Assembly and trafficking of nicotinic acetylcholine receptors (Review)

Abstract

Nicotinic acetylcholine receptors (nAChRs) are members of an extensive super-family of neurotransmitter-gated ion channels. In humans, nAChRs are expressed within the nervous system and at the neuromuscular junction and are important targets for pharmaceutical drug discovery. They are also the site of action for neuroactive pesticides in insects and other invertebrates. Nicotinic receptors are complex pentameric transmembrane proteins which are assembled from a large family of subunits; seventeen nAChR subunits (α1-α10, β1-β4, γ, δ and ε) have been identified in vertebrate species. This review will discuss nAChR subunit diversity and factors influencing receptor assembly and trafficking.

• Nicotinic receptor
• acetylcholine receptor
• assembly
• trafficking

Introduction

Nicotinic acetylcholine receptors (nAChRs) are members of a super-family of ligand-gated ion channels which also includes receptors for the neurotransmitters adenosine triphosphate (ATP), γ-aminobutyric acid (GABA), glutamate, glycine and 5-hydroxytryptamine (5-HT) [1], [2]. Nicotinic receptors are also members of a smaller sub-family of ligand-gated ion channels which also includes the ionotropic receptors for GABA, glycine and 5-HT and which is often referred to as the ‘Cys-loop’ family of receptors [1].

Nicotinic receptors are pentameric complexes assembled from an extensive family of subunits. In vertebrates, the 17 nAChR subunits (α1-α10, β1-β4, γ, δ and ε) can assemble into a variety of pharmacologically distinct receptor subtypes. Nicotinic receptors are expressed at the neuromuscular junction and also within the central and peripheral nervous system. More recently, nicotinic receptors have been identified in several non-neuronal cell types [3]. In humans, nAChRs have been implicated in several neuromuscular and neurological disorders [4–6], as a consequence of which they are major targets for pharmaceutical drug discovery [7], [8]. Nicotinic receptors are also expressed abundantly in insects and other invertebrates, where they are an important target for neuroactive pesticides [9], [10].

Nomenclature

The first nAChR subtype to be studied in detail was purified from the electric organ of fish such as the marine ray Torpedo and the freshwater eel Electrophorus (for a detailed review, see [11]). Four subunits were identified in the electric organ nAChR and were assigned the Greek letters α, β, γ and δ on the basis of their increasing apparent molecular weights when resolved on polyacrylamide gels. Since only the Torpedo α could be labelled by quaternary ammonium affinity-labelling reagents after polyacrylamide gel electrophoresis, it was assumed that this was the principal agonist binding site [12]. The subsequent molecular cloning of the Torpedo α subunit [13], [14] identified two adjacent cysteine residues (Cys192 and Cys 193), believed to be important in agonist binding. By convention, only nAChRs which contain two cysteine residues at positions analogous to Cys192 and Cys193 in the Torpedo α subunit have been classified as α-type subunits. It was assumed, for example on the basis of affinity-labelling experiments [12], that α subunits were agonist-binding subunits, whereas non-α subunits were ‘structural’ subunits. However, more recent data indicates that nicotinic agonists bind at subunit interfaces [15] and that both α and non-α subunits are able to contribute to the nicotinic agonist binding site [16]. As discussed in more detail below, seventeen nAChR subunits have been identified in vertebrates (α1-α10, β1-β4, γ, δ and ε). There is considerable subunit diversity amongst nAChR subtypes [17], [18] and, consequently, receptor subtypes are commonly referred to by their subunit composition. For example, α4β2 refers to a nAChR subtype containing only α4 and β2 subunits (even though the precise subunit stoichiometry may not be known). By convention, when the subunit composition of a nAChR subtype is less clearly defined this is indicated by an asterisk [19], [20]. For example, α4β2* indicates a nAChR which is known to contain α4 and β2 subunits but which may also contain additional subunit subtypes. Where both subunit composition and also subunit stoichiometry is known, the number of each subunit present in the assembled pentamer is indicated by subscript numbers, for example α7₅ or α1₂β1₁γ₁δ₁ (or, more commonly, α1₂β1γδ).

