Discoveries and future significance of research into amyloid-beta/α7-containing nicotinic acetylcholine receptor (nAChR) interactions

Abstract

Initiated by findings that Alzheimer’s disease is associated with a profound loss of cholinergic markers in human brain, decades of studies have examined the interactions between specific subtypes of nicotinic acetylcholine receptors and amyloid-β [derived from the amyloid precursor protein (APP), which is cleaved to yield variable isoforms of amyloid-β]. We review the evolving understanding of amyloid-β’s roles in Alzheimer’s disease and pioneering studies that highlighted a role of nicotinic acetylcholine receptors in mediating important aspects of amyloid-β’s effects. This review also surveys the current state of research into amyloid-β / nicotinic acetylcholine receptor interactions. The field has reached an exciting point in which common themes are emerging from the wide range of prior research and a range of accessible, relevant model systems are available to drive further progress. We highlight exciting new areas of inquiry and persistent challenges that need to be considered while conducting this research. Studies of amyloid-β and the nicotinic acetylcholine receptor populations that it interacts with provide opportunities for innovative basic and translational scientific breakthroughs related to nicotinic receptor biology, Alzheimer’s disease, and cholinergic contributions to cognition more broadly.

1. Alzheimer’s disease and Amyloid-β

Alzheimer’s disease (AD) is a devastating neurodegenerative disease that accounts for approximately 2/3 of all cases of dementia. The impact of AD is staggering: in the US alone, the cost of caring for individuals with diagnoses of dementia was estimated to exceed 300B USD in 2020. In the next three decades, as the population ages, this cost of care is expected to triple [153]. AD was originally diagnosed post mortem, using histological hallmarks. These include accumulation of fibrillar amyloid-β (fAβ) plaques and neurofibrillary tangles [10]. For many years, scientific efforts focused on understanding the complex relationship between neuropathological signs of Alzheimer’s disease (AD) and its clinical manifestation (i.e. cognitive impairment). These efforts largely focused on approaches that correlated late-stage clinical symptoms in living patients with these known AD-related biomarkers [amyloid deposition (insoluble amyloid plaques), neurofibrillary tangles (hyperphosphorylated tau), and neurodegeneration; [62]. This was the underlying basis of the “amyloid cascade hypothesis,” that aberrant processing of amyloid precursor protein led Aβ to aggregate into fAβ deposits (in particular, neuritic plaques), and that this was causally related to AD development and cognitive decline [123,126,52]. This hypothesis has developed into a useful conceptual framework that provides researchers and clinicians with testable hypotheses and diagnostic recommendations spanning different phases of the AD continuum [29]. However, as we will address next, continuing research has increasingly brought into question the role of amyloid plaques in AD pathological change.

2. Cognitive decline and oligomeric forms of Aβ

Studies over approximately the last two decades have deemphasized or shown marginal contributions of neuritic plaques and neurofibrillary tangles to dementia [15,159,84,96,98]. In addition, repeated failures of plaque-removing therapies during clinical trials have brought the classical amyloid hypothesis into further doubt or, at the least, indicated that other factors may play more important roles earlier during disease development [106,60,89,88]. While a significant amount of evidence indicates that Aβ plays a causative and initiating role in the development of AD, only a limited number of anti-Aβ agents have proven effective in clinical trials. For example, lecanumab-irmb (aka Laqembi), an immunoglobulin targeted against insoluble forms of Aβ, reduces markers of Aβ in early AD [111] but is associated with amyloid-related imaging abnormities (i.e. brain swelling or bleeding). Even the US Food and Drug Administration’s approval of Aducanumab, a monoclonal antibody treatment that reduces amyloid plaque loads [128], to treat AD has been controversial since patients receiving the drug did not see statistically-significant benefits over subjects receiving placebo and suffered significant side-effects [150]. Lastly, clinical trials using gantenerumab, a fully-human monoclonal IgG1 antibody designed to target and bind to aggregated forms of beta-amyloid, including oligomers, fibrils and plaques, exhibited an underwhelming level of Aβ removal and failed to improve the rate of cognitive and functional decline in a pair of phase 3 clinical trials [40]. These factors have all fueled the overall impression that other mechanisms than deposition of neuritic plaques may be more important [115].

As the primacy of the classical amyloid hypothesis was being challenged, several lines of evidence emphasized instead the important role of soluble, oligomeric forms of amyloid-β (oAβ), and their impact on the physiological processes that mediate cognition [21]. For example, oAβ begins to accumulate earlier during AD development than plaques [51,73,75]. The discovery of the Osaka familial AD mutation demonstrated that individuals could suffer from severe cognitive loss [130,138] while simultaneously exhibiting elevated soluble oAβ and extremely low levels of neuritic plaques [130,138,61,72]. Further studies have shown that plaque loads correlate poorly, while concentrations of soluble oAβ correlate better, with cognitive decline in both human AD patients and mouse models of AD [1,35,47,97].

The preceding findings raised questions about how circulating, soluble forms of amyloid play roles during the early stages of AD pathologic change. Specifically, how elevated levels of circulating oAβ alter neuronal and network-level function and, in turn, the development, integration, and retrieval of memory long before the expression of clinical symptoms [104]. Findings from animal models have provided new insights into causal upstream roles for oAβ in the alterations of neural circuits that are commonly affected in AD [16,105]. One key question is which form or forms of oAβ are most responsible for AD initiation and/or progression. Many in vivo and in vitro models support the hypothesis that oAβ₄₂ (oligomers of the Aβ amino acids 1–42) play a particularly critical role in neurodegeneration and subsequent memory deficits [120,48]. Together, the factors mentioned in this section have provided substantial momentum to inquiries into soluble oAβ in general, and oAβ₄₂, in particular. As we will address later, oAβ₄₂ likely interferes with stable neuronal function [104], and this understanding opens important avenues for understanding the basis of neuronal and network-level instability, neurodegeneration, and memory loss in AD (Table 1).

Table 1.

Summary of studies investigating the role of Aβ₄₂ on cellular and circuit-level function.

