Molecular function of α7 nicotinic receptors as drug targets

Abstract

Nicotinic acetylcholine receptors (nAChRs) are pentameric ligand‐gated ion channels involved in many physiological and pathological processes. In vertebrates, there are seventeen different nAChR subunits that combine to yield a variety of receptors with different pharmacology, function, and localization. The homomeric α7 receptor is one of the most abundant nAChRs in the nervous system and it is also present in non‐neuronal cells. It plays important roles in cognition, memory, pain, neuroprotection, and inflammation. Its diverse physiological actions and associated disorders have made of α7 an attractive novel target for drug modulation. Potentiation of the α7 receptor has emerged as a novel therapeutic strategy for several neurological diseases, such as Alzheimer's and Parkinson's diseases, and inflammatory disorders. In contrast, increased α7 activity has been associated with cancer cell proliferation. The presence of different drug target sites offers a great potential for α7 modulation in different pathological contexts. In particular, compounds that target allosteric sites offer significant advantages over orthosteric agonists due to higher selectivity and a broader spectrum of degrees and mechanisms of modulation. Heterologous expression of α7, together with chaperone proteins, combined with patch clamp recordings have provided important advances in our knowledge of the molecular basis of α7 responses and their potential modulation for pathological processes. This review gives a synthetic view of α7 and its molecular function, focusing on how its unique activation and desensitization features can be modified by pharmacological agents. This fundamental information offers insights into therapeutic strategies.

Introduction

In the human brain, billions of nerve cells use specific signals to communicate with each other and with other target cells. One of the main processes by which this specific communication takes place is chemical synaptic transmission. At chemical synapses, the neurotransmitter, which is released into a narrow synaptic gap after depolarization of the presynaptic terminal, binds to a postsynaptic receptor. Ligand‐gated ion channels (LGIC) include a family of receptors that play a central role in fast synaptic transmission because they convert the neurotransmitter signal into an electrical one by rapidly opening a pore that allows the flux of ions through the membrane. This process of opening and closing of a transmembrane ion channel, called gating, is the fundamental event underlying channel activity. Thus, moment‐to‐moment communication relies on rapid on and off responses of LGICs.

Pentameric LGICs (pLGICs) are a family of receptors that mediate fast communication. In vertebrates, pLGICs can be cation‐selective, as nicotinic (nAChRs) and 5‐hydroxytryptamine type 3 (5‐HT₃) receptors, or anion‐selective channels, as GABA_A and glycine receptors (Le Novère & Changeux, 2001; Bouzat, 2012). Because of their vital role in converting chemical recognition into an electrical impulse, these receptors are involved in a great variety of physiological and pathological processes. The study of pLGICs has led to the development of subtype‐selective agonists and antagonists. Allosteric ligands, which can enhance or decrease receptor function from sites different to those of the endogenous neurotransmitter, have offered new paradigms for modulating receptor function. The involvement of pLGICs in pathological processes and the presence of different drug target sites in these receptors make them potential therapeutic targets for a wide variety of medical conditions.

The nicotinic receptor family

The nAChR is widely distributed throughout the animal kingdom, from nematodes to humans. In vertebrates, there are 17 different nAChR subunits that combine to yield a variety of receptors with different pharmacology and localization (Fig. 1). nAChRs are expressed in the central and peripheral nervous system, and in non‐neuronal tissues. The muscle nAChR plays a major role in neuromuscular transmission and is the target of muscle relaxants (Sine, 2012). Different nAChR types are expressed in brain and other non‐neuronal tissues; their responses to endogenous ACh and choline, and to exogenous nicotine are involved in a number of physiological processes and pharmacological effects (Wonnacott, 2014).

Figure 1. Cladogram of vertebrate nAChR subunits.

Adapted from Le Novère et al. (2002).

nAChR subunits are classified as two types, α and non‐α, with the α‐type containing a disulphide bridge in the agonist binding site (ligand‐gated ion channel database, http://www.ebi.ac.uk/compneur-srv/LGICdb/cys-loop.php). Receptors can be homomeric, which are assembled from five identical α‐type subunits, such as α7 or α9, or heteromeric, which are assembled from different subunits, and at least two of them are α‐type subunits, such as the muscle (α2βε/γδ) and several neuronal receptors (Wonnacott, 2014).

During the last decade, important advances in the resolution of 3D‐structures of pLGICs were achieved (for review, see Nemecz et al. 2016). All pLGICs share a conserved organization with five subunits symmetrically arranged around a central ion pore (Fig. 2). Functional domains include: (i) the extracellular domain (ECD), which carries the agonist binding sites at subunit interfaces. Each binding site is formed by three different loops from an α‐type subunit (loops A, B, C) that form the principal face, and three regions provided by either α‐ or non α‐type subunits that form the complementary face (loops D, E and F); (ii) the transmembrane domain (TMD), which forms the ion pore and the gate; and (iii) the intracellular domain (ICD), which contains determinants of channel conductance and it is involved in intracellular signalling (Kelley et al. 2003; Kabbani et al. 2013). The interface between the ECD and TMD, also referred to as the coupling region, is important for coupling agonist binding to channel opening (Bouzat et al. 2004; Lee & Sine, 2005), as well as for determining open channel lifetime and rate of desensitization (Bouzat et al. 2008) (Fig. 2). This region contains the conserved Cys‐loop that has given the original name (Cys‐loop receptor) to the subfamily of pLGICs.

