GABA_B receptor promotes its own surface expression by recruiting a Rap1-dependent signaling cascade

For many GPCRs, signal transduction is controlled by their agonist-induced desensitization and subsequent internalization to protect cells against receptor over-stimulation (Moore et al., 2007). The best characterized such example is β2-adrenergic receptor (β2-AR). Upon agonist stimulation, GPCRs first undergo phosphorylation catalyzed by G-protein receptor kinases (GRKs) or other kinases. Subsequently, β-arrestin is recruited to the receptor to preclude receptor–G-protein interaction, leading to internalization of the receptor from the plasma membrane (Gainetdinov et al., 2004; Luttrell and Lefkowitz, 2002). Once internalized, receptors are dephosphorylated and recycled back to the plasma membrane or are targeted to lysosomes for degradation (Marchese et al., 2008).

By contrast, some other GPCRs such as GABA_B receptor maintain a rather stable expression level at the cell surface even after prolonged agonist stimulation (Couve et al., 2002; Fairfax et al., 2004; Kuramoto et al., 2007). Other examples include MrgX1, somatostatin-4, β3-AR, and κ-opioid receptors (Chu et al., 1997; Nantel et al., 1993; Rosenfeld et al., 2002; Schreff et al., 2000). GABA_B receptor belongs to the class C type of GPCRs and is the metabotropic receptor for γ-aminobutyric acid (GABA), the primary inhibitory neurotransmitter in the mammalian nervous system (Bettler et al., 1998). GABA_B receptor functions as a heterodimer composed of two subunits GB1 and GB2 (also known as GABA_BR1 and GABA_BR2) (Jones et al., 1998; Kaupmann et al., 1998; White et al., 1998). Each subunit contains an extracellular domain, a seven transmembrane domain, and an intracellular C terminal. Only the GB1 extracellular domain binds agonists, whereas the seven transmembrane domain of GB2 is responsible for the activation of Gi/o proteins (Galvez et al., 2001; Kniazeff et al., 2002). It is a common therapeutic target for a wide range of neurological disorders, including epilepsy, schizophrenia, addiction, depression, anxiety and chronic pain (Bettler et al., 2004; Gassmann et al., 2004; Thuault et al., 2004). Unlike GABA_A and GABA_C receptors, which are fast-acting ion channels, as a GPCR GABA_B receptor is best known to mediate the slow, prolonged inhibitory effect of GABA neurotransmitter (Couve et al., 2000). This feature demands persistent expression of functional GABA_B receptor at the cell surface. Thus, the unique cell surface stability of GABA_B receptor is well suited for its role in mediating long-term synaptic inhibition in the central nervous system (CNS).

However, the mechanism underlying the surface stability of GABA_B receptor upon prolonged agonist stimulation is unknown and somewhat controversial. Although GABA_B receptor is known to undergo basal endocytosis, it remains controversial whether agonist stimulation facilitates receptor internalization (Fairfax et al., 2004; Grampp et al., 2008, 2007). Thus, some studies attribute the cell surface stability of GABA_B receptor to a lack of agonist-induced internalization (Fairfax et al., 2004; Perroy et al., 2003), whereas others suggest that this might be due to re-insertion of the receptor into the plasma membrane by an unknown mechanism (Grampp et al., 2008, 2007; Wilkins et al., 2008).

Here, we sought to investigate the mechanism by which GABA_B receptor maintains its surface expression level under prolonged agonist stimulation. Through biochemical purification, we identified the small GTPase Rap1 (which has two isoforms, Rap1a and Rap1b) as a new GABA_B-receptor-binding partner. Activation of GABA_B receptor initiates a signaling cascade, including Gi/o, Rap1GAPII (also known as Rap1GAP1b, an isoform of Rap1GAP1) and Rap1 that binds to the receptor to promote its surface expression by facilitating receptor recycling. Our results identify a new mechanism that promotes GPCR surface expression, providing insights into the slow and long-lasting action of the neurotransmitter GABA.

The C-terminal region of GPCRs usually has many binding sites for soluble and scaffold proteins that are involved in various regulatory processes such as desensitization, internalization and recycling (Bockaert et al., 2004). As no proteins known to bind to the C-terminus of the GB1 or GB2 subunits appear to regulate the cell surface stability of GABA_B receptor (Benzing et al., 2000; Brock et al., 2005; Guetg et al., 2010; Kuramoto et al., 2007; Nehring et al., 2000; Schwenk et al., 2010; Tan et al., 2004), we set out to identify new GABA_B-receptor-interacting proteins. To this end, we chose to work with cerebellar granule neurons (CGNs) that express endogenous GABA_B receptor, as the function of GABA_B receptor has been well characterized by us and many others in these neurons (Lin et al., 2012; Tu et al., 2007, 2010). We stimulated CGNs with the prototypical GABA_B receptor agonist baclofen, and then used glutathione-S-transferase (GST) fusion proteins containing the C-terminal part of GB1 or GB2 (GST–GB1-C or GST–GB2-C) (Couve et al., 2004; Fairfax et al., 2004; Kuramoto et al., 2007; Nehring et al., 2000; Perroy et al., 2003) as a bait to pull down potential interacting proteins from the cell lysate (Fig. 1A). On the SDS-PAGE gel visualized by silver staining, several protein bands appeared to be pulled down by GST–GB1-C, which were characterized further by trypsin digestion followed by mass spectrometry (Fig. 1B; supplementary material Table S1). Using this strategy, we identified the small GTPase Rap1 as a binding partner of GB1.