Muscle-type nAChRs

Nicotinic receptors at the vertebrate neuromuscular junction are similar in subunit composition to nAChRs expressed in the electric organ of fish such as Torpedo and Electrophorus [11]. The electric organ has anatomical similarities to skeletal muscle tissue and is an extremely abundant source of nAChRs. Indeed, much of our knowledge about nAChR structure is a consequence of studies performed with electric organ nAChRs. The electric organ nAChR was the first neurotransmitter receptor to be biochemically purified [reviewed by 11] and was the first to be characterized by molecular cloning [13], [14]. It was also the first neurotransmitter receptor to yield functional receptors when expressed in heterologous expression systems such as the Xenopus oocyte [21] and cultured cells [22].

The high density and distribution of nAChRs within the post-synaptic membrane of Torpedo electric organ has made this preparation particularly suitable for electron diffraction studies; as a consequence, this receptor was one of the first ion channels for which high resolution structural information was available [23] (Figure 1). Such studies have established that the electric organ nAChR is a pentamer composed of two α subunits co-assembled with one each of the β, γ and δ subunits, the subunits being arranged in a clockwise order of α-γ-α-β-δ [23] (Figure 2). Studies on the Torpedo nAChR have also helped to determine the membrane topology of an individual nAChR subunit [23–25] (Figure 2). Each subunit contains a single polypeptide chain containing an N-terminal extracellular domain and four α-helical transmembrane domains (M1-M4), with the second of these transmembrane domains (M2) lining the channel pore [23], [26].

Figure 1.  The three-dimensional structure of the nAChR from Torpedo electric organ at 4Å resolution. This high resolution structure was obtained from electron images of helical tubes isolated from electric organ post-synaptic membranes from the marine ray Torpedo marmorata [23]. These images were derived from the Protein Data Bank file 2BG9, coloured using the Swiss-Pdb Viewer, Deep View (www.expasy.org/spdbv) and rendered using MegaPov (www.povray.org). Individual subunits have been coloured for identification (α, red; β, blue; γ, green and δ, yellow). (A) A view from the extracellular side of the receptor. (B) A view from the side.

Figure 2.  A model illustrating nAChR subunit topology and the pentameric arrangement of subunits in an assembled nAChR. A) A cartoon model of a nAChR subunit illustrating four transmembrane domains and the extracellular N- and C-termini. B) As discussed in the text, there is evidence that all assembled nAChRs (whether homomeric or heteromeric) are pentamers. One of the best characterized nAChRs is the heteromeric receptor expressed in the electric organ of the marine ray Torpedo. This representative cartoon image illustrates the known arrangement of subunits in the Torpedo electric organ nAChR. Individual subunits are illustrated by ovals and have been shaded to indicate the subunit type (α, red; β, blue; γ, green and δ, yellow).

Five nAChRs subunits are expressed in skeletal muscle (α1, β1, γ, δ and ε), two of which (γ and ε) are developmentally regulated (the γ subunit being expressed in embryonic muscle, whereas the ε subunit is expressed in adult muscle). As a consequence of this developmental switch in gene transcription, nAChRs in embryonic muscle have the subunit composition α1₂β1γδ, whereas those in adult muscle have the composition α1₂β1δε. The subunit composition and stoichiometry of the embryonic and adult forms of muscle-type nAChRs are highly constrained. As discussed previously [27], it is thought that cells achieve such a tightly controlled assembly process through subunit oligomerization occurring along a fixed and ordered pathway. While there is some debate about the precise order in which subunits interact [27–29], the ordered interaction of subunits is likely to provide a mechanism by which complex heteromeric proteins such as the muscle nAChR can be assembled in a fixed stoichiometry. In one model (often referred to as the ‘heterodimer’ model), the α subunits undergo extensive folding events prior to formation of αγ and αδ heterodimers, which subsequently associate with the β subunit to form an assembled pentamer [30–33]. A second (‘sequential’) model concludes that α, β and γ subunits rapidly assemble into trimers, followed by the addition of a δ subunit and second α subunit. A feature of the sequential model is that important subunit folding events occur only after initial subunit interactions [34–36] and, indeed, there is evidence for conformational changes to one subunit occurring after co-assembly with a partner subunit [37]. Recently, an attempt has been made to develop a model of nAChR subunit assembly based on computer homology modelling [38]. The computer model, which uses theoretical hydrophobic and electrostatic parameters for individual receptor subunits, does not agree exactly with either of the experimentally based assembly models for the muscle-type nAChR and it is likely that a consensus concerning the two opposing empirical assembly models will require further experimental data.