Figure 2. Structure of nAChR.

A, side view of the α4β2 receptor (PDB code: 5KXI) (Morales‐Perez et al. 2016). The receptor is formed by the combination of five subunits arranged around a central pore. Most of the ICD is not shown since it was removed to obtain well‐diffracting crystals (Morales‐Perez et al. 2016). B, schematic representation of the ACh binding site at the interface between two adjacent subunits in the ECD. The principal face is formed by loops A, B and C from one subunit, and the complementary face is formed by loops D, E and F from the adjacent subunit. Two key residues for ACh activation are shown. C, cartoon representation of the top view of the TMD. Each subunit shows four transmembrane segments (M1–M4), where the M2 segment of each of the five subunits form the channel pore.

Functional studies in concert with biophysical, structural and computational approaches are intensively performed to reveal how the neurotransmitter binding at the ECD is translated into opening of the transmembrane ion channel, which occurs at a ∼5 nm distance (Nemecz et al. 2016). The functional response of pLGICs can be interpreted as a selection between a few and discrete classes of conformational states: Closed, open and desensitized, the latter being a non‐conductive state showing high agonist affinity. Electrophysiological recordings have shown that for each receptor there are multiple closed, open and desensitized states, as well as intermediate states between closed and open states and between open and desensitized states (Lape et al. 2008; Mukhtasimova et al. 2009; Corradi et al. 2009; Cecchini & Changeux, 2015).

Physiological significance of α7

α7 is one of the most abundant nAChRs in the nervous system. It is highly expressed in hippocampus, cortex and several subcortical limbic regions (Lendvai et al. 2013). It contributes to cognition, sensory processing information, attention, working memory, and reward pathways. Significant reduction of α7 in the brain, particularly in the hippocampus, has been reported in Alzheimer's (Kadir et al. 2006) and schizophrenic patients (Dineley et al. 2015). The α7 gene, CHRNA7 on chromosome 15, is linked to several neurological disorders, including schizophrenia, intellectual disability, bipolar disorder, autism spectrum disorders, attention deficit hyperactivity disorder, epilepsy, Alzheimer's disease, and sensory processing deficit (Sinkus et al. 2015; Dineley et al. 2015; Deutsch et al. 2016). Activation of α7 by nicotine and selective α7 agents has been shown to enhance cognitive performance in patients and animal models of neurological disorders including Alzheimer's disease, schizophrenia, brain trauma and ageing (Uteshev, 2014).

α7 is also expressed on many non‐neuronal cells, such as immune cells, astrocytes, microglia, oligodendrocyte precursor cells, endothelial cells, where it plays a role in immunity, inflammation and neuroprotection (Wessler & Kirkpatrick, 2008; Zdanowski et al. 2015; Dineley et al. 2015; Egea et al. 2015; Zanetti et al. 2016; Corradi & Bouzat, 2016). It is an important player of the cholinergic anti‐inflammatory pathway where it modulates intracellular signal pathways that result in potent anti‐inflammatory effects (Egea et al. 2015). There is also a brain cholinergic pathway that regulates microglial activation through α7, which might be important in Parkinson's disease (Stuckenholz et al. 2013), and oxygen and glucose deprivation (Parada et al. 2013). In consequence, potentiation of α7 has emerged as an important strategy for the modulation of inflammation in different pathological contexts, including sepsis, ischaemia/reperfusion, rheumatoid arthritis and pancreatitis as well as for neurological and neurodegenerative disorders.

However, enhancement of α7 activity may not be always beneficial. High levels of α7 expression promote cancer cell proliferation and metastasis in lung, colon, and bladder tissues (Wu et al. 2011; Zhao, 2016; Pepper et al. 2017). Thus, interference with α7 protein expression or treatment with specific inhibitors can reverse proangiogenic effects of nicotine and inhibit cancer cell growth (Wu et al. 2011; Pepper et al. 2017).

Activation of α7 can turn on and off longer‐lived cellular signalling events. This α7‐dependent intracellular signalling has been related to regulation of inflammatory responses in immune cells, neurite growth and neuronal protection. α7 interacts with intracellular proteins, activates G proteins, modulates intracellular signal pathways, such as JAK2–STAT3 and PI3K–Akt, and enhances intracellular calcium released from intracellular stores (Dajas‐Bailador et al. 2002; Kabbani et al. 2013; Egea et al. 2015; Corradi & Bouzat, 2016). Thus, the ability of α7 to operate as a dual metabotropic/ionotropic receptor prolongs the transient electrical response, involving intracellular calcium influx and membrane depolarization, in a more prolonged and sustained response involving multiple intracellular events.