To verify this interaction, we probed the pulldown fraction with an anti-Rap1 antibody and found that it indeed interacted with the GB1 C-terminus but not with GST alone (Fig. 1C). By contrast, the same region in GB2 failed to pull down Rap1, suggesting that Rap1 specifically interacts with GB1 (Fig. 1D). A similar phenomenon was observed in HEK293 cells transfected with GABA_B receptor upon treatment with baclofen, a specific agonist for GABA_B receptor (Fig. 1E). Interestingly, no such interaction was detected in the absence of baclofen treatment, indicating that Rap1 was recruited to GABA_B receptor upon receptor activation. These results demonstrate that Rap1 was recruited to GABA_B receptor upon receptor activation by binding to the C-terminal end of the GB1 but not the GB2 subunit of the receptor.

As a molecular switch, Rap1 exists in at least two different forms: the active form Rap1-GTP and the inactive form Rap1-GDP. To ascertain whether GABA_B receptor binds to Rap1-GTP or Rap1-GDP, we tested the constitutive active mutant of Rap1 (Rap1^V12) which is GTP-bound, as well as its dominant negative mutant of Rap1 (Rap1^N17) which is GDP-bound. We found that GB1 C-terminus can interact with Rap1^V12 but not Rap1^N17 (Fig. 1F), suggesting that GABA_B receptor interacts with Rap1-GTP but not Rap1-GDP.

The observations that GB1 interacts with Rap1 in a baclofen-dependent manner and that it only interacts with the active form of Rap1 suggest that agonist stimulation of GABA_B receptor activates Rap1. To test this, we performed an RBD assay to detect the active form of Rap1 (Rap1-GTP). Briefly, the GST–RalGDS-RBD fusion protein is purified and used to bind the activated GTP-bound Rap1, but not inactivated GDP-bound, Rap1 (Tsukamoto et al., 1999), which can then be pulled down with glutathione resin. Rap1 activation levels are then determined by western blotting using an anti-Rap1 antibody. Indeed, baclofen induced a persistent activation of Rap1 in a time-dependent manner in both CGNs (Fig. 2A; supplementary material Fig. S1B) and HEK293 cells transfected with GABA_B receptor (Fig. 2B; supplementary material Fig. S1C). A similar phenomenon was observed upon treatment with CGP7930, a specific positive allosteric modulator of GABA_B receptor, which activates GABA_B receptor by binding to the HD domain of GB2 subunit (Binet et al., 2004) (supplementary material Fig. S1A,B). Furthermore, CGP54626, a specific antagonist of GABA_B receptor, could block baclofen-induced Rap1 activation (Fig. 2C). Taken together, these results demonstrate that stimulation of GABA_B receptor induces persistent activation of Rap1.

How does GABA_B receptor transmit signals to Rap1? GABA_B receptor is known to activate its downstream effectors through the Gi/o type of heterotrimeric G proteins (Mannoury la Cour et al., 2008). Activation of GABA_B receptor signals through Gαi/o and Gβγ subunits. We first tested whether Gi/o-type G proteins were involved in GABA_B-receptor-induced Rap1 activation by using pertussis toxin (PTX). We found that PTX inhibited baclofen-induced Rap1-GTP formation in CGNs (Fig. 2D). Furthermore, by using the previously characterized Gβγ-scavenger, consisting of the C-terminal region of GRK2 (βARK) fused to the extracellular and transmembrane domains of CD8, which then provides a membrane anchor for C-terminal tail of βARK (CD8–βARK), which is known to block Gβγ-mediated signaling such as GABA_B-receptor-induced phosphorylation of ERK1/2 (also known as MAPK3 and MAPK1, respectively) (Tu et al., 2007), failed to inhibit Rap1 activation induced by baclofen, whereas it did inhibit baclofen-induced ERK1/2 phosphorylation in HEK293 cells transfected with GABA_B receptor (Fig. 2E). These results indicate that GABA_B receptor induces the formation of Rap1-GTP by signaling through the Gαi/o subunit.