Subunit composition of neuronal nAChRs

In contrast to the homogeneous population of nAChRs found at the neuromuscular junction of embryonic or adult muscle, there is considerable diversity among the sub-family of neuronal nAChRs [17], [18]. In addition to the five subunits found in muscle, twelve vertebrate ‘neuronal’ nAChR subunits (α2–α10 and β2–β4) have been identified which, with some exceptions [3], are expressed predominantly in the central and peripheral nervous system.

Early studies of neuronal nAChRs immunoprecipitated from bovine, chicken, rat and human brain suggested that a major receptor subtype is assembled from only two types of subunit, α4 and β2 [39–41]. Similar studies of native neuronal nAChRs from chick ganglia have identified receptors containing either two or three different types of subunit [42], [43]. These findings provided an early indication that neuronal nAChRs might differ in terms of subunit stoichiometry from the better-characterized skeletal muscle nAChRs. Indeed, there is evidence for neuronal nAChRs containing only one type of subunit (homomeric receptors) and also for those containing two, three and four different subunit sub-types (heteromeric receptors) [17], [18]. Despite such heterogeneity in subunit composition and stoichiometry, several lines of evidence suggest that all nAChRs contain a total of five co-assembled subunits [44–47].

Neuronal nAChR subunits such as α2, α3, α4, β2 and β4 are able, at least in heterologous expression systems, to co-assemble into functional heteromeric nAChRs containing just two different types of subunit (i.e., in ‘pair-wise’ combinations). Heterologous expression studies in Xenopus oocytes demonstrate that functional heteromeric nAChRs are generated when any of the three α subunits α2, α3, α4 are co-expressed with β2 or β4 but not when any of these five subunits is expressed alone [16], [48], [49]. These findings are consistent with data obtained from heterologous expression studies in mammalian cell lines [50–53]. Heterologous expression studies have also demonstrated that, when these subunits are expressed alone, they are retained within the cell [37], [54].

The use of reporter mutations in oocyte expression studies suggests that nAChRs such as α4β2 and α3β4 contain two α subunits coassembled with three β subunits [45], [55]. This conclusion is supported by biochemical studies [44]. There is, however, evidence that some neuronal nAChRs (e.g., α4β2) can assemble into nAChRs of alternative stoichiometries (either α4₂β2₃ or α4₃β2₂), thereby influencing the functional properties of assembled receptors [56–59].

The α5 and β3 subunits have been described as ‘orphan’ subunits, largely because, unlike other previously-cloned neuronal nAChR subunits, they fail to generate functional recombinant nAChR even when expressed in pair-wise combinations with other subunits [60], [61]. However, both α5 and β3 can co-assemble into ‘triplet’ nAChRs (which contain three different types of subunit). In heterologous expression systems, subunit combinations containing α5 can be distinguished from ‘pair-wise’ subunit combinations on the basis of altered electrophysiological properties and include: α3α5β2 [62], [63], α3α5β4 [62–64] and α4α5β2 [65]. Co-assembly of β3 into functional α3β3β4 nAChRs has been demonstrated by heterologous expression in oocytes using a reporter mutation approach [66], leading to the proposal that the stoichiometry of this subunit combination is α3₂β3₁β4₂ [55]. As discussed below, there is also evidence for the preferential assembly of β3 into α6-containing nAChRs [67–69]. In contrast to such evidence for the functional assembly of β3 into triplet nAChRs such as α3β3β4 and α6β3β4, recent studies have shown instead a dominant-negative effect on nAChR functional expression when β3 co-assembles with several other nAChR subunit combinations [70]. A recently developed approach which may prove useful in determining the composition and stoichiometry of recombinant nAChRs is that of artificial subunit concatemers [58], [71], [72], although there are several potential problems associated with such approaches which may need to be considered [71], [73].

