DOI: 10.1111/dom.12959

REVIEW ARTICLE

Allosteric modulation as a unifying mechanism for receptor function and regulation†

Jean-Pierre Changeux1 | Arthur Christopoulos2

1Collège de France and CNRS URA 2182, Institut Pasteur, 75015 Paris, France Four major receptor families enable cells to respond to chemical and physical signals from their 2Drug Discovery Biology and Department of proximal environment. The ligand- and voltage-gated ion channels, G-protein-coupled receptors, Pharmacology, Monash Institute of nuclear hormone receptors and receptor tyrosine kinases are all allosteric proteins that carry Pharmaceutical Sciences, Monash University, multiple, spatially distinct, yet conformationally linked ligand-binding sites. Recent studies point VIC 3052 Parkville, Australia to common mechanisms governing the allosteric transitions of these receptors, including the Correspondence [email protected] (J.-P.C.), impact of oligomerization, pre-existing and functionally distinct conformational ensembles, [email protected] (A.C.) intrinsically disordered regions, and the occurrence of allosteric modulatory sites. Importantly, synthetic allosteric modulators are being discovered for these receptors, providing an enriched, yet challenging, landscape for novel therapeutics.

1 | INTRODUCTION spanning neurological disorders, cancers, cardiovascular, endocrino- logical, metabolic, and inflammatory diseases. One of the most important concepts in the biological sciences is the Signal transduction mediated by receptors involves, at the molec- notion of the receptor, or “receptive substance,” enunciated at the ular level, a “communication over a distance” between the activating turn of the 20th century by John Newport Langley in physiology and site and the locus of the biological response. An important conceptual Paul Ehrlich in immunology. It took, however, more than half a cen- advance was thus the proposal that the mechanism linking the acti- tury to biochemically identify and discover the structure and dynam- vating and response sites is allosteric. Typically, an allosteric interac- ics of these receptor molecules, and to critically evaluate the role tion2,3 is defined as an indirect interaction between topographically they play in the physiology of the organism, in particular the brain.1 distinct (non-overlapping) binding sites mediated by a discrete, About 2% of the human genome encodes these receptors, which reversible, conformational change in the protein structure referred to behave as the effective targets of nearly 60% of the medicines cur- as the allosteric transition (Figure 1). Moreover, many regulatory pro- rently used in clinical practice.. These essential cybernetic sensors teins function as discrete “cybernetic switches” exhibiting cooperative recognize chemical signals (such as neurotransmitters, hormones, binding interactions with and between substrates and regulatory odorants, and metabolites) and physical signals (including tempera- ligands. To account for this characteristic feature, it was proposed ture, electric fields, and physical stress), and transduce them into a that allosteric proteins are organized into symmetrical oligomers that cellular response. They include four major superfamilies of proteins, undergo, spontaneously, discrete cooperative changes of quaternary namely: the ligand-gated (LGICs) and voltage-gated ion channels structure between a minimum of two states that pre-exist in equilib- (VGICs), the G-protein-coupled receptors (GPCRs), the nuclear hor- rium to ligand binding – summarized conceptually and in terms of mone receptors (NHRs), and the receptor tyrosine kinases (RTKs). thermodynamic linkages in Figure 1B (top, middle). The signal trans- With the exception of the VGICs, which utilize physical activators duction mechanism would then operate through the selective stabili- such as voltage or temperature, the remaining canonical receptors zation of the particular state to which any ligand preferentially binds utilize chemical messengers as the initiators of signal transduction. and is referred to as the “Monod-Wyman-Changeux” (MWC) model.8 Given the fact that receptors are the preeminent mediators of inter- Subsequently, Koshland, Nemethy, and Filmer9 proposed a sequential cellular communication, they are also involved in numerous diseases induced-fit mechanism of the allosteric transition (“KNF” model), which involves a progressive change of conformation that excludes any conformational change of the protein in the absence of ligand. †Article reproduced with permission from Cell 2016; 166: 1084–1102, DOI: 10.1016/j.cell.2016.08.015. The reference citation style of the original article Although some debate remains about the two models, studies carried has been modified to match that of the present supplement. out with a large diversity of proteins have lent support to, and further

4 © 2016 Elsevier Inc. wileyonlinelibrary.com/journal/dom Diabetes Obes Metab. 2017;19(Suppl. 1):4–21. CHANGEUX AND CHRISTOPOULOS 5 extended, the MWC model, emphasizing in particular “population receptors, including the impact of oligomerization, of pre-existing and shifts” of discrete conformations within an energy landscape formal- functionally distinct conformational ensembles, a role for intrinsically ism (see10–14; Figure 1B, bottom). disordered regions, and the occurrence of allosteric modulatory sites The application of the MWC concept to membrane receptors, and their important pharmacological applications. specifically neurotransmitter receptors,15–17 considerably broadened its impact.1 From a pharmacological viewpoint18 (Figure 1A) receptors can thus be conceived as possessing three key binding loci, all of 2 | OLIGOMERIC VERSUS MONOMERIC which are linked in an allosteric manner. These are: (1) the orthosteric ORGANIZATION OF RECEPTORS site for endogenous or synthetic agonists, competitive antagonists, and physical signals; (2) the biologically active site, which includes the While receptor families share fundamental regulatory features, they channel pore for LGICs/VGICs; the G-protein-binding site for GPCRs; are structurally distinct (Figure 2). The original MWC formulation the tyrosine kinase domain of RTKs; the DNA response element for relates the cooperative behavior of regulatory proteins to their oligo- NHRs; and (3) more recently discovered allosteric modulatory sites, meric organization. Recent advances in structural biology have which are topographically distinct from the above mentioned revealed the functional architectures of many receptor families, dis- sites.18–21 In this Review, we examine recent advances in understand- closing their oligomeric organization. Yet, examples of monomeric ing the mechanisms governing the allosteric transitions of these proteins showing functionally important allosteric effects have been also highlighted,11 especially in GPCRs.22 Moreover, receptor oligo- mers are frequently made up of homologous subunits encoded by dif- ferent genes, thus introducing pseudo-symmetry in their multiple assemblies. The major subgroup of the LGICs (Figure 2A; 76 members in the human genome) first consists of the pentameric ligand gated ion channels (pLGICs),23 which include the stimulatory nicotinic acetyl-

choline receptors (nAChRs) and serotonin (5HT3) and the inhibitory

glycine, glutamate, and GABAA receptors; second, the tetrameric ionotropic glutamate receptors (NMDA, AMPA, kainate); and third, the trimeric ATP-gated P2X receptors and acid-sensing channels.24

FIGURE 1 All receptor superfamilies possess spatially distinct, but conformationally linked, binding sites. (A) Left: original models proposed for the feedback inhibition of L-Threonine deaminase by L-Isoleucine, including one of direct “steric” inhibition (model 1) or indirect “allosteric”, inhibition (model 2) from Changeux.2 Right: schematic summarizing the key conformationally linked sites common to all receptor superfamilies, namely, the “orthosteric site,” which recognizes the endogenous agonist or physical activator, the “biologically active site,” which propagates the initial stimulus imparted by the orthosteric ligand and one (or more) topographically distinct “allosteric sites” that modulate the interactive properties of the orthosteric and/or active sites via discrete conformational changes. (B) Schematic representations of the original Monod- Wyman-Changeux concerted model of allostery in terms of a minimum of two discrete, pre-existing states in equilibrium, “relaxed” (R) and “tense” (T), that can be differentially selected by various ligands. The three classes of molecules in the original conceptual model (top) were “activator” (A), “substrate” (S), and “inhibitor” (I) (from4). The cubic model underneath re-casts the MWC model in terms of thermodynamic equilibrium linkages comprising

unconditional isomerization (L) or dissociation constants (KA,KS), and cooperativity factors (α, β, γ, δ), which yield conditional dissociation constants and are direct thermodynamic measures of the allosteric linkage governing the different transitions.5,6 The graph on the bottom represents the MWC conformational selection model in terms of energy landscapes, both for a monomeric protein or a dimeric protein. Subscript numbers correspond to the number of ligand molecules bound. Arrows in gray indicate a hypothetical pathway for a Koshland-Nemethy-Filmer (KNF) induced fit model, including an intermediate (I) state for the dimer. Adapted with permission from Changeux and Edelstein.7 6 CHANGEUX AND CHRISTOPOULOS

Although some LGICs exist as homomeric assemblies, the majority absent in prokaryotic pLGICs. Complete atomic structures of homo- are hetero-oligomers comprising different subunits that share a com- meric pLGICs have been obtained for two prokaryotic receptors from mon architecture. Gloeobacter violaceus (GLIC)25 and Erwinia chrysanthemi (ELIC),26 and All pLGIC subunits consist of a large hydrophilic amino-terminal for four eukaryotic receptors: GluCl from Caenorhabditis elegans,27 28 extracellular domain (ECD), a transmembrane (TM) domain (TMD) the 5HT3 receptor from mouse, a human homopentameric β3 29 30 comprising four hydrophobic segments (M1-M4), and a variable GABAA receptor, and the glycine receptor from zebrafish. Overall, hydrophilic cytoplasmic or intracellular (IC) domain (ICD), which is the available structures of prokaryotic and eukaryotic pLGICs reveal strikingly similar 3D oligomeric arrangements with a 5-fold axis of symmetry perpendicular to the membrane plane. Available structural data from the ACh binding protein (AChBP), GLIC, ELIC, GluCl, and

GABAA validate the location of the orthosteric binding site in the ECD at the interface between subunits from three regions of a “prin- cipal” subunit (loops A, B, and C) and four from a “complementary” subunit (loops D, E, F, and G).23,31,32 The lies along the central axis of symmetry and is lined by the TM2 alpha helix.23,31 The

IC domain structure has been resolved in the case of the 5HT3 recep- tor. It consists of a bundle of five IC helices creating a closed vesti- bule where lateral portals are obstructed by loops.28 Orthosteric sites and channel pores are distant from each other with about 40 Å between different orthosteric sites on the same pLGIC, and about 60 Å between the orthosteric site and channel pore,33 thus making an “allosteric organization” that is remarkably preserved from prokar- yotes to eukaryotes. Structural analysis of rat homooligomeric GluA2 AMPA receptors (AMPARs) by X-ray34–36 and cryo-EM analyses37 revealed a tetra- meric organization with symmetry properties. The subunits form a string of modular domains, beginning with the amino-terminal domain (ATD), the ligand-binding domain (LBD), which binds the neurotrans- mitter glutamate, and the pore-forming TMD. The ATDs and LBDs are organized as “dimers of dimers”; the TMD exhibits ~4-fold

FIGURE 2 Schematic representation of the major receptor superfamilies. Panels on the left show the proposed quaternary arrangement of the different receptors, whereas the panels on the right highlight the key secondary features of the regions delineated by the dashed squares. (A) LGICs, which are composed of multiple subunits containing hydrophilic N-terminal regions and multiple hydrophobic M1-M4 TMD; the pore-forming M2 region is highlighted in red, and approximate location of the ECD and ICD is also indicated. (B) VGICs, which are comprised of a large, principal, α subunit and accessory β subunits. The α subunit is comprised of four domains, each consisting of six linked transmembrane segments (1–6), with the first four forming the voltage sensor, whereas segments 5 and 6 and the associated “P loop” contribute to the pore. (C) GPCRs, whose minimal functional unit is comprised of an extracellular N-terminal region, seven transmembrane-spanning domains (1–7) linked by three extracellular (e1-e3) and 3 intracellular (i1-i3) loops, and an intracellular C-terminal domain. Also shown in red is a heterotrimeric G protein. (D) RTKs have a highly modular structure comprised of subunits containing a large ECD, a TMD linked to a JMD, and the catalytically active intracellular TKD (red). (E) NHRs also exhibit a modular structure containing an N-terminal activator function 1 (AF1) ligand independent-transactivation domain, a DBD, a variable hinge region and a C-terminal domain that contains both the ligand-binding domain (LBD) and the ligand-dependent transactivation domain (AF2), which co-binds co-regulators. In all instances (A-E), representative orthosteric binding domains are denoted in yellow. CHANGEUX AND CHRISTOPOULOS 7 symmetry, yielding a symmetry mismatch between the LBD and TMD The orthosteric activator for VGICs is voltage itself, which modu- layers and giving rise to two non-equivalent subunit pairs, A/C lates positive gating charges at intervals of three amino acid residues and B/D.35 in the S4 TM segment, thus defined as the “orthosteric” site on these The NMDA receptor (NMDAR)38,39 heterotetramer (two GluN1 channels.18,49,50 This distinction is not entirely clear yet for TRP chan- and GluN2 subunits) shows a similar 3D organization as the AMPAR. nels. Both the heat-activated TRPV1 and cold-activated TRPM8 An important difference lies at the level of the glutamate binding channels are voltage sensitive51; however, voltage changes on their sites in the GluN1A-GluN2B heterotetrameric structure at the inter- own are not sufficient to open the pore and require additional allo- face between GluN1-GluN2B heterodimers, where the LBD loop 10 steric contributions by other potential activators, including tempera- in GluN2AR plays a major role in negative cooperativity between the ture, noxious chemicals, and various lipids.48,52 orthosteric glycine- and glutamate-binding “co-agonist” sites.40 In The largest receptor superfamily is the GPCRs (Figure 2C), which addition, the ATD in NMDARs contributes to the control of ion chan- comprise over 800 proteins in the human genome. Interestingly, nel opening and deactivation rates and carries binding sites for allo- these are also the receptors for which the role of oligomerization in steric modulators. Also, the TMD harbors a closed-blocked ion their functionality remains controversial and are the best example of channel (39; however, see38), a pyramidal central vestibule lined by the potential existence of “allosteric monomers.” Structurally, GPCRs residues implicated in binding ion channel blockers and Mg2+, and a are characterized by the presence of seven TM-spanning α-helical 2-fold symmetric arrangement of ion channel pore loops. Despite domains connected by three intra- and three extra-cellular loops, in these receptor-specific differences, however, AMPA and NMDARs addition to possessing variable length extracellular N and intracellular share a closely related oligomeric structure. C termini53 More than half of these GPCRs are olfactory receptors,54