In many non‐excitable cells, α7 channel‐independent signal transduction appears to be the dominant α7 activity, which has been related to its role in inflammation (Skok, 2009; Zanetti et al. 2016) and in neurite growth (Kabbani et al. 2013). In most immune cells, α7's ion channel activity cannot be detected, the cause of which still remains unknown. The only evidence that α7 may function also as an ion channel in immune cells was obtained in macrophages, from which single‐channel currents activated by physiological concentrations of choline, which is a full agonist of α7, were detected (Báez‐Pagán et al. 2015).

Heterologous expression of α7

The introduction of the patch clamp technique caused a revolutionary advancement in the field of ion channels, and, in particular, of nAChRs. The nAChR from skeletal muscle holds a special place in the history of its application, for it was the first biological ion channel from which unitary currents were recorded (Neher & Sakmann, 1976; Hamill & Sakmann, 1981; Hamill et al. 1981). The patch clamp technique allows currents to be registered in a wide variety of configurations, including unitary currents from cells and isolated membrane patches and macroscopic currents from the whole cell. Thus, it provides information about single‐channel amplitude and kinetics, as well as dose–response relationships and desensitization rate. The combination of mutagenesis and heterologous expression with patch clamp recordings has proved to be invaluable for understanding the structural basis of receptor function (Sine, 2012; Bouzat & Sine, 2017).

Information on α7 single‐channel properties has lagged behind that regarding other nAChRs mainly because of the difficulty of its heterologous expression and its fast kinetics. The discovery of chaperone proteins, which are involved in surface expression of α7, has largely facilitated its single‐channel characterization from heterologous expression in mammalian cells. The first identified chaperone protein was RIC‐3, an endoplasmic reticulum‐resident protein that facilitates folding of α7 subunits and their assembly into mature receptors (Halevi et al. 2002). RIC‐3 enhances the formation of surface functional α7 receptors in Xenopus oocytes (Halevi et al. 2002) and allows expression in mammalian cells that do not express it endogenously (Williams et al. 2005; Bouzat et al. 2008). Another chaperone protein, NACHO, was later discovered (Gu et al. 2016). NACHO is a transmembrane protein of endoplasmic reticulum that mediates assembly of α7 by promoting protein folding, maturation through the Golgi complex, and expression at the cell surface. The co‐expression of this chaperone with α7 in cells significantly increases its surface expression. Gu et al. (2016) showed that the level of expression and amplitude of macroscopic currents of α7 heterologously expressed in HEK 293 cells are negligible in the absence of NACHO and RIC‐3, and are larger with NACHO than with RIC‐3. The co‐expression of α7 with the two chaperone proteins produces the biggest increment in protein expression levels and peak currents. By measuring macroscopic currents in HEK cells, the authors showed that α7 kinetics, determined by desensitization and de‐activation rates and recovery rate from desensitization, are similar in cells transfected with RIC‐3 or NACHO cDNAs. Thus, it is now possible to reach high levels of surface α7 in mammalian cells by co‐expressing it with NACHO and RIC‐3 (authors’ personal observations).

α7 at the macroscopic current level

Macroscopic responses of α7 as a function of ligand concentration have yielded EC₅₀ values of ∼100–200 μm for ACh and 18–91 μm for nicotine. Choline is also a full α7 agonist, although it is relatively less potent than ACh. α7 is antagonized by α‐BTX (IC₅₀ ∼1–100 nm) and methyllycaconitine. A list of the effective concentration ranges for several natural and synthetic ligands that act as agonists or antagonists is available (Wonnacott, 2014).

A hallmark of α7 is its fast desensitization rate, which is the fastest among nAChRs. From macroscopic current recordings, this is evidenced as ACh‐elicited currents that decay very rapidly in the presence of the agonist (Fig. 3; Bouzat et al. 2008). Although desensitization rates are usually determined from the decay rates of macroscopic currents, this is not entirely certain for α7 since the experimental system and the temporal resolution of agonist exchange are limiting factors. In this respect, Pesti et al. (2014) showed a correlation between reported rise times or current decay time constants and solution exchange rates in α7. For faster solution exchange rates, the determined onset and offset kinetics were faster. Because the solution exchange is rate limiting, the intrinsic kinetics of α7 cannot be properly resolved. This is probably the main cause by which the reported desensitized rates for α7 show great variability and include values ranging from the sub‐millisecond to the second range (Pesti et al. 2014). Also, it is possible that posttranslational modifications, interactions with cell proteins or calcium concentration, which may differ among different cells, may contribute to variations in the desensitization rates among different systems.

Figure 3. Single channel and macroscopic recordings of human α7.

The receptor was expressed on mammalian BOSC‐23 cells, which are modified HEK293 cells. Top, schematic representations of patch clamp configurations used to obtain the electrophysiological recordings. Bottom left, macroscopic recordings obtained by fast perfusion of ACh or 4BP‐TQS (black traces) and ACh in the presence of PAMs (red traces). Bottom right, single channel recordings from human α7 receptors activated by ACh in the absence and presence of positive allosteric modulators (PAMs, type I: NS‐1738 and type II: PNU‐120596), and by 4BP‐TQS (allosteric agonist). Openings are shown as upward deflections. Membrane potential −70 mV.