Immunoprecipitation studies conducted on receptors purified from chick ciliary ganglion neurones have identified native nAChRs containing the α5 subunit [74–76]. Most, if not all, of the α5 subunits in this preparation are co-assembled with α3 and β4 and about 20% of α5-containing nAChRs also contain β2 [76], suggesting the formation of α3α5β4 and α3α5β2β4 subunit complexes. Immunoprecipitation studies with nAChRs native to chick optic lobe reveal high levels of α2-containing nAChRs, all of which also contain β2, with more than half also containing α5, suggesting the existence of α2α5β2 nAChRs [77]. Evidence has been obtained for the presence of α4α5β2 nAChRs in chick and rat brain [74], [78], [79], supporting electrophysiological data which demonstrates the co-assembly of these subunits in oocytes [65]. Further support for the formation of α4α5β2 nAChRs has come from immunoprecipitation studies performed in transfected mammalian cells [78]. Subunit deletion experiments, in which embryonic chick sympathetic neurones were treated with antisense oligonucleotides, support the conclusion that the α5 subunit can co-assemble into functional nAChRs, some of which appear also to contain the α7 subunit [80].

Like α5 and β3, the α6 subunit was also considered for several years to be an ‘orphan’ subunit, because of difficulties encountered in its functional expression in recombinant nAChRs. Heterologous expression in Xenopus oocytes has, however, produced functional nAChRs when either the chicken or rat α6 is co-expressed with the human β4 [81]. Despite forming functional nAChRs in combination with the human β4 subunit, the chicken α6 subunit failed to generate detectable whole-cell currents when co-expressed with chicken β4 [81]. Evidence obtained from radioligand binding studies in oocytes indicates that α6 is able to co-assemble with the β2 subunit but that α6β2 complexes do not generate functional nAChRs and are inefficiently expressed on the cell surface [67]. In contrast to these studies in oocytes, expression in transfected mammalian cells has provided evidence for the co-assembly of functional α6β2 nAChRs [82]. Heterologous expression studies have also demonstrated the preferential co-assembly of α6 into triplet subunit combinations. These include α6β3β4 [67], [68] and also α3α6β4 and α3α6β2 [67], [82]. Immunoprecipitation studies on nAChRs from chick retina are largely in agreement with the heterologous expression data. These studies indicate that almost all α6-containing nAChRs in chick retina also contain the β4 subunit [83]. About half also contain the α3 subunit, about half contain β3 and less than 10% contain the β2 subunit [83]. Immunoprecipitation studies with nAChRs expressed in rat and mouse dopamine neurones has identified α6-containing nAChR subtypes such as α6β2* and α4α6β2* [79], [84]. Similar studies with nAChRs expressed in rat retina identified a heterogeneous population of receptor subtypes, of which α4(non-α6)* nAChRs comprised 60% of the total, α6* nAChRs 26% and non-α4/non-α6 nAChRs (such as α2* and α3* nAChRs) 14% [85]. Recently, the assembly and subcellular trafficking of α6 and β3 subunits has been examined by use of Förster resonance energy transfer (FRET) [86]. This is a powerful experimental approach which has led to the conclusion that multiple subunit stoichiometries exist for α4- and α6-containing receptors, whereas only a single β3 subunit is incorporated into assembled nAChRs.