We then asked how GABA_B receptor induces Rap1 activation through Gαi/o. Rap proteins are active when bound to GTP and inactive when bound to GDP. Rap activation is mediated by specific guanine nucleotide exchange factors (GEFs), whereas inactivation of Rap1-GTP can be achieved with the aid of specific GTPase-activating proteins (GAPs). Notably, the Gαi/o subunit is known to inhibit the activity of Rap1GAPII, a GAP for Rap1 (Jordan et al., 1999; Mochizuki et al., 1999). Phosphorylation of Rap1GAPII controls Rap1 activity (McAvoy et al., 2009). We therefore tested the effect of baclofen activation of GABA_B-receptor-mediated Gαi/o signaling on Rap1GAPII phosphorylation. We found that baclofen treatment rapidly and persistently potentiated Rap1GAPII phosphorylation, indicating that activation of GABA_B-receptor–Gαi/o signaling inhibits Rap1GAPII in HEK293 cells transfected with GABA_B receptor and GST–Rap1GAPII (Fig. 2F).

To provide additional evidence, we reasoned that if inhibition of Rap1GAPII sustains Rap1 activity, then overexpression of Rap1GAPII should reduce Rap1 activation induced by stimulation of GABA_B receptor. Consistent with this prediction, overexpression of wild-type Rap1GAPII in GABA_B-receptor-expressing HEK293 cells blocked the formation of Rap1–GFP induced by baclofen. As a control, we showed that an inactive form of Rap1GAPII failed to elicit such an effect (Fig. 2G). As Rap1GAPII is a GAP for Rap1, these results suggest a model in which GABA_B receptor activates Rap1 by signaling through the Gαi/o subunit and Rap1GAPII.

To gain a better insight into how Rap1 interacts with GB1, we sought to map the sites that are required for the interaction. As a first step, we examined Rap1. Previous reports show that several negatively charged residues [aspartate (D) and glutamate (E)] in Rap1 mediate its interaction with the RBD domain of RalGDS, a known Rap1-binding protein (Fig. 3A) (Liu et al., 2011). As such, we mutated these negatively charged residues in Rap1 to alanine (i.e. D33A, E37A and D38A) to test whether they are important for mediating the interaction between Rap1 and GB1 C-terminus (supplementary material Table S2). D33A and E37A point mutations in Rap1 disrupted its interaction with the C-terminal end of GB1. A D38A point mutation in Rap1 also reduced this interaction. As a control, Rap1^N17, a GDP-bound form of Rap1, failed to bind to GB1 (Fig. 3B). The identification of D33, E37 and D38 in Rap1 as three crucial residues required for Rap1 and GB1 interaction provides further evidence that the observed interaction is specific.

We then sought to map the sites in GB1 C-terminus that are required for mediating its interaction with Rap1. To do so, we first aligned the sequences between GB-1 C-terminus and the RBD region of RalGDS, a domain known to interact with Rap1-GTP. We noted that a small region spanning from R947 to K960 in GB1 C-terminus is analogous to the key residues in the GalGDS-RBD domain that have been found to be important in mediating the interaction with Rap1-GTP in RalGDS (Nassar et al., 1995). We then aligned the sequences of this region in GB1 and GB2 from several organisms, and found that the residues 953–960 (SRVHLLYK) in GB1 are conserved in most organisms ranging from X. tropicalis to Homo sapiens but show no similarity with GB2-C terminus in this regions[residues 934–940 (FRVMVSGL)] (Fig. 3C). To test whether this short motif in GB1 is crucial for mediating the interaction, we made a series of deletions and point mutations in this motif and examined their effects (supplementary material Table S2; Fig. S1D). Deletion of S953 or S959 abolished the ability of GB1 C-terminus to interact with Rap1-GTP in CGNs (Fig. 3D). Two point mutations R954A and K960A also disrupted the interaction (Fig. 3D). Thus, the residues SRVHLLYK in GB1 appear to play an important role in mediating the interaction between GABA_B receptor and Rap1.

To provide further evidence, we designed and synthesized peptides (supplementary material Table S3), aiming to directly impair the interaction between Rap1-GTP and the GB1 C-terminus (Fig. 3E). These peptides are membrane permeable, making it possible to use them to disrupt Rap1-GB1 interaction in situ (supplementary material Fig. S2A). As expected, Pep (DGSRVHLLYK) and Pep-S7 (RVHLLYK) inhibited the interaction between GB1 C-terminus and Rap1-GTP both in CGNs (Fig. 3F) and HEK293 cells (supplementary material Fig. S2B). As negative controls, two unrelated peptides Pep-GB2 (FRVMVSGL) and PepN (PPDRLSCDGS) were unable to block the interaction (Fig. 3F; supplementary material Fig. S2B). Furthermore, two other peptides Pep-K960A (DGSRVHLLYA) and Pep-R954A (DGSAVHLLYK), which encompass point mutations, also failed to do so. This provides additional controls supporting the specificity of Pep and Pep-S7 in blocking the interaction between GB1 C-terminus and Rap1-GTP (Fig. 3F; supplementary material Fig. S2B). These analyses identify specific sites in Rap1 and GB1 C-terminus that are important for mediating the interaction between Rap1-GTP and GABA_B receptor, offering further evidence that the observed interaction is specific.

