An overview of trafficking and assembly of neurotransmitter receptors and ion channels (Review)

Abstract

Ionotropic neurotransmitter receptors and voltage-gated ion channels assemble from several homologous and non-homologous subunits. Assembly of these multimeric membrane proteins is a tightly controlled process subject to primary and secondary quality control mechanisms. An assembly pathway involving a dimerization of dimers has been demonstrated for a voltage-gated potassium channel and for different types of glutamate receptors. While many novel C-terminal assembly domains have been identified in various members of the voltage-gated cation channel superfamily, the assembly pathways followed by these proteins remain largely elusive. Recent progress on the recognition of polar residues in the transmembrane segments of membrane proteins by the retrieval factor Rer1 is likely to be relevant for the further investigation of trafficking defects in channelopathies. This mechanism might also contribute to controlling the assembly of ion channels by retrieving unassembled subunits to the endoplasmic reticulum. The endoplasmic reticulum is a metabolic compartment studded with small molecule transporters. This environment provides ligands that have recently been shown to act as pharmacological chaperones in the biogenesis of ligand-gated ion channels. Future progress depends on the improvement of tools, in particular the antibodies used by the field, and the continued exploitation of genetically tractable model organisms in screens and physiological experiments.

• Quality control
• multimerization
• assembly domain
• cell surface expression
• auxiliary subunits

Introduction

Ion channels provide an ion-conducting pore in the hydrophobic environment of the lipid bilayer [1]. The pore can be gated by different molecular events, e.g., the binding of a neurotransmitter ligand or a change in transmembrane potential. Ion channels are polytopic membrane proteins and in many cases complex heteromultimers. The species of heteromultimers present at the cell surface reflect many different (however not all) conceivable combinations of subunits, each arrangement conferring different functional properties. In the fully assembled channel complex, the pore is shielded from the lipid bilayer by the surrounding protein. However, during co-translational integration of the protein into the membrane or before completion of multimeric assembly of the channel, polar residues eventually contributing to the pore may be exposed. It has been shown that exposed polar residues in transmembrane segments can lead to the endoplasmic reticulum-associated degradation (ERAD) of membrane proteins [2]. The existence of energetically unfavourable assembly intermediates and the need to select only some of the possible multimers means that ion channels are demanding substrates of the secretory pathway: in addition to primary quality control processes at the endoplasmic reticulum (ER) they are subject to protein- or cell type-specific secondary quality control mechanisms [3]. Many have to pass trafficking checkpoints before leaving the ER-to-Golgi shuttle [3–5]. Ion channels are particularly interesting cargo proteins of the vesicular transport machinery because their numbers at the cell surface or in subcellular compartments are tightly controlled and sometimes regulated [6–13].

Complex heteromultimers need sophisticated quality control

Like membrane proteins of the immune system, neurotransmitter receptors and ion channels require particularly stringent quality control mechanisms [3] that ensure proper protein folding, subunit assembly, and functionality of the multimeric complex. All biophysical parameters of an ion channel can be influenced by events that occur early in the life of the protein. For instance, combination of alternative AMPA receptor subunits by a variety of mechanisms – e.g., differential expression of homologous and auxiliary subunits, differential splicing, and RNA-editing – gives rise to receptors of different ion selectivity or results in different kinetic properties [14–19]. In the case of ionotropic GABA_A receptors, in addition to the subunit composition, the positioning of subunits within the pentameric complex affects the pharmacology of the receptor [20], [21].

Classic approaches in this field combining molecular biology, biochemistry, and electrophysiology (mostly in heterologous expression systems like Xenopus oocytes or immortalized cell lines) still produce interesting results. Typical research in this area asks questions such as: which subunit combinations can reach the cell surface and what are their functional properties? Which domains drive the assembly of a given (hetero)multimer? How does the interaction between multimerization domains confer specificity to the assembly process? Which trafficking determinants contribute to cell surface expression or its regulation? At the same time, there are problems that the field often neglects because they are technically less tractable: how available are alternative subunits for assembly when one considers the timing of gene expression and spatial aspects of protein biosynthesis? Do the non-translated regions in the relevant messenger RNAs target translation of a given subunit to a specific subcellular localization or (a putative) assembly domain of the ER? Are concentrating mechanisms required to make assembly efficient under physiological conditions, e.g., when only small amounts of the assembling subunits are expressed?