The recent structures of two trimeric LGICs, the zebrafish P2X4 with the remaining GPCRs further subdivided into the class A, receptor in an apo, closed, state or in complex with the agonist, rhodopsin-like receptors, the class B secretin-like receptors, the class ATP,41,42 and the chicken acid-sensing channel 1a (ASIC1a) in a non- C metabotropic glutamate-like receptors, and the class F frizzled-like conducting, desensitized state,43 or in complex with receptors.55 GPCRs are undoubtedly “allosteric machines”,5,56 since (PcTx1)44 revealed surprising similarities in the architecture and their activation mechanism involves the recognition of an extracellu- mechanisms utilized by these two distinct cation channels.24 Both lar signal in the orthosteric site, and the transmission of the stimulus channels have six TM segments (two per subunit), short N and C ter- to the intracellular regions, at least ~40 Å away, resulting in the mini, similar TM helical arrangements in the non-conducting states, canonical interaction with G proteins or other transducers, such as and extracellular β sheets that “wrap” around 3-fold axes of symme- β-arrestins.22 try.24 In the agonist (ATP)- or gate-modifier (PcTX1)-bound structures, Since the turn of the millennium, there have been abundant high the putative orthosteric site is located at interfaces between adjacent resolution GPCR structural studies, in particular of class A receptors, subunits, ~40 Å (ATP) or 50 Å (PcTX1) from the membrane bilayer. In but structures of subdomains of some class B and class C, as well as short, all LGICs possess an oligomeric structure and show symmetry of the class F “smoothened” GPCR, have also been solved.53,57 Most properties with ligand binding sites mostly at subunit interfaces. of these structures have been of inactive states, although some par- VGICs (Figure 2B; 143 members in the human genome) comprise tial or fully active states have been solved, including rhodopsin/opsin, a large, channel-forming α subunit of ~260 kDa, and one or two β the β2 adrenergic receptor and the M2 muscarinic receptor using subunits of 30–40 kDa.45,46 The structure of the α subunit of ~2,000 either G- protein-mimetic peptides or nanobodies or, in the case of + amino acid residues from prokaryotic Na VGIC shows four homolo- the β2 adrenergic receptor, the landmark solution of an agonist- gous domains with a 4-fold axis of pseudo-symmetry.47 Each contains bound, activated receptor, coupled to the nucleotide-free heterotri- 58 six TM segments (S1–S6) and an additional membrane re-entrant seg- meric Gs protein complex. Some of these structures—e.g., the ment. Segments S1–S4 form the voltage-sensing module, whereas chemokine CXCR4 receptor59 and the μ opioid receptor60—revealed segments S5, S6, and the P loop between them form the pore. contacts proposed to contribute to dimeric interfaces, whereas the In addition to extensively studied examples of VGICs such as majority of other structures have not been interpreted in terms of those of the , sodium, and families, the transient obligate oligomeric assemblies. Nonetheless, there is evidence both receptor potential (TRP) channels are symmetrical homotetramers for and against GPCRs naturally existing as dimers or higher–order whose 3D structure resembles that of VGICs, wherein an ion permea- oligomers, and this varies with the receptor type. For instance, the tion pathway is formed by S5 and S6, and the re-entrant pore loop class A β2 adrenergic and μ opioid receptors have each been reconsti- region (S5–P–S6). This central pore is surrounded by four independ- tuted as functional monomers in high-density lipoprotein particles ently folded S1–S4 domains, which in the case of VGICs contain volt- and, in the presence of agonist and purified G proteins, retain their age sensors and undergo substantial movement during gating. In characteristic functional features.61,62 However, at the other extreme, mammals, the TRPV1 subtype,48 expressed by primary afferent noci- the class C GPCRs are well-validated to exist as obligate dimers and, ceptors, shows large IC N- and C-terminal domains, which together in some cases, higher-order oligomers,63 thus confirming that these account for 80% of the channel's mass and mediate extensive subunit GPCRs absolutely require an oligomeric architecture. Atomic force interactions. Intersubunit interactions are facilitated by discrete sub- microscopy studies of rhodopsin have also revealed an oligomeric structures within the cytoplasmic domain, which may regulate chan- arrangement.64 There is also a wealth of biochemical evidence of nel assembly and/or facilitate concerted conformational changes cooperativity in the actions of orthosteric ligands of both class A and after co-factor binding or agonist-evoked gating. class B GPCRs22,65–67 suggesting that many GPCRs participate in 8 CHANGEUX AND CHRISTOPOULOS oligomerization as part of their functional cycle, perhaps as a means It is well established that the majority of NHRs function as homo- of trafficking, compartmentalization or, for those receptors that are or hetero-dimers with only a subset of monomers, including many of not obligate dimers, a means of dynamically regulating their selectiv- the orphan NHRs.78 There have been abundant high-resolution struc- ity or signaling efficiency.22 tures of various LBDs and DBDs in different states, but far fewer of In humans, there are 58 RTKs (Figure 2D) that are further subdi- complete multi-domain complexes of NHRs bound to their ligands, vided into 20 families and function as receptors for polypeptide DNA, and any associated coregulators. The X-ray structures of the full growth factors, cytokines and related hormones.68 They all possess a eukaryotic nuclear PPARγ–RXRα (peroxisome proliferator-activated variable-length, multi-domain ECD that recognizes the orthosteric receptor–retinoid X receptor), the LXRβ/RXRα heterodimers and the ligand(s), a single-pass TM region, and an intracellular juxta- hepatocyte nuclear factor 4α (HNF4α) homodimer,81,82 and the cryo- membrane domain (JMD) that links to the tyrosine kinase domain EM structure of the vitamin D receptor/,83 each bound to DNA, show (TKD), which performs the catalytic function initiating signal trans- striking features in common, homologous to those of the much- duction. With a few key exceptions (e.g., the insulin receptor), most studied bacterial transcription factor, the lac repressor.84 The RTKs exist as monomers in the absence of ligand, but, in all instances, DNA-bound lac repressor forms a symmetrical homodimer.85 Despite dimerization is part of the activation process.69 To date, no full-length manifest sequence differences in the eukaryotic heterodimers, the rel- structure has been solved. At least four mechanisms, which vary by ative 3D arrangements of the proteins, the domain-domain interac- the extent to which the orthosteric ligand itself contributes to the tions, and the interfaces with DNA resemble that of the lac repressor, dimerization interface.68,70 have been hypothesized for the dimeriza- with a pseudosymmetrical organization around an axis perpendicular tion process. These include direct cross-linking of two receptors by a to the DNA response element. However, the homodimeric HNF-4α- bivalent orthosteric agonist, where the entire dimeric interface is pro- DNA structure adopts a surprisingly asymmetric organization, which vided by the ligand (e.g., nerve growth factor and its receptor TrkA71); is also evident when the overall domain architecture of this complex the combination of both bivalent ligand and receptor interfaces to is overlaid on that of the heterodimeric PPARγ–RXRα-DNA.82 form the activating dimer (e.g., stem cell factor and its receptor, In conclusion, a wealth of information from structural and func- KIT72); the requirement of multiple contacts provided by agonist, tional studies points to oligomerization playing an important role in receptor, and additional accessory proteins (e.g. FGF, heparin sul- the function of all receptor families with the LGICs and VGICs sitting phate and the FGF receptor73); and a conformational rearrangement at one end of the spectrum where multimerization is essential and by the agonist to “unmask” an otherwise-buried dimerization arm, the GPCRs at the other. which then forms the entire dimeric interface (e.g., EGF binding to the EGF receptor74). Given this multiplicity of hypothetical mechan- isms underlying RTK dimerization, both symmetric and asymmetric 3 | THE ALLOSTERIC TRANSITION AND arrangements of the extracellular domains have been suggested, DYNAMICS OF SIGNAL TRANSDUCTION depending on the receptor and its agonist.70,75 Similar considerations apply to the TK domain. High-resolution Perhaps the most fundamental question in the field of allostery structures have identified a putatively active conformation,76 which relates to the nature of the allosteric transition itself. Having estab- requires a precise orientation of the “activation loop” and αC helix in lished that oligomerization of these receptors is crucial for their regu- the N-lobe of the TK to ensure coordination of ATP and catalysis of latory functions, the issue then becomes understanding how changes phosphotransfer. The key differences between RTKs with regards to in the complexes, both substantial and subtle, impact their operation. the active site thus seem to arise from the transition of different inac- For pLGICs, their physiological responses are unambiguously tive states to the final (common) active state.68,70,77 The EGFR family mediated by several discrete conformations in reversible equilibrium, represents a special case, utilizing a separate mechanism whereby including resting (R), open channel (A), and slow (D), and fast two TK domains form an asymmetric dimer in which the C-lobe of (I) desensitized states.86,87 Recent structural and computational stud- one domain interacts with the N-lobe of the second domain to ies of GLIC (Figure 3A), ELIC, and GluCl consistently point to a model remove autoinhibitory constraints in the latter.68,70 of channel gating that follows a progressive stepwise isomerization Finally, the 48 human NHRs (Figure 2E) are ligand-regulated tran- (or conformational wave) that starts from the orthosteric binding site, scription factors with a conserved modular structure that comprises a propagates to the EC/TM domains interface via a rigid-body rear- disordered N-terminal domain, which contains a ligand-independent rangement of the extracellular β-sandwiches, and moves down to the AF1 transactivation domain, a central DNA binding domain (DBD), a TM helices to ultimately open the gate.88,90–92 It involves a sequence variable “hinge” region, and a C-terminal LBD that contains both the of two distinct quaternary transitions (Figure 3B). First, a radial con- orthosteric site and a second AF2 transactivation domain, which car- traction or (un)-blooming of the EC domain, as a large radial reorienta- ries the “coregulator” binding site.78 The LBD of all NHRs consists of tion or tilting of the extracellular β sandwiches that promotes the a three-layered helical sandwich with a hydrophobic pocket in which opening of the ion pore,88,90,93,94 followed by, a global un-twisting of lipophilic ligands are captured and shielded from the solvent environ- the receptor, best described by a concerted opposite-direction rota- ment.79 However, the ligand pockets of the NHRs exhibit a wide tion of the ECD relative to the TMD about the 5-fold symmetry axis, diversity of amino acid compositions such that they accommodate a to lock the channel in its active state.25,26,88,90,93–96 These “twist and broad variety of ligands within volumes that range from ~30 to nearly bloom” transitions mediate the dynamic reorganization of the inter- 1,500 Å3.80 faces between subunits, which host several binding sites (see below). CHANGEUX AND CHRISTOPOULOS 9