Most nAChR subunits appear to be unable to generate functional nAChRs unless co-assembled with at least one other type of subunit to generate a heteromeric complex. The first evidence that some nAChR subunits can generate homomeric nAChRs came from studies in which the α7 subunit was expressed in Xenopus oocytes [87]. An α8 nAChR subunit, with close sequence similarity to α7, has been isolated by molecular cloning from a chicken cDNA library [88] but an analogous subunit has not been identified in any mammalian species. However, like α7, the avian α8 subunit is able to generate functional homomeric nAChRs when expressed in Xenopus oocytes [89], [90]. Studies of native nAChRs suggest that homomeric α7 nAChRs are a major subtype in chicken brain, whereas homomeric α8 nAChRs are a major subtype in retina [91]. Native heteromeric nAChRs containing both α7 and α8 subunits have also been identified by immunoprecipitation, but they appear to be a minor population [78], [90–92]. Purification of native chick nAChRs (either with α-bungarotoxin or with antibodies specific for the α7 or α8 subunit) has revealed the existence of multiple (three or more) protein bands in SDS-polyacrylamide gel electrophoresis [93–95], suggesting that these may represent complexes containing other subunits in addition to α7 or α8. The identity of these co-precipitated proteins has not been established but they do not appear to be any of the known nAChR subunits [95]. In contrast, other studies of native nAChRs purified from chick optic lobe have indicated the presence of only a single major polypeptide [96]. The possibility of other (as yet unidentified) subunits co-assembling with α7 is supported by immunoprecipitation studies on receptors isolated from chick ciliary ganglia [97]. Further evidence for the co-assembly of other subunit subtypes with α7 has also been provided by experiments in which chick sympathetic neurones were treated with antisense oligonucleotides [98]. Studies with α7 nAChRs purified from rat brain have not, however, detected evidence for its co-assembly with other known nAChR subunits [99]. Further evidence supporting the conclusion that mammalian α7 nAChRs are homomeric comes from protein cross-linking studies with receptors expressed endogenously by the rat adrenal pheochromocytoma cell line, PC12 [47]. There have, however, been reports of native α7-containing nAChRs found in rat brain which differ in their electrophysiological properties from recombinant homomeric α7 nAChRs [100], [101] and there is evidence that this may result from the co-assembly of α7 with β2 into a heteromeric nAChRs [102].

The two most recently characterized nAChR subunits (α9 and α10) are somewhat atypical in terms of their pharmacological properties and in their anatomical distribution. These subunits are expressed predominantly within the hair cells of the cochlea [103], [104] and have been implicated in auditory processing [105]. As had been demonstrated previously with α7 and α8, the α9 subunit is able to generate functional homomeric nAChRs when expressed in Xenopus oocytes [103]. However, it seems likely that native α9-containing nAChRs are heteromeric complexes in which α9 is co-assembled with the α10 subunit. Indeed, the whole-cell currents generated when α10 is co-expressed with α9 are considerably larger than those seen when α9 is expressed alone [104]. Futhermore, the α10 subunit does not generate functional homomeric nAChRs when expressed alone in Xenopus oocytes [104]. Radioligand binding studies conducted with recombinant nAChRs support the conclusion that native α9-containing receptors are heteromeric complexes of the α9 and α10 subunits [106].