Taken together, our data suggest a model in which agonist stimulation of GABA_B receptor activates Rap1 by signaling through Gαi/o and Rap1GAPII, leading to recruitment of the active form of Rap1 to the receptor complex through a direct interaction between Rap1 and the GB1 subunit of the receptor.

Having demonstrated that agonist stimulation of GABA_B receptor activates and recruits Rap1 to the receptor through a direct protein–protein interaction, we then asked what functions such an interaction mediates. We therefore decided to disrupt the interaction between Rap1-GTP and the GB1 subunit to examine whether and how it affects the function of GABA_B receptor. Activity of GABA_B receptor was quantified by a commonly used functional assay to measure inositol trisphosphate (IP₃) production elicited by baclofen in HEK293 cells co-expressing GABA_B receptor and the chimeric G-protein Gαqi9 (the last nine C-terminal residues of Gαq protein were replaced by those from Gαi2), which facilitates the coupling of Gi-coupled receptors to the phospholipase C signalling pathway. Disruption of the Rap1–GB1 interaction by Pep treatment greatly suppressed the activity of GABA_B receptor in transfected HEK293 cells (Fig. 4A; supplementary material Fig. S3A). Interestingly, the EC50 value of baclofen was not affected by Pep treatment (Fig. 4A). This implies that disruption of Rap1–GB1 interaction might not affect the functional properties of GABA_B receptor, but instead might compromise the overall amount of functional receptors at the cell surface.

We thus performed a series of assays to explore whether the interaction between Rap1 and GB1 regulates the cell surface expression of GABA_B receptor. As a first step, we used an enzyme-linked immunosorbent assay (ELISA) to measure the amount of GABA_B receptor expressed at the cell surface. We took advantage of the peptides that can disrupt the Rap1–GB1 interaction (Fig. 3E,F; supplementary material Fig. S2B). Treatment with Pep, but not its control PepN, reduced the amount of baclofen-induced GABA_B receptor at the cell surface in a time- and concentration-dependent manner (Fig. 4B; supplementary material Fig. S3B). To gather additional evidence, we performed a flow cytometry assay to quantify the surface expression level of GABA_B receptor in transfected HEK293 cells, and observed a similar phenomenon (supplementary material Fig. S3C). In addition, disrupting the interaction between GB1 and Rap1 (i.e. GB1ΔS953) greatly reduced GABA_B receptor surface expression in this assay (supplementary material Fig. S3D). The results from both ELISA and flow cytometry assays demonstrate that agonist-induced recruitment of Rap1-GTP to the GABA_B receptor complex is important for the cell surface expression of the receptor.

To take a closer view at how GABA_B receptor behaves at the cell surface, we performed a biotinylation assay to follow the fate of the receptor expressed at the cell surface upon agonist stimulation. If the interaction between Rap1 and GB1 is important for the surface expression of GABA_B receptor, as suggested by the ELISA and flow cytometry analyses, then disruption of such interaction might lead to an accumulation of GABA_B receptor inside the cell. Indeed, Pep disruption of the Rap1–GB1 interaction gave rise to a dramatic increase in the amount of internalized GABA_B receptor upon baclofen stimulation (Fig. 4C). Another interesting observation is that baclofen stimulation alone in untreated CGNs did not increase the amount of internalized receptors. These two observations together pointed to an interesting model in which agonist stimulation of GABA_B receptor potentiates receptor internalization; however, as Rap1–GB1 interaction probably also promotes the recycling of internalized receptors, the net amount of the receptors at the cell surface then stays unchanged (Fig. 4C). In other words, both the internalization and the recycling of the receptor might be potentiated in response to agonist stimulation. This model explains the seemingly confounding observations that baclofen stimulation of GABA_B receptor did not affect the net amount of internalized receptors, yet blocking Rap1–GB1 interaction led to an accumulation of internalized receptors inside the cell.

To provide further evidence supporting the above model, we directly visualized the fate of GABA_B receptor present at the cell surface before and after agonist stimulation. To do so, we specifically labeled the GABA_B receptor present at the cell surface with primary antibodies and then followed its fate by immunofluorescence staining. Consistent with the results from our biotinylation assay, baclofen stimulation of GABA_B receptor did not cause a notable change in the net amount of the receptor at the cell surface (Fig. 4D). By contrast, disruption of Rap1–GB1 interaction by deleting the residues 953–960 in GB1 (GB1ΔS953) led to a dramatic decrease in the amount of GABA_B receptor at the cell surface (Fig. 4D). Notably, this effect was strictly baclofen-dependent, as in the quiescent state (i.e. no baclofen treatment), disruption of the Rap1–GB1 interaction did not cause a notable effect on the amount of the receptor found at the cell surface (Fig. 4D). Thus, the interaction between Rap1 and GB1 appears to regulate GABA_B receptor surface expression in an activity-dependent manner.