Additional recent studies, in agreement with the stepwise model protein structures in distinct conformations.48 Regions targeted by of channel gating, offer new insights on the transitions to desensi- proalgesic inflammatory agents stabilize conformational changes tized states that follow the opening of an ion channel. An ELIC closed associated with gating. These regions include the outer pore channel structure26 is consistent with a desensitized D state, with a domain, which is unusually dynamic compared to typical VGICs, as profound quaternary reorganization of the ECD together with a dif- well as the hydrophobic pocket defined by the external surface of ferent configuration of the ion pore.88 Similarly, the recently identi- the S3–S4 helices, S4–S5 linker, and S6 helix, thus exerting effects fied structure of the eukaryotic GABAA β2 homopentamer is on either of two restriction points defined by the selectivity filter consistent with a D state. A “locally closed” state of GLIC97 exhibits and lower gate. the conformation of the active state, but with a closed channel and In the case of GPCRs, the best-studied, and conceptually most plausibly corresponds to an I, fast-desensitised98 or intermediate99 straightforward model is that of rhodopsin, which, despite proceeding state. Thus, with an increase in structural models for discrete confor- through a series of very short-lived intermediate states, displays an mational channel states, a mechanistic understanding of the transi- almost “switch-like,” two-state behavior in going from the fully inac- tions between them is now becoming accessible. tive dark rhodopsin (which is pre-bound to the endogenous inverse Comparison of the structures of the AMPAR by X-ray agonist, 11-cis-retinal) to the fully all-trans-retinal bound active crystallography,39 EPR spectroscopy and cryoEM34 in resting, active metarhodopsin II.104 The characteristic features of the allosteric acti- and desensitized states reveal that transitions occur in the LBD clam- vation transition for monomeric GPCRs begin with a “pulling” of the shell and within the LBD dimer but also between LBD dimers and the extracellular sides of TMs 3, 5 and 7 (essentially contracting the ATD. AMPAR activation upon agonist binding to the apo/resting orthosteric binding pocket for hormone GPCRs) or an increased vol- state occurs with an intact D1-D1 LBD dimer interface, enhancing ume of the retinal binding pocket in rhodopsin. Both instances then the probability that domain closure in response to agonist is con- result in a rearrangement of highly conserved hydrophobic and aro- veyed to separation of the D2 lobes within a LBD dimer. Accompany- matic residues deeper in the receptor core (the “transmission switch”) ing the conformational changes within LBD dimers is a rotation and to ultimately result in the most dramatic change upon activation, a 6- translation of LBD dimers, thus resulting in a separation of the LBD 14 Å (depending on the receptor) rotation and outwards rigid body D2 lobes, pulling open the ion channel gate at or near the M3 bundle displacement of TM6 (Figure 3C), as well additional rearrangements crossing. Orthosteric agonist binding conveys a clamshell closure to of TMs 5 and 7 in the cytoplasmic region, thus opening the interface channel opening, while antagonists stabilize the cleft in a more “open” for interaction with the G protein.53 conformation. Under nondesensitizing conditions, the LBDs are Interestingly, there are numerous active state structures of organized as back-to-back dimers, and perturbations that weaken the metarhodopsin II with and without a peptide corresponding to the C dimer interface enhance receptor desensitization. In conclusion, as terminus of the Gα subunit of the Gt protein that are very similar, with pentameric LGICs, AMPARs undergo concerted changes of quat- indicating that the two key states of rhodopsin (inactive versus fully ernary organization of the molecule upon activation and active) are highly stable in their own right and represent preferred desensitization. lowest energy conformations, in agreement with the original For the trimeric P2X and ASIC1a channels, the key steps in the MWC.105 Studies of the partial and fully active states of the hormone allosteric transition are proposed to be an agonist-promoted outward class A GPCRs solved to date indicate that the complete transition to movement of the TM domains lining the channel pore that is the fully active state requires stabilization by the presence of both mediated by flexible β strands linking the (relatively rigid) amino ter- agonist and G protein, G-protein-mimetic (e.g., conformationally minal orthosteric ligand-binding domain to the channel.24 In the case selective nanobody), or other transducer protein.106 This need for of the P2X4 receptor, the transition involves a relatively simple “iris- dual interactions on the same receptor is still globally consistent with like” rotation and pore expansion that maintains symmetry, whereas the occurrence of discrete allosteric states12,89 (Figure 3D) but might the movements within ASIC1a are qualitatively different because possibly involve a broader dynamic sampling of functionally relevant they are sensitive to pH state, retaining symmetry at low pH but conformational substates. becoming asymmetric at high pH.44 In summary, these data are con- It should be noted that less is known about the allosteric transi- sistent with the general view that LGIC interfaces between subunits tions in other classes of GPCRs. NMR and crystallographic studies play a critical role in the allosteric transition. have thus far only provided atomic-level details on isolated domains For VGICs, early kinetic studies suggested that voltage sensor of class B and C GPCRs—e.g., the N-terminal domains versus the TM activation involves a concerted transition that opens the central ion domains.53 Interestingly, for the class C dimeric mGluR GPCRs, exam- pore,100–102 consistent with a MWC mechanism. The recent crystal ination of partial domain crystal structures, combined with elegant structure of the voltage-gated Na+ channel from Arcobacter butzleri structure-function, spectroscopic and computational methods, sup- (NavAb) is in accord with this model, revealing a “pre-open” state in ports a complex multi-step model, whereby the transition is triggered which all four voltage-sensing S4 domains are simultaneously acti- by the binding of orthosteric agonist to the “venus flytrap (VFT)” N- vated, in essence “poised” to propagate a concerted opening of the terminal lobe of the promoters, leading to VFT closure and conse- channel pore via the S4-S5 linker,47 which is ultimately known to be quent reorientation of the two lobes in a “scissor-like” movement transmitted to the pore-lining S6 segments.103 Interestingly, how- that brings the attached extracellular linker domains closer together, ever, cryo-EM studies with the TRPV1 channel reveal differential which then causes a rearrangement of the 7TM regions of the dimer gating mechanisms for TRPs compared to other VGICs, by capturing to asymmetrically activate the G protein.107 10 CHANGEUX AND CHRISTOPOULOS

FIGURE 3 Structural and dynamic insights into the allosteric transition of LGICs and GPCRs. (A) Superimposed crystal structures of the pentameric bacterial GLIC in closed (red, pH7, PDB 4NPQ) and open (green, pH4, PDB 4NPP) states. Left: side view, from the plane of the membrane, of the complete superimposed crystal structures. Right: upper view of the TMDs as they transition between states; reproduced with permission from Sauguet et al.88 (B) The proposed “bloom” and “twist” model for successive steps of the allosteric transition mediating gating of pLGICs, viewed from the top of the receptor; reproduced with permission from Cecchini and Changeux.31 (C) Superimposed crystal structures of the active (green; PDB: 4MQS) and inactive (red, PDB: 3UON) M2 muscarinic receptor (mAChR), highlighting the key movement underlying the allosteric transition, namely a rigid body movement of TM6 that closes the extracellular vestibule, contracts the orthosteric pocket, and opens the intracellular domains for interaction with G protein. (D) Fluorescence intensity histograms of a collection of β2 adrenergic receptor (β2 AR) molecules determined at the single-molecule level in the absence (top) or presence of a high efficacy agonist (middle) or antagonist (bottom). The dotted lines and percentages show Gaussian fits indicating transitions between two states (active, green; inactive, red), whereas the composite fit is shown by the bold red line. Adapted with permission, from Lamichhane et al.89

In terms of transcription-factor allosteric transitions, the X-ray clearly established, and involves the encapsulation of the orthosteric structures of the lac repressor in the induced and the repressed ligand within the hydrophobic LBD, receptor dimerization and inter- states reveal that the orientations of the subdomains change through action of the DBD with nuclear response elements, recruitment of a small hinge motion of the N-terminal subdomain relative to the C- coregulatory proteins, and association with transcription initiation terminal subdomain (with little, if any, change of tertiary structure) complexes.78 Numerous studies utilizing crystallography of apo and while preserving the 2-fold axis of the dimer. This conformational holo structures of the LBD, as well as hydrogen/deuterium exchange change links the effector site, through the dimer interfaces, to the (HDX) and NMR, have highlighted a pivotal role for movement of the hinge helices and the DNA-binding domains. Binding of the inducer C-terminal helix 12 (H12) in the LBD in both the opening-closing of stabilizes a subtle structural change in the N-terminal subdomain, the orthosteric pocket and the subsequent formation of the AF2 cor- which is sufficient to destabilize the repressor-operator complex and egulator binding surface, which represents the key allosteric transi- reduce the repressor's affinity for the operator by several orders of tion within the LBD of the NHRs.108 Interdomain allosteric magnitude, thus triggering the genetic switch. The overall data are communication is also critical for these receptors, as highlighted par- quantitatively fitted by the MWC model.84 ticularly by the PPARγ/RXRα and HNF-4α crystal structures bound to However, structural information concerning the allosteric transi- their DNA response element81,82 and also by studies showing how tion in human NHRs is still limited at this stage to isolated domains, response elements can modulate orthosteric ligand and coregulator in particular the LBD and DBD. Nonetheless, the basic mechanism is binding, how allosteric effects can be propagated between the LBD CHANGEUX AND CHRISTOPOULOS 11 and DBD, and through the computational identification of energeti- Mutations that disrupt key components of the allosteric transition in cally preferred communication “pipeline” residues that have coe- NHRs, such as ligand binding, dimerization and the ability to bind volved to link different surfaces across NHRs.109–111 DNA, have been implicated in cancer, metabolic dysfunction, and sex- Similarly, with a lack of complete high-resolution structures of ual and developmental disorders.123 human RTKs in different states, the mechanisms governing allosteric Hundreds of gain- or loss-of function mutations have been iden- transitions within these receptors remain an area of intense research. tified in GPCRs.124,125 The mutations cause diseases such as retinitis Even though it is well acknowledged that the formation, or reorgani- pigmentosa, autosomal dominant hypocalcemia, hyperthryroidism, zation, of dimeric interfaces is necessary for RTK activation, more male-limited precocious puberty, familial short stature, and even recent studies indicate that the extracellular RTK domains can be some forms of obesity.124,125 Several of them change the basal activ- directly coupled allosterically to mediate activating conformational ity of the receptor, which can be restored to the “resting activity” by changes to the TK domain across the plasma membrane, involving the use of positive or negative allosteric modulators, depending on active reorganization of both the TM helices and the juxtamembrane the context.126 These observations clearly highlight the potential for region.70 A striking illustration of this mechanism was proposed pharmacotherapeutic approaches to correct for the inappropriate through structural and long-timescale MD simulations that present a functionality of mutated (or overexpressed) receptors. model of how full-length inactive EGFR monomers can first associate into symmetric (but still inactive) dimers and how the subsequent binding of orthosteric ligand can actively reorganize the TM and jux- 5 | THE PHENOMENON OF “BIASED tamembrane regions to remove a steric block to formation of the AGONISM” active state, which is then ultimately completed by asymmetric TK domain assembly.112,113 The concept of “biased agonism”—i.e., the ability of a ligand to stabi- An important postulate of the allosteric transitions within the lize distinct conformations of a given receptor such that only a subset MWC model is the spontaneous occurrence of several of the possible signaling pathways mediated by that receptor are conformations—at least two—in reversible equilibrium in the absence engaged, to the relative exclusion of other pathways—was first of ligand. A variety of biophysical techniques applied to studies of explicitly defined in studies of GPCRs.127,128 For example, Azzi LGICs,31 GPCRs,22 RTKs,68 and, at a more preliminary stage, to et al.129 demonstrated that previously classified “beta-blockers” 105,114–116 NHRs have now unambiguously confirmed this concept for (i.e., clinically used competitive antagonists of the β2 adrenergic receptors, with the extension of this notion to multi-state ensembles receptor) could actually activate the MAP kinase pathway as agonists that are redistributed by the binding of ligand.10,11,13,115,117 The data in a β-arrestin-dependent but G-protein-independent manner, thus presently available in these diverse systems strongly support a con- highlighting the potential for the occurrence of multiple active con- formational selection scheme rather than any induced-fit mechanism formations that differentially recognize transducers. This phenome- elicited by ligand binding. non was extended to various classes of GPCRs,130 and biased agonism now represents a major paradigm in GPCR drug discov- ery.128 The same phenomenon may exist in even earlier studies of 4 | CONSTITUTIVE MUTATIONS AND NHRs, specifically in the context of compounds termed “selective RECEPTOR DISEASES NHR modulators,” with tamoxifen being a prototypical example at the estrogen receptor.78,131 This drug demonstrates either pro- An important prediction of the MWC model is that the shift of the estrogenic or anti-estrogenic actions in a tissue-specific manner, thus spontaneous equilibrium between active and inactive conformations being the first example of a “selective estrogen receptor modulator.” by mutations, causing loss or gain of function, might be associated The simplest allosteric mechanism to account for biased agonism with disease states.10 More than 1,000 activating and inactivating requires the existence of at least two—instead of one—active states, mutations in both LGICs and VGICs show that full channel openings each possessing different agonist-binding specificities and signal may take place spontaneously and with high frequency in the transduction properties.6,132,133 This view is consistent with the absence of agonist. These mutations lead to diseases that include broader notion of multi-state ensembles that are redistributed by congenital myasthenia and autosomal dominant nocturnal epilepsy, orthosteric or allosteric ligands binding not exclusively to one state, hyperplexia, ataxia and cardiac , including the “long-QT but to several states with different affinities.134 Studies employing 118,119 13 19 syndrome”. In the case of pLGICs these mutations are prefer- CH3ε-methionine and F-NMR spectroscopy to look at the confor- entially located either at the interface between subunits or within a mational landscape of the β2 adrenergic receptor have revealed highly given subunit at the interface between rigid domains of the receptor dynamic changes at both the extracellular and intracellular faces, con- protein,120 again highlighting the key role of oligomeric architecture sistent with the exploration of different conformational ensembles by on the function of allosteric proteins in both physiology and disease. ligands.135,136 Interestingly, the recent solution of the first crystal The overexpression of RTKs in the plasma membrane is well structure of a constitutively active form of rhodopsin bound to a pre- known to contribute to pathologies such as cancer due to inappropri- activated visual arrestin revealed substantial conformational differ- ate activation as a consequence of ligand independent dimeriza- ences in TM domains 1, 4, 5, and 7, in addition to the canonical tion.70,121 Similarly, naturally occurring mutations in RTKs increase movement of TM6 observed in the G- protein-binding state,137 con- catalytic activity and result in various clinical phenotypes.77,122 sistent with the notion that different GPCR transducer proteins 12 CHANGEUX AND CHRISTOPOULOS