Folding and assembly of nAChRs

As reviewed previously [27], the assembly of ion channels such as nAChRs is a relatively slow and inefficient process which is critically dependent upon appropriate subunit folding and requires trafficking of nAChRs from their site of synthesis in the endoplasmic reticulum (ER) to the cell surface. The importance of subunit folding is very clearly illustrated by the difficulties encountered in expressing recombinant α7 nAChRs in several non-neuronal cultured cell lines [107–109]. As discussed later, recent studies have demonstrated the requirement for a molecular chaperone RIC-3 [110–112], since the presence of RIC-3 enables α7 subunits to acquire their correct conformation instead of being retained within the ER [110–114]. In addition to achieving a correctly folded state, nAChR subunits must also assemble correctly into pentameric complexes before they are exported from the ER. This is most apparent in studies of the recombinant nAChR subunits which generate only heteromeric receptors. Several studies with recombinant nAChR subunits have demonstrated that, when expressed alone, such ‘heteromer-only’ subunits are retained within the cell [37], [54], [115]. Studies with mutated and chimeric subunits have demonstrated the importance of the N-terminal domain of nAChR subunits in mediating subunit-subunit interactions [31], [116], [117]. In addition, a series of studies employing subunit chimeras has illustrated the importance of sequences within the intracellular and transmembrane domains of nAChR subunits in determining efficient subunit folding and in influencing the level of cell-surface expression [54], [109], [118–123]. Other recent studies have identified conserved hydrophobic residues within the M3-M4 loop domain as being important for export of nAChRs from the ER [124].

Post-translational modification

After synthesis in the ER, nAChR subunits interact with a wide variety of proteins responsible for post-translational modifications such as glycosylation, disulphide-bond formation, palmitoylation, proline isomerization and proteolytic cleavage [27], [125]. Appropriate glycosylation has been shown to be required for correct assembly of both muscle-type and neuronal nAChRs [126–129]. Correct folding, assembly and cell-surface expression of nAChRs is also dependent upon appropriate disulphide-bond formation [35], [126], [127], [130], [131]. Studies with the neuronal nAChR α7 subunit suggest that inappropriate disulphide-bond formation may underlie the inefficient folding of this subunit which is observed in some mammalian cell lines [132]. An apparent link has also been demonstrated between the efficient folding of the nAChR α7 subunit in Xenopus oocytes and a requirement for the prolyl isomerase enzyme cyclophilin [133], [134]. This prompted studies to investigate whether the co-expression of cyclophilin proteins might alleviate the inefficient folding of α7 observed in some mammalian host cells but such experiments have not, so far, established this [108], [135].

Molecular chaperones and receptor-associated proteins

Nicotinic receptor subunits have been shown to associate with several molecular chaperones. In common with many transmembrane proteins, interactions have been identified between nAChR subunits and ER-resident chaperone proteins such as BiP [136], [137] and calnexin [138], [139]. Yeast two-hybrid studies with the intracellular loop region of the neuronal nAChR α4 subunit identified interactions with a cytoplasmic chaperone protein 14-3-3η and a peripheral membrane protein VILIP-1, both of which have been reported to influence cell-surface expression levels of α4β2 nAChRs [140], [141]. As mentioned earlier (and reviewed more extensively elsewhere [142]), maturation of nAChR α7 and α8 subunits has been shown to be critically dependent on a recently identified putative chaperone protein RIC-3 [110–114]. RIC-3 is an ER-resident transmembrane protein which was originally identified in genetic studies performed in the nematode C. elegans [143], [144] and which is conserved across vertebrate and invertebrate phyla [113], [114]. In contrast to chaperone proteins such as BiP, calnexin and 14-3-3, RIC-3 appears to be highly selective in interacting with nAChRs [142], although it has also been shown to interact with the 5-HT₃ receptor [110], [111], [114], [145]. RIC-3 has a strong facilitatory effect on the maturation of α7 nAChRs [110–114]. It has also been reported to have a modulatory effect on several heteromeric neuronal nAChRs [110], [111], [114], [146]. With some heteromeric nAChRs (such as α4β2), RIC-3 has been reported to cause either an enhancement or a suppression of agonist-induced reponses when examined in different expression systems [110], [111], [114], the reasons for which are unclear but may be due to differences in the host cell type [146]. UNC-50 is a transmembrane protein located within the Golgi apparatus which has been shown to have a selective effect on nAChR trafficking [147]. Like RIC-3, the gene encoding UNC-50 was originally identified in C. elegans through genetic screening for suppressor mutations [148]. Reduced binding of the nicotinic ligand levamisole in unc-50 mutants provided evidence that UNC-50 may be required for nAChR assembly [149]. Co-expression of recombinant nAChRs with a mammalian homologue of UNC-50 suggested that it can enhance levels of cell-surface nAChRs; however, this study concluded that this may be a consequence of an RNA-binding activity [150]. More recent studies with UNC-50 in C. elegans suggest that it is required for subtype-selective trafficking of nAChRs through the Golgi apparatus, thereby acting at a later stage in receptor maturation than RIC-3 [147].