Publication	Model and/or Aβ application	Site of Action	Phenotype
[41]	APP/PS1 mice; Basal forebrain slices; SH-EPI cells; oligomeric Aβ42 [100 nM]	α7- and a7b2-nAChR	Direct activation of α7 nAChR with preferential alteration of α7β2 nAChR kinetics; septal cholinergic hyperexcitation
[50]	Hippocampal CA1 pyramidal neurons; oligomeric Aβ42 [200 pM]	α7*-nAChR	Enhances mEPSP frequency; increased length of postsynaptic density; maintenance of hippocampal LTP
[58]	APP/PS1 mice	Unknown	Reduction of Purkinje cell intrinsic membrane excitability; altered GABAergic and glutamatergic release from climbing fiber interneurons
[119]	Hippocampal slices; exogenous Aβ42 [200 pM]	α7*-nAChR	Aβ regulation of presynaptic neurotransmitter release; maintenance of hippocampal LTP
[13]	PSAPP mice	Voltage dependent Na+ channels	CA1 pyramidal cell hyperexcitability
[64]	Hippocampal slices; 5XFAD mice	Carbachol-sensitive mAChR	decreased CA1 hippocampal neuronal excitability; enhancement of the afterhyperpolarization (AHP)
[139]	NG108–15 cells	α7*-nAChR	Pronounced and sustained increases in Ca2+ transients from axonal varicosities of NG108–15 cells
[66]	NG108–15 cells; oligomeric Aβ42 [100 pM to 100 nM]	α7*-nAChR	induces Ca2+ responses in NG108–15 neuritic varicosities
[158]	Drosophila giant fiber system; (Gal4)–UAS driven Aβ42	Unknown	Age-dependent depletion of presynaptic mitochondria and neurotransmission failure
[90]	Cortical and hippocampal synaptosomes; oligomeric Aβ42 [10PM to 100 nM]	α7*-nAChRs	Alterations in presynaptic neurotransmission; differential neuromodulation by Aβ of synapses in hippocampus and cortex
[34]	APPα7KO mouse model	α7*-nAChR	α7KO normalizes Aβ-induced LTP deficits; prevents learning and memory deficits observed in APP-overexpressing animals
[101]	Cultured hippocampal neurons; oligomeric Aβ42 [100 nM]	P/Q-type calcium channels	Inhibition of P/Q-Type Calcium Currents; suppression of spontaneous synaptic activity
[118]	Hippocampal slices; monomeric and oligomeric Aβ42 [200 pM]	α7*-nAChR	Enhancement of synaptic plasticity and enhancement of reference and contextual fear memory
[18]	Basal forebrain slices; Aβ42 [100 nM]	Non- α7*-nAChRs	Differential effects on mEPSC frequency but not amplitude in DBB neurons
[17]	Hippocampal slices	α7*-nAChR	Reduced neurotransmission and excitatory post-synaptic potentials (EPSPs); impairment of LTP
[33]	Aβ42 [1 pM to 100 nM]	α7*-nAChR	Sustained, increases in presynaptic Ca2+ via α7-containing and non- α7-containing nAChR
[39]	Basal forebrain cholinergic neurons; oligomeric Aβ42 [100 nM]	Non- α7*-nAChR	Activates non- α7 nAChR; Ab-induced DBB neuronal depolarizations; Ab-induced inward currents from DBB neurons
[30]	Acute hippocampal slices and Tg2576 mice	α7*-nAChRs	In vivo elevation of Aβ leads to the upregulation of α7*-nAChR protein.

3. AD and cholinergic signaling

Initial signs of cholinergic deficits in AD brain came from postmortem studies. These showed greater losses of cholinergic markers in cortical and hippocampal regions of AD patients than age-matched, cognitively-normal controls [110,11,149,28]. These findings gave rise to the “cholinergic hypothesis of AD” which “suggests that a dysfunction of acetylcholine containing neurons in the brain contributes substantially to the cognitive decline observed in those with advanced age and Alzheimer’s disease” [136]. The majority of cholinergic neurons in the mammalian brain are found in four regions. These include the brainstem pedunculo-pontine and lateral dorsal tegmental nuclei, a subset of thalamic nuclei, the striatum, and the basal forebrain cholinergic nuclei [6]. Basal forebrain cholinergic neurons (BFCNs) serve as the major sources of cholinergic projection neurons to neocortex, hippocampus, and amygdala [6], provide widespread innervation of the hippocampus and cerebral cortex, and important control of cognitive processing and memory acquisition [141]. Cholinergic projections from the medial septum and ventral diagonal band (MSDB) project to the hippocampus, parahippocampus, olfactory bulb, and midline cortical structures [6]. Based on their anterior-to-posterior and medial-to-lateral distribution, the horizontal limb of the diagonal band (HDB), the nucleus basalis (NB), and the substantia innominata (SI) project to medial frontal cortical targets, basal lateral amygdala, and the perirhinal cortex [6].

Early studies into the cholinergic hypothesis of AD provide correlative evidence that a key pathological characteristic of mild-cognitive impairment (MCI) and AD includes age-related reductions in basal forebrain volume and cholinergic transmission [59,80,79]. BFCN loss is an early feature both in humans [151] and in mouse models of AD [156,8]. Further support for an important role of cholinergic signaling in AD is provided by the fact that acetylcholinesterase inhibitors provide one of the few, albeit time-limited, therapeutic options for AD [44,45,7,9]. While BFCN signaling is clearly an important mediator of the cognitive decline associated with AD, the exact mechanisms by which the degeneration of BFCNs occurs in AD remain under active investigation. This review article focuses on mechanisms that involve interactions between oAβ and nicotinic acetylcholine receptors.

4. nAChR and AD

The initial findings of cholinergic deficits in postmortem AD brains were followed up with examinations of effects on cholinergic receptor expression. As reviewed in [102], changes in muscarinic acetylcholine receptor (mAChR) expression were not consistently seen. The story was very different for nAChR, however. Before mentioning these early findings, an overview will be provided of nAChR diversity.

nAChR subtypes exist as a diverse family of subtypes, each being a pentameric complex of specific, homologous, but genetically-distinct subunits [160,27,63]. These subtypes can, to a first approximation, be divided into three categories:

1. Muscle-subtype receptors are expressed only on muscle tissue, and are composed of α1, β1, δ, and either γ or ε subunits. We will not address them further in this review.