(e.g., β-arrestin versus G protein) stabilize structurally distinct active absence of agonist and is then referred to as an “allosteric agonist”.18 states. The extent to which biased agonism has been explored at Combinations of these properties are also observed.155 other receptor families beyond the GPCRs and NHRs remains unclear, but the ability of all receptors to undergo multiple allosteric – 7.1 | Allosteric sites on LGICs transitions suggests that the phenomenon is likely widespread.138 140 Several allosteric sites have been identified in pLGICs,19,88 from ECD interfaces to the TMD and, potentially, the cytoplasmic face. For 6 | THE CONTRIBUTIONS OF “INTRINSIC instance, Ca2+ ions bind as PAMs in the ECD at the interface DISORDER” TO RECEPTOR ALLOSTERY between subunits close to the TMD of α7 and α4β2 nAChRs.156–159 Ba2+ binds as a NAM to a similar site on ELIC,160 and GLIC shows a A major paradigm shift in the 1990s about the relationship of the homologous pocket.88 Given that most pLGICs are heterooligomers, amino acid sequence of a protein to its 3D architecture was the rec- the potential for non-agonist-binding ECD interfaces to harbor ognition that many eukaryotic proteins commonly contain regions hitherto-unappreciated modulatory sites was predicted to be a likely that are intrinsically disordered141–143 and dynamically fluctuate general feature in such receptors,161 and the best clinically validated through disorder-to-order transitions. The occurrence of intrinsic dis- examples of this are the benzodiazepines, which indeed bind as 162,163 order is now thought to provide a means of maximizing allosteric PAMs to a non-agonist ECD interface in the GABAA receptor. coupling in many protein families.13,144–146 The idea is that the In contrast, the anthelminthic PAM, ivermectin, binds within the absence of a unique “rigid” conformation confers a plasticity that α7nAChR TMD,164 as do synthetic PAMs, such as PNU-120596 and readily supports dynamic changes in response to changing local envir- LY2087101.165,166 In the GLIC structure, the general onments, including protein or ligand binding. (GAs), isoflurane and propofol, bind in the upper half of the TMD that Intrinsically disordered regions have been identified in all classes is accessible from the lipid bilayer.167 An ivermectin site (Figure 4) is of plasma membrane receptors, predominantly (although not exclu- also seen in the GluCl structure27 in the periphery of the TMD at sively) within cytosolic regions.147 Changes in order-to-disorder subunit interfaces. The intersubunit TMD cavity is conserved in organization (or vice versa) have been proposed as key mechanisms human glycine and GABAA receptors, involving residues that influ- underlying the allosteric modulation of G protein activation and ence ethanol and action.168–170 Computational analyses of potentially β-arrestin interaction with GPCRs,148,149 as well as the the cryo-EM structure of the Torpedo nAChR171 and a homology 77,150,151 172 activation of the EPHB2 RTK. Arguably, however, the best model of the GABAA receptor show the existence of several TMD example of the role of intrinsic disorder in receptor functionality is cavities complementary to , and possibly homologous to seen with NHRs. The N-terminal domain of NHRs, which contains the ivermectin site in GluCl. Cholesterol binding at these sites would the AF1 region, shows a high degree of intrinsic disorder. This region stabilize the channel open-pore conformation, suggesting a role is often more transcriptionally active than the better structurally char- (together with lipids in general) as putative endogenous allosteric acterized LBD AF2 region.152 Given such intrinsic disorder, no com- modulators in eukaryotic pLGICs. mon motif has yet been defined for coregulator binding proteins that Finally, in addition to the ECD and TMD allosteric modulator interact with AF1. NHR N-terminal domains contain most of the sites, the IC domain represents another potential region for allosteric phosphorylation sites for post-translational modifications,153 and thus modulation of pLGICs—e.g., by post-translational modifications and play a major contribution to the overall conformational dynamics interaction with scaffolding proteins—but remains the least pharma- that govern allosteric transitions between NHR domains. Post- cologically explored.20,28 translational modifications may influence the allosteric transition For the non-pLGIC NMDARs, the majority of allosteric sites have mediated by NHRs, as classically observed with eukaryotic regulatory been localized to the ATD173,174 and bind divalent cations, such as enzymes like phosphorylase b154 and many receptors (see above). Zn2+ and Mg2+, endogenous polyamines (e.g., spermine and spermi- dine) and various synthetic modulators, such as ifenprodil (Figure 4), which bind at the interface between two ATDs.175 Tentatively, any 7 | ALLOSTERIC MODULATORY SITES AND ligand that stabilizes the ATDs in an open-cleft configuration may THEIR PHARMACOLOGY behave as a PAM, while a ligand promoting ATD cleft closure acts as a NAM.

Since the early discovery of benzodiazepines as GABAA receptor allo- More recently, allosteric NMDAR modulators have been identi- steric activators, it was shown that, most, if not all, cellular receptors fied within the vicinity of the LBD, typified by compounds such as possess, in addition to the orthosteric site, spatially distinct sites that TCN-201, DQP-1105, UPB512, and QNZ46.174 Less is known about modulate their allosteric transitions (18,19; Figure 5). Pharmacologi- the potential for allosteric sites in the TMD or intracellular domains cally, allosteric ligands can be classed as “positive allosteric modula- of the NMDAR, although it has been suggested that the small mole- tors” (PAMs), which enhance the effect of the orthosteric ligand, cule, CIQ, and the endogenous neurosteroid, pregenolone sulphate, “negative allosteric modulators” (NAMs), which reduce the effect of are PAMs via the TMD.174 In contrast to the NMDAR, the LBD of the orthosteric ligand, and “neutral allosteric ligands” (NALs), which the AMPAR appears to be the major site for allosteric modulators, occupy the allosteric site, but do not modulate the effect of orthos- including the ions, Cl− and Na+, as well as synthetic PAMs that teric ligand. A PAM may sometimes activate the receptor in the include aniracetam, cyclothiazide, and biaryl propylsulfonamides.176 CHANGEUX AND CHRISTOPOULOS 13

Last but not least, recent structural and dynamic studies with revealed very little difference in the allosteric pocket in the absence LGICs demonstrate that, in agreement with their differential effects or presence of the PAM.187 This comparison suggests that the modu- on the allosteric equilibrium, the structure(s) of allosteric modulatory latory site is “pre-formed” upon transition to the active state, a mech- site(s) differ between resting and active states. For example, a close anism consistent with a MWC scheme whereby the effect of the comparison of the X-ray structures of active and resting conforma- PAM is to preferentially stabilize this state.188 It remains to be seen tions in GLIC88 and MD simulations with GluCl90 reveal that, during how generalizable the above mechanism is. Interestingly, mutagenesis activation in GLIC, a tertiary change significantly reshapes the subunit studies have also identified putative intracellular allosteric sites for interface and the binding pocket homolog of the PAM Ca2+ site and synthetic modulators of some chemokine receptors, and novel lipi- that, in GluCl, removal of ivermectin from its TM site shrinks substan- dated peptides (“pepducins”) have been designed to specifically mod- tially by receptor twisting.88 This observation has important conse- ulate the intracellular surfaces of different GPCRs.185 Finally, it is quences for the design of allosteric modulatory drugs. worth highlighting that there remain numerous pharmacological examples of GPCRs that display cooperativity in their modes of inter- action with both cognate and synthetic ligands, suggesting that the 7.2 | Allosteric sites on VGICs potential for interactions across GPCR oligomeric arrays is also There is a rich repertoire of and drugs that bind allosteri- possible.65,66 cally in VGICs. For sodium channels (NaV), there are at least six dis- tinct sites of action, as well as a site for local anaesthetics (LAs).18,177 LAs bind within the central pore cavity as blockers, 7.4 | Allosteric sites on RTKs whereas and μ- bind to the outer pore open- Although the catalytic TK domain has traditionally represented the ing. modulate voltage-dependent gating by binding to major focus of small molecules targeting RTKs, there are no examples the extracellular ends of the S3–S4 voltage sensor,50 while lipophilic yet of pure allosteric modulators for this region; all clinically used toxins (e.g., batrachotoxin) bind within the TMD of the S5-S6 small molecule inhibitors of the TK domain of RTKs are ATP-site regions.177 For L-type voltage-gated Ca2+ channels (CaV), there are at competitive to some extent.189,190 Better clinical examples of allo- least three allosterically coupled sites targeted by marketed steric targeting of RTKs involve the ECD, particularly with antibodies. drugs178–181 such as phenylalkylamines (e.g., ) or benzothia- For instance, although most ECD-directed antibody strategies target zepines (e.g., ), which act within the pore, or dihydropyri- the orthosteric ligand or its binding site,191 the clinically used trastu- dines (e.g., ), which act as gating modifiers (Table 1). There zumab (Hereptin) and pertuzumab (Perjeta) target domains 4 and are also neurotoxins, such as ω- GVIA and ω-agatoxin IVA, 2, respectively (Table 1), of the HER2/neu RTK to inhibit functionality which act as pore blockers or gating modifiers, respectively.18 either through promoting receptor downregulation (trastuzumab) or produce a multitude of peptide toxins that activate TRPV1 to elicit preventing dimerization (pertuzumab). Neither one of the RTK bind- pain as part of the 's mechanism. One such ing domains recognized by these antibodies constitute the orthosteric vanillotoxin (called “double-knot ,” DkTx), traps TRPV1 in its site,182,183 so the clinical effects of both agents are likely allosteric. open state with near-irreversible kinetics through attachment within Other allosteric RTK antibodies have been discovered, such as the S5–P–S6 pore region, consistent with the notion that the outer XMetA and XMetS, which bind to the ECD of the insulin RTK and pore of TRPV1 is conformationally dynamic and contributes directly activate the receptor or act as an insulin-sensitizing PAM.192,193 Allo- to gating.48 steric ECD interactions have also been identified at RTKs mediated by peptidomimetics.189 Most recently, the first ECD small-molecule 7.3 | Allosteric sites on GPCRs RTK inhibitor was identified (Figure 5). SSR128129E interacts with the D2D3 domain of multiple FGFRs194 to inhibit hormone signaling Most GPCRs possess at least one, if not more, allosteric modulatory in a clearly allosteric manner.195 Finally, a less-characterized RTK tar- sites.184,185 Furthermore, two GPCR allosteric modulators, maraviroc get is the juxtamembrane/TM region. However, recent work from (CCR5 NAM) and cinacalcet (calcium-sensing receptor PAM), are on Jang et al.196 identified an allosteric agonist, gambogic amide, which the market.18 The majority of synthetic GPCR allosteric modulators selectively binds to this region of the TrkA RTK. identified to date bind within the TMD region, but the location within the TMD cavity can vary dramatically depending on the receptor class. This variability is due to the high degree of both sequence and 7.5 | Allosteric sites on NHRs tertiary structure divergence in the orthosteric domain for these receptors.185,186 There is no current consensus on the importance of targeting allo- An important finding contributing to understanding the structural steric modulator sites on NHRs.197 Furthermore, virtually all clinically basis of GPCR allosteric modulatory sites was the solution of two relevant targeting of NHRs to date has been via orthosteric drugs.78 crystal structures of an active M2 muscarinic receptor, concomitantly Nonetheless, “alternate” pockets distinct from the orthosteric site bound to an orthosteric agonist and a G protein mimetic nanobody have been identified in some.78,197,198 The extent to which these and either the absence or presence of the small molecule PAM, sites unambiguously modulate orthosteric ligands remains largely LY2119620 (Figure 4187). These two structures clearly delineated the undetermined, however, and we thus refer to them as “putative allo- allosteric transition between active and inactive states but also steric sites.” 14 CHANGEUX AND CHRISTOPOULOS

FIGURE 4 Representative crystal structures of allosteric modulatory sites and their ligands. Shown are the homopentameric Caenorhabditis elegans glutamate-gated chloride channel (GluCl) (PDB: 3RHW), where the allosteric modulator, ivermectin, sits within the transmembrane regions of the channel; the

human M2 mAChR (PDB: 4MQT), where the allosteric modulator, LY2119620, sits within the extracellular vestibular entrance to the receptor, above the orthosteric agonist, iperoxo; the rat NMDA R (PDB: 4PE5), where the allosteric modulator, ifenprodil, is located at interfaces between amino terminal domain subunits, whereas the orthosteric co-agonists, L-glutamate and glycine, sit within the ligand binding domain.