Because of the large size and slow drug application, Xenopus oocytes are not a good expression system to determine α7 kinetics. Even more so, current decay rates of α7 expressed in HEK cells are different if macroscopic currents are recorded from whole‐cell or from smaller outside‐out patches (Bouzat et al. 2008). Under fast drug application, the desensitization time constant of α7 is below the millisecond range although the real value could be even lower (Bouzat et al. 2008).

The onset and decay rates of the current are so fast that the peak amplitude occurs already during the early onset phase of the agonist pulse, well before the solution exchange is complete. Thus, the measured response does not correspond to the nominal concentration, but to a lower one and, therefore, EC₅₀ values are not fully accurate (Papke & Thinschmidt, 1998; Papke et al. 2000; Pesti et al. 2014). More accurate EC₅₀ values can be obtained if the net charge, which represents the time integration of all channel activation, rather than the peak current is used for the analysis of the concentration–effect relationship (Corradi & Bouzat, 2016).

Given the fast desensitization onset of α7, an important issue is how rapidly the receptor returns to the activatable state after removal of the agonist, since this time will affect the inter‐response latency at a synapse at which the pulse of agonist is transient. By using a protocol of double application of a 150 ms‐pulse of 1 mm ACh, it was found that desensitized α7 receptors recover in the absence of agonist with a time constant of ∼1 s (Bouzat et al. 2008). However, recovery is delayed with longer pulses of agonist, indicating the existence of more than one desensitized conformational state (Pesti et al. 2014).

In α7, desensitization appears to be important to terminate the synaptic response. Evidence for this is supported by experiments showing that inhibitors of acetylcholinesterase (AChE) do not affect the amplitude and decay rate of α7 in both slices of rat hippocampal CA1 interneurons and oocytes (Fayuk & Yakel, 2004) and that α7‐mediated EPSCs in the mouse cortex are insensitive to perturbations of AChE (Bennett et al. 2012).

α7 at the single‐channel level

Each nAChR has its own kinetic signature, which is given by the activation pattern, mean durations of openings, bursts and closures, and open probability. The determination of these parameters is key for understanding how the receptor operates.

The single‐channel activity pattern of human α7 consists of very brief single‐channel openings that exhibit a broad distribution of current amplitudes (Mike et al. 2000; Bouzat et al. 2008; Andersen et al. 2013; daCosta & Sine, 2013). When longer time resolution is imposed during the construction of the amplitude histogram, a Gaussian distribution of amplitudes with a mean of about 10 pA (−70 mV membrane potential, 140 mm K⁺ in the extracellular solution) is observed, thus indicating that the broad distribution mainly arises from the lack of full resolution of the very brief opening events (Andersen et al. 2013).

The typical α7 activity elicited by ACh includes isolated brief opening events (0.1–0.3 ms) flanked by long closed periods and, less often, a few brief openings in quick succession, which are called bursts; each one corresponding to an activation episode of a single receptor molecule (Fig. 3). For human α7 expressed in HEK cells, open‐channel lifetime is similar to the mean burst duration and to the decay time constant of macroscopic currents (determined from outside out patches), with all values in the sub‐millisecond range. Also, the temporal pattern of single‐ACh‐activated currents does not show any significant agonist concentration dependence (Bouzat et al. 2008). This concentration‐insensitive single‐channel pattern is very different from that of the muscle nAChR, which shows well‐defined concentration‐dependent activation episodes. Each episode begins when a receptor recovers from desensitization and undergoes repeated cycles of channel closing, agonist dissociation, agonist rebinding, and channel opening until it reenters the desensitized state. Dwell times in the closed state within these episode intervals become progressively briefer with increasing ACh concentration, in contrast to the lack of agonist concentration dependence of dwell times in α7 (Akk & Auerbach, 1999; Sine, 2012). For α7, this lack of concentration dependence, together with the fact that most receptor activation episodes consist of a single opening with a duration similar to the desensitization time constant, suggests that desensitization is the predominant pathway for channel closing. However, there is still not any consensual kinetic model for explaining α7 activation (Papke et al. 2000; Mike et al. 2000; McCormack et al. 2010; Pesti et al. 2014). Among the different proposed models, the desensitized state was assumed to arise mainly either from the open (Papke et al. 2000; Mike et al. 2000; Bouzat et al. 2008) or from the resting closed state (McCormack et al. 2010; Pesti et al. 2014). The fast desensitization and brief open duration may be tuned to avoid cell toxicity caused by increased intracellular Ca²⁺ due to overstimulation since α7 is, among nAChRs, the one that shows the highest Ca²⁺/Na⁺ permeability ratio.

α7 as a drug target

The discovery of new drug targets is a key issue for the pharmaceutical industry and academic biomedical research. Several orthosteric agonists, full or partial, have been tested as candidate drugs for neurological and inflammatory processes and some of them have reached phase II in clinical assays (for reviews, see Lendvai et al. 2013; Wonnacott, 2014; Dineley et al. 2015).

