Trafficking of 5-HT₃ and GABA_A receptors (Review)

Abstract

The 5-HT₃ and GABA_A receptors are members of the Cys-loop family of neurotransmitter-gated ion channels that also include receptors for glycine and acetylcholine. The 5-HT₃ and acetylcholine receptors (cationic ion channels) and the GABA_A and glycine receptors (anionic ion channels) generally depolarize or hyperpolarize, respectively, the neuronal membrane. Within the amino-terminal extracellular region, all members of this family exhibit a similar architecture of ligand binding domains and a number of key residues are completely conserved. The molecular characterization of their ligand binding and gating characteristics has benefited from the existence of a large repertoire of individual subunits that contribute to the pentameric ion channel. Although differences do exist, advances in our knowledge of one member offers valuable insight into the family as a whole. Each member of the Cys-loop receptors (and all other multimeric ion channels) must face the same challenges: How to assemble individual subunits into an ion channel and which subunits to use? How are assembled receptors distinguished from those that are unassembled or misassembled, then exported from the endoplasmic reticulum and delivered to the cell surface? How are they targeted to, and anchored at synaptic and extrasynaptic sites? How and when are they to be removed from these sites to provide long-term regulation of neuronal activity? In this review, we summarize our current knowledge for the 5-HT₃ and GABA_A receptors that have provided complementary information and helped us build an overall picture of how receptor biogenesis and trafficking occurs.

• ER export
• trafficking
• synaptic localization
• endocytosis
• neurotransmitter
• receptor
• membrane biogenesis
• targeting

Introduction

The Cys-loop family of neurotransmitter-gated ion channels consists of excitatory receptors for 5-HT, and nicotine/acetylcholine (nACh), and inhibitory receptors for glycine and GABA (GABA_A). Each member of this family is thought to share a conserved molecular architecture within their amino-terminal extracellular domain [1]. This structural similarity has enabled much experimental data to be interpreted by homology modelling, based on the crystal structure of acetylcholine binding protein from the snail, Lymnaea stagnalis [2]. Moreover, findings from one receptor type may be extrapolated to other members of the family and tested directly. This has enabled a more rapid progress of our understanding for the whole Cys-loop family.

5-HT₃ receptors

The 5-HT₃ receptors are unique amongst the other 5-HT (serotonin) receptors in that they are ligand gated ion channels, with the actions of the other 5-HT receptors being mediated via metabotropic G-protein coupled signalling. The 5-HT₃ receptors are localized within both the peripheral and central nervous systems where they play important roles in gut motility, emesis, anxiety and cognition [3]. Within the central nervous system, 5-HT₃ receptors have been localized to many brain areas including the hippocampus, cortex, substantia nigra, and the brain stem [4]. Despite the wide distribution of the 5-HT₃ receptors and the diverse range of clinical conditions with which they have been implicated, their therapeutic use is currently limited to the treatment of emesis and irritable bowel syndrome.

5-HT₃ receptors are constructed as pentameric ion channels, with the 5-HT_3A homomeric receptors and 5-HT_3A/3B heteromeric receptors being well characterized [3]. When expressed alone, 5-HT_3B does not produce functional receptors, but is retained in the endoplasmic reticulum (ER) [5]. However, when rescued by the co-expression of 5-HT_3A, 5-HT_3A/3B heteromers are formed, with a stochiometry of 3(3B):2(3A) and the arrangement in the pentamer of B-B-A-B-A [6]. Several studies have implied the existence of native receptors that cannot be explained by the formation of 5-HT_3A and 5-HT_3A/3B receptors and their promiscuous assembly with nicotinic acetylcholine receptor subunits has been postulated. More recently, 3 novel 5-HT₃ receptor subunits, 5-HT_3C-E have been cloned. Like the 5-HT_3B subunit, the 5-HT_3C-E subunits do not reach the cell surface or function as homomers, but they are rescued from the ER and expressed on the cell surface by co-expression with 5-HT_3A and appear to exhibit functional receptors with altered efficacy for 5-HT [7]. Although it appears that the incorporation of the 5-HT_3A subunit is a prerequisite for receptor formation, no investigation into the existence of other di-heteromeric (for example, 5-HT_3B/3C) or tri-heteromeric (for example, 5-HT_3A/3B/3C) combinations have been reported.