Another interesting observation is that although baclofen treatment did not notably affect the net amount of internalized GABA_B receptor, once the Rap1–GB1 interaction was disrupted by the GB1ΔS953 deletion, then the same baclofen treatment greatly increased the amount of the receptor found inside the cell (Fig. 4D). This data strongly suggests that baclofen stimulation of GABA_B receptor promotes receptor internalization. These results, together with those from ELISA and biotinylation assays, support a model in which agonist stimulation of GABA_B receptor promotes receptor internalization and also recruits Rap1 to the receptor, which in turn might potentiate receptor recycling. A combination of such two actions would then lead to a relatively stable expression of the receptor at the cell surface. Our data also help clarify a controversy over whether agonist stimulation of GABA_B receptor facilitates receptor internalization.

The above model suggests that agonist stimulation of GABA_B receptor promotes recycling of internalized receptors back to the cell surface. If this is the case, then disrupting receptor recycling should reduce the surface expression of the receptor. To test this, we treated the cells with monensin, which blocks vesicle recycling, and found that it greatly reduced the surface expression level of GABA_B receptor upon agonist stimulation. As a control, we treated the cells with brefeldin A (BFA), which inhibits de novo delivery of GABA_B receptor by impairing ER–Golgi trafficking of newly synthesized membrane receptors. BFA had no notable effect on the surface expression of GABA_B receptor (Fig. 5A). These results suggest that agonist stimulation of GABA_B receptor promotes receptor recycling.

It might be argued that Rap1–GB1 interaction promotes GABA_B receptor surface expression by blocking agonist-induced receptor internalization rather than by promoting receptor recycling. If this is the case, then simultaneous inhibition of the Rap1–GB1 interaction and receptor recycling should have an additive effect. In other words, impairing the Rap1–GB1 interaction would reduce the surface expression of GABA_B receptor when receptor recycling is absent. However, this is not the case, as disruption of the Rap1–GB1 interaction with Pep treatment cannot further reduce the surface expression of GABA_B receptor after receptor recycling was blocked by monensin (Fig. 5A). Thus, it appears that the Rap1–GB1 interaction and receptor recycling act in the same pathway, providing additional evidence that agonist stimulation of GABA_B receptor promotes receptor recycling. As a positive control, we found that disruption of Rap1–GB1 interaction with Pep treatment can still suppress the surface expression of GABA_B receptor in BFA-treated cells, further suggesting that recycling, rather than de novo delivery, of GABA_B receptor contributes to its stable surface expression upon agonist stimulation (Fig. 5A).

What is the fate of internalized GABA_B receptor after its recycling was blocked? To address this question, we first examined to which subcellular compartments internalized GABA_B receptor is localized. Internalized receptors are primarily directed to one of two distinct destinations – either to lysosomes for degradation or they are recycled back to the cell surface. We found that ∼60% of GABA_B receptor found inside the cell colocalized with EEA1, an early endosome marker, whereas merely 15% of the receptor colocalized with LAMP1, which marks late endosomes and lysosomes (Fig. 5B–D). By using monensin, which inhibits the fusion of intracellular vesicles with the plasma membrane, then blocks recycling pathway (Grampp et al., 2008), we analyzed whether recycling or degradation is the main pathway of endocytosed GABA_B receptors. We found that monensin treatment drastically reduced the amount of GABA_B receptor localized to early endosomes while greatly increased its presence in late endosomes or lysosomes (Fig. 5B–D). This data suggests that blockade of GABA_B receptor recycling re-routes the receptor to late endosomes and lysosomes for degradation.

We also found that disruption of Rap1–GB1 interaction with Pep treatment substantially reduced the amount of intracellular GABA_B receptor localized to early endosome while it increased its presence in late endosomes and lysosomes (Fig. 6A,B). A similar observation was made with the GB1ΔS953 mutation, which disrupts the interaction between Rap1 and GB1 (supplementary material Fig. S4A,B). Furthermore, our results also show that disruption of the Rap1–GB1 interaction with Pep treatment substantially reduced the amount of intracellular GABA_B receptor localized to recycling endosome (Fig. 7A,B). These findings are very similar to those obtained with monensin, consistent with our model that the interaction between Rap1–GB1 is important for GABA_B receptor recycling.

One prominent feature of many types of membrane receptors, particularly GPCRs, is the downregulation of their surface expression following prolonged agonist treatment, a mechanism adopted by cells to protect them from receptor over-stimulation (Ferguson, 2001). How the surface expression of these receptors is downregulated in an activity-dependent manner has been very well characterized (Calebiro et al., 2010; Premont and Gainetdinov, 2007; Shenoy and Lefkowitz, 2011). However, some other GPCRs such as GABA_B receptor remain relatively stable at the cell surface even after prolonged agonist treatment (Benke, 2010; Chu et al., 1997; Nantel et al., 1993; Schreff et al., 2000; Solinski et al., 2010). GABA is the primary inhibitory neurotransmitter in the mammalian CNS. The relative stable surface expression of GABA_B receptor is well suited for its role in mediating the slow, long-lasting inhibitory function of GABA (Benke, 2010). Despite extensive studies, the mechanisms underlying the stable surface expression of GABA_B receptor is unclear.