The presence of drug target, topographically distinct sites has offered new paradigms for small molecules to modulate receptor function (Wenthur et al. 2014). For α7, several compounds bind to allosteric sites: (i) positive allosteric modulators (PAMs), which potentiate currents only in the presence of the orthosteric agonist; (ii) allosteric agonists, which activate receptors from non‐orthosteric sites; (iii) negative allosteric modulators (NAMs), which block the open channel, allosterically inhibiting activation, or increasing desensitization; and (iv) silent allosteric modulators (SAMs), which have no effect on orthosteric agonist responses but block allosteric potentiation (Gill‐Thind et al. 2015; Chatzidaki & Millar, 2015) (Fig. 4).

Figure 4. Scheme of modulation of α7 by different types of allosteric ligands.

Cartoon representation of the α7 receptor and the sites of action of different compounds. The response elicited by the natural agonist (ACh) can be increased, decreased or remain unaffected by the binding of the different allosteric modulators PAMs, NAMs and SAMs, respectively.

When compared to agonists, PAMs are better therapeutic tools because they: (i) better maintain the temporal spatial characteristics of endogenous activation since they only act in the presence of the neurotransmitter; (ii) show higher selectivity since the orthosteric site is more conserved among nAChR subtypes than allosteric sites; (iii) allow a higher structural diversity and final effects; and (iv) reduce tolerance due to α7 desensitization. Therefore, it is increasingly accepted that targeting α7 allosteric sites can provide promising therapeutic tools. Possible therapeutic uses of PAMs include the treatment of cognitive deficits, schizophrenia, pain and inflammatory processes (Uteshev, 2014; Dineley et al. 2015). Several PAMs and agonists have been tested in clinical trials, some of which are still underway (Hashimoto, 2015). Furthermore, endogenous compounds act as α7 PAMs, including Ca²⁺ ions (Galzi et al. 1996), SLURP‐1 (secreted mammalian Ly‐6/uPAR‐related protein 1) (Chimienti et al. 2003), and serum albumin peptides (Conroy et al. 2003).

Antagonism of α7 can be also pursued as a therapeutic strategy. Since enhanced α7 activity has been shown to play an important role in mediating oncogenic signal transduction during cancer development, inhibition of α7 activity would provide a novel therapeutic strategy (Wu et al. 2011; Pepper et al. 2017). Thus, selective antagonists or NAMs that could specifically target the overexpressed α7 would be good candidate drugs.

Electrophysiological characterization of allosteric modulators

Based on macroscopic effects, PAMs have been classified as type I PAMs, including 5‐HI (Zwart et al. 2002) and ivermectin (Krause et al. 1998), which mainly enhance agonist‐induced peak currents without significantly affecting current decay, and type II PAMs, which enhance peak currents and also delay the current decay rate (Bertrand & Gopalakrishnan, 2007; Chatzidaki & Millar, 2015). When applied to α7 desensitized receptors, type II but not type I PAMs can resensitize receptors from desensitized states (Gronlien et al. 2007; Andersen et al. 2016). This property can be clinically exploited for therapies that combine an agonist with a PAM or for enhancing activation of α7 by endogenous choline or ACh.

Probably the best characterized PAM is PNU‐120596 (N‐(5‐chloro‐2,4‐dimethoxyphenyl)‐Nʹ‐(5‐methyl‐3‐isoxazolyl)‐urea), an urea‐derived compound that is the most efficacious PAM reported so far (Hurst et al. 2005). In the presence of ACh, PNU‐120596 (1–10 μm) gives rise to significantly prolonged openings, grouped in bursts of openings separated by brief closings (200–300 μs), which in turn coalesce into long activation periods, named clusters, that can last for several seconds (daCosta et al. 2011, 2015; Palczynska et al. 2012; Andersen et al. 2016) (Fig. 3). Not all type II PAMs produce such a dramatic effect. PAM‐2 (3‐furan‐2‐yl‐N‐p‐tolylacrylamide), PAM‐3 (3‐furan‐2‐yl‐N‐o‐tolylacrylamide), and PAM‐4 (3‐furan‐2‐yl‐N‐phenylacrylamide) are amide‐derived compounds more recently introduced that behave macroscopically as type II PAMs (Arias et al. 2011). Although they are relatively potent (EC₅₀ 5–25 μm; Arias et al. 2011), the increase in open duration is significantly smaller than that elicited by PNU‐120596. Also, they produce short bursts of ∼10–40 ms instead of the typical episodes that last several seconds in the presence of PNU‐120596 (Andersen et al. 2016) (Fig. 3).