2. Non-muscle subtypes can be subdivided into two categories, heteromeric and homomeric. Heteromeric, non-muscle nAChR subtypes also contain multiple different subunits, typically at least one of a set of α subunits (in mammalian CNS; α2, α3, α4, or α6) and at least one of the β2 or β4 subunits. These α and β subunits host orthosteric agonist binding pockets at the subunit interfaces, with a “principal” face provided by the α subunit, and a “complementary” face provided by the β subunit. Further diversity can be produced by incorporation of an accessory subunit (α5 or β3) that modifies receptor functional properties but does not directly participate in agonist binding [160,27,63] For example, the most commonly expressed subtype in mammalian brain is the α4β2*-nAChR [85], where * indicates the potential presence of additional subunits [82] such as the α5 subunit [14]. In mammalian brain, heteromeric nAChR, especially the α;4β2*-nAChR subtype, are associated with high-affinity agonist binding sites as detected in radioligand binding and autoradiography experiments [85].

3. Homomeric subtypes contain five identical subunits, and include the α7-only-nAChR subtype which is among the most common and widespread within mammalian brain [85]. In a homomeric nAChR context, both the principal and the complementary faces of agonist binding sites are provided by opposite sides of the same subunit [160,27,63]. As we will explore later most, but not all, α7*-nAChRα7* -nAChR are homomeric α7-only-nAChR. In mammalian brain, α7*-nAChR are associated with high-affinity binding of the selective antagonist α-bungarotoxin (α-Bgt) in radioligand binding and autoradiography experiments [85].

A large number of investigations used combinations of autoradiographic, immunochemical, and radioligand binding approaches to probe for changes in nAChR expression. As carefully reviewed by [23], the overall conclusions were that widespread, significant, and consistent reductions of high-affinity agonist binding site expression (i.e., of heteromeric nAChR, likely predominantly corresponding to α4β2*-nAChR) occurred in AD brain. In addition, significant losses of α-Bgt-binding sites (i.e., α7*-nAChR) also occurred in AD brain, although in a more restricted set of locations.

5. Early studies of nAChR-mediated protection against the cellular toxicity of Aβ

Inspired by the discovery that the brains of AD patients exhibit deficits of nAChR expression (see preceding section), a wide range of studies examined the ability of nAChR pharmacological agents to protect cells (often neurons) from the toxic effects of Aβ. A comprehensive overview of all of these studies is not the focus of this review, so an abridged outline will be provided here.

In an early example, rat cultured cortical neurons were shown to be protected from cell death induced by exposure to amyloid-β peptide (amino acids 25–35; Aβ_25–35) by co-incubation with nicotine, in a concentration dependent manner [68]. This same study showed that nicotine’s protective effect could be replicated using a selective α7*-nAChR agonist (DMXB), and blocked by the α7*-nAChR-selective antagonist α-Bgt. The same research team also demonstrated a similar protective effect, in the same preparation, by exposure to the α4β2*-nAChR-selective agonist cytisine, that could be blocked by the α4β2-selective antagonist DHβE [69]. However, partial protection of cultured mouse against Aβ₄₂-induced toxicity by the α7*-nAChR-selective antagonist MLA was also reported [86]. This observation questioned whether activation of nAChR, or persistent desensitization due to prolonged exposure was the cause of agonist-mediated protection [86]. Effects of compounds outside of typical competitive nAChR agonists/antagonists were also reported. Examples include protection by β-estradiol of rat PC12 cells from Aβ_25–35-induced toxicity (an effect blocked by the α7*-nAChR-selective competitive MLA and the less subtype-selective non-competitive nAChR antagonist mecamylamine [134], RNA-interference knockdown of α7*-nAChR expression (which increased vulnerability of SH-SY5Y cells to Aβ_25–35-induced toxicity; [122], and protective effects of acetylcholinesterase inhibitors such as tacrine and donepezil (protection at PC12 cells blocked by the poorly subtype-selective nAChR inhibitors mecamylamine and d-tubocurarine [135]) or galantamine [protection against Aβ₄₀-induced cell death, reversed by α-Bgt, in SH-SY5Y cells [4].

Similarly, a wide range of studies examined potential mechanisms through which nAChR-mediated protection against Aβ-induced cell death might occur. One of the earliest such studies demonstrated nicotine inhibition of Aβ-evoked activation of phospholipases A2 and D [131]. Other examples include nAChR-mediated modulation of amyloid precursor protein (APP) processing and tau phosphorylation [127,55], α7*-nAChR-mediated signaling through the phosphoinositol-3-kinase signaling [70], and Janus kinase 2 pathways [129], and nicotine-evoked changes in oxidative stress responses [46].

The many and varied outcomes of the preceding publications (and of the many other related studies not mentioned here) are hard to integrate in detail, likely due in large part to the wide range of reagents, systems, and forms of Aβ used. Nonetheless, a clear overall message does emerge: manipulation of nAChR, often α7*-nAChR in particular, can impact the toxic effects of Aβ. However, few of these studies examined whether Aβ directly interacts with nAChR, the subject of our next section.

6. Early studies of direct Aβ/nAChR interactions

A pair of studies claimed to show direct binding of amyloid-β protein (either amino acids 1–40 or 1–42; Aβ₄₀ or Aβ₄₂, respectively) to α7*-nAChR [148,147]. These studies should be treated with caution since the claimed affinities were both exceptionally high, and apparently variable across experiments (K_i values claimed at α7*-nAChR ranged from 4.1 or 5.0 pM in [147], to as low as 8 fM in [148]. Nevertheless, two subsequent studies showed rapid and reversible antagonism of whole-cell α7*-nAChR function on rat hippocampal neurons by rat Aβ₄₂. Potencies reported in both cases were in the nanomolar range (no specific IC₅₀ value; [112], IC₅₀ = 7.5 nM; [76]). Of interest, [76] also reported that inhibition was selective towards α7*-nAChR over other subtypes expressed in the preparation, and appeared to be non-competitive in nature. These observations were reinforced by [42], who demonstrated non-competitive block of α7*-nAChR with an IC₅₀ of 90 nM, and by [121], who found that nM concentrations of either Aβ₄₀ or Aβ₄₂ were able to inhibit function of α7*-nAChR, transiently enhanced that of α4β2*-nAChR, and were without effect on that of α3β4*-nAChR. An interesting counterpoint is provided by [32], in which pM concentrations of Aβ₄₂ were shown to activate α7*-nAChR expressed X. laevis oocytes, an effect diminished at higher concentrations (nM range). Contemporaneous studies did not replicate this activation of α7*-nAChR function by sub-nM Aβ₄₂ [for example [121] and [42], although the concentration dependent effects reported by [32] closely match those reported in a recent single-channel study [74]. To complicate matters further, one study [39] showed Aβ activation of non--α7*-nAChR (putatively of the α4*-nAChR subtype) in rat basal forebrain neurons.

