Allosteric modulation of nicotinic acetylcholine receptors

Abstract

Allosteric modulation refers to the concept that proteins could exist in multiple conformational states and that binding of allosteric ligands alters the energy barriers or “isomerization coefficients” between various states. In the context of ligand gated ion channels such as nicotinic acetylcholine receptors (nAChRs), it implies that endogenous ligand acetylcholine binds at the orthosteric site, and that molecules that bind elsewhere on the nAChR subunit(s) acts via allosteric interactions. For example, studies with the homomeric α7 nAChRs indicate that such ligand interactions can be well described by an allosteric model, and that positive allosteric effectors can affect energy transitions by (i) predominantly affecting the peak current response (Type I profile) or, (ii) both peak current responses and time course of agonist-evoked response (Type II profile). The recent discovery of chemically heterogeneous group of molecules capable of differentially modifying nAChR properties without interacting at the ligand binding site illustrates the adequacy of the allosteric model to predict functional consequences. In this review, we outline general principles of the allosteric concept and summarize the profiles of novel compounds that are emerging as allosteric modulators at the α7 and α4β2 nAChR subtypes.

Introduction

Nicotinic acetylcholine receptors (nAChRs) are widely characterized transmembrane proteins involved in the physiological responses to acetylcholine. These receptors are composed of homologous subunits encoded by a large multigene family, and are expressed in muscle, nerve and sensory cells where they play critical roles in neuronal transmission and modulation. The aim of this work is to review current knowledge on allosteric modulation at nAChRs. We first briefly discuss general principles of allosteric modulation and the properties of the α7 nAChR as a prototype of allosteric ligand gated channels and then describe the profiles of novel allosteric modulators acting at the α7 and α4β2 nAChR subtypes that are increasingly being explored in terms of therapeutic potential.

Introduced by Wymann, Monod and Changeux in 1965, the allosteric concept proposes that a protein can have multiple conformations (or states) and that binding of a molecule preferentially stabilizes the protein in a given conformation [1]. Although initially developed for enzymatic reactions, this concept was soon proposed to adequately describe properties of membrane proteins such as the endplate nAChR [2]. A key feature of the allosteric concept is that, due to thermal agitation, the protein spontaneously changes conformation in absence of external influences. Another feature of the allosteric concept is that binding of a ligand stabilizes the protein in a preferential conformation (see for review [3]). For example, in the case of nAChRs, agonists are ligands that preferentially stabilize the receptor in the active open state, whereas competitive antagonists are ligands that stabilize the protein in a closed conformational state. Fig. 1A (upper panel) schematically depicts the concept of a protein with two conformational states. By definition, the site occupied by the natural ligand, which is typically at the interface between subunit protomers is called the orthosteric site. Allosteric sites are distinct from the orthosteric site and can be localized elsewhere on the protein. Binding of the ligand at the orthosteric site stabilizes the protein in the active state, whereas binding of an effector at an allosteric site alters the overall properties by modifying the energy barriers, represented by isomerization coefficients, between one or more states (Fig. 1B). While a two state model may be sufficient to account for the properties of some proteins, it is insufficient for the description of nAChRs and the minimal model should include at least three states such as described in Fig. 1B.

nAChRs are involved in a range of synaptic and extra synaptic functions [4]. In the peripheral nervous system, nAChRs mediate ganglionic neurotransmission whereas in the CNS, nicotinic cholinergic innervation mediates fast synaptic transmission and regulates processes such as transmitter release, synaptic plasticity and neuronal network integration by providing modulatory input to a range of other neurotransmitter systems. Over the years, genetic and pharmacological studies have implicated nAChR subtypes in a range of physiological and pathophysiological functions related to cognitive function, learning and memory, reward, motor control, arousal, analgesia and immune function. These and other observations have, in recent years, attracted considerable attention to nAChR subtypes as drug targets, the working hypothesis being that compounds capable of enhancing nAChR function can ameliorate pathophysiological deficits in many disease states. Many systematic reviews published on nAChRs agonists and antagonists are available that cover these aspects in greater detail [5], [6].

nAChR subunits also host multiple allosteric sites. Binding of an allosteric ligand can affect the energy barrier between transitions resulting in a displacement of the equilibrium between states. Consequently, functional outcomes of the binding of an allosteric ligand (generally referred to as effector) can differ as a function of the transitions that are affected. Allosteric effectors that lower the energy barrier between the resting and active states increase the agonist-evoked response and are referred as positive allosteric modulators (PAM). In contrast, effectors that increase this energy barrier will cause a reduction of the agonist response and are termed negative allosteric modulators (NAM). Fig. 2A illustrates effects of lowering or increasing the L₀ coefficient on the agonist concentration–response curves. Note that for low L₀ values, a fraction of receptors remains activated even in absence of agonist. This corresponds to the spontaneous probability of transition from the resting to the active state. In other words, this implies that if a receptor displays a low isomerization coefficient between the resting and active state, a significant fraction of the receptors may be constitutively active.