There have been relatively few well-characterized examples of interactions between nAChRs and cytoskeletal proteins. The clearest evidence for such an interaction is between muscle nAChRs and rapsyn (or 43K protein), the discovery of which long predates the advent of recombinant proteomic techniques. The interaction between muscle nAChRs and rapsyn has been shown to be important in the clustering of postsynaptic nAChRs at the neuromuscular junction [151], [152]. Further evidence of a role for rapsyn in neuromuscular synaptogenesis has been provided by studies with transgenic mice lacking rapsyn, in which nAChRs fail to cluster at the neuromuscular junction [153]. The importance of rapsyn in anchoring the muscle-type nAChR within the post-synaptic membrane at the neuromuscular junction has been well documented [154], [155]. As well as being expressed in muscle, rapsyn has been detected in brain [156] and in peripheral neuronal cells such as mouse sympathetic neurones [157] and chick ciliary ganglion neurones [158]. However, it appears that rapsyn is not required for clustering of neuronal nAChRs such as those located at ganglionic synapses [157], [159]. Studies using transfected cells indicate that rapsyn is able to cause clustering of neuronal nAChRs such as α3β2, α3β4, α4β2 and α7 [157], [160], but suggest that, at least for some of these subunit combinations, the clustered neuronal nAChRs are retained within the cell and are not transported to the cell surface [157]. Evidence suggesting that rapsyn is not required for nAChR clustering in sympathetic ganglia has come from studies with rapsyn-deficient transgenic mice, in which the number, size and density of nAChR clusters in superior cervical ganglia are similar to those in control animals [157]. Despite the presence of rapsyn mRNA in chick ciliary ganglia, rapsyn protein is either absent or is present at a much lower ratio (relative to nAChR levels) in ganglia than in muscle [159].

In addition to rapsyn, several other cytoskeletal proteins in have been implicated in the regulation of nAChR distribution and function. For example, there is evidence that actin is important in regulating such events for α7nAChRs located in somatic spines of chick ciliary ganglia and on spinal cord neurones [161–163]. Recent studies have identified interactions between neuronal nAChRs and PDZ-domain cytoskeletal proteins such as PSD-93, PSD-95 and PICK1 [164–166]; interactions which, in some cases, influence events such as receptor clustering [165], [166]. As discussed previously [164], the interaction of nAChRs with PDZ-domain proteins is perhaps unexpected, given that interactions with PDZ-domain proteins have generally been identified with receptors containing intracellular C-terminal domains. In addition, recent studies have identified a role for the tumour-suppressor protein APC in synaptic localization of both muscle and neuronal nAChR subtypes [167–169]. A ‘proteomics’ approach has been used to identify 21 proteins associated with the β2 subunit purified from mouse brain, several of which are implicated in regulation of sub-cellular trafficking [170].

Up-regulation and trafficking of nAChRs

There is extensive evidence to indicate that chronic exposure to nicotine, such as that which occurs during tobacco smoking, causes an up-regulation of brain nAChRs [171–173]. This is brought about by a post-translational mechanism [174] and can be mimicked in cultured cell lines by exposure to low concentrations of nicotine for several hours [175–178]. With nAChR subtypes such as α4β2 there is evidence that chronic nicotine treatment can enhance subunit folding and assembly, and induce conformational changes within subunits [37], [179]. Indeed, it has been proposed that nicotine causes receptor up-regulation by acting as a pharmacological chaperone [180], [181] via a direct action at the agonist binding site [182], thereby promoting receptor maturation [183]. There have, however, been several explanations proposed for the mechanism of nicotine-induced nAChR up-regulation, as reviewed elsewhere [184]. Recent studies have revealed that chronic nicotine has a differential effect on up-regulation of functional α4-containing nAChRs in different brain regions, providing a possible explanation for the differential effect of chronic nicotine (sensitization and tolerance) observed in distinct regions of the brain [185]. Nicotine-induced up-regulation has been reported for nAChR subtypes other than α4β2, although there is evidence from studies both in brain and in cultured cell lines that nAChR subtypes differ in the extent to which they are up-regulated by nicotine [186–191].