To date, no research has yet been reported to establish and identify the existence of amino-terminal assembly signals within any of the 5-HT₃ receptor subunits. This is in contrast to the extensive studies performed on the GABA_A receptors (Sieghart, this issue), where a number of assembly signals have been identified. Given the conserved architecture of the Cys-loop receptors [2], combined with the common formation of the neurotransmitter binding sites occurring at subunit interfaces [3], it is reasonable to assume that they will be analogous to those of the GABA_A receptors. Of course, this is a risky strategy and the actual assembly signals within 5-HT₃ receptors need to be determined empirically.

Arguably, more important than the identity of the assembly signals within the 5-HT₃ receptors, is an understanding of how this process is regulated. For example, if a cell is capable of assembling 5-HT_3A receptors, how can it handle the opportunity to incorporate 5-HT_3B subunits? Temporal gene expression may contribute to whether both subunits are present. However, when both subunits are available a cell would have to either select subunits randomly or show a preference. For 5-HT_3A and 5-HT_3B co-expression, a preference appears to be the option taken as 5-HT_3A/3B receptors appear to be formed preferentially [6]. How could a preference be executed? This could be achieved by a differential affinity for assembly signals. Alternatively, signals might be exposed at different points during subunit folding, giving a temporal advantage to some signals. Finally, it is possible that spatial segregation of unassembled subunits may contribute to actual subunit availability. It is not known which (if any) of these mechanisms actually operate.

In addition to the contribution of assembly signals, post-translational modifications appear to play a role in receptor biogenesis [1], [3], [8]. An interesting example of this has been reported for the N-linked glycosylation of murine 5-HT_3A receptors. N-glycosylation at N109 [1], [3] is required for 5-HT_3A receptor assembly, as indicated by the failure to produce ligand binding sites when this site is mutated [8]. This glycosylation site is variably present in other members of the Cys-loop family and is located within a region containing multiple GABA_A receptor assembly signals responsible for inter-subunit interactions [1]. Interestingly, N-glycosylation at N190, appears to be required, not for receptor assembly (as ligand binding sites are formed), but for transport to the cell surface. In contrast, N-glycosylation at N174 is not required for receptor assembly or transport to the cell surface, but does appear to inhibit receptor function [8]. Thus, N-glycosylation of 5-HT_3A may participate at multiple stages of receptor maturation, transport and function. However, a similar study on human 5-HT_3A determined that all these N-glycosylation sites were required for receptor assembly (ligand binding) and cell surface expression [9].

Extrapolation from our knowledge of assembly signals within the large extracellular amino-terminus of GABA_A receptors (Sieghart, this issue), it seems reasonable to assume that 5-HT_3B is unable to assemble in the absence of the 5-HT_3A subunit. However, the use of subunit chimaeras between 5-HT_3B and 5-HT_3A revealed that the 5-HT_3B subunit appears to be able to assemble, but not traffic beyond the ER due to the existence of cytoplasmic ER retrieval signals (ERS) [10]. The ERS are responsible for returning proteins from the cis-Golgi to the ER and so prevent forward transport to the cell surface. One ERS was identified within the first cytoplasmic loop (between TMI-II) of 5-HT_3B and shown to cause ER retention unless masked by the co-expression of 5-HT_3A [10]. A similar situation is observed for the γ2 subunit of GABA_A receptors, in which a short splice variant (γ2S) is capable of exiting the ER and reaching the cell surface when expressed alone. In contrast, the long splice variant (γ2L), containing an 8 amino acid insert, is retained in the ER unless co-expressed with both α1 and β2 [11]. Thus, in addition to a strict quality control mechanism operating on receptor folding and assembly within the lumen of the ER, an additional mechanism exists on the cytoplasmic side of the ER membrane. Together, these mechanisms combine to determine which surface expressed receptor compositions are permitted.