There are at least five potential alternate binding surfaces on Although the pursuit of allosteric modulators of NHRs remains a NHRs (Figure 5). The best-studied is the AF2 coregulator site.198–200 nascent field for drug discovery, recent studies of non-hormone tran- An exemplar AF2-targeting molecule, HPPE, binds covalently at the scription factors have identified synthetic modulators aimed at target- interface between the LBD-AF2 binding interface.201 A second LBD ing the transcription factor dimeric interface. One such example is a surface identified adjacent to AF2 on the androgen receptor, termed series of small molecules targeting the OLIG2 transcription factor, “binding function 3”,202 recognizes molecules such as 3,3,5- which show promise as potential inhibitors of glioblastoma.205 triiodothyroacetic acid and may be found on other NHRs.203 Hughes et al.197 identified a third LBD pocket in the PPARγ receptor that binds synthetic molecules concomitantly to the orthosteric site. Other 8 | CONCLUDING REMARKS molecules have been described that interact with nuclear response elements and with zinc fingers in the DBD, such as the estrogen- 8.1 | Allosteric interactions in receptors receptor-targeting compound DIBA.198 Finally, at least one small mol- ecule that blocks transactivation of the androgen receptor, EPI-001, Abundant structural data show that the original concept of allosteric interacts within the intrinsically disordered AF1 region.204 interactions as indirect interactions established between CHANGEUX AND CHRISTOPOULOS 15

FIGURE 5 A multiplicity of allosteric modulatory sites across all receptor superfamilies. Shown schematically are different regions that have been proposed to interact with allosteric modulator ligands for LGICs, VGICs, GPCRs, RTKs, and NHRs, including representative examples for each of the sites (where validated; otherwise they are referred to as “putative” allosteric sites). Note, for ion channels, black circles denote sites of interaction for molecules that directly interact with the endogenous agonist site(s) or sterically hinder the channel. For GPCRs, the nature/ location of the orthosteric site varies with the class of GPCR (A, B, C, or F). ATD, amino terminal domain; RE, response element. All other abbreviations as defined in the legend for Figure 2.

topographically distinct sites and mediated by a discrete protein con- transducer binding. Although complete atomistic details remain to be formational change successfully applies to signal-transducing recep- resolved for other receptors, it is worth noting that oligomerization, tors. The signal transduction mechanism mediated by these cooperativity, and long-range conformational changes mediated receptors, in many instances, occurs in agreement with the MWC through allosteric transitions have been demonstrated for other model via the selective stabilization by ligands of a few (frequently classes of receptors, such as the cell-adhesion-mediating integrin more than two) pre-existing discrete states in conformational equilib- receptors and the pathogen-targeting pattern-recognition (e.g., Toll- rium. Organization into symmetrical oligomers, a feature that was ini- like) receptors,206–209 further highlighting the universality of the allo- tially proposed to specify the cooperativity of the structural steric transition mechanism. transition that mediates cooperative ligand binding, is also found in Finally, it is worth noting that the allosteric concept, as applied to many receptor categories. Yet, several well-documented cases have regulatory proteins and now receptors, can also be extended to even been described, typified by the GPCRs, in which signal transduction larger protein assemblies,10,16 creating important functional links takes place in non-oligomeric structures, nevertheless, via pre- between the molecular and cellular levels through large-scale allo- existing conformational states. steric transitions. This represents fertile ground for future studies of In several receptor systems, in particular LGICs and GPCRs, the “supramolecular” protein assemblies. conformational change engaged in the signal transduction mechanism has been resolved at the atomic scale. In the case of pLGICs, it 8.2 | Toward a new pharmacology includes a stepwise quaternary change that includes blooming and twisting with reorganization of the interfaces between subunits, Until relatively recently, drugs were classically designed following the which host orthosteric and allosteric binding sites (Figure 4). In concept of direct competitive interactions between ligands for a com- GPGRs, the conformational transition involves an inward contraction mon “rigid” site.210,211 The discovery of the intrinsic flexibility of allo- of the extracellular regions of the TMD that is propagated as an out- steric proteins, and of allosteric modulatory sites on all classes of ward opening of the intracellular domains to expose the interface for receptor (Figure 5), supersedes the notion of “rigid site” 16 CHANGEUX AND CHRISTOPOULOS

TABLE 1 Representative examples of marketed allosteric medicines

Name Company Indication Target Rilpivirine Janssen HIV NNRT NAM Etravirine Janssen HIV NNRT NAM Temsirolimus Pfizer Cancer mTOR NAM Thrombomodulin α Asahi Kasei Pharma Blood clotting Thrombin PAM Trametinib GlaxoSmithKline Cancer MEK1/2 NAM Perampanel Eisai Epilepsy AMPA receptor NAM Aniracetam Roche Cognitive disorder; Stroke AMPA receptor PAM Diltiazem Mitsubishi Tanabe Angina; Essential hypertension Calcium channel NAM Verapamil Elan Essential hypertension Calcium channel NAM Novartis Angina; Atherosclerosis; Hypertension Calcium channel NAM Lomerizine Pfizer; Santen Glaucoma; Migraine Calcium channel NAM Bayer Angina; Essential hypertension Calcium channel NAM Elan Pain Calcium channel NAM

Diazepam Roche Anxiety; Epilepsy GABAAR PAM

Zolpidem Sanofi Insomnia GABAAR PAM

Triazolam Pfizer Insomnia GABAAR PAM

Alprazolam Pfizer Anxiety & Panic disorder GABAAR PAM

Eszopiclone Sunovion Insomnia GABAAR PAM

Topiramate Janssen Epilepsy; Lennox Gastaut syndrome; Migraine GABAAR PAM; NMDAR NAM; NAM Pirmenol Dainippon Heart Sodium channel NAM Novartis Bipolar disorder; Epilepsy; Trigeminal neuralgia Sodium channel NAM GlaxoSmithKline Bipolar disorder; Epilepsy Sodium channel NAM Pfizer Epilepsy; neuropathic pain Calcium channel NAM Cinacalcet Amgen Hyperparathyroidism Calcium sensing GPCR PAM Maraviroc Pfizer HIV CCR5 GPCR NAM Trastuzumab Genentech Breast cancer HER2/Neu receptor NAM Pertuzumab Genentech Breast cancer HER2/Neu receptor NAM

Data collated from the Thomson Reuters Cortellis Database, using the search term “allosteric,” followed by manual inspection and literature-based valida- tion of mechanism of each entry by the authors. The only exceptions were trastuzumab and pertuzumab, the mechanism of which was taken from Cho et al.182 and Franklin et al.183 “PAM” denotes “postitive allosteric modulator” and “NAM” denotes “negative allosteric modulator.” pharmacology and opens the field to the design of new categories of maximize the detection of allosteric effects.217 Furthermore, as dis- “conformation-specific” drugs that interact à distance with topograph- cussed above, most receptors are oligomers, and thus their subunit ically distinct active sites. This concept applies to both orthosteric assembly and stoichiometry can have a dramatic effect on screening ligands and allosteric modulators because the binding pockets for for allosteric modulator activity.212 Finally, it is important to note the either class concomitantly change between states during the allo- potential for endogenous allosteric substances that may play impor- steric transition that underlies receptor signal transduction.31,88,212 tant roles in both health and disease.218 However, a specific advantage of targeting allosteric modulator sites Such drug discovery challenges can be better addressed as more over orthosteric sites is that the former can exhibit a higher degree structural data on allosteric sites and transitions become available. of sequence diversity between receptor subtypes, thus providing a Interestingly, at the time of writing, a curated online “allosteric data- degree of selectivity in receptor targeting that is simply unattainable base”219 classifies nearly 72,000 substances as allosteric modulators with many orthosteric agonists/antagonists.20,185 It should also be (http://mdl.shsmu.edu.cn/ASD/), which attests to the ongoing growth noted that allosteric targeting need not be achieved solely through and interest of the phenomenon and even allows one to envisage the the design of synthetic small molecules, but also can also be reached concept of an “allosterome” as a valuable resource for new chemical, via conformationally specific allosteric antibodies, which represents cellular, molecular and structural biology breakthroughs. As high- an important field of future research. There are already clear exam- lighted by the selected examples in Table 1, there is already clear evi- ples of monoclonal antibodies that allosterically target ion dence for success in allosteric pharmacology for receptors and channels,213 GPCRs,214 and RTKs,189 as well as cytokine and integrin enzymes. receptors.215,216 In conclusion, it is now clear that substantial opportunities exist However, because allosteric modulators can display a diverse for allosteric targeting of all receptor families. This is a critical pursuit spectrum of effects, usually in a “context-dependent” manner,18 tradi- because allosteric medicines have the potential to display tional biochemical approaches are suboptimal for studying such mole- conformation- and site-based selectivity, pathophysiologically rele- cules, and functional screening methods are still being developed to vant “context-sensitivity,” and greater mechanism-based on-target CHANGEUX AND CHRISTOPOULOS 17 safety than traditional orthosteric activators or blockers. Despite 17. Karlin A. On the application of “a plausible model” of allosteric proteins being recognized over 50 years ago, the concept of allosteric interac- to the receptor for acetylcholine. J Theor Biol. 1967;16:306-320. 18. Christopoulos A, Changeux JP, Catterall WA, et al. International tion, as illustrated in this Review, continues to blossom with its Union of Basic and Clinical Pharmacology. XC. multisite pharmacol- extension to all receptors and is yielding new and fertile avenues for ogy: recommendations for the nomenclature of receptor allosterism research across the life sciences. and allosteric ligands. Pharmacol Rev. 2014;66:918-947. 19. Changeux JP. Allostery and the Monod-Wyman-Changeux model after 50 years. Annu Rev Biophys. 2012;41:103-133. 20. Changeux JP. The concept of allosteric modulation: an overview. ACKNOWLEDGMENTS Drug Discov Today Technol. 2013;10:e223-e228. 21. Christopoulos A. Allosteric binding sites on cell-surface receptors: The authors are grateful to Professor Pierre Paoletti for helpful dis- novel targets for drug discovery. Nat Rev Drug Discov. cussions, Dr. Christopher Langmead for collating the data shown in 2002;1:198-210. 22. Canals M, Sexton PM, Christopoulos A. Allostery in GPCRs: ‘MWC’ Table 1 from the Thomson Reuters Cortellis Database, and Dr. David revisited. Trends Biochem Sci. 2011;36:663-672. Thal for assistance in preparing Figures 2 and 3. A.C. is a Senior Prin- 23. Corringer PJ, Poitevin F, Prevost MS, Sauguet L, Delarue M, cipal Research Fellow of the National Health and Medical Research Changeux JP. Structure and pharmacology of pentameric receptor channels: from bacteria to brain. Structure. 2012;20:941-956. Council of Australia. J.-P.C. contributed to this Review while Interna- 24. Baconguis I, Hattori M, Gouaux E. Unanticipated parallels in archi- tional Faculty, Kavli Institute for Brain & Mind, University of Califor- tecture and mechanism between ATP-gated P2X receptors and acid nia, San Diego and a visiting scientist at Woods Hole Marine sensing ion channels. Curr Opin Struct Biol. 2013;23:277-284. Biological Laboratory, Massachusetts. J.-P.C. has a contract of con- 25. Bocquet N, Nury H, Baaden M, et al. X-ray structure of a pentameric ligand-gated ion channel in an apparently open conformation. sultancy with Servier Laboratories France for the design of nicotinic Nature. 2009;457:111-114. agents against Alzheimer's disease. A.C. has a research contract and 26. Hilf RJ, Dutzler R. X-ray structure of a prokaryotic pentameric consultancy with Les Laboratoires Servier. ligand-gated ion channel. Nature. 2008;452:375-379. 27. Hibbs RE, Gouaux E. Principles of activation and permeation in an anion-selective Cys-loop receptor. Nature. 2011;474:54-60. 28. Hassaine G, Deluz C, Grasso L, et al. X-ray structure of the mouse REFERENCES serotonin 5-HT3 receptor. Nature. 2014;512:276-281. 1. Changeux, J.-P. (1990). Functional architecture and dynamics of the 29. Miller PS, Aricescu AR. Crystal structure of a human GABAA recep- nicotinic acetylcholine receptor: an allosteric ligand-gated ion chan- tor. Nature. 2014;512:270-275. nel. In Fidia Resarch Foundation Neuroscience Award Lectures, 30. Du J, Lü W, Wu S, Cheng Y, Gouaux E. Glycine receptor mechanism R. Llinas, D. Purves, and F. Bloom, eds. (Raven Press), pp. 21–168. elucidated by electron cryo-microscopy. Nature. 2015;526:224-229. 2. Changeux JP. The feedback control mechanisms of biosynthetic L- 31. Cecchini M, Changeux JP. The nicotinic acetylcholine receptor and threonine deaminase by L-isoleucine. Cold Spring Harb Symp Quant its prokaryotic homologues: Structure, conformational transitions & Biol. 1961;26:313-318. allosteric modulation. Neuropharmacology. 2015;96(Pt B):137-149. 3. Monod J, Jacob F. Teleonomic mechanisms in cellular metabolism, 32. Fourati Z, Sauguet L, Delarue M. Genuine open form of the penta- growth, and differentiation. Cold Spring Harb Symp Quant Biol. meric ligand-gated ion channel GLIC. Acta Crystallogr D Biol Crystal- 1961;26:389-401. logr. 2015;71:454-460. 4. Changeux JP. Allosteric interactions interpreted in terms of quater- 33. Herz JM, Johnson DA, Taylor P. Distance between the agonist and nary structure. Brookhaven Symp Biol. 1964;17:232-249. noncompetitive inhibitor sites on the nicotinic acetylcholine recep- 5. Christopoulos A, Kenakin T. G protein-coupled receptor allosterism tor. J Biol Chem. 1989;264:12439-12448. and complexing. Pharmacol Rev. 2002;54:323-374. 34. Dürr KL, Chen L, Stein RA, et al. Structure and dynamics of AMPA 6. Edelstein SJ, Changeux JP. Relationships between structural dynam- receptor GluA2 in resting, pre-open, and desensitized states. Cell. ics and functional kinetics in oligomeric membrane receptors. 2014;158:778-792. Biophys J. 2010;98:2045-2052. 35. Sobolevsky AI, Rosconi MP, Gouaux E. X-ray structure, symmetry 7. Changeux, J.P., and Edelstein, S. (2011). Conformational selection or and mechanism of an AMPA-subtype glutamate receptor. Nature. induced fit? 50 years of debate resolved. F1000 Biol. Rep. Published 2009;462:745-756. online September 1, 2011. https://doi.org/10.3410/B3-19. 36. Yelshanskaya MV, Li M, Sobolevsky AI. Structure of an agonist- 8. Monod J, Wyman J, Changeux JP. On the Nature of Allosteric Tran- bound ionotropic glutamate receptor. Science. 2014;345:1070-1074. sitions: A Plausible Model. J Mol Biol. 1965;12:88-118. 37. Meyerson JR, Kumar J, Chittori S, et al. Structural mechanism of glu- 9. Koshland DE Jr, Némethy G, Filmer D. Comparison of experimental tamate receptor activation and desensitization. Nature. binding data and theoretical models in proteins containing subunits. 2014;514:328-334. Biochemistry. 1966;5:365-385. 38. Karakas E, Furukawa H. Crystal structure of a heterotetrameric 10. Changeux JP. 50 years of allosteric interactions: the twists and turns NMDA receptor ion channel. Science. 2014;344:992-997. of the models. Nat Rev Mol Cell Biol. 2013;14:819-829. 39. Lee CH, Lü W, Michel JC, et al. NMDA receptor structures reveal sub- 11. Cui Q, Karplus M. Allostery and cooperativity revisited. Protein Sci. unit arrangement and pore architecture. Nature. 2014;511:191-197. 2008;17:1295-1307. 40. Regalado MP, Villarroel A, Lerma J. Intersubunit cooperativity in the 12. Horst R, Stanczak P, Stevens RC, Wüthrich K. b2-Adrenergic recep- NMDA receptor. Neuron. 2001;32:1085-1096. tor solutions for structural biology analyzed with microscale NMR 41. Hattori M, Gouaux E. Molecular mechanism of ATP binding and ion diffusion measurements. Angew Chem Int Ed Engl. 2013;52:331-335. channel activation in P2X receptors. Nature. 2012;485:207-212. 13. Motlagh HN, Wrabl JO, Li J, Hilser VJ. The ensemble nature of allos- 42. Kawate T, Michel JC, Birdsong WT, Gouaux E. Crystal structure of tery. Nature. 2014;508:331-339. the ATP-gated P2X(4) ion channel in the closed state. Nature. 14. Tsai CJ, Nussinov R. A unified view of “how allostery works”. PLoS 2009;460:592-598. Comput Biol. 2014;10:e1003394. 43. Jasti J, Furukawa H, Gonzales EB, Gouaux E. Structure of acid- 15. Changeux J-P. On the allosteric properties of biosynthetic L- sensing ion channel 1 at 1.9 A resolution and low pH. Nature. threonine deaminase. Vi General Discussion Bull Soc Chim Biol. 2007;449:316-323. 1965;47:281-300. 44. Baconguis I, Gouaux E. Structural plasticity and dynamic selectivity 16. Changeux JP, Thiéry J, Tung Y, Kittel C. On the cooperativity of bio- of acid-sensing ion channel-spider toxin complexes. Nature. logical membranes. Proc Natl Acad Sci U S A. 1967;57:335-341. 2012;489:400-405. 18 CHANGEUX AND CHRISTOPOULOS

