Gα_s protein binds ubiquitin to regulate epidermal growth factor receptor endosomal sorting

Significance

Gα_s protein is classically known as a signaling component of cell surface G protein-coupled receptor transduction pathways. However, Gα_s has also been localized on endosomes, where it plays a role in receptor sorting to lysosomes. How Gα_s regulates this trafficking step is unclear. Although Gα_s structure has been known for years, we found a motif in Gα_s that allows its interaction with ubiquitin, a key signal for cargo sorting to the lysosomal pathway. We show that disrupting the ubiquitin-interacting motif (UIM) in Gα_s impairs its interaction with components of the endosomal sorting machinery and the lysosomal degradation of epidermal growth factor receptor. This study provides molecular insights on the mechanism by which Gα_s is involved in this noncanonical function.

Abstract

The Gα_s subunit is classically involved in the signal transduction of G protein-coupled receptors (GPCRs) at the plasma membrane. Recent evidence has revealed noncanonical roles for Gα_s in endosomal sorting of receptors to lysosomes. However, the mechanism of action of Gα_s in this sorting step is still poorly characterized. Here, we report that Gα_s interacts with ubiquitin to regulate the endosomal sorting of receptors for lysosomal degradation. We reveal that the N-terminal extremity of Gα_s contains a ubiquitin-interacting motif (UIM), a sorting element usually found in the endosomal sorting complex required for transport (ESCRT) machinery responsible for sorting ubiquitinated receptors into intraluminal vesicles (ILVs) of multivesicular bodies (MVBs). Mutation of the UIM in Gα_s confirmed the importance of ubiquitin interaction for the sorting of epidermal growth factor receptor (EGFR) into ILVs for lysosomal degradation. These findings demonstrate a role for Gα_s as an integral component of the ubiquitin-dependent endosomal sorting machinery and highlight the dual role of Gα_s in receptor trafficking and signaling for the fine-tuning of the cellular response.

Heterotrimeric G proteins are classically known for their role in transducing signals from the extracellular environment across the plasma membrane (1). However, it is becoming increasingly apparent that these signaling molecules also have noncanonical roles in other cellular locations (2). Recent findings have shown the presence of the Gα_s subunit on endosomes, where it plays a role in receptor signaling and trafficking (3–5). In fact, some G protein-coupled receptors (GPCRs), in addition to activating Gα_s at the plasma membrane, appear to elicit a “second wave” of Gα_s activation after ligand-induced internalization in endosomes (6). Along with its endosomal signaling function, Gα_s has also been reported to regulate the endosomal trafficking and down-regulation of single and seven transmembrane receptors. Indeed, cellular depletion of Gα_s specifically delays the lysosomal degradation of epidermal growth factor receptor (EGFR) and GPCRs (CXCR4 and DOPr) by disrupting their transfer into the intraluminal vesicles (ILVs) of multivesicular bodies (MVBs) (4, 5). However, the precise and likely intricate mechanisms by which Gα_s regulates the endosomal sorting of receptors remain unresolved.

The endosomal sorting of receptors to lysosomes is coordinated by the endosomal sorting complex required for transport (ESCRT) machinery, which consists of four distinct complexes, ESCRT-0, ESCRT-I, ESCRT-II, and ESCRT-III, together with several accessory components (7). The ESCRT complexes act in a sequential and coordinated manner to bind and sort receptors from the early endosome limiting membranes to ILVs of MVBs (7, 8). The MVBs eventually fuse with lysosomes to degrade the internal vesicles and their protein cargo (9). For most receptors (including receptor tyrosine kinases and GPCRs), the covalent attachment of ubiquitin serves as a key sorting signal for entry into the MVB/lysosomal pathway (10, 11). In fact, components of the three first ESCRT complexes (ESCRT-0, ESCRT-I, and ESCRT-II) harbor ubiquitin-binding domains (UBDs) that interact with ubiquitin moieties and specifically catalyze the sorting of ubiquitinated receptors to lysosomes (12, 13). Other contenders for ubiquitin-sorting receptors have also been identified, including eps15b, Tom1, Bro1, and GGA3, which bind ubiquitin and ESCRT components and act conjointly with the ESCRT machinery to sort receptors into MVBs (10, 14).