As may be seen, the detailed results of these early studies are mixed and, in some instances, contradictory. This may reflect technical differences across the studies in how Aβ₄₀ and Aβ₄₂ solutions were prepared [and therefore the forms of Aβ aggregates present [132], and of how nAChR currents were recorded [α7*-nAChR responses are particularly sensitive to the timing and concentration of ligand application [144]. Nevertheless, an overall picture emerges that α7*-nAChR are particularly sensitive to the effects of Aβ₄₀ and Aβ₄₂, and that high concentrations of these ligands inhibit or negatively modulate function of α7*-nAChR, possibly through a non-competitive mechanism. It has been generally considered that α7 nAChR subunits assemble into a homomeric (α7-only) subtype. However, as we will explore next, evidence for expression of heteromeric α7*-nAChR emerged contemporaneously with evidence of α7-only-nAChR expression.

7. Evidence for the native expression of heteromeric α7*-nAChR

As reviewed in [20], it was only with the cloning and expression in 1990 of what is now known as the nAChR α7 subunit [124,24] that the existence of a functional, non-muscle, α-Bgt sensitive nAChR subtype was widely accepted. Even at this early stage, immunochemical experiments [124] indicated that a minority of chick α7-nAChR contained another subunit (now named α8). This initial finding of heteromeric avian α7*-nAChR expression was quickly supported by immunohistochemical evidence of co-location in chick brain of expression of α7 with α8 or β2 subunits [12], although caution must be exercised regarding the specificity of immunocytochemical labeling of nAChR α7 subunits [56]. Further circumstantial evidence for native heteromeric α7*-nAChR expression was found in differences in functional pharmacology between α7-only- and putatively α7α8-nAChR isolated from chick retina [3], and subtle differences of agonist pharmacology between artificially-expressed α7-only-nAChR and α7*-nAChR immunoisolated from chick brain [2]. However, the α8 subunit is not expressed in mammalian species, and findings suggestive of heteromeric α7*-nAChR expression in mammals were not obtained for several more years. Whole-cell patch-clamp recordings from isolated rat intracardiac and superior cervical ganglion neurons identified α7*-nAChR with unusually slow desensitization properties and the ability to recover rapidly from α-Bgt antagonism [25,26]. In the rat CNS, dual-labeling in situ mRNA hybridization demonstrated co-expression of α7 and β2 subunits, frequently in the absence of the α4 subunit that typically combines with β2, across cholinergic cells within the basal forebrain, mesopontine tegmentum, and striatum [5]. Some GABAergic neurons associated with cholinergic nuclei were also found in this study to have similar co-expression patterns of α7 with β2 but not α4 nAChR subunits. This finding was highly reminiscent of the earlier immunohistochemistry finding of α7 and β2 nAChR subunit co-expression in chick brain [12]. More recently, we and others [137,94] used affinity purification and immunoprecipitation methods to isolate α7*-nAChR and probe them for the presence of an α7β2-nAChR subpopulation. A consistent finding between both studies is that western blotting, using antibodies whose specificity has been verified in subunit-null mutant mice, identified co-assembly of β2 subunits into α7β2-nAChR in both mouse and human forebrain. This region, of course, contains the basal forebrain cholinergic neurons (BFCNs) identified in [5]. The evidence to date indicates that α7β2-nAChR are the most prominent, perhaps only, heteromeric α7*-nAChR population expressed in mammalian brain, and that their expression is most pronounced in cholinergic nuclei (and cholinergic neurons in particular). This makes BFCNs an ideal native system in which to study α7β2-nAChR, and their effects on normal and disease pharmacology.

8. Functional properties of α7β2-nAChR

However, functional pharmacology profiles of α7β2-nAChR have been most extensively assessed using artificial expression systems that allow better control over nAChR subtype expression compared to native neurons. In an early study, α7 and β2 subunits were co-expressed in Xenopus laevis oocytes, resulting in a nAChR population with significantly slower desensitization properties than α7-only-nAChR [67]. These same authors confirmed assembly of β2 subunit into functional α7β2-nAChR using a reporter-mutation approach. They also showed co-immunoprecipitation of α7 and β2 subunits, as well as reduced efficacy of a small number of agonists at α7β2-compared to α7-only-nAChR. In a later example we used a fluorescence resonance energy transfer (FRET) approach to demonstrate directly that fluorescently-tagged human α7 and β2 subunits assemble into α7β2-nAChR when stably transfected into SH-EP1 cells [95]. Initial functional studies in this paper indicated that agonist and antagonist profiles were generally similar between α7-only- and α7β2-nAChR, although possible increased sensitivity of α7β2-nAChR to the competitive antagonist dihydr-β-erythroidine (DHβE) was noted. We later performed more extensive pharmacological profiling of α7-only- vs. α7β2-nAChR expressed in X. laevis oocytes using a linked-subunit approach to ensure full control of subunit assembly and association [94]. The resulting concatenated nAChR subtypes again showed little difference in agonist or antagonist potencies, although some differences in agonist efficacy again were noted, a finding further emphasized in α7β2-nAChR formed from unlinked subunits [161]. Broadly similar findings were reported from α7-only- vs. α7β2-nAChR expressed in X. laevis oocytes using unlinked subunits [137], although this study did note somewhat increased potency of DHβE antagonism at α7β2- over α7-only-nAChR, and reduced potency of Compound B (an α7*-nAChR selective partial agonist compound).

