Prolonged stimulation of the β2‐adrenergic receptor (β2AR) leads to receptor ubiquitination and downregulation. Using a genome‐wide RNA interference screen, we identified arrestin domain‐containing 3 (ARRDC3) as a gene required for β2AR regulation. The ARRDC3 protein interacts with ubiquitin ligase neural precursor development downregulated protein 4 (NEDD4) through two conserved PPXY motifs and recruits NEDD4 to the activated receptor. The ARRDC3 protein also interacts and co‐localizes with activated β2AR. Knockdown of ARRDC3 expression abolishes the association between NEDD4 and β2AR. Furthermore, functional inactivation of ARRDC3, either through small interfering RNA (siRNA)‐mediated knockdown or overexpression of a mutant that does not interact with NEDD4, blocks receptor ubiquitination and degradation. Our results establish ARRDC3 as an essential adaptor for β2AR ubiquitination.
The β2‐adrenergic receptor (β2AR) is a prototypic member of the G protein‐coupled receptor (GPCR) family (Pierce et al, 2002). The binding of agonists to β2AR stimulates G proteins that alter the intracellular levels of cyclic AMP and the activity of protein kinase A, resulting in important physiological changes such as relaxation of airway smooth muscles and bronchodilation (Penn & Benovic, 2008). The attenuation of β2AR signalling is equally crucial for the maintenance of normal physiology and homeostasis (Premont & Gainetdinov, 2007). This is achieved through receptor desensitization and is dependent on the duration of exposure to the agonist. Immediately after agonist activation, β2AR becomes phosphorylated and recruits cytosolic β‐arrestins (DeWire et al, 2007). The interaction of β‐arrestins with β2AR uncouples the activated G protein complex from the receptor and terminates signalling. By contrast, prolonged stimulation is attenuated by receptor downregulation through endocytosis and protein degradation (Hanyaloglu & von Zastrow, 2008; Marchese et al, 2008).
A crucial event that precedes degradation of β2AR is receptor ubiquitination (Shenoy et al, 2001). Protein ubiquitination is mediated by multiple enzymatic events that culminate with the covalent addition of one or more ubiquitin moieties to a substrate by an E3 ligase (Hershko & Ciechanover, 1998; Rotin & Kumar, 2009). Neural precursor development downregulated protein 4 (NEDD4), which is a member of the HECT (homologous to E6AP carboxyl terminus) family of E3 ubiquitin ligases, can ubiquitinate β2AR (Shenoy et al, 2008). The NEDD4 E3 ligase contains WW domains, which can interact directly with proline‐rich regions, such as the well‐characterized PPXY motif, in target or adaptor proteins (Rotin & Kumar, 2009). However, neither β2AR nor a putative adaptor protein β‐arrestin 2 contains such a motif. Thus, the mechanism by which NEDD4 recruits to activated β2AR remains unclear. In this study, we report a new arrestin domain‐containing protein 3 (ARRDC3), as an essential adaptor that recruits the NEDD4 E3 ligase to mediate β2AR ubiquitination.
ARRDC3 is a novel regulator of β2AR
We performed a genome‐wide RNA interference‐based genetic screen to identify genes required for β2AR downregulation (Fig 1A). The screen was carried out in a cell line that expresses a distally truncated form (after residue 385) of β2AR (β2ARt), which undergoes fast and efficient receptor degradation on agonist stimulation (Cao et al, 1999; Gage et al, 2001). Briefly, β2ARt cells were transduced by using a lentiviral genome‐wide short hairpin RNA (shRNA) library and subjected to agonist (isoproterenol; ISO) stimulation for 16 h. We then performed several rounds of fluorescence‐activated cell sorting to enrich for cells that, after ISO stimulation, maintained high levels of β2AR, indicative of possible defects in β2AR degradation. shRNAs that potentially interfere with β2AR downregulation were identified from the sorted cell population by genomic PCR and sequencing.