Whereas it would go beyond the scope of this review to address properties of α7 nAChR in detail, it is important to recall that the α7 is the only mammalian subunit known thus far, capable of forming functional nAChRs alone. The CHRNA7 gene, which encodes for the α7 subunit, is thought to be the ancestral gene from which have progressively emerged, by genetic duplication and variations, additional genes encoding the entire nAChR family [7]. In spite of the proposition that α7 can co-assemble with other subunits to make heteromeric nAChRs [8], [9], it is widely accepted that majority of α7 nAChRs exist as homomeric receptors. The initial functional characterization of the chick α7 nAChR revealed that this homomeric receptor displays a rather low sensitivity to ACh and an unusually fast desensitization ([10], illustrated in Fig. 1B). Subsequently, numerous studies have confirmed that native α7 nAChRs of the mammalian central nervous system share similar functional properties [11], [12], [13], [14]. Another salient feature is the high calcium permeability of the α7 nAChR [12], [15], [16].

In addition to viewing the α7 nAChR as an ancestral subunit found across species ranging from C. elegans to higher mammals, this receptor can also be viewed as a prototypical allosteric protein [3], [17]. Functionally, α7 nAChRs are characterized by low agonist sensitivity, rapid activation and fast desensitization kinetics in the continued presence of the agonist. Kinetically, the receptor must have at least three possible states: (i) a resting closed state which is populated in absence of agonist, (ii) an active open state that is transiently stabilized upon exposure to the agonist and (iii) a desensitized closed state in which the receptor is stabilized by the sustained presence of the agonist. As functional studies reveal no rebound activity when the agonist is removed, it must be concluded that the return from desensitized-to-resting state occurs without detectable re-activation of the receptor. This suggests that the receptor can change conformation from the desensitized to the resting state without passing through the active state. The three state allosteric model initially proposed by Heidmann and Changeux for the neuromuscular junction receptor is therefore a minimal model that ought to be considered to describe properties of nAChRs [18], [19].

A clear enigma in understanding the functioning of ligand gated channels is how binding of a ligand provokes the opening of the ion conducting pore formed by the transmembrane segments. Analysis of the crystal structure of the acetylcholine binding protein (AChBP) resolved, at atomic resolution, has begun to provide clues about protein ligand interactions [20], [21], [22], [23]. These results confirmed previous studies indicating that the orthosteric ligand binds in the extracellular domain of the receptor at the interface formed by two adjacent subunits. Moreover, structure function studies carried out at the nAChRs and GABA_A receptors revealed that the short amino acid segment between the second and third transmembrane domain is an important determinant in the gating process that takes place during receptor activation [24], [25], [26], [27]. These studies indicate that binding of the ligand stabilizes the structure of the extracellular domain in a given conformational state and that coupling between the channel domain (CD) and the ligand binding domain (LBD) is facilitated by these amino acid interactions. The coordinated contribution of the LBD and CD are well illustrated by the effects observed in site directed mutagenesis studies. Amino acid exchanges in the LBD where shown to alter both receptor affinity and pharmacological properties. Mutations in the CD, however, yield more complex effects that can also affect both biophysical and pharmacological properties of the receptor. While it is easily understood that mutations in the selectivity filter of the ionic pore should yield differences in the ionic conductance, the pleiotropic effects caused by a single amino acid exchange (L247T) in the chick α7 nAChR are clearly of a different nature. Namely, mutation of a single amino acid facing the ionic pore causes (i) suppression of receptor desensitization, (ii) some 200-fold increase in apparent affinity for the ligand, and (iii) altered pharmacological profile converting the effects of some competitive antagonists into full agonists [28], [29]. Interestingly, however, all properties of the L247T mutant can be explained on the basis of the allosteric model assuming that this mutation renders a desensitized conductive state (see Fig. 1B). According to this hypothesis, properties of the L247T mutant should resemble those of the desensitized state rather than the active state, which also explains that competitive antagonists who stabilize the desensitized state activate the mutant receptor. The fact that competitive antagonists, such as methyllycaconitine or α-bungarotoxin, remain antagonists at the L247T mutant indicates that they stabilize the resting closed state. This hypothesis is supported by the fact that α-bungarotoxin binding is not affected by the L247T mutation [30].