Overall, there is consensus that type II PAMs increase the open‐channel duration, the number of detectable open states, the burst or cluster duration and the open probability (Hurst et al. 2005; daCosta et al. 2011, 2015; Palczynska et al. 2012; Andersen et al. 2013). The suggested underlying mechanisms involve the increase in the energetic barrier for desensitization and/or reverse of some forms of desensitized states induced by agonists (Williams et al. 2011; Szabo et al. 2014; Bouzat & Sine, 2017). Analysis at the single‐channel level has revealed that type I PAMs, such as 5‐HI (Zwart et al. 2002) and NS‐1738 (Timmermann et al. 2007), significantly change channel kinetics by leading to prolonged openings (6‐ to 10‐fold longer than in its absence) in quick succession forming short bursts (∼4 and ∼12 ms for 5‐HI and NS‐1738, respectively) (Andersen et al. 2016, Fig. 3). Thus, all PAMs prolong open channel lifetime and elicit activation in episodes that cover a wide spectrum of different durations. The basis for the distinct effects of type I and type II PAMs is still uncertain. It could be accounted by differences in the chemical structure of compounds that are interacting with a common or overlapping binding site, to the presence of different binding sites, or to different allosteric effects.

An unequivocal way to identify drug binding sites is the co‐crystallization of the receptor with the drug, but no crystal structure is available for α7. The location of α7 PAM binding sites has been proposed mainly by evaluating their effects on mutant or chimeric receptors. However, changes in amino acids can alter either the binding site or the transduction pathway, and therefore the location of the binding site cannot be unequivocally determined. A chimeric receptor composed of the α7 sequence from the N‐terminus to the start of the first transmembrane segment followed by the 5‐HT_3A receptor sequence forms functional homomeric receptors when expressed in mammalian cells and has served as a prototype for investigating the pharmacology of α7 (Eiselé et al. 1993; Rayes et al. 2005). A comparison of the ability of a given PAM to potentiate α7, 5‐HT_3A and the chimeric α7‐5HT_3A has allowed the identification of the main domain involved. This strategy has been used for 5‐HI, PNU‐120596, TQS, NS‐1738, PAM‐2, ivermectin, and genistein, among others (Gronlien et al. 2007; Bertrand et al. 2008; Young et al. 2008; Collins & Millar, 2010; Andersen et al. 2016). Domains reported to be involved in PAM potentiation were the ECD at a site distinct from the ACh binding site for galantamine (Ludwig et al. 2010), the ECD for 5‐HI (Gronlien et al. 2007) and NS‐1738 (Bertrand et al. 2008), and the M2–M3 domain for NS‐1738 and genistein (Gronlien et al. 2007, 2010; Bertrand et al. 2008).

By introducing mutations at PAM sites discovered by molecular docking studies, several residues located in the transmembrane domain involved in PNU‐120596 potentiation were identified (Young et al. 2008; Collins & Millar, 2010; daCosta et al. 2011). A quintuple mutant receptor completely insensitive to PNU‐120596 resulted in a good tool for comparing the structural determinants of potentiation among PAMs. It was also found to be insensitive to another type II PAM, PAM‐2, and to the type I PAM NS‐1738, indicating common structural determinants for their actions (Andersen et al. 2016). In contrast, 5‐HI potentiates the quintuple mutant, indicating that it requires different structural determinants, which could be either different binding sites or different residues involved in its allosteric action.

Ago‐allosteric modulators are ligands that mediate receptor responses in the absence of an orthosteric ligand while also producing a potentiating effect in the presence of an orthosteric ligand (Wenthur et al. 2014). For α7, the compound 4‐(4‐bromophenyl)‐3a,4,5,9btetrahydro‐3H‐cyclopenta[c]quinoline‐8‐sulfonamide (4BP‐TQS), which is similar in chemical structure to the type II PAM TQS, behaves as an allosteric agonist (Gill et al. 2011). At the macroscopic level, responses evoked by 4BP‐TQS display a relatively slow onset and very slow decay rates (Fig. 3). At the single‐channel level, this agonist elicits prolonged openings grouped in very long‐duration clusters that resemble those elicited by ACh and PNU‐120596 (Fig. 3). It was shown that the α7M253L mutation at the M2 domain abolishes completely both the agonist and the potentiating effects of 4BP‐TQS (Gill et al. 2011), although a second extracellular site required for its activation has been proposed (Horenstein et al. 2016). The built‐in agonist signalling profiles of ago‐allosteric modulators may represent an attractive alternative therapy for degenerative diseases in which endogenous ligand tone becomes attenuated over the course of the disease.

Temperature dependence of α7 potentiation

Although most in vitro studies regarding PAM effects are performed at room temperature (RT), preclinical studies and clinical use obviously take place at physiological temperatures. At the macroscopic level, it has been shown that α7 currents potentiated by PAMs are markedly reduced if the temperature is increased (Sitzia et al. 2011). The single‐channel activation pattern in the presence of PNU‐120596 changes dramatically at 34°C with respect to RT: bursts do not coalesce into the long clusters observed at RT; they instead appear isolated, thus leading to significantly shorter activation episodes when compared to the super‐long activation episodes at RT (Andersen et al. 2016; Bouzat & Sine, 2017). As for PNU‐120596, the longest duration bursts elicited by PAM‐2 (also a type II PAM) are not detected at higher temperatures. In contrast, for 5‐HI (type I PAM) the maximal open and burst durations do not change with temperature, although the frequency of the potentiated events is higher at RT. The differences in the temperature sensitivity among PAMs could be due to different mechanisms of potentiation or that dissociation of PAMs from their binding sites is differently affected by temperature. Also, the desensitization rate increases with increasing temperature in the absence of PAMs (Sitzia et al. 2011; Jindrichova et al. 2012). Thus, potentiation by PAMs that mainly affect desensitization, such as type II PAMs, may be more sensitive to temperature.