The fact that Gα_s is involved in endosomal sorting of ubiquitinated receptors into MVBs and has been shown to interact with the ESCRT-0 component (4, 5) led us to postulate that Gα_s also participates in receptor endosomal sorting through an interaction with ubiquitin. Here, we report seminal data demonstrating that Gα_s binds directly to ubiquitin through an N-terminal ubiquitin-interacting motif (UIM), an event that is crucial for Gα_s interaction with the ESCRT-0 component and EGFR sorting into ILVs of MVBs for lysosomal degradation. These findings demonstrate that Gα_s is an integral component of the ubiquitin-dependent endosomal sorting machinery and provide concrete molecular insights on the molecular mechanism by which Gα_s could be involved in this noncanonical endosomal sorting step.

Results and Discussion

Gα_s Binds Ubiquitin.

Because ubiquitin is key in the endosomal sorting pathway, we reasoned that Gα_s might interact with ubiquitin. To investigate whether human Gα_s has the ability to bind ubiquitin, pull-down assays with ubiquitin-agarose beads were performed on extracts of HEK293 cells expressing endogenous Gα proteins. Gα_s, but not Gα_i3, Gα_q, Gα_o, Gα₁₂, or Gβ₁, was pulled down by ubiquitin beads (Fig. 1A), demonstrating that ubiquitin interacts specifically with Gα_s. Consistent with these results, Gα_s and ubiquitinated proteins colocalized extensively on Rab5-labeled endosomes (Fig. S1). We next took advantage of the enlarged endosomes produced by expressing Rab5QL, a GTPase-defective mutant of Rab5 (15), to better visualize the spatial relationship of Gα_s and ubiquitin on endosomal membranes (Fig. 1B). A strong colocalization between Gα_s and conjugated ubiquitin proteins was observed in subdomains of enlarged endosomes (Fig. 1 B and C). Thus, Gα_s not only interacts with ubiquitin but also colocalizes with ubiquitinated proteins in microdomains of early endosomes.

Fig. 1.

Gα_s interacts directly with ubiquitin and colocalizes with ubiquitinated proteins on endosomal microdomains. (A) Gα_s is specifically pulled down by ubiquitin beads. HEK293 cell extracts were incubated with ubiquitin-agarose beads (Ub) or control protein G-agarose beads (Ctl). (B) Gα_s colocalizes with ubiquitinated proteins on endosomal microdomains. HeLa cells transfected with Gα_s and Rab5-QL (to create enlarged endosomes) were subjected to confocal immunofluorescence microscopy with antibodies specific for ubiquitin and Gα_s (also Fig. S1). Insets represent enlarged views of the boxed regions (magnification: 4×). (C) Quantification of the fraction of Gα_s colocalizing with ubiquitin (Gα_s/Ubi) and the fraction of ubiquitin colocalizing with Gα_s (Ubi/Gα_s) on enlarged endosomal membranes (as shown in B). Data are represented as the mean ± SEM of n = 53 endosomes. (D) Interaction of Gα_s with ubiquitin is altered by Gα_s activation status. HEK293 cell extracts pretreated with GDP (to mimic the inactive state) or either GTP_γS or GDP/AIF₄⁻ (to mimic the active state) were incubated with Ub or Ctl beads (also Fig. S2). (E) Comparison of the direct interaction between ubiquitin and the active and inactive forms of Gα_s. In vitro-translated ³⁵S-labeled Gα_s short (S) and long (L) forms were pretreated with GDP or GTP_γS and incubated with Ub or Ctl beads. HRS, a known ubiquitin-binding protein, served as a positive control.

