol. 12, 88. Mulle, C., Lena, C., Changeux, J.P., 1992. Neuron 8, 937. Neher, E., 1983. J. Physiol. 339, 663. Nemecz, A., Taylor, P., 2011. J. Biol. Chem. 286, 42555. Newell, J.G., McDevitt, R.A., Czajkowski, C., 2004. J. Neurosci. : Off. J. Soc. Neurosci. 24, 11226. Nury, H., Poitevin, F., Van Renterghem, C., Changeux, J.P., Corringer, P.J., Delarue, M., Baaden, M., 2010. Proc. Natl. Acad. Sci. U. S. A. 107, 6275. Nury, H., Van Renterghem, C., Weng, Y., Tran, A., Baaden, M., Dufresne, V., Changeux, J.P., Sonner, J.M., Delarue, M., Corringer, P.J., 2011. Nature 469, 428. Ochoa, E.L., Dalziel, A.W., McNamee, M.G., 1983. Biochim. biophys. acta 727, 151. Ochoa, E.L., Chattopadhyay, A., McNamee, M.G., 1989. Cell. Mol. Neurobiol. 9, 141. Olsen, R.W., Sieghart, W., 2009. Neuropharmacology 56, 141. Olsen, R.W., Li, G.-D., Wallner, M., Trudell, J.R., Bertaccini, E.J., Lindahl, E., Miller, K.W., Alkana, R.L., Davies, D.L., 2014. Alcohol. Clin. Exp. Res. 38, 595. Pan, J., Chen, Q., Willenbring, D., Mowrey, D., Kong, X.P., Cohen, A., Divito, C.B., Xu, Y., Tang, P., 2012. Structure 20, 1463. Prevost, M.S., Sauguet, L., Nury, H., Van Renterghem, C., Huon, C., Poitevin, F., Baaden, M., Delarue, M., Corringer, P.J., 2012. Nat. Struct. Mol. Biol. 19, 642. Prevost, M.S., Moraga-Cid, G., Van Renterghem, C., Edelstein, S.J., Changeux, J.P., Corringer, P.J., 2013a. Proc. Natl. Acad. Sci. U. S. A. 110, 17113. Prevost, M.S., Delarue-Cochin, S., Marteaux, J., Colas, C., Van Renterghem, C., Blondel, A., Malliavin, T., Corringer, P.J., Joseph, D., 2013b. J. Med. Chem. 56, 4619. Purohit, P., Auerbach, A., 2013. J. Gen. Physiol. 141, 467. Purohit, P., Mitra, A., Auerbach, A., 2007. Nature 446, 930. Purohit, P., Bruhova, I., Auerbach, A., 2012. Proc. Natl. Acad. Sci. U. S. A. 109, 9384. Purohit, P., Gupta, S., Jadey, S., Auerbach, A., 2013. Nat. Commun. 4, 2984. Rendon, G., Kantorovitz, M.R., Tilson, J.L., Jakobsson, E., 2011. Channels (Austin) 5, 325. Sakmann, B., Patlak, J., Neher, E., 1980. Nature 286, 71. Sauguet, L., Poitevin, F., Murail, S., Van Renterghem, C., Moraga-Cid, G., Malherbe, L., Thompson, A.W., Koehl, P., Corringer, P.J., Baaden, M., Delarue, M., 2013a. EMBO J. 32, 728. Sauguet, L., Howard, R.J., Malherbe, L., Lee, U.S., Corringer, P.J., Harris, R.A., Delarue, M., 2013b. Nat. Commun. 