β subunits typically make important contributions to the pharmacological profiles of competitive agonists and antagonists [81]. For this reason, the very similar pharmacological profiles of α7-only- and α7β2-nAChR suggested that both subtypes might be activated only by agonist binding at their common α7/α7 subunit interfaces. Indeed, we tested this hypothesis using a cysteine accessibility mutant approach [95]. Altering a critical residue on the complementary face of α7 (L119C) and subsequent covalent modification by the sulfhydryl reagent methanethiosulfonate ethylammonium (MTSA) blocks access of agonists to the competitive binding site of α7-only-nAChR [109]. We replicated this finding for α7β2-nAChR (please see the earlier section “Nicotinic Receptors (nAChR) and AD” for an explanation of the terms “complementary” and “principal” faces of nAChR competitive binding sites). However, when the equivalent mutation was made in the complementary face of the β2 subunit (L121C), agonist (ACh) activation of α7β2-nAChR was unaffected. We concluded that these results showed that complementary faces of α7 subunits are critical contributors to agonist binding at both α7-only- and α7β2-nAChR, while those of β2 subunits are not [95]. This conclusion was reinforced by an elegant single-channel electrophysiology study that used an “electrical fingerprinting” approach [100]. Here, α7 subunits were tagged using mutations that significantly reduced single-channel amplitude and conductance. The effect is additive, meaning that each “electrically fingerprinted” α7 subunit that is incorporated into an α7*-nAChR complex reduces open-channel amplitude by an incremental amount, allowing α7:β2 subunit ratios to be counted. This technique demonstrated that α7β2-nAChR incorporating up to three β2 subunits can be functional, a finding compatible with the concept that two adjacent α7 subunits are required to form an agonist activation site. This requirement was confirmed using a further mutation to the complementary face of the α7 nAChR subunit that prevents agonist activation via the α7/α7 subunit-interface binding site (in this case, α7[W55T]). Incorporation of this mutation into α7β2-nAChR prevented all function, showing that loss of function at α7/α7 sites could not be rescued by agonist activation via any α7/β2 sites that were present.

Two prominent exceptions have been seen to the trend of similar pharmacology between α7-only- and α7β2-nAChR. The first exception is in the case of positive allosteric modulator (PAM) compounds. This presumably arises from the fact that PAM compounds, by definition, have sites of action outside of the conventional competitive agonist binding pocket. Differences were seen for two Type I PAM compounds [100]. Most notably, 5-hydroxyindole (5-HI) had much less ability to potentiate macroscopic current responses of α7β2-than α7-only-nAChR. This finding was matched by decreased ability of 5-HI to increase open and burst durations of single-channel responses of α7β2- compared to α7-only-nAChR, with increased incorporation of β2 subunits producing progressively lower enhancement by 5-HI. In the case of the second Type I PAM (NS-1738) the ability to prolong single-channel open and burst durations was again lessened in α7β2- compared to α7-only-nAChR, although significant enhancement of macroscopic responses by NS-1738 was seen. Interestingly, the sole Type II PAM tested (PNU-120596) had indistinguishable effects at α7-only and α7β2-nAChR (regardless of the α7:β2 subunit ratio). The second exception is in responses to soluble oligomers of amyloid-β (amino acids 1–42; oAβ₄₂). We demonstrated that oAβ₄₂ (100 nM; within the range of pathophysiologically-relevant concentrations described in humans) can evoke single-channel responses in both α7-only- and α7β2-nAChR [41]. Measured in the presence of the PAM PNU-120596, burst durations fell into two categories for both subtypes (short- and long-duration bursts). For α7-only-nAChR, short-duration bursts induced by ACh or oAβ₄₂ were indistinguishable from each other (in terms of mean duration or open amplitude), and the same was true for long-duration bursts. However, the situation was more complex in the case of α7β2-nAChR. While short-duration bursts induced by ACh or oAβ₄₂ at α7β2-nAChR were again indistinguishable from each other, long-duration bursts were approximately 5 times longer when induced by oAβ₄₂ than ACh. Applying ACh and oAβ₄₂ simultaneously to α7β2-nAChR resulted in outcomes that were indistinguishable from those induced by applying oAβ₄₂ alone, suggesting a lack of competition between oAβ₄₂ and ACh (i.e., that OAβ₄₂ likely is producing activation through an allosteric mechanism, at least at α7β2-nAChR). A further publication determined that the effects of oAβ₄₂ are concentration dependent: concentrations as low as 100 pM (applied alone) produce single-channel activation of α7-only-nAChR, while high-nanomolar concentrations predominantly act as negative modulators of ACh-induced function [74]. This range of effects and concentration ranges closely matches that seen in the study of macroscopic function by [32]. Application of a spectroscopic approach also suggested that OAβ₄₂ binds outside of the ACh binding pocket [74], further reinforcing the concept that oAβ₄₂ may act in an allosteric manner at α7*-nAChR.

The overall conclusion from this section is that functional and pharmacological properties of α7-only- and α7β2-nAChR are very similar. This is due to activation of both α7*-nAChR subtypes through conserved α7/α7 subunit interface ligand binding sites, at least in the case of conventional competitive agonists and antagonists. Despite this, some differences in agonist efficacy are seen consistently and there is some evidence of slower desensitization and increased sensitivity to DHβE antagonism of α7β2- compared to α7-only-nAChR). However, there is accumulating evidence from multiple studies that some allosteric ligands are able to distinguish convincingly between these closely-related subtypes.

9. Evidence for mammalian α7β2-nAChR function and interactions with oAβ in mammalian CNS

As detailed earlier, there has been substantial evidence over the last two decades that heterologous α7*-nAChR populations may occur in both peripheral and CNS locations. The use of artificial expression systems has provided critical information regarding the functional properties of α7β2*-nAChR in particular (see preceding section), allowing comparison to the properties recorded from native α7*-nAChR that do not conform to those typically associated with α7-only-nAChR. Earlier studies provided hints that these “atypical” α7*-nAChR might be α7β2-nAChR. For example, the Yakel laboratory identified α7*-nAChR with slow desensitization properties in rat hippocampal interneurons together with a significant correlation between expression of α7 and β2 subunits in these neurons [133]. This finding of slow-desensitizing α7*-nAChR function in hippocampal interneurons was extended in [77], which also showed high sensitivity of these responses to oAβ₄₂ and DHβE. The same group had earlier [78] identified similar responses in mouse BFCNs, a population highlighted in [5] as being likely to host α7β2-nAChR. We found further evidence of α7β2-nAChR function contributions in specific populations of mouse BFCNs. Chronic application of oAβ₄₂ (100 nM, 9 days) to cultured basal forebrain slices induced hyperexcitation in BFCNs of the medial septal diagonal band (MSDB) and horizontal diagonal band (HDB). This hyperexcitation phenotype was abolished both by application of α7*-nAChR-selective antagonists, and in recordings from slices from nAChR β2 subunit-null mice, indicating dependence on α7β2-nAChR expression [41]. Interestingly, BFCNs of the nucleus basalis (NB) did not show the same excitation phenotype. This may be explained by the earlier in situ hybridization study [5], which showed that α4 mRNA expression was far more prominent in NB BFCNs than the populations found in the HDB and MSDB. This may result in a higher proportion of α4β2- instead of α7β2-nAChR expression in NB BFCNs. Interestingly, AD and MCI (mild cognitive impairment) patients exhibit up-regulation of α7*-nAChR messenger RNA expression in NB BFCNs, with no differences found for other nAChR subtypes [22]. The usual caveat must be considered in this case, that changes in mRNA expression do not necessarily correspond to proportionate changes in protein expression (see for example the [23] paper cited earlier in this review).