To determine whether the GTPase activity of Gα_s affects its binding with ubiquitin, pull-down assays were performed with HEK293 cell lysates pretreated with GDP (to mimic the inactive state of Gα_s) or either GDP/AlF₄⁻ or GTPγS (to mimic the active state of Gα_s). The interaction between endogenous Gα_s and ubiquitin was increased with the latter treatment compared with GDP treatment or no treatment (Fig. 1D). Similar results were obtained using cell lysates expressing constitutively active and inactive Gα_s mutants (Fig. S2A), suggesting that ubiquitin binds preferentially to the active form of Gα_s. These results were confirmed using ³⁵S-labeled in vitro-translated long or short form of Gα_s (Fig. 1E). This in vitro pull-down assay demonstrated that both ³⁵S-labeled Gα_s isoforms interact directly with ubiquitin, which was confirmed using purified Gα_s proteins (Fig. S2B), and that this interaction is increased with the active (GTPγS-pretreated) form of Gα_s (Fig. 1E)_. Together, these findings suggest that ubiquitin binds directly and preferentially to the active form of Gα_s.

The N-Terminal Extremity of Gα_s Contains a UIM.

To delineate the ubiquitin-binding region in Gα_s, ubiquitin-agarose pull-down assays were performed with a series of truncation mutants of Gα_s. Removal of the first 21 residues (Met1 to Glu21) abolished Gα_s binding to ubiquitin (Fig. 2A). In concordance, a fusion protein of these 21 residues (Gα_s^1-21) to GFP enabled it to bind ubiquitin (Fig. 2B), indicating that this region (residues 1–21) of the Gα_s N terminus is necessary and sufficient for ubiquitin binding.

Fig. 2.

Gα_s contains a UIM. (A) Mapping the Gα_s domain responsible for binding to ubiquitin. HEK293 cell extracts expressing various Gα_s N-terminal–truncated mutants were incubated with ubiquitin-agarose beads (Ub) or control protein G-agarose beads (Ctl). (B) Gα_s N-terminal residues 1–21 are sufficient for ubiquitin binding. Pull-downs were performed on HEK293 cells expressing GFP alone or fused to the 1–21 region of Gα_s. (C) NMR chemical shift mapping of the residues of a Gα_s^1-34 peptide interacting with ubiquitin. The ¹H-¹⁵N-HSQC spectrum of Gα_s^1-34 in the absence (black) and presence (red) of ubiquitin at a 1:10 ratio is shown. Fully assigned spectra are shown in Fig. S3A. (D) Normalized weighted CSDs observed in Gα_s^1-34 upon ubiquitin binding. Significant displacements are assigned to cross-peaks that have CSD values of 1 SD above the mean (dashed line). (E) Alignment of the 1–14 region of Gα_s with the sequences of the UIMs found in various ubiquitin-binding proteins (also Fig. S4). e, acidic residue; x, any amino acid; Φ, large hydrophobic residue. Nearly invariant Ala and Ser positions are colored red, and highly similar amino acids are colored blue and green. Hs, Homo sapiens. (F) Structure of Gα_s^1-34 peptide modeled by Swiss-PdbViewer and SWISS-MODEL, showing in red the residues with the greatest changes in chemical shift upon addition of ubiquitin.

To define this interaction at a residue-specific level, NMR spectroscopy was employed. Given that the addition of flanking residues to ubiquitin-binding regions had been shown to stabilize the conformation of other ubiquitin-binding peptides and increase ubiquitin-binding efficiency (16), a peptide containing residues 1–34 of Gα_s (Gα_s^1-34) was double-labeled with ¹³C and ¹⁵N and used for NMR analysis. The ubiquitin-binding residues of Gα_s^1-34 were defined by detection of the ¹H and ¹⁵N amide chemical shift displacements (CSDs) in 2D heteronuclear single quantum coherence spectroscopy (HSQC) spectra upon the addition of unlabeled ubiquitin (Fig. 2C and Fig. S3A). Consistent with the results obtained on the Gα_s truncation mutants, residues displaying the most disturbances upon binding are localized exclusively at the N-terminal region of the Gα_s peptide, including, more specifically, the Cys3-Ser7 region (Fig. 2D). In particular, Cys3 and Ser7 display the highest CSDs, suggesting their importance in the interaction with ubiquitin.