Although, the identification of assembly and ERS signals provides insight into the receptor combinations that are capable of generating cell surface functional ion channels, they provide no information regarding how this process is orchestrated. Although a number of generalized ER chaperone molecules, such as Immunoglobulin binding protein (BiP), calnexin and protein disulphide isomerase, have been shown to orchestrate receptor folding and glycosylation and return immature subunits to the ER, they operate indiscriminately [5], [12]. The recent discovery of RIC-3 (resistance to inhibitors of acetylcholinesterase 3) has opened the doors to the existence of discriminatory ER chaperone molecules. Like the general ER chaperones, RIC-3 is ER-localized (Figure 1). RIC-3 is a transmembrane protein that exhibits chaperone activity on 5-HT₃ and nicotinic acetylcholine receptors, but not on glutamate or GABA_A receptors (at least for the subunit combinations examined). RIC-3 promotes the folding, assembly and/or exit from the ER of 5-HT_3A receptors [13]. Similar properties have been reported for its effects on nicotinic acetylcholine receptors [14]. Although the 5-HT_3B subunit is preferentially incorporated into the pentamer [6] at sites that the 5-HT_3A subunit could occupy, upon the co-expression of 5-HT_3A and 5-HT_3B with RIC-3, homomeric 5-HT_3A receptor assembly is favoured [15]. Therefore, RIC-3 expression appears to reverse the normal preference for 5-HT_3B subunits and so manipulate 5-HT₃ receptor composition.

Figure 1.  COS7 cells expressing RIC3(isoform d)-DsRed (red) and 5-HT_3A-CFP (blue). RIC3 is restricted to the ER and associated vesicular structures that overlap with 5-HT_3A receptors. In addition, 5-HT_3A receptors are observed on the cell surface in microspikes and lamellopodia. Scale bar = 15 µm.

A large amount of controversy exists around RIC-3, with significant species differences being reported [14]. Mammalian RIC-3 has been shown to possess a cleavable signal sequence [15]. However, in RIC-3 from invertebrates, this region is predicted to form a non-cleaveable transmembrane domain. Moreover, whereas RIC-3 was shown to inhibit the functional expression of certain (α3β4 and α4β2) nACh receptors in Xenopus oocytes, in mammalian cells, a paradoxical enhancement of these receptors was observed [16]. Similarly, human RIC-3 has been reported to abolish the functional expression of murine 5-HT_3A receptors in oocytes [17], yet enhance the functional expression of human 5-HT_3A in mammalian cells [13], [15]. Finally, drosophila RIC-3 promotes the functional expression of nACh receptors (mammalian or drosophila receptors) more efficiently in drosophila cells, than in human cells [18]. The reciprocal is also true for human RIC-3. Together, these results point to the existence of other, host cell specific, accessory cofactors.

It remains to be established whether RIC-3 expression constitutively drives the formation of 5-HT_3A receptors or if it may operate to dynamically regulate receptor composition. Moreover, how RIC-3 would handle the expression of 5-HT_3C-E is currently unknown. Similar to that of RIC-3, cyclophilin A has been reported to promote the functional expression of 5-HT_3A receptors. In the case of cyclophilin A this has been attributed to its peptidyl prolyl isomerase activity [19]. Again, little is known regarding how this operates on 5-HT_3B-E receptors.

To study receptor trafficking beyond the ER a combination of a fluorescently labelled antagonist and a fluorescent chimaera with 5-HT_3A receptors were utilized [20]. Using a fluorescently labelled 5-HT_3A construct (5-HT_3A-CFP), receptors were detectable within the ER within 3 h, the Golgi within 4 h and the surface within 4.5 h. As expected, ligand binding sites were detected within the ER, Golgi and on the cell surface. In the presence of cycloheximide, to block new protein synthesis, the majority of receptors were localized to the cell surface. The 5-HT_3A receptor is trafficked to the cell surface within vesicles associated with microtubules and delivered to actin-rich surface domains such as lamellipodia and microspikes [20], [21] (Figure 1). This localization may be relevant to the targeting of receptors to actin-rich synapses (see below). Overall, these findings are consistent with a rate-limiting step existing at the ER [22] where subunits are folded and assembled prior to rapid transport to the cell surface.