Characterization at physiological temperatures of the potentiation by a given PAM is necessary for better extrapolating to its potential in vivo effects. High temperature sensitivity may make a drug less efficacious than predicted from the responses at RT. For some very efficacious PAMs, reduced potentiation could be beneficial to avoid toxicity.

Functional stoichiometry of homomeric α7 receptors

Homomeric receptors are the simplest structural class of pLGICs and contain five identical subunits, but most present‐day receptors are heteromeric. How could the contribution of each of the five subunits to a given receptor function be determined? In heteromeric receptors, this can be determined by replacing or omitting subunits during transfection or by using chimeric subunits. For homomeric receptors, the strategy is more complex. In principle, one can install a mutation that disables a certain function and co‐express the mutant with a non‐mutant subunit. The resulting population of receptors will consist of a mixture of receptors with zero to five mutant subunits, but the individual receptors will be indistinguishable in single‐channel recording. What is needed is a means to directly assign a kinetic signature to each individual receptor stoichiometry.

One way to define the functional stoichiometry of the homomeric α7 is the use of concatemeric receptors, which allows control of stoichiometry and limits expression to exactly one receptor subtype (Fig. 5). This strategy has been used for deciphering functional contributions of subunits of α4β2 receptors (Carbone et al. 2009; Mazzaferro et al. 2011, 2017). We recently generated a homomeric α7 receptor by linking five subunits with the tripeptide alanine‐glycine‐serine as described for α4β2. The single‐channel properties (amplitude, and open and burst duration) and PNU‐120596 potentiation of the concatemeric receptor are indistinguishable from those of wild‐type α7 (B.E. Nielsen, T. Minguez, I. Bermudez and C. Bouzat, unpublished observations). These results make concatemeric technology promising for defining the stoichiometry of a given α7 function (Fig. 5). However, studies with concatemeric receptors require strict controls, and the results should be verified with unlinked subunits to ensure that the function of native receptors is fully recapitulated.

Figure 5. Scheme showing two different ways to determine the functional stoichiometry of α7.

One strategy uses concatemeric receptors that allow the introduction of changes in any of the five identical subunits. The other corresponds to the electrical fingerprinting strategy that combines two subunits, one of which contains a reporter triple mutation (Q428R, E432R, S436R) that decreases the amplitude of single channel events to undetectable levels under the present recording conditions without affecting receptor kinetics. The mutant subunit is called α7 low conductance (α7LC). By co‐expressing α7LC with α7 subunits, receptors of different stoichiometries are formed. The amplitude of each single‐channel opening is the signature of the stoichiometry of the receptor that originated that event. Typical single‐channel traces are shown for each strategy. Channels were recorded from α7 activated by 100 μm ACh in the presence of PNU‐120596. Openings are shown as upward deflections. Membrane potential −70 mV.

Another approach to define functional stoichiometry, which has been developed in our laboratory and extensively used for α7, is the electrical fingerprinting strategy (Rayes et al. 2009; Andersen et al. 2011, 2013; daCosta & Sine, 2013) (Fig. 5). The strategy is based on the use of an α7 subunit that contains three mutations at the intracellular M3–M4 loop region (α7LC, which decreases the single‐channel conductance of the receptor to undetectable levels. The conductance mutations were originally described in 5‐HT₃ receptors. In elegant studies, Kelley et al. (2003) showed that the replacement of three arginine residues at the M3–M4 domain of 5‐HT_3A receptors by the equivalent ones in human 5‐HT_3B subunit (R432Q, R436D and R440A) gives rise to the high conductance form of the homomeric 5‐HT_3A receptor, in which single channel currents can be detected (Kelley et al. 2003; Bouzat et al. 2008).

α7 shows a very large conductance and does not contain the arginine residues shown to govern the low conductance of the 5‐HT_3A receptor; instead it contains negatively charged residues. The substitution of the homologous residues by three arginine residues (Q428R, E432R, S436R; Andersen et al. 2013) leads to an α7 receptor whose macroscopic responses can be recorded but whose single channels cannot be detected due to the low channel conductance (α7LC). These mutations are kinetically silent and only affect amplitude. Instead of the homogenous amplitude population of α7 channels in the presence of a potentiator (that allows full amplitude resolution), channels of different amplitudes are detected if α7 is expressed together with α7LC. The analysis shows that channels can be grouped into five discrete amplitude classes, and each class corresponds to pentameric receptors containing a specified number of LC subunits. Thus, the amplitude of a single opening event is the signature of the stoichiometry of the receptor that originated the event, and this is the basis of the fingerprinting strategy (Fig. 5) (Andersen et al. 2013). The second requirement is to find a mutation that affects the function under study, for example, mutations that make α7 receptors insensitive to ACh activation (W55T at the complementary face and Y190T at the principal face of the binding site; Rayes et al. 2009), or insensitive to α‐BTX inhibition (Sine et al. 2013) or to PNU‐120596 potentiation (daCosta et al. 2011). The last step involves single‐channel recordings from cells transfected with α7 and α7LC, one of which carries the additional mutations that make receptors insensitive to the function under study. The number of subunits in the pentameric receptor required for the function under study is revealed through the detected channel amplitude classes.