While there is consistent evidence for α7β2-nAChR expression on hippocampal interneurons and, especially, BFCNs, it has to be noted that α7β2-nAChR may be expressed elsewhere in the brain. Certainly there is suggestive evidence that α7β2-nAChR also may be expressed in cholinergic neurons in the striatum and mesopontine tegmentum [5], but the question of whether α7β2-nAChR might be expressed elsewhere in the brain has not been addressed systematically. This likely reflects the difficulty in distinguishing α7β2-nAChR from the larger population of α7-only-nAChR expressed across the brain. However, even our current knowledge of the locations of α7β2-nAChR expression indicates that they can have profound physiological impacts. Network function within the hippocampus is critically regulated by hippocampal interneurons [38]. Similarly, BFCNs project widely across hippocampal and cortical regions and play prominent roles in modulating neuronal circuits that mediate cognitive processing [113,157,91]. Our recent finding that abolishing α7β2-nAChR function can prevent BFCN hyperexcitation induced by chronic oAβ₄₂ exposure and rescue cognitive deficits observed in the APP/PS1 transgenic mouse model of Alzheimer’s disease [41] illustrates the physiological importance of α7β2-nAChR function and oAβ₄₂/α7β2-nAChR interactions.

10. Impacts of soluble forms of Aβ on neuronal, synaptic, and circuit function

It is critical to acknowledge that not all effects of soluble Aβ are deleterious. Various forms of soluble Aβ (including Aβ₄₂ and Aβ₄₀) are observed within the central nervous system throughout an individual’s lifetime [43,92,103]. Deficits in synaptic transmission and synaptic plasticity (including those that mediate learning and memory) are observed with pathologically-relevant levels of soluble Aβ (Snyder et al., 2005; Chapman et al., 1999; [146]; D’Amelio et al., 2011 [140]. However, lower, physiologically-relevant levels of soluble Aβ have been shown to enhance specific engrams of memory, including LTP (Puzzo et al., 2008; [119]; Gulisano et al., 2019). Interestingly, enhancement of both reference and contextual fear memory, and the mechanism(s) of action of picomolar levels of Aβ₄₂ on both synaptic plasticity and memory, involves α7-containing nicotinic acetylcholine receptors (Puzzo et al., 2008). This extraordinary sensitivity fits neatly with imaging studies of isolated rat hippocampus presynaptic nerve endings that demonstrated directly-evoked, sustained, increases in presynaptic Ca²⁺ via α7-containing and non-α7-containing nAChR by picomolar concentrations of Aβ₄₂ (Dougherty et al., 2003), and with the even earlier claims of exceptionally-high affinity binding to α7-only nAChR [148,147] noted earlier in this review. The apparently contradictory effects of soluble Aβ can be explained by investigations into concentration-dependent effects of Aβ on neural function. These have clearly shown that low concentrations play a positive, modulatory role on the physiological process that govern neurotransmission and memory [119], and, paradoxically, higher concentrations of Aβ can have an inverse effect by reducing potentiation in in neural circuits that are involved in memory formation (Puzzo et al., 2008). Another important consideration is that production of soluble Aβ and neuronal/synaptic activity are not independent. For example, Aβ levels increase during the induction of long-term potentiation (LTP; Palmeri et al., 2017) and many studies have demonstrated that Aβ can modulate neurotransmission and synaptic plasticity in neural circuits implicated in the acquisition and consolidation of memory (Araki et al., 1991; Mucke et al., 1994; Ishida et al., 1997; Meziane et al., 1998; Dougherty et al., 2003; Puzzo et al., 2008). Further, Aβ is regulated within presynaptic terminals and is implicated in vesicle cycling (Cirrito et al., 2005, 2008).

Our recent investigation illuminates the importance of soluble Aβ/nAChR interactions in the context of neuronal function. As described earlier, we showed that oAβ₄₂ exposure leads to persistent activation of α7*-nAChR, with a selective enhancement of α7β2-nAChR function in particular. We also showed hyperexcitation of BFCNs from the medial septum diagonal band (MSDB) and horizontal diagonal band (HDB), but not the nucleus basalis (NB) that was dependent on interactions of α7β2-nAChR with a pathophysiologically-relevant concentration of oAβ₄₂ [41]. The hyperexcitation phenotype (enhanced neuronal intrinsic excitability and action potential firing rates) resulted from a reduction in action potential afterhyperpolarization (AHP) and alterations in the maximal rates of voltage change during BFCN spike depolarization and repolarization. We also demonstrated that aged male and female APP/PS1 transgenic mice, genetically null for the β2 nAChR subunit gene, showed improved spatial reference memory compared with APP/PS1 aged-matched littermates. Indeed, genetic deletion of the α7-nAChR subunit (which will eliminate expression of both homomeric α7-only and heteromeric α7β2-nAChR subtypes) causes an impairment of hippocampal synaptic plasticity and memory at 12 months of age, paralleled by an increase of amyloid precursor protein (APP) expression and Aβ levels [140]. Combined, findings from this study provide a molecular mechanism supporting a role for α7β2-nAChR in mediating the effects of oAβ₄₂ on excitability of specific populations of cholinergic neurons, and provide a framework for understanding the role of α7β2-nAChR in oAβ₄₂-induced cognitive decline (Fig. 1).

Fig. 1.