Two of the identified shRNAs corresponded to a new gene ARRDC3, also known as thioredoxin‐binding protein‐2‐like inducible membrane protein (TLIMP; Oka et al, 2006). To confirm the effect of ARRDC3 shRNA‐mediated knockdown on β2AR downregulation, we transduced β2ARt cells with lentiviral particles carrying the identified ARRDC3 shRNA sequence. Flow cytometry analysis showed that cells expressing the ARRDC3‐specific shRNA exhibited a higher (>20%) level of cell‐surface receptor after agonist stimulation, compared with the control non‐targeting shRNA‐expressing cells (Fig 1B). To ascertain that the stabilization of β2AR was due to ARRDC3 knockdown and was not caused by a potential off‐target effect of the shRNA, we tested the effect of an ARDDC3 small interfering RNA (siRNA) targeting a sequence different from that of the identified shRNA. The ARRDC3 protein expression was efficiently knocked down (>75%) by the siRNA and, as a result, ISO‐induced degradation of β2ARt was markedly inhibited: the level of undegraded receptor in ARRDC3‐knockdown cells was approximately six‐fold higher than that in the scrambled control siRNA‐transfected cells (Fig 1C). This result demonstrates that ARRDC3 is required for agonist‐stimulated β2AR degradation. The ARRDC3‐knockdown cells also exhibited a slight increase (less than two‐fold) in the basal level (before ISO stimulation) of the receptor, suggesting that ARRDC3 might also have a role in the slow, yet constant, steady‐state receptor degradation in the absence of an agonist (Morrison et al, 1996). To confirm that the effect of ARRDC3 knockdown on β2AR degradation is not limited to the modified β2ARt, we determined the effect of siRNA knockdown on the degradation of the wild‐type β2AR. Consistent with a previous study (Cao et al, 1999), wild‐type β2AR exhibited slow degradation (∼10%) on ISO stimulation (Fig 1D). Nevertheless, such slow degradation of β2AR was blocked in ARRDC3 siRNA‐transfected cells (Fig 1D). Together, our data identify ARRDC3 as a new gene that is required for the efficient degradation of β2AR.
The ARRDC3 protein belongs to a protein family that includes five other ARRDC proteins (supplementary Fig S1A online; Alvarez, 2008; Aubry et al, 2009). Most of the ARRDC proteins have no annotated biological functions. However, two of the ARRDCs (ARRDC1 and ARRDC3), when expressed as green fluorescent protein (GFP) fusion proteins, localize to the plasma membrane (supplementary Fig S1B online), suggesting membrane‐related functions. To determine whether ARRDC1, similarly to ARRDC3, has a role in ISO‐induced β2AR degradation, we examined the effect of ARRDC1 knockdown. Despite efficient knockdown (>90%) of ARRDC1 using siRNA, β2AR degradation was not affected (supplementary Fig S1C online).
ARRDC3 is required for β2AR ubiquitination
As ubiquitination is a prerequisite for efficient agonist‐induced receptor degradation (Shenoy et al, 2001), we next tested whether ARRDC3 has a role in β2AR ubiquitination. We determined the level of β2AR ubiquitination by immunoprecipitation of β2AR followed by anti‐ubiquitin immunoblot analysis. As shown in Fig 1E, ISO stimulation led to a significant increase in the amount of ubiquitinated species of β2ARt in both scrambled control siRNA‐ and ARRDC1 siRNA‐transfected cells, but the increase in ubiquitination was strongly inhibited in ARRDC3‐knockdown cells (∼70% reduction in normalized amount of ubiquitinated β2ARt compared with the controls). Such an effect was specific for β2AR ubiquitination as depletion of ARRDC3 did not grossly alter total ubiquitination levels (data not shown). Similarly, knockdown of ARRDC3 also diminished agonist‐dependent ubiquitination of wild‐type β2AR (∼30% reduction in ubiquitinated β2AR compared with ISO‐treated control; Fig 1F). Consistent with the degradation data (supplementary Fig S1C online), knockdown of ARRDC1 did not inhibit β2AR ubiquitination (Fig 1E,F). These observations demonstrate that ARRDC3 is required for agonist‐dependent ubiquitination of β2AR.