45. Catterall WA, Goldin AL, Waxman SG. International Union of Phar- 70. Endres NF, Barros T, Cantor AJ, Kuriyan J. Emerging concepts in the macology. XLVII. Nomenclature and structure-function relationships regulation of the EGF receptor and other receptor tyrosine kinases. of voltage-gated sodium channels. Pharmacol Rev. 2005;57: Trends Biochem Sci. 2014;39:437-446. 397-409. 71. Wiesmann C, Ultsch MH, Bass SH, de Vos AM. Crystal structure of 46. Catterall WA, Perez-Reyes E, Snutch TP, Striessnig J. International nerve growth factor in complex with the ligand-binding domain of Union of Pharmacology. XLVIII. Nomenclature and structure- the TrkA receptor. Nature. 1999;401:184-188. function relationships of voltage-gated calcium channels. Pharmacol 72. Yuzawa S, Opatowsky Y, Zhang Z, Mandiyan V, Lax I, Schlessinger J. Rev. 2005;57:411-425. Structural basis for activation of the receptor tyrosine kinase KIT by 47. Payandeh J, Scheuer T, Zheng N, Catterall WA. The crystal structure stem cell factor. Cell. 2007;130:323-334. of a voltage-gated sodium channel. Nature. 2011;475:353-358. 73. Schlessinger J. Cell signaling by receptor tyrosine kinases. Cell. 48. Cao E, Liao M, Cheng Y, Julius D. TRPV1 structures in distinct con- 2000;103:211-225. formations reveal activation mechanisms. Nature. 74. Burgess AW, Cho HS, Eigenbrot C, et al. An open-and-shut case? 2013;504:113-118. Recent insights into the activation of EGF/ErbB receptors. Mol Cell. 49. Catterall WA. Ion channel voltage sensors: structure, function, and 2003;12:541-552. pathophysiology. Neuron. 2010;67:915-928. 75. Hubbard SR. Structural analysis of receptor tyrosine kinases. Prog 50. Catterall WA, Swanson TM. Structural basis for pharmacology of Biophys Mol Biol. 1999;71:343-358. voltage-gated sodium and calcium channels. Mol Pharmacol. 76. Huse M, Kuriyan J. The conformational plasticity of protein kinases. 2015;88:141-150. Cell. 2002;109:275-282. 51. Nilius B, Talavera K, Owsianik G, Prenen J, Droogmans G, Voets T. 77. Hubbard SR. Juxtamembrane autoinhibition in receptor tyrosine Gating of TRP channels: a voltage connection? J Physiol. kinases. Nat Rev Mol Cell Biol. 2004;5:464-471. 2005;567:35-44. 78. Burris TP, Solt LA, Wang Y, et al. Nuclear receptors and their selec- 52. Liao M, Cao E, Julius D, Cheng Y. Structure of the TRPV1 ion chan- tive pharmacologic modulators. Pharmacol Rev. 2013;65:710-778. nel determined by electron cryo-microscopy. Nature. 79. Renaud JP, Rochel N, Ruff M, et al. Crystal structure of the RAR- 2013;504:107-112. gamma ligand-binding domain bound to all-trans retinoic acid. 53. Venkatakrishnan AJ, Deupi X, Lebon G, Tate CG, Schertler GF, Nature. 1995;378:681-689. Babu MM. Molecular signatures of G-protein-coupled receptors. 80. Li Y, Lambert MH, Xu HE. Activation of nuclear receptors: a per- Nature. 2013;494:185-194. spective from structural genomics. Structure. 2003;11:741-746. 54. Buck L, Axel R. A novel multigene family may encode odorant recep- 81. Chandra V, Huang P, Hamuro Y, et al. Structure of the intact PPAR- tors: a molecular basis for odor recognition. Cell. 1991;65:175-187. gamma-RXR- nuclear receptor complex on DNA. Nature. 55. Perez DM. From to man: the GPCR “tree of life”. Mol Pharma- 2008;456:350-356. col. 2005;67:1383-1384. 82. Chandra V, Huang P, Potluri N, Wu D, Kim Y, Rastinejad F. Multido- 56. Maguire ME, Van Arsdale PM, Gilman AG. An agonist-specific effect main integration in the structure of the HNF-4a nuclear receptor of guanine nucleotides on binding to the beta adrenergic receptor. complex. Nature. 2013;495:394-398. Mol Pharmacol. 1976;12:335-339. 83. Orlov I, Rochel N, Moras D, Klaholz BP. Structure of the full human 57. Wang C, Wu H, Katritch V, et al. Structure of the human smooth- RXR/VDR nuclear receptor heterodimer complex with its DR3 target ened receptor bound to an antitumour agent. Nature. DNA. EMBO J. 2012;31:291-300. 2013;497:338-343. 84. Lewis M. The lac repressor. C R Biol. 2005;328:521-548. 58. Rasmussen SG, DeVree BT, Zou Y, et al. Crystal structure of the b2 85. Lewis M. Response: DNA looping and lac repressor–CAP interaction. adrenergic receptor-Gs protein complex. Nature. 2011;477:549-555. Science. 1996;274:1931-1932. 59. Wu B, Chien EY, Mol CD, et al. Structures of the CXCR4 chemokine 86. Changeux JP, Edelstein SJ. Allosteric mechanisms of signal transduc- GPCR with small-molecule and cyclic peptide antagonists. Science. tion. Science. 2005;308:1424-1428. 2010;330:1066-1071. 87. Katz B, Thesleff S. A study of the desensitization produced by ace- 60. Manglik A, Kruse AC, Kobilka TS, et al. Crystal structure of the m- tylcholine at the motor end-plate. J Physiol. 1957;138:63-80. opioid receptor bound to a morphinan antagonist. Nature. 88. Sauguet L, Shahsavar A, Poitevin F, et al. Crystal structures of a pen- 2012;485:321-326. tameric ligand-gated ion channel provide a mechanism for activation. 61. Kuszak AJ, Pitchiaya S, Anand JP, Mosberg HI, Walter NG, Proc Natl Acad Sci U S A. 2014;111:966-971. Sunahara RK. Purification and functional reconstitution of mono- 89. Lamichhane R, Liu JJ, Pljevaljcic G, et al. Single-molecule view of meric mu-opioid receptors: allosteric modulation of agonist binding basal activity and activation mechanisms of the G protein-coupled by Gi2. J Biol Chem. 2009;284:26732-26741. receptor b2AR. Proc Natl Acad Sci U S A. 2015;112:14254-14259. 62. Whorton MR, Bokoch MP, Rasmussen SG, et al. A monomeric G 90. Calimet N, Simoes M, Changeux JP, Karplus M, Taly A, Cecchini M. protein-coupled receptor isolated in a high-density lipoprotein parti- A gating mechanism of pentameric ligand-gated ion channels. Proc cle efficiently activates its G protein. Proc Natl Acad Sci U S A. Natl Acad Sci U S A. 2013;110:E3987-E3996. 2007;104:7682-7687. 91. Grosman C, Zhou M, Auerbach A. Mapping the conformational wave 63. Kniazeff J, Prézeau L, Rondard P, Pin JP, Goudet C. Dimers and of acetylcholine receptor channel gating. Nature. 2000;403:773-776. beyond: The functional puzzles of class C GPCRs. Pharmacol Ther. 92. Purohit P, Mitra A, Auerbach A. A stepwise mechanism for acetyl- 2011;130:9-25. choline receptor channel gating. Nature. 2007;446:930-933. 64. Fotiadis D, Liang Y, Filipek S, Saperstein DA, Engel A, Palczewski K. 93. Althoff T, Hibbs RE, Banerjee S, Gouaux E. X-ray structures of GluCl Atomic-force microscopy: Rhodopsin dimers in native disc mem- in apo states reveal a gating mechanism of Cys-loop receptors. branes. Nature. 2003;421:127-128. Nature. 2014;512:333-337. 65. Agnati LF, Guidolin D, Cervetto C, Borroto-Escuela DO, Fuxe K. 94. Nury H, Poitevin F, Van Renterghem C, et al. One-microsecond Role of iso-receptors in receptor-receptor interactions with a focus molecular dynamics simulation of channel gating in a nicotinic recep- on dopamine iso-receptor complexes. Rev Neurosci. 2016;27:1-25. tor homologue. Proc Natl Acad Sci U S A. 2010;107:6275-6280. 66. Bouvier M. Oligomerization of G-protein-coupled transmitter recep- 95. Hilf RJ, Dutzler R. Structure of a potentially open state of a proton- tors. Nat Rev Neurosci. 2001;2:274-286. activated pentameric ligand-gated ion channel. Nature. 67. Smith NJ, Milligan G. Allostery at G protein-coupled receptor homo- 2009;457:115-118. and heteromers: uncharted pharmacological landscapes. Pharmacol 96. Taly A, Delarue M, Grutter T, et al. Normal mode analysis suggests a Rev. 2010;62:701-725. quaternary twist model for the nicotinic receptor gating mechanism. 68. Lemmon MA, Schlessinger J. Cell signaling by receptor tyrosine Biophys J. 2005;88:3954-3965. kinases. Cell. 2010;141:1117-1134. 97. Prevost MS, Sauguet L, Nury H, et al. A locally closed conformation 69. Ullrich A, Schlessinger J. Signal transduction by receptors with tyro- of a bacterial pentameric proton-gated ion channel. Nat Struct Mol sine kinase activity. Cell. 1990;61:203-212. Biol. 2012;19:642-649. CHANGEUX AND CHRISTOPOULOS 19