Oligomeric amyloid-β (oAb₄₂) interactions and α7*-nAChR. (A) Under neuronormal conditions, both homomeric α7- and heteromeric α7β2-nAChRs are activated by the endogenous ligand acetylcholine (ACh; yellow circle). Homomeric α7-only nAChR harbor competitive agonist binding pockets at the interfaces between “principal” (+) faces of each α7 subunit and the opposing “complementary” faces (−) provided by adjacent α7 subunits. Heteromeric α7β2-nAChRs host competitive agonist binding pockets only at the interfaces between adjacent α7 subunit interfaces. (B) During periods of elevated oAb₄₂, oAb₄₂ binds to and activates both α7-only- and α7β2-nAChR subtypes. However, as described in [41] oAβ₄₂ selectively enhances the single-channel open-dwell times of α7β2-containing nAChRs. The location(s) through which oAβ produce α7*-nAChR activation are not yet defined (as indicated by a question mark, although there is evidence for a non-competitive / allosteric mode of action (please see text).

The studies mentioned in this section predominantly focus on the ability of soluble Aβ to bind and functionally modulate α7*-nAChR and, in turn, alter the cellular properties of specific neuronal populations implicated in cognition. Yet, there is more to study. It is important to consider not only 1) how soluble Aβ/nAChR interactions directly alter the functional output of individual neurons (i.e. cell autonomous effects), but also 2) how oAβ₄₂/α7*-nAChR interactions indirectly affect neuronal function by altering the functional output of local inhibitory (i. e. GABA) and/or excitatory (i.e. glutamate) neurons. Some work assessing the effects of extracellular OAβ₄₂ administration on presynaptic and postsynaptic mechanisms underlying synaptic function and memory has been described [50]. However, these observations have largely been limited to an analysis of homogenous populations of neurons or extracellular electrophysiological field recordings and measurements of hippocampal long-term potentiation. As described in [142], investigation into the effects of Aβ₄₂ on GABA-mediated synaptic transmission within the somatosensory cortex revealed reduced GABAergic input within layer V pyramidal neurons resulting from GABA_R internalization [142]. Conversely, using an oligomeric form of Aβ_1–42, Nimmrich et al. reported depressed vesicular release at GABAergic and glutamatergic synapses, due to Aβ-induced decrease in the probability of vesicle exocytosis from presynaptic terminals [101]. Given these findings, it is evident that in order to fully define the effects of oAβ₄₂ on neural function, one must consider the effect(s) of oAβ₄₂/α7*-nAChR interactions on local inhibitory and excitatory cell populations and how those local interactions can alter the balance of excitation and inhibition within local circuits, in turn changing the functional output of distinct brain nuclei (Fig. 2).

Fig. 2.

Local excitatory and inhibitory cellular processes that mediate basal forebrain cholinergic (BFCN) excitability. Within the medial septum, glutamatergic (light blue) and GABAergic (yellow) inputs regulate BFCN (green) excitability through α7*-nAChR-mediated synaptic transmission. Given the potential expression of α7*-nAChR (α7-only and/or (α7β2-nAChR; indicated by green and red cylinders, respectively) on glutamatergic and GABAergic terminals, the observed Aβ-induced BFCN hyperexcitability [41] may occur through enhanced glutamatergic synaptic transmission directly onto BFCNs or through disinhibition of GABAergic transmission onto BFCNs (i.e., circuit-based effects) In addition, oAβ₄₂/α7*-nAChR interactions directly on BFCNs produce enhanced BFCN excitability. We have shown that BFCN hyperexcitation occurs through alterations in the intrinsic properties of BFCNs [likely via alterations in function of Ca²⁺ mediated or voltage-dependent potassium (K⁺) channels; blue triangles] [41].

11. Challenges and future opportunities

In this review, we have explored early findings that indicated an important role for studies of cholinergic biology, and of nAChR in particular, in AD research. These prior studies consistently indicate the importance interactions between soluble oligomeric forms of Aβ and nAChR and consistently highlight that oAβ/α7*-nAChR are especially of interest. More recent developments have shown that while most α7*-nAChR are of the α7-only (homomeric) subtype, the less-common heteromeric α7β2-nAChR population is particularly sensitive to the effects of oAβ AND is located on neuronal populations known to be particularly vulnerable early in the development of AD. These insights give us, as a field, accessible and relevant model systems to work with and thus opportunities to make further scientific breakthroughs. We will conclude by outlining some of the specific challenges and opportunities associated with ongoing studies of oAβ/nAChR interactions, informed by the studies performed to date:

Differentiation of α7-only from α7β2-nAChR:

Extensive pharmacological characterization has proved the pharmacological profiles of α7-only and α7β2-nAChR are very similar, likely due to the fact that agonism and antagonism by conventional competitive compounds appears to only occur at α7/α7 subunit interfaces that are found in both α7*-nAChR subtypes. There is some evidence that some non-competitive/allosteric ligands are able to discriminate between α7-only and α7β2-nAChR, however. Identification of ligands that are highly-selective for α7β2-over α7-only-nAChR would be an exceptionally significant achievement, allowing definitive assignment of roles to α7β2- and/or α7-only-nAChR in mediating oAβ-induced outcomes, and permitting exploration of all locations in which the rarer (but functionally important) α7β2-nAChR population is found.

Forms of oAβ:

Especially in early studies, different Aβ fragments were studied. More recently, studies have tended to emphasize the use of Aβ₄₀ and/or oAβ₄₂ fragments. Even if focusing on these particular peptides, however, it is critically important to know the aggregation state present in the experimental system. Careful control of time and solution conditions is absolutely critical to having known, consistent, distributions of oligomeric Aβ forms [132,99]. It is also important to note that other Aβ fragments are produced in the CNS and that these can also be biologically active [114,37,36]. Careful investigations of Aβ fragments outside of Aβ₄₀ and/or oAβ₄₂ are likely to be highly informative.

oAβ concentrations:

As highlighted in earlier sections of this review, studies support both α7*-nAChR activation and inhibition by oAβ. The dissimilarities in results across prior studies appear to be directly related to the model system used to study receptor function (including receptor subunit stoichiometry) and the aggregation state and concentration of Aβ₄₂ used due to differences in preparation and handling of Aβ stocks (see preceding paragraph and [31,53]. The consideration of oAβ concentration is particularly difficult to address satisfactorily since estimates of physiologically-relevant concentrations of oAβ in normal control and AD brain vary widely from low picomolar to high micromolar [154,19]. Accordingly, both a firmer understanding of what concentrations of oAβ are pathophysiologically-relevant, and studies that examine effects across a range of oAβ concentrations are likely required to gain consistency across the field.