Further analysis of the three-state model predicts that allosteric effectors that affect the energy barrier between resting and active states (L₀) should cause little or no modification of the agonist-evoked response kinetics. However, allosteric effectors affecting other energy transitions may cause detectable effects on the time course of the agonist-evoked response. Accordingly, two types of PAM may be recognized – Type I that predominately affects the apparent peak current, and Type II that increases both the apparent peak current and changes the desensitization profile of the agonist response. Other allosteric effectors, affecting additional properties of the receptors may also be expected. It should be noted that observations made at the α7 nAChR subtype could apply to heteromeric receptors as well, and a diversity of allosteric effects may be anticipated.

The anthelminthic agent, ivermectin (IVM) was identified as the first positive allosteric modulator of the α7 nAChR [31] (Fig. 3). Ivermectin increases maximal ACh-evoked current, modestly reduces desensitization time course, reduces the EC₅₀ value of ACh and increases the slope of the concentration–activation curve (Hill coefficient). Multiple effects caused by ivermectin can be explained by a reduction of the isomerization coefficient L₀ between resting and active states. A fundamental consequence for such reduction of the energy barrier is that the pharmacological profile of the receptor will also be modified, as for example, increasing the efficacy of partial agonists. Indeed, dimethylphenylpiperazinium (DMPP), a partial agonist at the α7 nAChR, became almost a full agonist following exposure to ivermectin [31].

Subsequent to this initial observation, several molecules have been reported to enhance the ACh-evoked currents at the α7 nAChR [32], [33], [34], [35], [36]. For example, 5-hydroxyindole (5-HI) causes a significant increase of subsequent ACh-evoked current (Fig. 3). However, 5-HI is nonselective and weak, requiring high concentrations (1–20 mM) for potentiating α7 nAChR mediated current. 5-HI also modulates the 5-HT₃ receptor [32]. Genistein, a well-known tyrosine kinase inhibitor was found to act as an α7 PAM that like 5-HI, predominantly affected apparent peak current [33], [37] (Fig. 3). As presence of Ca²⁺ in the extracellular medium was found to be necessary for proper functioning of many nAChR subtypes, it was suggested that these divalent cations could act as allosteric modulators. Confirmation of the allosteric nature of Ca²⁺ was made using site directed mutagenesis and construction of chimeras between the α7 nAChR and the 5HT₃ receptor, the latter insensitive to the extracellular Ca²⁺[38]. These studies revealed that Ca²⁺ binds in the extracellular domain of the α7 nAChR and interacts with the glutamate residue at position 172 localized in the complementary part of the ACh binding site.

The definition of an allosteric effector is also applicable to peptides and/or other proteins. This is best illustrated by the finding that SLURP-1 (secreted mammalian Ly-6/uPAR related protein 1) secreted by human keratinocytes, is a positive allosteric modulator at the human α7 nAChR [39]. Exposure of the human α7 nAChR to SLURP-1, in the nanomolar range, causes significant increases in ACh-evoked currents, accompanied by a reduction in the EC₅₀ and an increase of the slope of the concentration–activation curve [39]. The increase of the response amplitude without alteration of desensitization kinetics indicates that exposure to SLURP-1 reduces the L₀ coefficient but, otherwise, does not alter the transition equilibrium. Modulation of the native α7 nAChRs by a peptide present in the bovine serum albumin (BSA) was demonstrated in chick ciliary ganglia and rat hippocampus [40]. Fig. 2B depicts the effects of three different PAMs on concentration–activation effects of ACh at the human α7 nAChRs. Note the correspondence between theoretical (Fig. 2A) and experimental (Fig. 2B) data.

The past 3–4 years has witnessed the emergence of novel α7 nAChR PAMs although available biological data in terms of in vitro activity and selectivity profiles are somewhat limited especially those reported in patent publications. Fig. 3 presents structures of prototypical PAMs along with current traces depicting how selected PAMs modify ACh responses. (for a review, see [41]).