ARRDC3 recruits NEDD4 to ubiquitinate β2AR
The human ARRDC3 protein contains two PPXY motifs at its carboxyl terminus, which are conserved among other orthologues (Fig 2A). As PPXY motifs are known to mediate protein–protein interactions through interaction with the WW domains (Rotin & Kumar, 2009), we hypothesized that ARRDC3 might interact with the WW domain‐containing NEDD4, which was characterized recently as an ubiquitin ligase for β2AR (Shenoy et al, 2008), to facilitate β2AR ubiquitination. Indeed, we showed that ARRDC3 interacts with NEDD4 in a co‐immunoprecipitation experiment (Fig 2B). To determine whether the PPXY motifs in ARRDC3 are required for the observed interaction with NEDD4, we generated PPXY mutants (for either single or double motifs) and showed that, although the first PPXY motif has a main role in the interaction, both PPXY motifs had to be mutated to fully abrogate the interaction of ARRDC3 with NEDD4 (Fig 2B). These results demonstrate that ARRDC3 interacts with NEDD4 through the conserved PPXY motifs.
To explore further the ARRDC3–NEDD4 interaction, we examined whether ARRDC3 and NEDD4 co‐localize with each other. We expressed two fluorescent fusion proteins (ARRDC3–GFP and mCherry–NEDD4) in 293β2AR cells. The ARRDC3–GFP fusion protein, when expressed alone, is localized mostly at the plasma membrane (supplementary Fig S1B online). However, co‐expression with mCherry–NEDD4 led to a cytosolic localization of ARRDC3 (Fig 2C, upper panel). This is probably owing to steady‐state association between ARRDC3 and NEDD4, resulting in the redistribution of ARRDC3 to the cytosol where NEDD4 resides. On addition of agonist, both mCherry–NEDD4 and ARRDC3–GFP co‐localized at the vicinity of the plasma membrane and in discrete bodies in the cytosol (Fig 2C). However, relocalization of NEDD4 was not very marked and most NEDD4 remained in the cytoplasm (Fig 2C). This probably reflects a transient recruitment of NEDD4 to the plasma membrane. Unlike the wild‐type ARRDC3, the double PPXY motif mutant (ARRDC3ΔΔPPXY) showed little co‐localization with mCherry–NEDD4 either before or after agonist stimulation (Fig 2C, lower panels). These data reinforce the idea that ARRDC3 interacts with NEDD4 through the PPXY motifs.
To determine whether the ARRDC3–NEDD4 interaction is required for the degradation of β2AR, we next tested the effect of overexpression of ARRDC3ΔΔPPXY, which does not interact with NEDD4. As shown in Fig 2D, expression of the double PPXY ARRDC3 mutant inhibited ISO‐dependent β2AR degradation, whereas the control GFP or wild‐type ARRDC3–GFP did not. Moreover, expression of ARRDC3ΔΔPPXY blocked β2AR ubiquitination, whereas vector or ARRDC3‐transfected controls displayed robust receptor ubiquitination after ISO stimulation (Fig 2E). These data strongly suggest that the ARRDC3–NEDD4 interaction is required for β2AR ubiquitination and subsequent degradation.
ARRDC3 interacts with activated β2AR
The amino‐terminal region of ARRDC3 is homologous to the arrestin proteins, which interact with activated (phosphorylated) GPCRs (Gurevich & Gurevich, 2006) and have important roles in attenuating GPCR signalling (DeWire et al, 2007). We envisioned that ARRDC3 might also interact with the receptor on agonist stimulation. We therefore investigated whether ARRDC3 associates with β2AR. As shown in the co‐immunoprecipitation experiments (Fig 3A), haemagglutinin (HA)‐tagged ARRDC3 was pulled down with immunoprecipitated β2AR. Although a weak association was observed before ISO addition, the ARRDC3–β2AR association was enhanced greatly after agonist stimulation. This interaction does not require the PPXY motifs in ARRDC3, as the double PPXY mutant was also associated, though less robustly, with activated β2AR (Fig 3A).
We investigated further the ARRDC3–β2AR interaction by examining the co‐localization of the two proteins. Cells expressing FLAG–β2AR were transfected with control GFP or ARRDC3–GFP, and β2AR was visualized by immunostaining with anti‐FLAG. On ISO stimulation, β2AR was internalized from the plasma membrane into cytosolic endosomal vesicles, as indicated by co‐immunostaining with the early endosomal marker early endosomal antigen 1 (EEA1; Fig 3B). This was evident in both GFP and ARRDC3–GFP transfected cells. Prior to ISO stimulation, ARRDC3–GFP localized almost exclusively to the plasma membrane where β2AR also resides. On agonist stimulation, some of the ARRDC3 signal internalized into cytosolic vesicles (supplementary Fig S2A online) that co‐localize with both β2AR and EEA1 (Fig 3B, right panels). This is consistent with the ARRDC3–β2AR interaction observed in the co‐immunoprecipitation experiment, and suggests that ARRDC3 might continue to associate with β2AR after agonist stimulation and during receptor endocytosis. Similarly, co‐localization of ARRDC3 with the fast‐degrading receptor β2ARt was also observed (supplementary Fig S2B online). The co‐localization of ARRDC3 with the receptors was also observed in the PPXY mutant (supplementary Fig S2A,B online), although less clearly as it occurred at the plasma membrane. Together, these data show that ARRDC3 associates with β2AR and that such an association is enhanced on agonist stimulation.