Preclinical model systems:

Somewhat related to the preceding is the choice of preclinical models for oAβ/AD-related studies. The use of AD animal models has given investigators valuable insight into the pathology underlying AD. While the wide range of transgenic mouse models mimic specific aspects of human AD [93,105], overexpression of specific genes [e.g. human amyloid precursor protein (APP) and presenilin 1 (PS1)] needed to reproduce AD neuropathology, including cognitive and behavioral deficits commonly associated with AD, correlate poorly to behavioral phenotypes observed in the majority of humans with AD [117,16]. Careful selection of animal models that mimic applicable aspects of aging, AD pathology, and expression of soluble oAβ concentrations appropriate to particular sets of experiments will be an important consideration for investigators interested in using these animal models toward understanding the role of cholinergic transmission within the context of AD.

Neuron / circuit model systems:

Findings to date highlight the importance of BFCNs. They are a neuronal population with proved, significant, roles in early AD. BFCNs also are the neuronal population for which the best evidence of α7β2-nAChR expression has been collected, and are readily identified [41]. The local and projection circuitry of BFCN nuclei is also well understood [6] As a result, BFCNs provide a superb native system with proven AD relevance in which to study effects of oAβ/α7β2-nAChR interactions on cell-autonomous neuronal, and local and projection circuit outcomes.

Cellular excitability:

Aβ’s ability to influence synaptic function has been well-described [116,119,125,146,53,87] and more recent findings using mouse models describe the exogenous administration of Aβ exerting its effects on synaptic plasticity through α7*-nAChR [49]. As described in [50], synaptic plasticity (driven by oAβ/α7*-nAChR interactions) modulates the cellular mechanisms that underlie transmitter release (presynaptic) and, concomitantly, treatment with picomolar concentrations of oAβ₄₂ before a weak tetanic stimulation converts the early, induction phase of hippocampal long-term potentiation (LTP) into a longer, maintained phase of LTP [49]. These recent findings underscore the need to identify the precise cellular mechanism(s) that contribute to changes in neuronal excitability in context of oAβ/α7*-nAChR associations. At the presynaptic level, the effects of Aβ on the cellular mechanisms that regulate neurotransmitter release have been described [101]. More recent studies, including our own, have focused on Aβ-induced alterations in the intrinsic excitability of single neurons. These changes have been reported in both excitatory and inhibitory neurons [13,145,41,54,58,64,65,93]. Clearly, in order to understand how elevated levels of Aβ can influence short and/or long-term changes in neuronal plasticity, Aβ/nAChR interactions occurring at both sides of the synapse must be considered. Further, more recent findings demonstrate α7*-nAChR- calcium transients mediated by a G-protein-coupled mechanism that functions synergistically to augment calcium release in response to α7*-nAChR activation [71]. These findings may provide a novel intracellular signaling cascade in which Aβ/α7*-nAChR interactions mediate not only short-term synaptic plasticity, but also downstream changes in postsynaptic neuronal excitability. Understanding the relationship between Aβ and α7*-nAChR and its effects on ion channels that mediate specific intrinsic processes governing neuronal excitability may deepen our understanding of how Aβ-induced long-term changes in neuronal function could potentially explain the increase in epileptiform activity commonly observed in AD patients [105].

Technical difficulty of measuring α7*-nAChR function:

As extensively studied by the Papke laboratory, studies of macroscopic function of α7*-nAChR are significantly complicated by the complex relationship between agonist occupation, receptor activation, and receptor desensitization [109,152]. The exceptionally rapid desensitization (in some circumstances) of α7*-nAChR also makes correct assessment of pharmacological parameters technically challenging when considering macroscopic responses [108,107,143]. As pointed out in our recent publication [41], these considerations may be especially challenging in the case of an unconventional α7*-nAChR ligand such as oAβ and there may be significant advantages to studying α7*-nAChR function at the single-channel level, instead.

12. Significance and concluding remarks

Studies of oAβ/nAChR interactions have reached an exciting phase in which earlier findings are beginning to coalesce into a more-coherent picture of disease relevance and disease-related processes. Addressing carefully and effectively the items listed in the preceding section should result in opportunities to build an internally consistent and scientifically significant research effort on this foundation. Such progress has promise to provide further insights into cholinergic control of learning and memory, and may provide us with further (and direly needed) clinical / drug-development targets for AD. Critically, investigating the cellular and circuit-level mechanisms altered by oAβ₄₂/nAChR interactions has the potential to provide a common understanding applicable to future scientific and clinical work in other conditions in which amyloidogenic processes are prominently involved. For example, similar mechanisms are known to be in play in aspects of Down syndrome [155,57,83] and studies in both areas may define common threads spanning multiple disorders. These shared mechanisms could then be targeted for therapeutic intervention by, e.g., manipulating oAβ₄₂/nAChR associations directly or through intervention at newly-defined pathways altered by these interactions.

The realization that α7β2-nAChR may be especially-important partners in oAβ/nAChR interactions also provides significant opportunities. This subtype appears to have restricted distribution and physiologically-significant roles. Accordingly, it is a potentially-promising target for the kind of therapeutic interventions just mentioned. In addition, the need to develop α7β2-nAChR using ligands that interact outside of the conventional α7/α7 agonist binding pocket seems likely to yield invaluable insights regarding how best to manipulate nAChR functional outcomes across a wide range of subtypes and associated normal and disease processes.

Abbreviations:

5-HI

5-hydroxyindole

α

Bgt

α

bungarotoxin

Aβ

amyloid-β

AD

Alzheimer’s disease

BFCN

basal forebrain cholinergic neuron

DHβE

dihydro-β-erythroidine

fAβ

fibrillar amyloid-β

FRET

Förster/fluorescence resonance energy transfer

HDB

horizontal limb of the diagonal band

mAChR

muscarinic acetylcholine choline receptor

MCI

mild cognitive impairment

MLA

methyllycaconitine

MSDB

medial septum and ventral diagonal band

nAChR

nicotinic acetylcholine receptor

NB

nucleus basalis

oAβ

oligomeric amyloid-β

PAM

positive allosteric modulator

Data availability

Data will be made available on request.