This overview to the thematic issue of trafficking, assembly, and regulation of neurotransmitter receptors and ion channels attempts to summarize progress and emerging themes across the superfamilies of ligand-gated and voltage-gated channels that are addressed in the individual reviews.

Assembly is at minimum a bimolecular reaction

Since by definition assembly is at least a bimolecular reaction, the process depends on the concentration of the co-assembling subunits. In the case of neurotransmitter receptors and ion channels, these parameters are usually unknown. Our knowledge of ion channel assembly depends largely on experiments employing in-vitro translation or heterologous expression (in Xenopus oocytes or tissue culture cells). This means that key results have been obtained under vast over-expression conditions. Whilst many conclusions, e.g., about assembly domains and assembly intermediates should not be affected, these experimental conditions may have skewed our picture of the kinetic aspects of assembly [22–30]. In principle, assembly could be a reaction of higher molecular order since many ion channels and ionotropic receptors are trimers, tetramers, or pentamers. Based on their studies of Kv1.3 assembly, Tu and Deutsch have proposed an assembly pathway involving a dimerization of dimers [30] – sequential bimolecular reactions resulting in the final tetrameric assembly.

This sequential dimerization concept has proven useful for understanding the biogenesis of tetrameric AMPA receptors [31], [32]. The dose-response curve obtained for the dominant negative effect of a chimeric subunit on AMPA receptor heteromeric channel assembly suggested dimeric assembly intermediates. This chimeric subunit had been engineered to be compatible only in the N-terminal assembly domain and not in other regions of the protein required for the formation of functional tetramers. The assembly process of AMPA receptors utilizing the chimeric subunit was stalled, presumably as a dimeric intermediate [22]. Subsequent biochemical work [15] was able to detect the proposed dimeric assembly intermediate directly by cross-linking and by blue native polyacrylamide gel electrophoresis of endogenous neuronal AMPA receptors.

A dimer of dimers is also apparent in the structure of a bacterial inward rectifier potassium channel [33]. Thus, these channels might assemble from dimeric intermediates. However, the situation in cyclic-nucleotide gated (CNG) channels is an important reminder that it might be dangerous to deduce assembly pathways from the structure of the fully assembled complex: a low-resolution structure of the retinal cGMP-gated channel suggests arrangement of the four subunits as a pair of dimers [34]. But beautiful biochemical work addressing the subunit stoichiometry of the same native rod CNG channel found that the channel is a tetramer composed of three A1 and one B1 subunit [35], [36]. Interestingly, Zhong et al. identified a trimer-forming multimerization domain that is the molecular determinant of a subunit stoichiometry of 3:1 for A and B type subunits in the assembled tetramer [36], [37]. Therefore, assembly of this channel is unlikely to involve a dimerization of dimers despite the tetrameric nature of the mature channel and the intriguing two-fold rotational symmetry of the intracellular domains observed at low resolution [34]. In conclusion, the symmetry of multimers might change during assembly and a proposal for an assembly pathway is more plausible if based on the detection of assembly intermediates.

Shaker-type voltage-gated potassium channels like Kv1.3 are capable of forming a tetramer via their N-terminal T1 domains (Figure 1a) while the protein is still being translated and translocated [25], [27]. However, this conclusion is based on in-vitro translation in rabbit reticulocyte lysate. It is unknown if coupled tertiary folding and oligomerization of endogenous Kv channels occurs under physiologically relevant conditions. The question is whether the concentration of nascent channel subunits would suffice to make an encounter during the initial steps of biosynthesis probable.

Figure 1.  N- and C-terminal assembly domains found in the superfamily of voltage-gated cation channels. (a) Schematic representation of a Kv pore-forming subunit. The N-terminal location of the T1 domain is indicated and the corresponding crystal structure obtained for four assembled T1 domains of Kv1.2 [109] is shown. (b) Schematic representation of the KCNQ7 pore-forming subunit. The location of the C-terminal A-domain tail, a self-assembling, parallel, four-stranded coiled-coil, is shown along with the corresponding crystal structure of four assembled coiled-coil forming domains [39]. Structure reproduced from [109] and [39] with permission from Elsevier.