Interestingly, sequence alignment shows that the Gα_s N-terminal region (Cys3 to Glu10) has homology to the UIMs found in other ubiquitin-binding proteins, such as the ESCRT-0 component HRS and the proteasome subunit 5A (S5a) (17) (Fig. 2E). The UIM in Gα_s is conserved among different species (Fig. S4A) but absent in other Gα proteins (Fig. S4B). Classically, UIMs comprise a 15-residue consensus sequence e-e-x-x-Φ-x-x-Ala-Φ-x-Φ-Ser-x-x-e (where e, x, and Φ represent acidic residue, any amino acid, and large hydrophobic residue, respectively) (17, 18) in which two conserved residues, Ala and Ser, are crucial for ubiquitin binding (16, 18, 19). In the Gα_s UIM, Ser is conserved, whereas Ala is replaced by Cys, which has similar hydrophobic features. In concordance, NMR results indicated that Cys3 and Ser7 are indeed the main ubiquitin-interacting residues in the Gα_s UIM. Another difference in the Gα_s UIM is the absence of the upstream Φ and negatively charged residues, which have been proposed to mediate favorable interactions (16, 20, 21), and may thus result in lower affinity for ubiquitin. In summary, the N-terminal extremity of Gα_s harbors a UIM that is shorter than the classical UIMs but contains the central conserved hydrophobic residues (Cys3 and Ser7) critical for the ubiquitin-binding capacity of these motifs.

Most UIMs consist of a single α-helix structure (22). Even though the 3D structure of Gα_s has been elucidated, no discernable structure was observed in the first eight residues of the N-terminal region, indicating that this region is highly flexible and does not adopt a fixed structure in the crystal (23, 24). Therefore, a structure of the Gα_s^1-34 peptide was modeled using Swiss-PdbViewer, in conjunction with SWISS-MODEL (25). The predicted structure exhibits a high α-helical content and similarities to the helical secondary structure of the UIM of S5a protein (Fig. 2F). Interestingly, Cys3 and Ser7 of the Gα_s UIM are located on the same side of the α-helix surface (Fig. 2F), as previously reported for Ala219 and Ser223 in the UIM of S5a (26) or Ala266 and Ser270 in the UIM of HRS (27), enabling their interaction with the conserved Leu8-Ile44-Val70 hydrophobic patch of ubiquitin (18, 19, 27, 28).

The UIM of Gα_s Interacts with the Ile44 Hydrophobic Patch of Ubiquitin.

NMR spectroscopy was subsequently employed to define the corresponding Gα_s-binding surfaces in ubiquitin. The ¹⁵N-labeled ubiquitin was titrated with unlabeled Gα_s^1-34 peptide, and the resulting ¹H-¹⁵N-HSQC spectra were analyzed to detect localized CSDs (Fig. 3A and Fig. S3B) and identify specific residues on ubiquitin involved in the interaction (Fig. 3B). The residues with the most shifted cross-peaks are clustered around the C-terminal strands of the ubiquitin β-sheet, corresponding to the Ile44 hydrophobic surface of ubiquitin (Fig. 3C). In addition to basic residues Lys6, Arg42, and His68, which may interact with the acidic residues of Gα_s UIM, the Gα_s surface binding on ubiquitin comprises mainly exposed hydrophobic residues (Leu8, Ile44, Phe45, Leu67, Val70, Leu71, and Leu73) and suggests a complementary hydrophobic surface interaction with Gα_s (Fig. 3D). The polar residue Gln49 also displays considerable displacement upon Gα_s binding. However, the CSDs of remote residues like Gly10 and Ile13 most likely originate from differential interactions with residues disturbed by the direct interaction. Importantly, the identified Gα_s-binding surfaces in ubiquitin correspond very closely to ubiquitin-binding sites that were mapped for other UIMs (e.g., HRS, hS5a) (27, 28) (Fig. S5). This similarity lends further credence to the idea that Gα_s contains a UIM that binds ubiquitin in a very similar way.

Fig. 3.