With this strategy, it was possible to decipher fundamental properties of α7 function. (i) Occupancy of only one ACh binding site allows activation of α7 (Andersen et al. 2013). Thus, by having four additional binding sites, α7 is adapted to function with submaximal occupancy of its sites. This property may be important for α7 activity since this receptor is found, in addition to the post‐synapsis, in pre‐ and extra‐synaptic locations as well as in many non‐neuronal cells, where it is probably exposed to low, non‐saturating, ACh concentrations (Lendvai & Vizi, 2008). (ii) Occupancy of a single site by α‐bungarotoxin is enough to prevent channel opening (daCosta et al. 2015). This finding indicates that α‐BTX binding produces a conformational arrest in which ACh occupancy of the other four sites cannot lead to activation. (iii) Four PNU‐120596 sensitive subunits are required for full potentiation of α7 (daCosta & Sine, 2013).

Perspectives

Despite decades of active investigation, α7 receptor has still to reveal some of its most intimate secrets and very fundamental questions remain unanswered. A high‐resolution crystal structure of α7 is obviously needed. In complex with α7 allosteric modulators, it will unequivocally reveal their binding site locations and will allow the identification of potential leads by virtual High‐throughput screening. There will still be much to learn about the allosteric movements involved in activation and drug modulation.

Although the ionotropic activity is of significance for α7 physiological roles, its metabotropic actions are also of importance. The transient calcium response through the ion channel is further sustained by the release of calcium from intracellular sources, and several signalling pathways are also activated. Understanding the cross‐talk between ionotropic and metabotropic activities will provide an integral vision of α7 actions and will point to novel sites for α7 modulation.

Another issue to be considered is the presence of α7 heteromeric receptors. In particular, α7β2 heteromeric receptors have been detected in several brain areas (Moretti et al. 2014; Thomsen et al. 2015). Thus, the functional distinction between α7 and heteromeric α7‐containing receptors is required.

α7 PAMs are good candidate drugs to improve cognition and to reduce inflammatory processes. However, they cover a wide spectrum of degrees of potentiation and mechanisms and, therefore, different PAMs may be used in different pathological situations. Several further points to explore include: (i) the potential toxic effects due to increased calcium in the presence of non‐desensitizing efficacious PAMs, which is an issue of controversy (Guerra‐Álvarez et al. 2015; Uteshev, 2016); (ii) the fact that increased α7 activity is associated with several types of cancer for which NAMs should be the preferred therapeutic drugs; (iii) the fact that due to the ubiquitous presence of α7 its modulation may produce beneficial effects in some tissues but adverse effects in others; (iv) the long‐term effects mediated by potentiated α7 due to its metabotropic action; (v) how PAMs affect α7 heteromeric receptors or intracellular receptors such as those found in mitochondria (Gergalova et al. 2014), and (vi) the fact that PAMs targeting multiple receptors might show better efficacy (Möller‐Acuña et al. 2015; Iturriaga‐Vásquez et al. 2015), indicating that for each pathological context their actions at other involved receptors should be tested.

Thus, although much is known about α7, a lot more remains unknown, which reminds us of Socrates’ famous phrase: “I only know that I know nothing”, and invites us to keep on working.

Additional information

Competing interests

The authors declare that there are no competing interests.

Author contributions

C.B. wrote the paper; M.L., B.E.N., J.C., and M.C.E. contributed to the writing and made the figures. All authors approved the final version of the manuscript, all persons designated as authors qualify for authorship, and all those who qualify for authorship are listed.

Funding

This work was supported by grants from Universidad Nacional del Sur (UNS), Agencia Nacional de Promoción Científica y Tecnológica (ANPCYT) and Consejo Nacional de Investigaciones Científicas y Técnicas de Argentina (CONICET).

Biography

Cecilia Bouzat is currently a research member of the National Research Council of Argentina (CONICET), associate professor of pharmacology at the National University of the South (UNS), Bahía Blanca, Argentina, Deputy Director of the Instituto de Investigaciones Bioquímicas de Bahía Blanca (INIBIBB), Argentina, and Chair of the Latin American Committee of the International Brain Research Organization (IBRO LARC). She carried out her PhD studies at UNS and postdoctoral studies at Mayo Clinic (MN, USA). Her laboratory at INIBIBB focuses on the molecular basis of function and drug modulation of pentameric ligand‐gated ion channels, particularly nicotinic and serotonin receptors.