Here, we identified a mechanism by which GABA_B receptor promotes its own surface expression under prolonged agonist stimulation (Fig. 7C). Our data show that it does so by recruiting a signaling cascade including Gi/o, Rap1GAPII and Rap1, with the latter protein recruited to the GB1 subunit of the receptor through direct protein–protein interactions in an activity-dependent manner. Specifically, in response to agonist stimulation, GABA_B receptor activates the Gi/o protein, which in turn triggers phosphorylation of Rap1GAPII, a GAP for Rap1, to inhibit Rap1GAPII activity. This leads to an increase in half-life of the active form of Rap1 (Rap1-GTP). Rap1-GTP then binds to the C-terminal end of GB1 subunit of GABA_B receptor to promote its recycling to the plasma membrane. This effect counteracts agonist-induced internalization of GABA_B receptor, leading to a relatively stable surface expression of the receptor. Blocking this signaling cascade by inhibiting Rap1 activity, disrupting the interaction between Rap1 and the receptor, or repressing vesicle recycling greatly compromised the surface expression of GABA_B receptor. To the best of our knowledge, this represents a new mechanism by which membrane receptors regulate their own surface expression.

Our results show that the relatively stable surface expression of GABA_B receptor under chronic agonist stimulation is not a static but rather a dynamic process. Agonist stimulation of GABA_B receptor potentiates its internalization, which is supported by multiple lines of evidence, including ELISA analysis, flow cytometry, biotinylation assay and immunofluorescence staining. In the meantime, agonist stimulation also facilitates recycling of intracellular GABA_B receptors. A combination of these two actions leads to a relatively stable surface expression of the receptor. These results also help clarify a long-standing controversy over whether GABA_B receptor potentiates internalization in response to agonist stimulation.

It is not clear why agonist stimulation of GABA_B receptor facilitates both internalization and recycling (Benke, 2010; Grampp et al., 2008). On the one hand, this seems to constitute a futile cycle, which is energy inefficient. On the other hand, agonist-induced internalization of membrane receptors is considered a general phenomenon and represents a default pathway for most, if not all, GPCRs (Marchese et al., 2003). In this case, to maintain a relatively stable surface expression of the receptor, it is logical for the cell to promote recycling of internalized receptor and/or increase de novo delivery of newly synthesized receptor to the cell surface (Tsao and von Zastrow, 2000). Our results suggest that GABA_B receptor mainly adopts the former mechanism under our conditions (a 2-h window). Nevertheless, we do not exclude the possibility that the latter mechanism also plays a role under other conditions, for example, in a longer time frame.

We identified Rap1 as a new GABA_B-receptor-binding partner through a classic biochemical purification approach. Rap1 is a small GTPase and regulates GABA_B receptor activity by promoting its surface expression. Conversely, GABA_B receptor regulates Rap1 activity by promoting the formation of its active form Rap1-GTP. This constitutes an interesting positive-feedback loop that regulates GPCR function through small GTPases, a phenomenon that has not been reported. In addition, although several types of GPCR-interacting proteins have been identified, no small GTPases have been shown to directly bind to GPCRs. Rap1 represents the first such small GTPases associated with GPCRs. Small GTPases, such as Rab proteins, are known to regulate vesicle trafficking by acting as a sorting signal (Bhattacharya et al., 2004; Rosenfeld et al., 2002). It is possible that Rap1 might function through a similar mechanism. Nevertheless, future studies are needed to uncover exactly how Rap1 promotes recycling of GABA_B receptor. As Rap1 and GABA_B receptor are evolutionarily conserved from worms to humans, our findings raise the possibility that a similar phenomenon might occur in other organisms.

All experiments were approved by the Animal Experimentation Ethics Committee of School of Life Science and Technology in Huazhong University of Science and Technology and were specifically designed to minimize the number of animals used. Culturing of primary cerebellar granule neurons (CGNs) was as described previously (Tu et al., 2010). Briefly, the cerebella were dissected from 1-week-old KunMing mice of either sex from Hubei Provincial Center for Disease Control and Prevention, and neuronal cells were maintained in a 1:1 mixture of Dulbecco's modified Eagle's medium (DMEM) and F-12 nutrients (Invitrogen), supplemented with 30 mM glucose, 2 mM glutamine, 3 mM sodium bicarbonate, 5 mM HEPES buffer, 30 mM KCl and 10% fetal calf serum to improve neuronal survival. Pharmacological treatments were performed in Ca²⁺-free HEPES-buffered solution (HBS). HEK293 cells were cultured in DMEM medium supplemented with 10% fetal bovine serum (FBS).