Ubiquitin-binding surface of Gα_s. (A) NMR chemical shift mapping of the residues in ubiquitin interacting with the Gα_s^1-34 peptide. The ¹H-¹⁵N-HSQC spectra of ubiquitin in the absence (black) and increasing ratios (1:1, 1:2.5, 1:5, 1:10, 1:25, 1:50, and 1:100) of Gα_s peptide (blue to purple) are shown. Fully assigned spectra are shown in Fig. S3B. (B) Normalized weighted CSD of ubiquitin residues observed upon binding to Gα_s^1-34 peptide. (C) Surface/worm representation of ubiquitin showing in red the residues with the greatest CSD upon addition of Gα_s^1-34 peptide. Worm radii are proportional to CSD values, and regions of higher perturbations (red) have a thicker backbone (also Fig. S5). (D) Interaction model mapping the interaction residues on ubiquitin structure and Gα_s^1-34 peptide. Worm radii are proportional to CSD values and are coded in a white-to-red (Gα_s peptide) or white-to-green (ubiquitin) gradient.

The NMR analysis also indicated that the ubiquitin–Gα_s peptide interactions are in the fast exchange regime, with an estimated dissociation constant (K_d) ranging from 0.49 to 1.32 mM (average of 1.10 ± 0.25 mM) (Fig. S6 and Table S1). Such binding affinity is in good agreement with the K_d values previously reported for other UIM–ubiquitin complexes (varying from 0.1 to 2 mM) (10, 22).

Ubiquitin Binding Is Important for the Role of Gα_s in Endosomal Sorting.

To determine the functional relevance of ubiquitin binding by Gα_s, a UIM-deficient Gα_s was generated. Based on the NMR data (Fig. 2 C and D), we selected the Leu4-to-Ser7 sequence for mutagenesis and ubiquitin-binding studies. Even if Cys3 was identified as a potential key determinant for ubiquitin binding (Fig. 2D), it was not substituted, given that it is a palmitoylation site responsible for Gα_s membrane recruitment (29, 30). In concordance with our NMR results and previous studies mutating the UIM consensus sequence (16, 31), which predict Ser7 as a key ubiquitin-binding residue, mutating Ser7 to Glu (^S7EGα_s) dramatically reduced the ability of Gα_s to bind ubiquitin (Fig. 4A). Although Leu4 and Asn6 have only a modest CSD on the HSQC spectrum (Fig. 2D), their mutations (^L4EGα_s and ^N6AGα_s) clearly alter Gα_s binding to ubiquitin (Fig. 4A). A possible explanation is that these changes affect the adjacent residues Cys3 and Ser7, and prevent their interaction with ubiquitin. Moreover, previous mutational studies suggested that the bulky hydrophobic residue usually located on the C-terminal side of Ala in the UIM motif is an important affinity determinant (20, 21). In contrast, mutation of Gly5 (^G5EGα_s), whose CSD was similar to Asn6, did not alter ubiquitin binding (Fig. S7). Thus, Gly5 CSD must be caused by an indirect effect, such as stabilization of the α-helical motif upon ubiquitin binding. Indeed, based on the Gα_s orientation in complex with ubiquitin (Figs. 2F and 3D), Gly5 is oriented on the opposite side of the binding interface and may not be involved in the interaction. In summary, the single mutations of Leu4, Asn6, and Ser7 reduced Gα_s binding to ubiquitin, but residual binding still remained (Fig. 4A), possibly due to incomplete inactivation of the UIM. Therefore, a mutant of the three residues Leu4, Asn6, and Ser7 was generated (^LNS/EAEGα_s), which showed a stronger defect in ubiquitin association (Fig. 4A) and was thus defined as UIM-deficient. Importantly, given that previous reports showed the importance of the Gα_s N-terminal region in regulating Gα_s activity (32), we used a bioluminescence resonance energy transfer-based assay to confirm that ^LNS/EAEGα_s did not alter the rate of cAMP production (Fig. S8). Therefore, we ruled out that the effect of this Gα_s mutant on ubiquitin binding is due to an alteration in Gα_s activity.

Fig. 4.