In addition to compounds like 5-HI and genestein described above, a number of compounds have been recently reported as positive allosteric modulators with little effect on desensitization kinetics. Ng et al. [42] recently described compound6, N-4-chlorophenyl)-α-[[(4-chloro-phenyl)amino]methylene]-3-methyl-5-isoxazole acetamide, that was derived from a chemical class previously exploited as GABA_A modulators, to enhance α7 nAChR currents evoked by ACh without affecting desensitization kinetics [42] (Fig. 3). In vivo efficacy was demonstrated in models such as sensory gating, MK-801 evoked hyper-locomotion and eight-arm radial maze suggesting that compounds with this profile can exhibit efficacy across various cognitive domains. NS-1738, a prototypical analog from the biarylurea series was reported to enhance the potency of ACh as well as the maximal efficacy [43] (Fig. 3). NS-1738 improved recognition memory performance in rat, and further, in Morris water maze, a hippocampal-dependent model of learning and memory, the compound also reversed learning impairment induced by the muscarinic antagonist scopolamine.

The acetylcholine esterase inhibitor galantamine, currently used in the symptomatic treatment of Alzheimer's disease, showed as a function of the concentration, both a potentiating effect and an inhibitory effect of type IA currents evoked by choline in cultured hippocampal neurons [34] (Fig. 3). As these currents are thought to be mediated by the activation of α7 containing receptors it was proposed that galantamine effects may, in part, be mediated by the allosteric potentiation of α7 nAChRs. It should be noted, however, that potentiatory effects by galantamine is rather modest as compared to that observed with PAMs such as 5-HI or NS-1738. The interaction of galantamine at the Aplysia AchBP, was recently resolved at the atomic level revealed that this compound binds in a region distinct from acetylcholine [44].

PNU-120596 remains one of the better characterized urea analogs [45] (Fig. 3). PNU-120596 not only increased maximal amplitude and potency of ACh-evoked α7 nAChR current by several fold, but also almost suppressed desensitization. Another striking feature of PNU-120596 is its ability to restore responses in an otherwise desensitized receptor. In other words, while continuous exposure to agonist (nicotine) desensitizes the α7 nAChR and reduces the response to a non-detectable level, PNU-120596 application during continued exposure to the agonist is able to restore a current even larger than the peak current evoked by the agonist alone (see Fig. 4 in [45]). This phenomenon does not appear to be unique to PNU-120596, but may be a common feature of Type II PAMs as revealed by analysis of structurally diverse compounds [37]. These data further supports the multiple state transformations of nAChRs described earlier, and indicate that PAMs with a Type II profile preferentially stabilizes a state of the receptor that differs from the state normally stabilized by agonists. It is interesting to note that although both PNU-120956 and NS-1738 are urea derivatives (see Fig. 3), NS-1738 exhibited only marginal effects on the desensitization kinetics of α7 nAChRs unlike PNU-120596 suggesting that these molecules may have distinct mechanisms or sites of action. In vivo studies indicate that PNU-120596 improves sensory gating deficits induced by amphetamine in rats, a model thought to reflect neuronal network alterations associated with schizophrenia. In a separate study, it was shown that PNU-120596 improved performance in a rodent short-term recognition memory model and further increased the phosphorylation of cAMP response element-binding protein (CREB), a well-recognized biochemical process implicated in learning and memory [46]. These observations collectively demonstrate that Type II PAMs also exhibit in vivo efficacy in models of cognition.

Analogs belonging to the tetrahydroquinoline series have also been described as α7 nAChR PAMs useful for the treatment of conditions associated with reduced nAChR transmission. Like PNU-120596, a prototype analog 4-naphthalene-1-yl-3a,4,5,9b-tetrahydro-3-H-cyclopenta[c]quinoline-8-sulfonic acid amide (TQS), increased peak current and slowed desensitization of current responses [37] (Fig. 3). Concentration-responses to various agonists (ACh, GTS-21 and PNU-282987) were potentiated along with leftward shift in the EC₅₀ values. In the same study, when applied at desensitized α7 nAChRs, TQS, like PNU-120596 was able to reactivate currents, unlike 5-HI or genistein.