For many other channels of the voltage-gated cation channel superfamily including CNG channels [36], [37], ether á go-go (eag) channels [23], intermediate conductance (IK) calcium-activated channels [38], KCNQ channels [24], [26], [28], [39–41], and TRPM channels [29], assembly domains that form coiled-coil structures are found in the C-terminus of the polypeptide chain (Figure 1b). Based on the recent discovery of these C-terminal assembly domains Tsuruda et al. concluded: ‘In the case of a C-terminally located assembly domain, the temporal order of assembly must be very different’ [29] from the assembly of Kv channels via the T1 domain (Figure 1). This determination holds true if the physiological concentration of nascent subunits is of sufficient magnitude to allow for co-translational interactions between subunits presenting N-terminal assembly domains. Careful structural and biophysical characterization of assembly domains continues to be a prerequisite for further investigation of assembly pathways in the cellular environment.

Properly assembled – and functional!

Ellgaard and Helenius distinguish between primary (e.g., generic, Figure 2a) and secondary (e.g., cargo-specific, Figure 2b) quality control processes [3]. Depending on their topology, channels and receptors interact with different subsets of the chaperones involved in primary quality control [42–44].

Figure 2.  Major mechanisms involved in general and protein-specific quality control. Following the terminology introduced by Ellgaard and Helenius [3] primary (a; closed triangles) and secondary (b; open triangles) quality control processes are shown. The assembling subunits of a heteromultimeric channel are shown as rectangular shapes. Large dashed arrows indicate forward transport and ER retrieval. (a) General chaperones assist the folding of channels with luminal domains and monitor core glycosylation [3]. Assembly intermediates exposing polar residues in the plane of the membrane can be eliminated by ERAD [2]. Alternatively, they can be prevented from leaving the ER-Golgi shuttle by the retrieval factor Rer1 [45], [46]. (b) Herrmann et al. divided protein-specific chaperones into outfitters, escorts and guides [81]. Many auxiliary subunits of ion channels can perform these functions for the specific pore-forming subunit(s) that they interact with. Beyond quality control, they usually contribute to the functional properties of the channel at the cell surface. Adapted with permission from [5].

An emerging new aspect of primary quality control is the recognition by the ER retrieval factor Rer1 of membrane-spanning segments that are energetically unstable in the lipid bi-layer (Figure 2a). This type of quality control is highly relevant to multimeric receptors and ion channels because they may expose polar residues in the plane of the membrane until fully assembled. The mechanism was first studied in Saccharomyces cerevisiae where Sato et al. demonstrated that in the Golgi, the membrane protein Rer1 binds to cargo proteins via transmembrane segments containing polar residues and sorts them back to the ER in coatomer (COPI)-coated vesicles [45], [46]. The same group went on to show that recognition of a polar residue within a transmembrane segment by Rer1 and subsequent retrieval to the ER was also the basis for the retention of an unassembled iron transporter subunit from the cell surface [45], [46]. This work and studies addressing the role of mammalian Rer1 in the multimeric assembly of γ-secretase (a protease involved in intramembrane proteolysis) [47], [48] has raised an interesting possibility: Rer1 might be involved in the biogenesis of multimeric neurotransmitter receptors and ion channels since assembly intermediates of these complexes are likely to expose polar residues.

Insights into the function of Rer1 might also be relevant to the study of channelopathies where mutations in genes coding for ion channels cause human inherited disorders [49], often because the mutated gene gives rise to a channel that cannot reach the cell surface. For example, in familial hyperinsulinism, mutations in the ABCC8 gene encoding the regulatory subunit of ATP-sensitive potassium (K_ATP) channels, SUR1, can cause trafficking defects where SUR1 is localized to the ER instead of the plasma membrane [50–56]. Some of the relevant mutations introduce a polar residue into a transmembrane segment [54]. Hence one may speculate that the mammalian homologue of Rer1 could be involved in the ER localization of these disease-associated channel variants.

Consistent with the idea that changes in the transmembrane domain might confer a dominant additional retrieval signal (e.g., a Rer1 recognition site) to SUR1, mutation of an arginine (R)-based ER localization motif in SUR1 did not suppress trafficking defects found for variants containing A116P in the third and V187D in the fifth transmembrane segment of SUR1 [54]. However, the same manipulation fully rescued the cell surface expression of another disease-associated SUR1 variant, L1544P [53]. Variants SUR1 A116P and V187D could hypothetically be recognized and retrieved by Rer1 whereas L1544 maps to the distal cytosolic C-terminus of SUR1. Thus, it might be worthwhile to experimentally test the hypothesis that mammalian Rer1 participates in the quality control of disease-associated channel variants. If true, Rer1 would have a role at the intersection of primary, ER-based quality control and protein sorting [3].