The detection of native 5-HT₃ receptors in neurons is difficult due their low expression levels. Nevertheless, native 5-HT_3AS receptors (a short splice variant) have been identified to exist as dendritic clusters that colocalize with actin in cultured hippocampal neurons [21]. When 5-HT_3AS receptors were expressed recombinantly in these cells, receptors were preferentially localized to both filamentous and clustered (1–2 µm²) actin structures. The maintenance of these clusters appears to depend on the actin cytoskeleton, as the disruption of the cytoskeleton lead to a dispersal of the clusters into smaller structures (<0.5 µm²) [21]. Similarly, 5-HT₃ receptors have been localized to the tips of micropodia (HEK cells) and on dendritic spines [23]. To date, no information is available regarding the existence of synaptic and/or actin anchoring proteins that might be responsible for 5-HT₃ receptor targeting and retention at synaptic sites.

In addition to their postsynaptic localization [21], [23], 5-HT_3A receptors have been localized to presynaptic GABAergic (inhibitory) nerve terminals in the amygdala [24] and CA1 region of the hippocampus [25]. Within the superficial dorsal horn, 5-HT_3A receptors are localized exclusively to presynaptic glutamatergic (excitatory) synapses [26]. Similarly, 5-HT_3A receptors have been localized to neuronal terminals in axons and dendrites, as well as in glial cell membranes within the medial nucleus of the solitary tract [27]. These findings are consistent with a role of presynaptic 5-HT_3A receptors in the modulation of neurotransmitter (including GABA, 5-HT, glutamate, dopamine and acetylcholine) release, as well as in postsynaptic responses. Moreover, the localization of 5-HT_3A receptors to glial processes apposing 5-HT_3A receptor containing axons and dendrites suggests that the receptors may participate in glial signalling pathways in response to the same 5-HT release events.

Unusually, a large amount of intracellular 5-HT_3A receptors have been observed [27] suggesting that 5-HT_3A receptor-containing synapses may be dynamically regulated. Although some intracellular receptors exist within the biosynthetic pathway (ER and Golgi), indicating the delivery of newly synthesized receptors, others may be derived from the endocytic removal of surface receptors from the somatodendritic (endosomes) or axonal (dense core vesicles) membranes [27]. Indeed, intracellularly retained, unassembled NR1 subunits are thought to serve as substrates for the assembly of NMDA receptors [28]. Moreover, intracellular AMPA receptors residing within the biosynthetic and endosomal pathways are thought to contribute to trafficking events leading to synaptic targeting and the modification of synaptic strength [29].

Using novel live imaging techniques [30], it was determined that a large variability in cell surface mobility of 5-HT_3A receptors exists. This variability most likely reflects a heterogeneous population of receptors that are either freely diffusable or anchored to the actin cytoskeleton [20], [21]. Interestingly, 5-HT_3A receptor mobility decreases in response to exposure to the receptor agonist, 5-HT. This raises the possibility that synaptic 5-HT₃ receptor anchoring may be triggered at sites of 5-HT release. However, in support of the endocytic regulation of 5-HT_3A receptors, recombinantly expressed surface receptors were shown to be dynamically removed when exposed to the continued presence of agonist, with ∼50% of surface receptors being internalized within 5 min [20]. Similar receptor internalization has been observed for native receptors in myenteric neurons [31]. Interestingly, actin appears to play a role in the recruitment of 5-HT_3A receptors to the cell surface via the stimulation of PKC activity [32]. The PKC-stimulated 5-HT_3A receptor delivery to the cell surface appears to be inhibited by polymerized actin and it has been hypothesized that actin may provide an inhibitory barrier to receptor recruitment [32]. Interestingly, these processes are reminiscent of GABA_A receptor internalization [33]. Together, these studies suggest a possible scenario for the synaptic targeting of 5-HT₃ receptors, in which receptors are delivered to cell surface extrasynaptic sites and diffuse into synapses where they may be anchored to actin in response to 5-HT release. Such a scenario has been reported for the GABA_A receptors (see below). To date, there is little detail regarding how 5-HT₃ receptors are trafficked to the cell surface or synapses, or how they are removed. Extrapolation from the evidence obtained on other Cys-loop receptors provides some clues as to where to look. However, no trafficking or anchoring molecules have been identified that operate on 5-HT₃ receptors and no receptor interacting proteins (other than RIC-3) have been reported.