Recent studies have demonstrated up-regulation of cell-surface α7 nAChRs in hippocampal neurones by tyrosine dephosphorylation [192] and also by brain-derived neurotrophic factor (BDNF) [193]. In the case of tyrosine dephosphorylation, there is evidence that this may occur via SNARE-dependent trafficking [192]. Evidence has also been obtained to indicate that SNARE-dependent trafficking is important for maintaining functional coupling between α7–responses and downstream signalling in somatic spines [194].

Influence of subunit domains upon receptor trafficking

Assembled nAChRs are targeted to both pre- and post-synaptic sites. Within the brain, pre-synaptic receptors are important in, for example, modulating neurotransmitter release [195], [196]. Post-synaptic nAChRs are important in mediating synaptic transmission at the neuromuscular junction and in autonomic ganglia [197] and, more recently, post-synaptic nAChRs have been identified in the mammalian brain [198]. The differential targeting of nAChRs (for example to pre- or post-synaptic sites) has prompted a search for signals which may regulate receptor targeting.

The influence of the M3-M4 cytoplasmic loop region of neuronal nAChR subunits upon receptor targeting has been examined by the expression of subunit chimeras in chick ciliary ganglion neurones by retrovirus-mediated gene transfer [199]. Whereas the α7 subunit is located peri-synaptically, chimeric subunits in which the M3-M4 cytoplasmic loop of the α7 subunit is replaced by the analogous loop region of the α3 subunit are targeted to post-synaptic sites. In contrast, synaptic targeting is not observed with similar chimeric subunits containing the cytoplasmic loop domain from either the α5 or β4 subunit [199], [200]. These studies implicate sequences within the M3-M4 cytoplasmic loop of the α3 subunit in the targeting of receptors to post-synaptic regions in ciliary ganglion neurones. More recently, motifs responsible for differential targeting to axons and dendrites have been identified within the M3-M4 loops of the α4 and α7 subunits [201]. The influence of the nAChR M3-M4 loop upon folding, cell-surface expression and targeting has also been examined in a recent study in which a series of fourteen subunit chimeras was constructed, each containing a different M3-M4 loop domain (α1-α10 and β1-β4) [202]. This has confirmed that M3-M4 loop domain can influence subunit folding, cell-surface expression and receptor targeting [202]. Mutagenesis studies have demonstrated the importance of hydrophobic amino acids within the M3-M4 loop domain and also of amino acids within the M1 transmembrane domain in regulating export of nAChRs from the endoplasmic reticulum and trafficking to the cell surface [124], [203].

Conclusion

Studies conducted with both native and recombinant receptors have contributed to our current understanding of nAChR assembly and trafficking, but our knowledge of these complex cellular events is still far from complete. Whereas the subunit composition of nAChRs found at the neuromuscular junction is well established, an important task is to identify the extent of subunit diversity amongst neuronal nAChR subtypes. However, as discussed in this review, a variety of experimental approaches are now helping to answer such questions. Recent work has also revealed the influence of subunit domains upon events such as receptor assembly and trafficking and the interaction of nAChRs with several classes of intracellular proteins. In particular, the role of scaffolding proteins in nAChR targeting and in determining the functional state of these receptors is emerging, as is the importance of molecular chaperones such as RIC-3 in the modulation of nAChR maturation.

Acknowledgements

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.