4, 1697. Sauguet, L., Shahsavar, A., Poitevin, F., Huon, C., Menny, A., Nemecz, A., Haouz, A., Changeux, J.P., Corringer, P.J., Delarue, M., 2014. Proc. Natl. Acad. Sci. U. S. A.111, 966. Sauguet, L., Shahsavar, A., Delarue, M., 2014. Biochim. biophys. acta. Sawyer, G.W., Chiara, D.C., Olsen, R.W., Cohen, J.B., 2002. J. Biol. Chem. 277, 50036. Smit, A.B., Syed, N.I., Schaap, D., van Minnen, J., Klumperman, J., Kits, K.S., Lodder, H., van der Schors, R.C., van Elk, R., Sorgedrager, B., Brejc, K., Sixma, T.K., Geraerts, W.P., 2001. Nature 411, 261. Smith, G.B., Olsen, R.W., 2000. Neuropharmacology 39, 55. Sooksawate, T., Simmonds, M.A., 2001. Neuropharmacology 40, 178. Spurny, R., Ramerstorfer, J., Price, K., Brams, M., Ernst, M., Nury, H., Verheij, M., Legrand, P., Bertrand, D., Bertrand, S., Dougherty, D.A., de Esch, I.J., Corringer, P.J., Sieghart, W., Lummis, S.C., Ulens, C., 2012. Proc. Natl. Acad. Sci. U. S. A. 109, E3028. Spurny, R., Billen, B., Howard, R.J., Brams, M., Debaveye, S., Price, K.L., Weston, D.A., Strelkov, S.V., Tytgat, J., Bertrand, S., Bertrand, D., Lummis, S.C., Ulens, C., 2013. J. Biol. Chem. 288, 8355. Taly, A., Delarue, M., Grutter, T., Nilges, M., Le Novere, N., Corringer, P.J., Changeux, J.P., 2005. Biophys. J. 88, 3954. Taly, A., Corringer, P.J., Guedin, D., Lestage, P., Changeux, J.P., 2009. Nat. Rev. Drug Discov. 8, 733. Taly, A., Henin, J., Changeux, J.P., Cecchini, M., 2014. Allosteric regulation of pentameric ligand-gated ion channels: an emerging mechanistic perspective. Channels 8 (4), 350e360. http://dx.doi.org/10.4161/chan.29444. Tasneem, A., Iyer, L.M., Jakobsson, E., Aravind, L., 2005. Genome Biol. 6, R4. Thompson, A.J., Lester, H.A., Lummis, S.C., 2010. Q. Rev. Biophys. 43, 449. Unwin, N., 2005. J. Mol. Biol. 346, 967. Unwin, N., 2013. Q. Rev. Biophys. 46, 283. Velisetty, P., Chalamalasetti, S.V., Chakrapani, S., 2012. J. Biol. Chem. 287, 36864. Vernino, S., Amador, M., Luetje, C.W., Patrick, J., Dani, J.A., 1992. Neuron 8, 127. Wang, J., Kuryatov, A., Lindstrom, J., 2015. Expression of cloned a6* nicotinic acetylcholine receptors. Neuorpharmacology 96, 194e204. Weng, Y., Yang, L., Corringer, P.J., Sonner, J.M., 2010. Anesth. Analg. 110, 59. Yamodo, I.H., Chiara, D.C., Cohen, J.B., Miller, K.W., 2010. Biochemistry 49, 156. Yan, D., Schulte, M.K., Bloom, K.E., White, M.M., 1999. J. Biol. Chem. 274, 5537. Zhong, W., Gallivan, J.P., Zhang, Y., Li, L., Lester, H.A., Dougherty, D.A., 1998. Proc. Natl. Acad. Sci. U. S. A. 95, 12088. Zhou, Y., Morais-Cabral, J.H., Kaufman, A., MacKinnon, R., 2001. Nature 414, 43. Zimmermann, I., Dutzler, R., 2011. PLoS Biol. 9, e1001101. Zimmermann, I., Marabelli, A., Bertozzi, C., Sivilotti, L.G., Dutzler, R., 2012. PLoS Biol. 10, e1001429. Zoli, M., Pistillo, F., Gotti, C., 2015. Diversity of native nicotinic receptor subtypes in mammalian brain. Neuropharmacology 96, 302e311.