GABA_A receptors

GABA_A receptors are pentameric ion channels constructed from a repertoire of subunits that include α1-6, β1-3, γ1-3, δ, ε, π and θ. A detailed account of GABA_A receptor assembly is presented elsewhere in this issue (Sieghart, this issue). With respect to the most common composition of GABA_A receptors (α1β2γ2) in the brain, the minimal requirement for the production of functional, GABA gated, ion channels is the incorporation of α1 and β2 subunits and inclusion of the γ2 subunit is required for sensitivity to benzodiazepines [12]. It has been thought that the incorporation of subunits from the remaining classes (δ, ε, π and θ) occurs as a direct replacement for the γ subunit, but this has now been questioned for the ε [34], [35] and θ [36] subunits.

Although no proteins, such as RIC-3, have been reported to manipulate receptor composition, there is one potential candidate, that of GABARAP (GABA_A receptor associated protein). GABARAP is a molecule that acts as a linker between the γ2 subunit and microtubules and promotes (but is not essential for) the synaptic clustering of GABA_A receptors [37]. Although GABARAP enhances the surface expression (and so synaptic localization) of GABA_A receptors, it is not localized to inhibitory synapses [38], ruling out a role in synaptic anchoring. Instead, GABARAP is enriched in ER and Golgi structures. GABARAP has been reported to exert a number of influences on GABA_A receptors. These include decreased desensitization, a rightward shift in EC₅₀ for GABA, increased conductance and increased synaptic targeting. All of these features are consistent with the encouragement of γ2-containing receptors. Intriguingly, when the GABA_A receptor composition is constrained (by physically linking subunits into concatemers) to the production of αβγ, such that αβ receptors cannot be formed, GABARAP no longer exhibits any effects on promoting a γ2 physiological phenotype [39]. Likewise, when the γ2 subunit is present in large excess (1:1:10 ratio), GABARAP appears to have no effect. Therefore, it is possible that an important function of GABARAP is to promote the incorporation of γ2 into αβγ receptors. A role for GABARAP in promoting the formation of αβγ, over αβ, GABA_A receptors is reminiscent of the role RIC-3 plays in promoting the formation of 5-HT_3A, at the expense of 5-HT_3A/3B, receptors [15].

An alternative role for GABARAP is suggested by its homology to GATE-16 (Golgi-associated transport enhancer of 16KDa), pointing to a possible role in vesicular trafficking. Interestingly, P130 competitively inhibits GABARAP binding to γ2, yet P130 is required for the efficient expression of γ2-containing receptors [40]. This would be consistent with a transient requirement for the interaction of γ2 with GABARAP (during its assembly?). Subsequent trafficking steps might then require the interaction with other molecules, such as P130. GABARAP has also been shown to interact with NSF (N-ethylmalemide-sensitive factor) and GRIP-1 (glutamate receptor interacting protein 1) [38], [41]. Both these molecules have been shown to play a role in the trafficking and synaptic localization of AMPA receptors [29], but their relevance to these processes in GABA_A receptor trafficking is not established.

Once GABA_A receptors have assembled in the ER, they are transported through the Golgi stacks and trans-Golgi network, prior to transport to the cell surface. This may involve Plic-1, which is associated with intracellular membranes including the ER and membranes at the edges of Golgi, but is not concentrated at sites of surface clustered GABA_A receptors [42]. Plic-1 interacts with the cytoplasmic domains of α and β, but not γ2 or δ subunits. Interestingly, inhibiting the Plic-1 interaction with α subunits, leads to a decrease in surface receptor levels (but no enhancement of receptor internalization). Thus, Plic-1 may operate by enhancing the recruitment of receptors to the cell surface. The observation that Plic-1 increases receptor stability is consistent with the proposed role for Plic-1 as a negative regulator of the proteasome and it has been proposed that Plic-1 traffics with the receptors, along the biosynthetic pathway, protecting them from degradation. However, an enhancement of subunit folding, receptor assembly or transport out of the ER would all lead to protection against proteasomal degradation. Indeed, these are the most likely mechanisms by which RIC-3 operates on 5-HT₃ and nACh receptors. Therefore, it is possible that Plic-1 may exhibit chaperone-like activity on newly synthesized GABA_A receptors that is independent of proteasomal activity. At the cell surface, it is possible that Plic-1 provides protection against proteasomal/lysosomal degradation within the endocytic pathway and so permit the recycling of internalized receptors back to the cell surface [42].