The plasmids pRK5-HA-GB1 and pRK5-Flag-GB2 encoding the wild-type GB1 and GB2 subunits were described previously (Rondard et al., 2008). pSRα-SK-T7-Rap1^V12 and pSRα-SK-T7-Rap1^N17, pEBG-Rap1GAPII (wild-type) and its negative control plasmid were gifts from N. Minato (Department of Immunology and Cell Biology, Kyoto University, Japan). The GST fusions of GB1-C (encoding amino acids 854–960 of GB1a C-terminus) and GB2-C (encoding amino acids 744-941 of GB2 C-terminus) were cloned between BamHI and XhoI of pGEX-6p1. The plasmids pRK5-HA-GB1, pSRα-SK-T7-Rap1 and pGEX-6p1-GB1-C-terminus were used for site-directed mutagenesis (stratagene). Antibodies were as follows: anti-Rap1 (Sc-65) and anti-Rap1GAPII (G-17) were from Santa Cruz Biotechnology; anti-Src, anti-HA-tag (C29F4) and anti-β-actin were from Cell Signaling Technology, and anti-GB1 (ab55051) was from Abcam. Peptides utilized in this study were synthesized and purified by the Proteintech Group Inc. (CA, USA). The quality of peptides was assessed by high-performance liquid chromatography analysis, and the expected molecular mass was observed using matrix-assisted laser desorption (MDLC) mass spectrometry. Peptides were dissolved in DMSO and diluted in phosphate-buffered saline (PBS, pH 7.4) to a concentration of 5 mg/ml and stored at −80°C.

Mass spectrometry analysis was carried out as previously described (Lee et al., 2010; Perkins et al., 1999). Briefly, 60 µg GST–GB-C protein was used to pulldown cell lysates from four 100-mm dishes of cultured CGNs. Protein bands at 20–24 kDa were excised from SDS-PAGE gel and digested with trypsin, and peptides were separated and analyzed by MDLC-dependent tandem mass spectrometry (MS/MS). We used the EttanTM MDLC system (GE Healthcare), and desalted tryptic peptide mixtures were loaded onto the columns for separation. A FinniganTM LTQTM linear ion trap MS (Thermo Electron) equipped with an electrospray interface was connected to the liquid chromatography setup for detection of eluted peptides. Data were obtained simultaneously based on MS/MS spectra against the non-redundant International Protein Index (IPI) human protein database (version 3.26, 67,687 entries) using BioworksBrowser rev. 3.1.

The GST pulldown assay for Rap1-GTP detection was performed as previously described (Guetg et al., 2010; Kuramoto et al., 2007; White et al., 2000). Bacterial GST fusion proteins were immobilized on glutathione–Sepharose beads. After incubation with cell lysate, the beads were recovered and washed four times with RBD buffer (150 nm NaCl, 50 nM Tris-HCl pH 7.6, 0.5% Triton X-100, 1 mM phenylmethylsulfonyl fluoride, 1 mM Na₃VO₄, 10 mM NaF, 2 μg/ml leupeptin, 2 μg/ml aprotinin). Protein bound to the beads was eluted with SDS sample buffer and analyzed by western blotting. The RBD pulldown assay for detecting intracellular Rap1-GTP was performed as described previously (Raaijmakers and Bos, 2009). In brief, cell lysate in RBD buffer was incubated with glutathione–Sepharose beads conjugated to GST–RalGDS protein for 60 min at 4°C. After four washes, Rap1-GTP was analyzed by western blotting using a Rap1 antibody.

Immunoblotting analysis was performed as previously described (Baloucoune et al., 2012). Briefly, after determination of protein concentrations using Bradford reagent (Bio-Rad Laboratories), equal amounts of protein were resolved by SDS-PAGE and transferred onto nitrocellulose membrane. The blots were incubated with primary antibodies, then with horseradish peroxidase (HRP)-linked secondary antibodies, detected by enhanced chemiluminescence reagents (Pierce) and visualized by using X-ray film.

Inositol phosphate (IP) accumulation in HEK293 cells co-transfected with GABA_B receptor and Gαqi9 chimeric proteins was measured after stimulation with baclofen for 30 min in 96-well microplates as previously described (Couve et al., 2000). After incubation in the presence of LiCl (10 mM, 30 min) and termination of the reaction with 0.1 M formic acid, the supernatant was recovered and purified by ion exchange chromatography using DOWEX resin. Radioactivity was measured using a Wallac 1450 MicroBeta microplate liquid scintillation counter (Perkin Elmer, Waltham, MA, USA).

Cell surface GB1 expression was detected by ELISA and flow cytometry. HA-tagged GB1 and Flag-tagged GB2 were co-transfected into HEK293 cells for 24 h, and then the cells were seeded into 96-well microplates. Cell surface GB1 expression was detected with a monoclonal rat anti-HA antibody (3F10, Roche) and a goat anti-rat second antibody coupled to HRP (Jackson Immunoresearch, West Grove, PA) as previously described (Rondard et al., 2008). Bound antibody was detected by chemoluminescence using SuperSignal substrate (Pierce) and a 2103 EnVision™ Multilabel Plate Reader (Perkin Elmer, Waltham, MA, USA). Flow cytometry analysis was performed on FACSAria (Beckman). HEK293 cells co-transfected with GB1 and GB2 were incubated with a mouse monoclonal anti-GB1 antibody ab55051 (Abcam) and FITC-conjugated IgG. The surface mean green fluorescence intensity (MFI) was acquired with CXP software (Beckman Coulter, CA, USA).