98. Prevost MS, Moraga-Cid G, Van Renterghem C, Edelstein SJ, 124. Tao YX. Constitutive activation of G protein-coupled receptors and Changeux JP, Corringer PJ. Intermediate closed state for glycine diseases: insights into mechanisms of activation and therapeutics. receptor function revealed by cysteine cross-linking. Proc Natl Acad Pharmacol Ther. 2008;120:129-148. Sci U S A. 2013;110:17113-17118. 125. Vassart G, Costagliola S. G protein-coupled receptors: mutations 99. Lape R, Colquhoun D, Sivilotti LG. On the nature of partial agonism and endocrine diseases. Nat Rev Endocrinol. 2011;7:362-372. in the nicotinic receptor superfamily. Nature. 2008;454:722-727. 126. Leach K, Conigrave AD, Sexton PM, Christopoulos A. Towards 100. Kuzmenkin A, Bezanilla F, Correa AM. Gating of the bacterial tissue-specific pharmacology: insights from the calcium-sensing sodium channel, NaChBac: voltage-dependent charge movement receptor as a paradigm for GPCR (patho)physiological bias. Trends and gating currents. J Gen Physiol. 2004;124:349-356. Pharmacol Sci. 2015;36:215-225. 101. Zagotta WN, Hoshi T, Aldrich RW. Shaker gating. 127. Kenakin T. Agonist-receptor efficacy. II. Agonist trafficking of recep- III: Evaluation of kinetic models for activation. J Gen Physiol. tor signals. Trends Pharmacol Sci. 1995;16:232-238. 1994;103:321-362. 128. Kenakin T, Christopoulos A. Signalling bias in new drug discovery: 102. Zhao Y, Yarov-Yarovoy V, Scheuer T, Catterall WA. A gating hinge detection, quantification and therapeutic impact. Nat Rev Drug Dis- in Na + channels; a molecular switch for electrical signaling. Neuron. cov. 2013;12:205-216. 2004;41:859-865. 129. Azzi M, Charest PG, Angers S, et al. Beta-arrestin-mediated activa- 103. Catterall WA, Cestèle S, Yarov-Yarovoy V, Yu FH, Konoki K, tion of MAPK by inverse agonists reveals distinct active conforma- Scheuer T. Voltage-gated ion channels and gating modifier toxins. tions for G protein-coupled receptors. Proc Natl Acad Sci U S A. Toxicon. 2007;49:124-141. 2003;100:11406-11411. 104. Deupi X, Standfuss J, Schertler G. Conserved activation pathways 130. Violin JD, Lefkowitz RJ. Beta-arrestin-biased ligands at seven- in G-protein-coupled receptors. Biochem Soc Trans. 2012; transmembrane receptors. Trends Pharmacol Sci. 2007;28:416-422. 40:383-388. 131. Jordan VC. Chemoprevention of breast cancer with selective 105. Manglik A, Kobilka B. The role of protein dynamics in GPCR func- oestrogen-receptor modulators. Nat Rev Cancer. 2007;7:46-53. tion: insights from the b2AR and rhodopsin. Curr Opin Cell Biol. 132. Galzi J-L, Edelstein SJ, Changeux J. The multiple phenotypes of allo- 2014;27:136-143. steric receptor mutants. Proc Natl Acad Sci U S A. 1996;93:1853- 106. DeVree BT, Mahoney JP, Vélez-Ruiz GA, et al. Allosteric coupling 1858. from G protein to the agonist-binding pocket in GPCRs. Nature. 133. Leff P, Scaramellini C, Law C, McKechnie K. A three-state receptor 2016;535:182-186. model of agonist action. Trends Pharmacol Sci. 1997;18:355-362. 107. Rondard P, Pin JP. Dynamics and modulation of metabotropic gluta- 134. Rubin MM, Changeux J-P. On the nature of allosteric transitions: mate receptors. Curr Opin Pharmacol. 2015;20:95-101. implications of non-exclusive ligand binding. J Mol Biol. 108. Mackinnon JA, Gallastegui N, Osguthorpe DJ, Hagler AT, Estébanez- 1966;21:265-274. Perpiñá E. Allosteric mechanisms of nuclear receptors: insights from 135. Bokoch MP, Zou Y, Rasmussen SG, et al. Ligand-specific regulation computational simulations. Mol Cell Endocrinol. 2014;393:75-82. of the extracellular surface of a G-protein-coupled receptor. Nature. 109. Hall JM, McDonnell DP, Korach KS. Allosteric regulation of estrogen 2010;463:108-112. receptor structure, function, and coactivator recruitment by differ- 136. Liu JJ, Horst R, Katritch V, Stevens RC, Wüthrich K. Biased signaling ent estrogen response elements. Mol Endocrinol. 2002;16:469-486. pathways in b2-adrenergic receptor characterized by 19 F-NMR. 110. Meijsing SH, Pufall MA, So AY, Bates DL, Chen L, Yamamoto KR. Science. 2012;335:1106-1110. DNA binding site sequence directs glucocorticoid receptor structure 137. Kang Y, Zhou XE, Gao X, et al. Crystal structure of rhodopsin and activity. Science. 2009;324:407-410. bound to arrestin by femtosecond X-ray laser. Nature. 111. Shulman AI, Larson C, Mangelsdorf DJ, Ranganathan R. Structural 2015;523:561-567. determinants of allosteric ligand activation in RXR heterodimers. 138. Ellis IR, Schor AM, Schor SL. EGF AND TGF-alpha motogenic activ- Cell. 2004;116:417-429. ities are mediated by the EGF receptor via distinct matrix- 112. Arkhipov A, Shan Y, Das R, et al. Architecture and membrane inter- dependent mechanisms. Exp Cell Res. 2007;313:732-741. actions of the EGF receptor. Cell. 2013;152:557-569. 139. Rowlinson SW, Yoshizato H, Barclay JL, et al. An agonist-induced 113. Endres NF, Das R, Smith AW, et al. Conformational coupling across conformational change in the growth hormone receptor determines the plasma membrane in activation of the EGF receptor. Cell. the choice of signalling pathway. Nat Cell Biol. 2008;10:740-747. 2013;152:543-556. 140. Suh JM, Jonker JW, Ahmadian M, et al. Endocrinization of FGF1 114. Didenko T, Liu JJ, Horst R, Stevens RC, Wüthrich K. Fluorine-19 produces a neomorphic and potent insulin sensitizer. Nature. NMR of integral membrane proteins illustrated with studies of 2014;513:436-439. GPCRs. Curr Opin Struct Biol. 2013;23:740-747. 141. Dunker AK, Garner E, Guilliot S, et al. Protein disorder and the evo- 115. Nussinov R, Tsai CJ. Free energy diagrams for protein function. lution of molecular recognition: theory, predictions and observa- Chem Biol. 2014;21:311-318. tions. Pac Symp Biocomput. 1998;473–484. 116. Sharon M, Horovitz A. Probing allosteric mechanisms using native 142. Dunker AK, Silman I, Uversky VN, Sussman JL. Function and struc- mass spectrometry. Curr Opin Struct Biol. 2015;34:7-16. ture of inherently disordered proteins. Curr Opin Struct. Biol. 117. Weber G. Ligand binding and internal equilibria in proteins. Biochem- 2008;18:756-764. istry. 1972;11:864-878. 143. Wright PE, Dyson HJ. Intrinsically disordered proteins in cellular sig- 118. Hübner CA, Jentsch TJ. Ion channel diseases. Hum Mol Genet. nalling and regulation. Nat Rev Mol Cell Biol. 2015;16:18-29. 2002;11:2435-2445. 144. Eginton C, Cressman WJ, Bachas S, Wade H, Beckett D. Allosteric 119. Steinlein OK. Ion channel mutations in neuronal diseases: a genetics coupling via distant disorder-to-order transitions. J Mol Biol. perspective. Chem Rev. 2012;112:6334-6352. 2015;427:1695-1704. 120. Taly A, Corringer PJ, Grutter T, Prado de Carvalho L, Karplus M, 145. Hilser VJ, Thompson EB. Intrinsic disorder as a mechanism to opti- Changeux JP. Implications of the quaternary twist allosteric model mize allosteric coupling in proteins. Proc Natl Acad Sci U S A. for the physiology and pathology of nicotinic acetylcholine recep- 2007;104:8311-8315. tors. Proc Natl Acad Sci U S A. 2006;103:16965-16970. 146. Simon AJ, Vallée-Bélisle A, Ricci F, Plaxco KW. Intrinsic disorder as a 121. Zwick E, Bange J, Ullrich A. Receptor tyrosine kinase signalling as a generalizable strategy for the rational design of highly responsive, target for cancer intervention strategies. Endocr Relat Cancer. 2001; allosterically cooperative receptors. Proc Natl Acad Sci U S A. 2014; 8:161–173. 111:15048-15053. 122. Blume-Jensen P, Hunter T. Oncogenic kinase signalling. Nature. 147. Minezaki Y, Homma K, Nishikawa K. Intrinsically disordered regions 2001;411:355-365. of human plasma membrane proteins preferentially occur in the 123. Tenbaum S, Baniahmad A. Nuclear receptors: structure, function cytoplasmic segment. J Mol Biol. 2007;368:902-913. and involvement in disease. Int J Biochem Cell Biol. 1997;29:1325- 148. Flock T, Ravarani CN, Sun D, et al. Universal allosteric mechanism 1341. for Ga activation by GPCRs. Nature. 2015;524:173-179. 20 CHANGEUX AND CHRISTOPOULOS