Within the Golgi, GABA_A receptors interact with a range of molecules involved in further trafficking steps. A palmitoyltransferase, GODZ (Golgi specific protein with the DHHC zinc finger domain) is able to interact with the γ2 subunit and trap the subunit within the Golgi unless it is assembled with α and β subunits [43]. This may be relevant, given the ability of γ2S to escape the ER and traffic to the cell surface prior to assembling with α1 and β2 [11]. In the presence of assembled receptors (αβγ), GODZ can be observed beneath, but not on, the cell surface [43]. Thus GODZ may also be involved in the transport of GABA_A receptors to the cell surface. GODZ does not form a stable interaction with the γ2 subunit, but has been shown to palmitoylate the γ2 subunit and this has been reported to be a requirement for the synaptic localization of receptors [44], [45]. It is important to note that surface trafficking is a prerequisite to synaptic targeting, and this has also been shown to require palmitoylation [45]. Palmitoylation does not appear to be restricted to the GABA_A receptor members of the Cys-loop family, both α7 nACh and 5-HT_3A receptors are palmitoylated [46]. In the case of these receptors, it appears that palmitoylation occurs during receptor assembly and is required for the formation of ligand binding sites and cell surface expression.

Another molecule primarily localized to the Golgi and is involved in GABA_A receptor targeting is BIG2 (brefeldin A-inhibited GDP/GTP exchange factor 2) that interacts with β subunits [47]. Although it is primarily localized to the trans Golgi network and at the ends of Golgi cisterae, it has also been identified within presynaptic and postsynaptic, inhibitory and excitatory, synapses, where it is associated with the synaptic membrane and intracellular vesicles. In addition, the overexpression of BIG2 has been reported to increase GABA_A receptor release from the ER, a compartment that is upstream in the biosynthetic pathway. Therefore, BIG2 may be more widely distributed, dynamically moving throughout the biosynthetic and endocytic pathways and may play a general role in protein trafficking.

GABA_A receptors may be selected for forward transport, or simply guided, through the biosynthetic pathway by their interaction with GRIF1 (GABA_A receptor interacting factor-1, otherwise known as TRAK2). GRIF-1 is a cytoplasmically localized protein that is thought to be an adaptor, linking organelles to kinesin 1 motor proteins, to facilitate their anterograde transport [48]. GRIF-1 also interacts with α and β subunits of GABA_A receptors [49] and may be involved in their surface and/or synaptic targeting.

At the cell surface, GABA_A receptors may exist at synaptic or extrasynaptic sites where they participate in phasic (synaptically induced, by presynaptic neurotransmitter release) and tonic (intrinsic, non-synaptic) inhibition. Synaptic targeting of GABA_A receptors requires the incorporation of the γ subunit [50], while δ-containing receptors are found predominantly extrasynaptically [51]. The GABA_A receptor expression level on the neuronal surface is a major contributor to the level of neuronal inhibition. This is well illustrated in heterozygotes of γ2 knockout mice. These animals have reduced synaptic GABA_A receptor levels and exhibit anxiety that may be relieved by enhancing GABA_A receptor activity with benzodiazepine treatment [52].

Importantly, GABA_A receptors are not thought to be static portals of neuronal inhibition. Rather, they represent a dynamic pool of receptors undergoing rapid recruitment to, and removal from, the neuronal surface and synapses. Therefore, the exocytic and endocytic pathways that regulate GABA_A receptor trafficking are directly relevant to the inhibitory control of neuronal activity in the brain. Further detail regarding GABA_A receptor trafficking is available in recent reviews [33], [37], [53].

To date, the only molecule implicated in the synaptic anchoring of GABA_A receptors is gephyrin. Interestingly, it appears that both the γ2 subunit and gephyrin are mutually required for synaptic localization at most, but not all, synapses [54]. At extrasynaptic sites GABA_A receptor clusters appear to anchor to the actin cytoskeleton via an ERM (ezrin/radixin/moesin) family protein, radixin [54].