CGNs grown on poly-L-lysine-coated dishes were incubated for 1 h in culture medium with Pep (100 µg/ml) or PepN (100 µg/ml) in presence of leupeptin (100 µg/ml). Dishes were placed on ice and washed twice with ice-cold buffer. Sulfo-NHS-SS-Biotin (Pierce) was freshly dissolved at a concentration of 0.5 mg/ml in ice-cold buffer A (25 mM HEPES, pH 7.4, 119 mM NaCl, 5 mM KCl, 2 mM CaCl₂, 2 mM MgCl₂, 30 mM glucose), and cells were incubated with biotin for 15 min on ice. After three washes with buffer A, cells were incubated with baclofen (100 µM; 120 min; 37°C) in the presence of Pep (100 µg/ml; 120 min; 37°C) or PepN (100 µg/ml; 120 min; 37°C) followed by incubation on ice. Cell surface biotin was cleaved off by incubation with glutathione solution (75 mM glutathione, 75 mM NaCl, 10 mM EDTA, 1% BSA) twice for 15 min each on ice. Finally, cells were washed twice in ice cold PBS, lysed in 300 µl of ice cold RIPA buffer (50 mM Tris-HCl pH 7.4, 5 mM EGTA, 5 mM EDTA, 50 mM NaF, 1% Nonidet P-40, 0.1% SDS, 2.5 mM sodium pyrophosphate, 1 mM sodium orthovanadate, 1 mM PMSF, 10 µg/ml leupeptin, 10 µg/ml pepstatin) followed by rotating for 30 min at 4°C. Cell lysates were collected and centrifuged at 13,400 g for 15 min at 4°C to remove nuclear and cellular debris. The supernatant containing equal amounts of protein was precipitated with 50 µl of strepavidin–Sepharose (GE Healthcare) overnight at 4°C. Sepharose beads were then washed twice in RIPA buffer containing 500 mM NaCl and once in RIPA buffer containing 150 mM NaCl. Beads were resuspended in 40 µl of SDS sample buffer and boiled for 5 min at 95°C followed by SDS-PAGE and western blotting using a GB1 antibody.

HEK293 cells transiently transfected with HA–GB1 or HA–GB1ΔS953 and Flag–GB2 were incubated with an HA antibody in buffer A (25 mM HEPES, pH 7.4, 119 mM NaCl, 5 mM KCl, 2 mM CaCl₂, 2 mM MgCl₂, 30 mM glucose) containing 10% normal goat serum and 100 µg/ml leupeptin for 40 min at 4°C. After extensive washes with ice-cold buffer A, cells were incubated for 120 min at 37°C in the presence or absence of baclofen as indicated. Control cultures were left at 4°C, to prevent receptor internalization. For the immunofluorescence-based internalization assay, after washing the cells with ice-cold buffer A, we incubated cells with secondary antibodies coupled to Alexa Fluor 488 at 4°C. Then cells were fixed with 4% paraformaldehyde, 4% sucrose for 10 min and permeabilized for 5 min with 0.1% Triton X-100 in PBS. Cells were incubated with secondary antibodies coupled to Cy3 (1:500; Invitrogen) for 1 h at room temperature. After three washes with PBS, cells were analyzed by a Spinning-disk confocal imaging system (CSU-X1 Nipkow Yokogawa, Japan) with Andor IQ 1.8 software (Andor Technology plc, Springvale Business Park, UK). Images were processed and fluorescence was measured by Image J (NIH).

For colocalization, cells were fixed with 4% paraformaldehyde for 10 min and permeabilized for 5 min with 0.1% Triton X-100 in PBS. Then cells were incubated either with EEA1 antibody or LAMP1 antibody for 1 h at room temperature. After washing with PBS, we incubated cells with secondary antibodies coupled either to Alexa Fluor 488 or Cy3 (1:500; Invitrogen) for 1 h at room temperature. After three washes with PBS, cells were analyzed by a spinning-disk confocal imaging system (CSU-X1 Nipkow Yokogawa, Japan) with Andor IQ 1.8 software. Images were processed and fluorescence was quantified using Image J.

Data expressed as mean±s.e.m. were analyzed by GraphPad Software Prism 6.0 and comparisons were made by one-way ANOVA analysis of variance with Bonferroni's multiple comparison tests or Student's t-test. Values of P<0.05 were taken as being indicative of statistical significance.

We thank Philippe Rondard and X.Z. Shawn Xu for critically reading an early version of the manuscript.