ARRDC3 mediates the NEDD4 and β2AR association
The NEDD4 protein was found previously to associate with β2AR on agonist stimulation and it was suggested that β‐arrestin 2 might mediate the NEDD4–β2AR association (Shenoy et al, 2008). We asked whether ARRDC3 is required for the association between NEDD4 and activated β2AR. We transfected β2ARt‐expressing cells with siRNAs targeting ARRDC3, β‐arrestin 2 or a non‐targeting sequence, and then assessed the NEDD4–β2AR interaction using a co‐immunoprecipitation assay. As shown in Fig 4A, NEDD4 associated with β2AR in scrambled control siRNA‐transfected cells, and such association was augmented moderately on ISO stimulation. The knockdown of ARRDC3 by siRNA completely abrogated the association of NEDD4 with the receptor in the presence or absence of ISO stimulation. By contrast, a 60% reduction in the level of β‐arrestin 2 by siRNA transfection did not affect the interaction between NEDD4 and β2AR (Fig 4A). These data establish ARRDC3 as the adaptor that mediates the association between NEDD4 and β2AR.
In this study, we report the discovery of a new ARRDC protein, ARRDC3, as a critical component of the β2AR downregulation process. Our data demonstrate that ARRDC3 functions as an adaptor for β2AR ubiquitination by recruiting the NEDD4 E3 ligase to activated receptor (Fig 4B). The ARRDC3 protein interacts with NEDD4 through its conserved PPXY motifs and with β2AR, particularly after β‐agonist stimulation. Through the interaction with ARRDC3, NEDD4 is recruited to the proximity of the receptor, allowing the E3 ligase to recognize and ubiquitinate the receptor. Although our study characterizes how ARRDC3 interacts with NEDD4, it remains to be determined whether the arrestin‐like region of ARRDC3 mediates the interaction with activated (phosphorylated) β2AR in a manner similar to arrestins (Mittal & Mcmahon, 2009). Nevertheless, our study represents the first demonstration of the interaction of an ARRDC protein with a prototypical GPCR and its crucial role in receptor downregulation, indicating that structural similarities between ARRDCs and arrestins yield interesting functional parallels.
An arrestin protein, β‐arrestin 2, has been shown previously to interact with β2AR and to be required for receptor downregulation (Shenoy et al, 2001, 2008). Although both ARRDC3 and β‐arrestin 2 are required for efficient β2AR downregulation, our data show that only ARRDC3 is required for the recruitment of NEDD4 to the activated receptor. β‐Arrestin 2 has been shown to have an important role in receptor internalization by recruiting components of the internalization machinery, such as clathrin and assembly protein AP‐2 (Goodman et al, 1996; Santini et al, 2002). Our data do not suggest a direct role for ARRDC3 in receptor internalization. Thus, ARRDC3 and β‐arrestin 2 seem to have distinct roles in β2AR downregulation. Whether and how the two proteins coordinate during β2AR downregulation remains to be determined.