The function of R-based ER localization motifs in assembly-dependent forward transport is a well-established example for the interplay between the cellular sorting and quality control machinery [5], [57–67]: the Golgi-localized COPI vesicle coat complex recognizes these peptide-sorting motifs presented by ion channel and neurotransmitter receptor subunits. COPI recognition of the ER localization motifs leads to the ER retrieval of unassembled subunits or partially assembled complexes. Steric masking and binding of 14-3-3 proteins can overcome ER retrieval leading to the cell surface expression of properly assembled complexes. The presence of several ER localization and 14-3-3 binding motifs in multimeric K_ATP and acid-sensitive potassium TASK channel complexes makes it difficult to propose a detailed mechanistic model for the relevant binding and masking events [58], [63], [68]. In order to understand the interplay between COPI and 14-3-3 protein binding precisely, high-resolution structures for these channels will have to become available.

Both ER localization signals and forward trafficking information on ion channel subunits contribute to the rate with which the fully assembled complex can leave the ER and reach the cell surface [4]. Current studies characterized di-acidic sorting motifs in the acid-sensitive potassium channel TASK-3 and the plant KAT1 potassium channel [69], [70]. It is plausible to think that these sorting motifs are recognized by the COPII vesicle coat involved in the budding of anterograde transport vesicles at the ER. It remains to be shown which cellular machinery recognizes other recently identified forward-trafficking determinants like a RSRYW motif found in the α-subunit of epithelial sodium channels [71].

Unlike most other ion channels, ionotropic neurotransmitter receptors contain a large luminal domain that is stringently monitored to ensure proper folding of this crucial part of the protein. A number of recent reports demonstrate the correlation between the ability of glutamate receptors to leave the ER and their functionality, e.g., ability to bind glutamate and respond to it with the appropriate conformational changes [14], [15], [72–75]. Specifically, mutations that affect both glutamate binding and desensitization kinetics result in receptor subunits that are competent to tetramerize but almost completely fail to pass a post-assembly trafficking checkpoint [74]. This failure results in drastically reduced cell surface expression of the mutant variants although their functional properties can be analyzed in over-expressing systems. Priel et al. suggest that this intracellular retention of variants with an altered response to glutamate is irrespective of channel activity in the ER membrane by combining two mutations: one that blocks ion flow and one that does not desensitize after glutamate binding and causes ER retention [74]. The authors conclude from this experiment that glutamate binding affects the structure of the luminal domain in a way that enables forward transport of the complex. Coupling between the functionally relevant conformational changes and ER exit is an elegant mechanism to prevent the appearance of improperly regulated receptors at the cell surface. (See a recent review by Greger et al. [31] for an excellent discussion of functional quality control in glutamate receptors.)

If glutamate is required for receptor maturation it becomes important to ask whether glutamate is present in the lumen of the ER. The amino acid has been detected in the ER of magnocellular endocrine cells of the supraoptic nucleus in the rat hypothalamus employing an anti-glutamate antibody [76]. Work on metabotropic glutamate receptors in the ER and nuclear envelope of striatal neurons suggests that glutamate can enter the ER via cystine-glutamate exchanger and other secondary active transporters [77]. Gating motions of glutamate receptors in the ER [78] might come as a surprise but are consistent with the picture of the ER as a metabolic compartment [79] studded with transport proteins running on coupled gradients [80]. It is now clear that this transport activity provides small molecules that perform the function of pharmacological chaperones to support the biogenesis of ligand-binding membrane proteins.

Dedicated chaperones or regulatory subunits?

In many cases it is difficult to assign subunits of channels and receptors that are not pore-forming subunits neatly to one category or the other – particularly if one further divides the factors involved in secondary quality control into categories such as outfitters, escorts, and guides [81] to distinguish between chaperones that promote the folding of specific substrates, ER exit factors, and associating proteins that provide targeting information beyond the ER (Figure 2b).

The well characterized effect of different auxiliary subunits (e.g., beta-subunits of the oxido-reductase fold [82], Ca^+ +-binding proteins of the KChIP family [83], or the most recently discovered dipeptidyl aminopeptidase-like protein DPPX [84]) on the cell surface expression of voltage-gated potassium (Kv) channels is a good example to illustrate the blurring of these concepts (without disputing their usefulness): acting as escorts, they promote forward transport in the secretory pathway [84–87] and acting as guides they determine axonal targeting [85], [88]. Each type of auxiliary subunit can also modulate the functional properties of Kv channels at the cell surface. This suggests that trafficking competence and certain functional properties may be different aspects of the same structural states and is reminiscent of the correlation found between trafficking and glutamate binding or desensitization for glutamate receptors (compare above).