149. Shukla AK, Manglik A, Kruse AC, et al. Structure of active 173. Traynelis SF, Wollmuth LP, McBain CJ, et al. Glutamate receptor ion b-arrestin-1 bound to a G-protein-coupled receptor phosphopep- channels: structure, regulation, and function. Pharmacol Rev. tide. Nature. 2013;497:137-141. 2010;62:405-496. 150. Shan Y, Eastwood MP, Zhang X, et al. Oncogenic mutations coun- 174. Zhu S, Paoletti P. Allosteric modulators of NMDA receptors: multi- teract intrinsic disorder in the EGFR kinase and promote receptor ple sites and mechanisms. Curr Opin Pharmacol. 2015;20:14-23. dimerization. Cell. 2012;149:860-870. 175. Karakas E, Simorowski N, Furukawa H. Subunit arrangement and 151. Wybenga-Groot LE, Baskin B, Ong SH, Tong J, Pawson T, Sicheri F. phenylethanolamine binding in GluN1/GluN2B NMDA receptors. Structural basis for autoinhibition of the Ephb2 receptor tyrosine Nature. 2011;475:249-253. kinase by the unphosphorylated juxtamembrane region. Cell. 176. Partin KM. AMPA receptor potentiators: from drug design to cogni- 2001;106:745-757. tive enhancement. Curr Opin Pharmacol. 2015;20:46-53. 152. Simons SS Jr, Edwards DP, Kumar R. Minireview: dynamic structures 177. Cestèle S, Catterall WA. Molecular mechanisms of neurotoxin action of nuclear hormone receptors: new promises and challenges. Mol on voltage-gated sodium channels. Biochimie. 2000;82:883-892. Endocrinol. 2014;28:173-182. 178. Spedding M. Activators and inactivators of Ca++ channels: new per- 153. Hill KK, Roemer SC, Churchill ME, Edwards DP. Structural and func- spectives. J Pharmacol. 1985;16:319-343. tional analysis of domains of the progesterone receptor. Mol Cell 179. Spedding M. Competitive interactions between Bay K 8644 and Endocrinol. 2012;348:418-429. nifedipine in K+ depolarized smooth muscle: a passive role for Ca2+? 154. Fischer, E.H. (1992). Protein phosphorylation and cellular regulation. Naunyn Schmiedebergs Arch Pharmacol. 1985;328:464-466. Nobel lecture. http://www.nobelprize.org/nobel_prizes/medicine/ 180. Spedding M, Berg C. Interactions between a “calcium channel ago- laureates/1992/fischer-lecture.html. nist”, Bay K 8644, and calcium antagonists differentiate calcium 155. Keov P, Sexton PM, Christopoulos A. Allosteric modulation of antagonist subgroups in K + −depolarized smooth muscle. Naunyn G protein-coupled receptors: a pharmacological perspective. Neuro- Schmiedebergs Arch Pharmacol. 1984;328:69-75. pharmacology. 2011;60:24-35. 181. Striessnig J. Pharmacology, structure and function of cardiac L-type 156. Galzi JL, Bertrand S, Corringer PJ, Changeux JP, Bertrand D. Identifi- Ca(2+) channels. Cell Physiol Biochem. 1999;9:242-269. cation of calcium binding sites that regulate potentiation of a neu- 182. Cho HS, Mason K, Ramyar KX, et al. Structure of the extracellular ronal nicotinic acetylcholine receptor. EMBO J. 1996;15:5824-5832. region of HER2 alone and in complex with the Herceptin Fab. 157. Le Novère N, Grutter T, Changeux JP. Models of the extracellular Nature. 2003;421:756-760. domain of the nicotinic receptors and of agonist- and Ca2 + −bind- 183. Franklin MC, Carey KD, Vajdos FF, Leahy DJ, de Vos AM, ing sites. Proc Natl Acad Sci U S A. 2002;99:3210-3215. Sliwkowski MX. Insights into ErbB signaling from the structure of 158. Mulle C, Léna C, Changeux JP. Potentiation of nicotinic receptor the ErbB2-pertuzumab complex. Cancer Cell. 2004;5:317-328. response by external calcium in rat central neurons. Neuron. 184. Christopoulos A. Advances in G protein-coupled receptor allostery: 1992;8:937-945. from function to structure. Mol Pharmacol. 2014;86:463-478. 159. Vernino S, Amador M, Luetje CW, Patrick J, Dani JA. Calcium modu- 185. Gentry PR, Sexton PM, Christopoulos A. Novel allosteric modulators lation and high calcium permeability of neuronal nicotinic acetylcho- of G protein-coupled receptors. J Biol Chem. 2015;290:19478- line receptors. Neuron. 1992;8:127-134. 19488. 160. Zimmermann I, Dutzler R. Ligand activation of the prokaryotic pen- 186. Conn PJ, Christopoulos A, Lindsley CW. Allosteric modulators of tameric ligand-gated ion channel ELIC. PLoS Biol. 2011;9:e1001101. GPCRs: a novel approach for the treatment of CNS disorders. Nat 161. Galzi JL, Changeux JP. Neurotransmitter-gated ion channels as uncon- Rev Drug Discov. 2009;8:41-54. ventional allosteric proteins. Curr Opin Struct Biol. 1994;4:554-565. 187. Kruse AC, Ring AM, Manglik A, et al. Activation and allosteric modu- 162. Sawyer GW, Chiara DC, Olsen RW, Cohen JB. Identification of the lation of a muscarinic acetylcholine receptor. Nature. bovine gamma-aminobutyric acid type A receptor alpha subunit resi- 2013;504:101-106. dues photolabeled by the imidazobenzodiazepine [3H]Ro15-4513. 188. Kruse AC, Kobilka BK, Gautam D, Sexton PM, Christopoulos A, J Biol Chem. 2002;277:50036-50045. Wess J. Muscarinic acetylcholine receptors: novel opportunities for 163. Smith GB, Olsen RW. Deduction of amino acid residues in the drug development. Nat Rev Drug Discov. 2014;13:549-560. GABA(A) receptor alpha subunits photoaffinity labeled with the ben- 189. De Smet F, Christopoulos A, Carmeliet P. Allosteric targeting of zodiazepine flunitrazepam. Neuropharmacology. 2000;39:55-64. receptor tyrosine kinases. Nat Biotechnol. 2014;32:1113-1120. 164. Krause RM, Buisson B, Bertrand S, et al. Ivermectin: a positive allo- 190. Fabbro D. 25 years of small molecular weight kinase inhibitors: steric effector of the alpha7 neuronal nicotinic acetylcholine recep- potentials and limitations. Mol Pharmacol. 2015;87:766-775. tor. Mol Pharmacol. 1998;53:283-294. 191. Hicklin DJ, Witte L, Zhu Z, et al. Monoclonal antibody strategies to 165. Bertrand D, Gopalakrishnan M. Allosteric modulation of nicotinic block angiogenesis. Drug Discov. Today. 2001;6:517-528. acetylcholine receptors. Biochem Pharmacol. 2007;74:1155-1163. 192. Bhaskar V, Goldfine ID, Bedinger DH, et al. A fully human, allosteric 166. Chatzidaki A, Millar NS. Allosteric modulation of nicotinic acetylcho- monoclonal antibody that activates the insulin receptor and line receptors. Biochem Pharmacol. 2015;97:408-417. improves glycemic control. Diabetes. 2012;61:1263-1271. 167. Nury H, Van Renterghem C, Weng Y, et al. X-ray structures of gen- 193. Corbin JA, Bhaskar V, Goldfine ID, et al. Improved glucose metabo- eral anaesthetics bound to a pentameric ligand-gated ion channel. lism in vitro and in vivo by an allosteric monoclonal antibody that Nature. 2011;469:428-431. increases insulin receptor binding affinity. PLoS ONE. 2014;9: 168. Li GD, Chiara DC, Sawyer GW, Husain SS, Olsen RW, Cohen JB. e88684. Identification of a GABAA receptor anesthetic binding site at subu- 194. Herbert C, Schieborr U, Saxena K, et al. Molecular mechanism of nit interfaces by photolabeling with an etomidate analog. J Neurosci. SSR128129E, an extracellularly acting, small-molecule, allosteric 2006;26:11599-11605. inhibitor of FGF receptor signaling. Cancer Cell. 2013;23:489-501. 169. Olsen RW. Analysis of g-aminobutyric acid (GABA) type A receptor 195. Bono F, De Smet F, Herbert C, et al. Inhibition of tumor angiogene- subtypes using isosteric and allosteric ligands. Neurochem Res. sis and growth by a small-molecule multi-FGF receptor blocker with 2014;39:1924-1941. allosteric properties. Cancer Cell. 2013;23:477-488. 170. Sauguet L, Howard RJ, Malherbe L, et al. Structural basis for poten- 196. Jang SW, Okada M, Sayeed I, et al. Gambogic amide, a selective tiation by and anaesthetics in a ligand-gated ion channel. agonist for TrkA receptor that possesses robust neurotrophic activ- Nat Commun. 2013;4:1697. ity, prevents neuronal cell death. Proc Natl Acad Sci U S A. 2007; 171. Brannigan G, Hénin J, Law R, Eckenhoff R, Klein ML. Embedded 104:16329-16334. cholesterol in the nicotinic acetylcholine receptor. Proc Natl Acad Sci 197. Hughes TS, Giri PK, de Vera IM, et al. An alternate binding site for USA. 2008;105:14418-14423. PPARg ligands. Nat Commun. 2014;5:3571. 172. Hénin J, Salari R, Murlidaran S, Brannigan G. A predicted binding site 198. Moore TW, Mayne CG, Katzenellenbogen JA. Minireview: Not pick- for cholesterol on the GABAA receptor. Biophys J. 2014;106: ing pockets: nuclear receptor alternate-site modulators (NRAMs). 1938–1949. Mol Endocrinol. 2010;24:683-695. CHANGEUX AND CHRISTOPOULOS 21

199. Arnold LA, Estébanez-Perpiñá E, Togashi M, et al. Discovery of small and that of the neuromuscular transmission. In molecule inhibitors of the interaction of the thyroid hormone recep- Nobel Lectures, Physiology or Medicine 1942–1962 (Elsevier), tor with transcriptional coregulators. J Biol Chem. 2005;280:43048- pp. 552–578. 43055. 212. Taly A, Corringer PJ, Guedin D, Lestage P, Changeux JP. Nicotinic 200. Kojetin DJ, Burris TP, Jensen EV, Khan SA. Implications of the bind- receptors: allosteric transitions and therapeutic targets in the nerv- ing of tamoxifen to the coactivator recognition site of the estrogen ous system. Nat Rev Drug Discov. 2009;8:733-750. receptor. Endocr Relat Cancer. 2008;15:851-870. 213. Lee JH, Park CK, Chen G, et al. A monoclonal antibody that targets 201. Estébanez-Perpiñá E, Arnold LA, Jouravel N, et al. Structural insight a NaV1.7 channel voltage sensor for pain and itch relief. Cell. into the mode of action of a direct inhibitor of coregulator binding to 2014;157:1393-1404. the thyroid hormone receptor. Mol Endocrinol. 2007;21:2919-2928. 214. Mukund S, Shang Y, Clarke HJ, et al. Inhibitory mechanism of an 202. Estébanez-Perpiñá E, Arnold LA, Nguyen P, et al. A surface on the allosteric antibody targeting the glucagon receptor. J Biol Chem. androgen receptor that allosterically regulates coactivator binding. 2013;288:36168-36178. Proc Natl Acad Sci U S A. 2007;104:16074-16079. 215. Rizk SS, Kouadio JL, Szymborska A, et al. Engineering synthetic anti- 203. Buzón V, Carbó LR, Estruch SB, Fletterick RJ, Estébanez-Perpiñá E. body binders for allosteric inhibition of prolactin receptor signaling. A conserved surface on the ligand binding domain of nuclear recep- Cell Commun Signal. 2015;13:1. tors for allosteric control. Mol Cell Endocrinol. 2012;348:394-402. 216. Schwarz M, Meade G, Stoll P, et al. Conformationspecific blockade 204. Andersen RJ, Mawji NR, Wang J, et al. Regression of castrate- of the integrin GPIIb/IIIa: a novel antiplatelet strategy that selec- recurrent prostate cancer by a small-molecule inhibitor of the tively targets activated platelets. Circ Res. 2006;99:25-33. amino-terminus domain of the androgen receptor. Cancer Cell. 217. Wootten D, Christopoulos A, Sexton PM. Emerging paradigms in 2010;17:535-546. GPCR allostery: implications for drug discovery. Nat Rev Drug Dis- 205. Tsigelny IF, Mukthavaram R, Kouznetsova VL, et al. Multiple spa- cov. 2013;12:630-644. tially related pharmacophores define small molecule inhibitors of 218. van der Westhuizen ET, Valant C, Sexton PM, Christopoulos A. OLIG2 in glioblastoma. Oncotarget. Published online October Endogenous allosteric modulators of G protein-coupled receptors. 30, 2015. https://doi.org/10.18632/oncotarget.5633. J Pharmacol Exp Ther. 2015;353:246-260. 206. Arnaout MA, Mahalingam B, Xiong JP. Integrin structure, allostery, 219. Huang Z, Zhu L, Cao Y, et al. ASD: a comprehensive database of and bidirectional signaling. Annu Rev Cell Dev Biol. 2005;21:381-410. allosteric proteins and modulators. Nucleic Acids Res. 2011;39:D663- 207. Gay NJ, Symmons MF, Gangloff M, Bryant CE. Assembly and locali- D669. zation of Toll-like receptor signalling complexes. Nat Rev Immunol. 2014;14:546-558. 208. Hynes RO. Integrins: bidirectional, allosteric signaling machines. Cell. 2002;110:673-687. How to cite this article: Changeux J-P, Christopoulos A. 209. Schürpf T, Springer TA. Regulation of integrin affinity on cell sur- Allosteric modulation as a unifying mechanism for receptor faces. EMBO J. 2011;30:4712-4727. function and regulation. Diabetes Obes Metab. 2017;19:4–21. 210. Black J. Drugs from emasculated hormones: the principle of syntopic https://doi.org/10.1111/dom.12959 antagonism. Science. 1989;245:486-493. 211. Bovet, D. (1964). The relationship between isosterism and competi- tive phenomena in the field of drug therapy of the autonomic