The finding that steroids inhibit the chick nAChRs in a non-competitive and nonsteric manner provided the first evidence of allosteric modulation of heteromeric nAChRs (reviewed [47]) and prompted further investigation of the underlying mechanisms [48]. Characterization of steroid inhibition revealed that progesterone exerts, in the micromolar range, a negative allosteric effect at the α4β2 nAChR [47], [49]. These observations, later confirmed in many laboratories, have shown that progesterone and neurosteroids equally inhibit rat and human α4β2 nAChRs. A difference, however, was observed when examining the effects of a broader series of steroids at the human α4β2 nAChR where it was found that 17β-estradiol actually increases ACh-evoked current [50]. Studies carried out using the rat and human α4β2 nAChRs revealed an important species difference in the effects of 17β-estradiol. While 17β-estradiol consistently increased ACh-evoked currents at the human α4β2 nAChR, this molecule was ineffective at the rat α4β2 nAChRs [51]. Studies with the water-soluble form of 17β-estradiol pointed out an interaction between this compound and the extracellular domain of the α4β2 nAChR. Subsequently, it was subsequently found that a short segment of amino acid residues (WLAGMI) in the α4-subunit is responsible for the differences observed between the rat and human receptors [50], [51]. Although these studies could not demonstrate without ambiguity that steroids interact with the short C-terminal sequence, they nonetheless highlight the determinant role of this segment of the protein.

Divalent cations also modulate both homomeric and heteromeric receptors but their profile display significant differences. For example, zinc, which is co-released with neurotransmitters in the central nervous system, was shown to have differential effects depending on the receptor composition. While zinc, in the micromolar range, inhibited α3β2 containing receptors it causes a significant increase in ACh-evoked current at the α4β2 or α4β4 nAChRs [52]. Further analysis of zinc potentiation and inhibition revealed a determinant role of amino acids in the extracellular domain and a contribution of the histidine residue at position 162 in the N-terminal domain of the α4 subunit [53].

Developments in the identification of PAMs selective for various heteromeric nAChRs have been somewhat limited. Broad et al., recently reported 2-amino-5-ketothiazole analogs as allosteric modulators of human nAChRs [36]. For example, thiazole analogs such as LY-2087101 increased α4β2 currents evoked by low ACh concentrations by about 8-fold. This compound showed, however, little subtype selectivity as it also increased the amplitudes of α4β4 and α7 nAChRs. Selecting LY-2087101 as an prototype and using a chimeric α7/5HT₃ receptor hosting the N-terminal extracellular domain and the transmembrane and C-terminal regions of the 5HT₃ receptor, it was revealed that the site of allosteric potentiation lies in regions downstream from the N-terminus.

Although identification of the first allosteric modulators occurred almost in a serendipitous manner, targeted methods are increasingly being utilized to screen for novel positive allosteric modulators. A range of high-throughput screening approaches based on two-electrode voltage clamp, whole cell patch clamp and fluorescence-based approaches have enabled identification of different classes of molecules acting as allosteric effectors. As noted above, two types of PAM profiles have generally been observed – one that predominately affects the apparent peak current, and the other that increases both the apparent peak current and changes the desensitization profile of the agonist response. The statement “one finds what (s)he is looking for” certainly applies for the screening of allosteric effectors. Consider for example, Type II PAMs such as PNU-120596. Recalling that the ACh-evoked current at the α7 nAChR is transient, calcium influx associated with ACh-evoked responses are typically not detected using conventional fluorometric imaging plate readers such as the FLIPR. Therefore, screening assays using such technologies aimed at identifying compounds that cause an increase of the calcium influx are bound to identify molecules that predominantly modify receptor desensitization. Compounds that enhance the amplitude of the response without substantially modifying the desensitization time course are not likely to be revealed by such approaches unless modifications to the receptor or assay techniques are implemented. As positive allosteric effectors evoke larger effects at low agonist concentrations, it is to be emphasized that functional studies need to be conducted using lower range of ACh concentrations. Further, normalization of the concentration–activation curves to unity obscures an important point, i.e., variations in current amplitude responses. Finally, it is also important to determine early on that the putative effector alone does not activate the nAChR per se, and exhibit substantial selectivity versus other nAChRs and amongst other LGICs. Screening for negative allosteric modulators may prove somewhat more difficult than for positive modulators, as the criteria involves reduction of the agonist-evoked current and shift to the right of the concentration-response curve.