The Stargazin or transmembrane AMPA receptor regulatory protein (TARP) family is yet another example of a family of auxiliary subunits with multiple effects: these proteins act as outfitters, escorts, and guides and modify the function of AMPA receptors at the cell surface (reviewed in [89], [90]). Other interesting auxiliary proteins that have recently been identified include RIC-3 [91–99], a membrane protein with cytosolic domains bearing features of a coiled-coil domain, and its homologues. This factor has been shown to enhance the cell surface expression and affect the functional properties of nicotinic acetylcholine and serotonine receptors. Elegant work performed by Eimer et al. in Caenorhabditis elegans [100] identified the conserved UNC-50 protein as a membrane protein involved in the subtype-specific sorting of nicotinic receptors within the Golgi apparatus. Also, two specific β-subunits for members of the CLC family of chloride transport proteins have been identified [101], [102]: Barttin is a β-subunit that is required for the cell surface expression of CLC-Ka and −Kb in the kidney and the inner ear [101] and Ostm1 is a highly glycosylated membrane protein that travels with ClC-7 to lysosomes where it stabilizes the chloride transport protein [102]. The precise molecular mechanism by which any of these auxiliary proteins exert their function in protein trafficking is unclear.

Perspectives – better tools and the power of model organisms

Progress towards understanding the trafficking of neurotransmitter receptors and ion channels in a physiological setting is frequently complicated by the fact that these cargo proteins are difficult to detect and to manipulate in their native environment. This situation can be improved by at least two approaches. One has recently been taken by the NeuroMab facility at the University of California, Davis (supported by a grant from NINDS and NIMH) that pursues the goal ‘to generate high quality mAbs that are validated … and to make these validated NeuroMabs available to the research community’ [103]. This move to end the ubiquitous use of ‘reagents of mass distraction’ [103] can only be met with the greatest enthusiasm because only good antibodies will allow the field to move away from heterologous expression and over-expression into model organisms. It will require the context of the physiological system to understand the sequence and orchestration of the many protein-protein interactions underlying assembly, trafficking, scaffolding, regulation, and internalization of neurotransmitter receptors and ion channels. As there is redundancy in most of these processes, one major challenge is to understand and distinguish the physiological roles of closely related proteins [104]. Elias et al. combined genetic engineering of the mouse with RNA interference methodology to investigate the role of closely related scaffolding proteins of the PSD-95-like membrane-associated guanylate kinase family (PSD-MAGUK) in the targeting of AMPA receptors [104]. They combine these two complementary methods of controlling gene expression with detailed functional analyses and biochemical experiments to show that the two closely related MAGUKs PSD-95 and PSD-93 affect AMPA receptor localization in a synapse-specific and developmentally regulated manner. In addition to synaptic targeting, dendrite growth and local dendritic protein biosynthesis are areas that require physiologically integrated approaches. Using organotypic hippocampal slice and dissociated neuronal primary cultures Raab-Graham et al. discovered a link between dendritic translation and surface expression of the voltage-gated potassium channel Kv1.1 and synaptic activity [105]. The presence and function of secretory organelles in dendrites is a particularly exciting frontier of cellular neuroscience: recent work in mammalian neurons [106] and a revealing genetic screen in Drosophila melanogaster [107] provide the first glimpses of how a spatially distinct secretory pathway dedicated to a specific dendrite might eventually contribute to neuronal plasticity. Another genetically tractable model organism, C. elegans, has allowed for the discovery of three novel auxiliary proteins involved in the biogenesis, post-ER trafficking, and scaffolding of nicotinic acetylcholine receptors, namely RIC-3, UNC-50, and LEV-10 [96], [100], [108]. These examples illustrate convincingly that the classical model organisms continue to contribute original and groundbreaking insights into the trafficking of ionotropic neurotransmitter receptors and ion channels.

Acknowledgements

I thank all members of my group – past and present – for sharing excitement and frustrations whilst working on ion channel trafficking. Work in my laboratory was funded by the ZMBH (Universität Heidelberg) Deutsche Forschungsgemeinschaft, Landesstiftung Baden-Württemberg, EMBO Young Investigator Programme, Fonds der Chemischen Industrie, and is currently funded by The Wellcome Trust. I am indebted to Nikolai Braun, Thomas Mrowiec, and Volker Schmid for valuable discussions and very helpful comments on the manuscript.