The role of ARRDC3 as an ubiquitination adaptor is remarkably similar to that of arrestin‐related trafficking (ART) adaptor proteins in yeast. ART proteins have been shown to recruit the yeast HECT domain E3 ligase Rsp5 to ubiquitinate target plasma membrane proteins (Lin et al, 2008; Nikko et al, 2008; Nikko & Pelham, 2009). Although the sequence similarity is low between ARRDC3 and the yeast ART proteins (supplementary Fig S3 online; Alvarez, 2008), they share a similar functional domain organization: an arrestin‐like domain at the N‐terminus and PPXY motifs at the C‐terminus. The conservation of structural domains and the analogous functions of the ARTs and ARRDC3 indicate that the ubiquitin ligase adaptor functions have been conserved throughout evolution. Similar to yeast ARTs, there are several ARRDC proteins in humans (supplementary Fig S3 online). The fact that the other plasma membrane‐localized ARRDC, ARRDC1, has no obvious role in β2AR ubiquitination indicates little functional redundancy among the ARRDC family members. Unlike yeast, which has only one HECT domain ligase (Rsp5), mammalian cells have at least nine such ubiquitin ligases (Rotin & Kumar, 2009), all of which (including NEDD4) contain WW domains that might interact with PPXY motifs. Indeed, ARRDC3 interacts with at least three other HECT domain E3 ligases (supplementary Fig S4 online). It is thus possible that ARRDC3, through its interactions with other members of the NEDD4 E3 ligase family, might function as an ubiquitination adaptor for multiple receptor protein substrates.
Our study identifies ARRDC3 as an essential mediator of β2AR downregulation. Meticulous control of receptor levels through downregulation ensures appropriate levels of signalling in tissues such as the lung airways, whereas excessive downregulation can lead to the reduction or loss of functional β2AR, thus severely limiting therapies (such as β‐agonists for asthma treatment) that rely on efficient β2AR activation and signalling (Shore & Moore, 2003). Future studies on the role of ARRDC3 in physiological sites of β2AR activity might provide new insights into in vivo β2AR regulation and lead to new approaches that improve β2AR‐based therapies.
β2AR ubiquitination and degradation assays. Cells expressing β2AR or β2ARt were subjected to 10 μM ISO (Sigma) stimulation, lysed in 0.5% Nonipet‐40 lysis buffer and immunoblotted with anti‐FLAG–horseradish peroxidase (Sigma). For β2AR ubiquitination, β2ARt‐ or wild‐type β2AR‐expressing cells were first incubated with 10 μM alprenolol antagonist (Sigma) for 15 min (to reduce constitutive background ubiquitination), stimulated with 10 μM ISO and lysed in buffer supplemented with 10 mM N‐ethylmaleimide (Sigma) to inhibit deubiquitinase activity. Anti‐FLAG EZview beads (Sigma) used for the immunoprecipitation were subjected to three 10 min washes with lysis buffer and resuspended in 50 μl 2 × lithium dodecyl sulphate sample buffer (Invitrogen) containing 10% β‐mercaptoethanol (Sigma) and analysed by immunoblotting.
Co‐immunoprecipitations. 293T cells were transfected using Fugene 6 (Roche) or Turbofect (Fermentas). After 48 h, cells were lysed, pre‐cleared with protein A agarose, and HA‐ or FLAG‐ tagged proteins were immunoprecipitated using anti‐HA EZview or anti‐FLAG EZview beads. Washed beads were resuspended in sample buffer and analysed by immunoblotting. Where noted, crosslinking was achieved by incubation for 40 min with 2 mM crosslinker dithio‐bismaleimidoethane (Pierce) in phosphate‐buffered saline containing 10 mM HEPES (pH 7.5), as previously described (Shenoy et al, 2008) with minor modifications. The immunoprecipitation complexes were washed three times consecutively in lysis, high‐salt and low‐salt buffers before being resuspended in sample buffer for immunoblotting.
Immunofluorescence assays and confocal microscopy. The β2AR‐expressing cells seeded on glass cover slips (VWR) were transfected and then incubated with 10 μg per ml anti‐FLAG M2 polyclonal or monoclonal antibodies (Sigma) before stimulation with ISO. Cells were washed and fixed with 3.8% paraformaldehyde and permeabilized with 0.1% Triton X‐100 buffer in Tris‐buffered saline supplemented with 3% bovine serum albumin. Primary and secondary antibody incubations were carried out as indicated, before visualization. Image acquisition was carried out using a Leica TCS‐NT laser scanning confocal microscope (Leica) fitted with air‐cooled argon and krypton lasers.
Details of other experimental methods can be found in the supplementary information online.
Supplementary information is available at EMBO reports online (http://www.emboreports.org).
Conflict of Interest
The authors declare that they have no conflict of interest.
We thank L. Kobzik, G. Hotamisligil and S. Cohen for critical reading of the paper. This study was supported in part by the Milton Fund and by a startup fund from the Harvard School of Public Health.
- Copyright © 2010 European Molecular Biology Organization