Once on the cell surface, GABA_A receptors exhibit significant movement as a result of lateral diffusion and constitutive exocytosis and endocytosis. By engineering a binding site for an activity-dependent, irreversible inhibitor, it has been possible to selectively inactivate receptors that were stimulated by synaptic activity [55]. Using this approach to inactivate synaptic receptors, it was discovered that synaptic activity recovered rapidly, within minutes. No recovery occurred when all surface receptors were blocked. In contrast, synaptic recovery was not blocked when the exocytosis of intracellular receptors was inhibited. Therefore, these findings provide strong evidence that synaptic GABA_A receptors are recruited from an extrasynaptic surface pool. This conclusion is supported by studies that determined that synaptic receptors are preferentially stabilized at synapses [56]. Moreover, extrasynaptic GABA_A receptors are constitutively recycled by endocytosis and exocytosis. Both the machinery for clathrin-dependent endocytosis and the sites of delivery of newly inserted receptors were found to be extrasynaptic [57]. Together, these studies show that surface GABA_A receptor levels are regulated at extrasynaptic sites. Synapses then draw on this mobile pool of receptors as required.

In addition to the surface extrasynaptic pool, a significant intracellular pool of receptors is also available. Indeed, there are both constitutive and regulated pathways for GABA_A receptor endocytosis, and receptors may be either recycled to the cell surface or degraded [58–60]. The constitutive endocytosis of GABA_A receptors occurs via clathrin-coated vesicles to which GABA_A receptors are recruited by a novel AP2 binding motif (KTHLRRRSSQLK) [60], which is conserved in all β subunits. This process appears to require the recruitment of PRIP (phopholipase C-related catalytically inactive protein) to β subunits [59]. In addition, γ2-containing GABA_A receptors can undergo regulated, PKC-activated, endocytosis [58]. This regulated GABA_A receptor endocytosis utilizes a distinct second AP2 binding domain is based on a di-leucine motif present in β subunits [61]. Interestingly, the constitutive endocytosis signal binding to AP2 is blocked by phoshorylation. Given that this motif carries a PKC phosphorylation site, constitutive endocytosis of GABA_A receptors may be blocked, and regulated endocytosis potentiated, upon PKC activation. Such an event might deplete both the extrasynaptic and intracellular pools of receptors by endocytosis and degradation and so limit the normal maintenance of synaptic receptors. This process may regulate the functions of HAP-1 (Huntingtin-associated protein) and NSF in controlling GABA_A receptor surface expression. HAP-1 appears to interact with β subunits and protect GABA_A receptors from lysosomal degradation and so facilitate receptor recycling to the cell surface [62]. NSF has been shown to interact with β subunits at, or near the cell surface and lead to a decrease in surface receptor levels [63]. This is in contrast to its interaction with GABARAP that presumably (this has not been shown) promotes the delivery of GABA_A receptors to the surface. The mechanism(s) by which NSF may play such opposing roles has not been established, but it may enable the recruitment of synaptic (γ2-containing) receptors via NSF-GABARAP interactions in the biosynthetic pathway, and the removal of extrasynaptic (non γ2-containing) receptors via a direct interaction with the β subunits in the endocytic pathway. These mechanisms may play important pathological roles in the reduced phasic and/or tonic inhibition observed as a result of reduced surface GABA_A receptor expression during anxiety [52], epilepsy [64–67] and ischemia [68].

Concluding remarks

It is clear that the pooling of knowledge from both 5-HT₃ and GABA_A receptors, as well as that of nACh (Millar, this issue) and glycine (Becker, this issue) receptors, strengthens our understanding of Cys-loop receptor trafficking as a whole. Clearly, more research is needed to identify the array of accessory proteins (that undoubtedly exist) involved in the regulation of 5-HT₃ receptor biogenesis, trafficking and synaptic anchoring. Moreover, given the localization of several of the identified proteins to multiple points on the biosynthetic and endocytic pathways, it will be important to investigate their dynamic mobility using time-lapse microscopy. If this could be performed with respect to the receptors that they traffic, as well as the movement of the other accessory molecules, a dynamic picture of the entire process may emerge. This is an exciting prospect, especially if it could then be investigated during pathological states. This may identify critical points at which the process is perturbed in conditions such as ischemia, epilepsy, anxiety, schizophrenia and the neurodegenerative diseases. A full understanding of how these receptor trafficking events are orchestrated and regulated may provide novel therapeutic opportunities.