Loss of ubiquitin binding inhibits Gα_s interaction with HRS and delays EGFR sorting into ILVs of MVBs for lysosomal degradation. (A) Comparison of the contributions of the different UIM residues to ubiquitin binding using pull-down assays (also Fig. S7). Ctl, control protein G-agarose beads; Ub, ubiquitin-agarose beads. (B) Gα_s-UIM mutation delays EGFR degradation. Levels of EGFR in Gα_s-depleted HeLa cells transfected with siRNA-resistant forms of ^wtGα_s or UIM-deficient Gα_s (^LNS/EAEGα_s) and incubated with or without 100 nM EGF are shown. (C) Quantification of EGFR degradation (as shown in B). Data are presented as the percentage of EGFR expression compared with unstimulated cells (T0) and as the mean ± SEM from three independent experiments. *P < 0.05; **P < 0.01 (one-way ANOVA). (D, Top) Mutating the Gα_s UIM alters the transfer of EGFR to the ILVs of MVBs. Gα_s-depleted HeLa cells were transfected as described in B, together with Rab5QL-GFP. Cell-surface EGFRs labeled by Alexa Fluor 647-EGF were internalized for 30 min and then processed for confocal microscopy. (D, Bottom) Expression of Gα_s proteins were checked by Western blot. (E) Representative line scan analysis to quantify EGFR localization to the ILVs of endosomes. The normalized diameter represents the diameter of the endosome, where 0 and 100 correspond to the pixel distances with the first and second maximum pixel intensities of Rab5QL-GFP, corresponding to the limiting membranes of the endosomes. The hatched box shows the normalized fluorescence values of pixels from 40 to 60% of the normalized diameter that were used to determine the mean intraluminal fluorescence for each endosome. (F) Compiled results of the line scan analysis for ^wtGα_s or ^NLS/EAEGα_s. Data are represented as the mean ± SEM of n ≥ 180 endosomes. ***P < 0.001 (Student’s t test). (G) Gα_s interacts preferentially with ubiquitinated HRS. HEK293 cell extracts expressing GFP-Gα_s and HRS were immunoprecipitated with either anti-GFP or HRS antibodies (first IP). Bound material was eluted and immunoprecipitated with either antiubiquitin or anti-HRS antibodies (second IP). (H) Quantification of the level of ubiquitinated HRS in the Gα_s-GFP IP eluate and the HRS IP eluate (as shown in G). Data are presented as the percentage of ubiquitinated HRS versus total HRS and as the mean ± SEM from five independent experiments. **P < 0.01 (Student’s t test). (I) Gα_s interaction with HRS is inhibited by Gα_s UIM mutation. HEK293 cell extracts expressing ^wtGα_s or UIM-deficient Gα_s (^LNS/EAEGα_s) were incubated with HRS or TSG101 antibodies and analyzed by Western blotting using specific antibodies.

The functional role of the UIM-deficient Gα_s in endosomal sorting was next evaluated on the degradation of EGFR, the prototypical receptor used to study ubiquitin-dependent endosomal sorting. Gα_s-depleted HeLa cells were transfected with siRNA-resistant forms of wild-type Gα_s (^wtGα_s) or ubiquitin-binding mutant (^LNS/EAEGα_s) (Fig. 4 B and C). As previously reported (5), Gα_s depletion delays EGFR degradation. Reintroducing ^wtGα_s in these cells restored the level of EGFR degradation to levels not significantly different from those of control cells. In contrast, reintroducing ^LNS/EAEGα_s did not reverse the effect of Gα_s depletion on the levels of EGFR degradation (Fig. 4C). These results indicate that EGFR was not efficiently delivered to the lysosomes in cells expressing the UIM-deficient ^LNS/EAEGα_s.

Based on the colocalization of Gα_s with ubiquitinated proteins on early endosome membranes (Fig. 1B) and our previous findings that Gα_s plays a crucial role in the sorting of receptors in the ILVs of MVBs (4), we examined whether UIM-deficient Gα_s altered the sorting of EGFR into ILVs of Rab5QL-enlarged endosomes. Cell-surface EGFRs were labeled with Alexa Fluor 647-EGF and internalized for 30 min (33). In Gα_s-depleted HeLa cells rescued with ^wtGα_s, Alexa Fluor 647-EGF was mainly located in ILVs of enlarged endosomes (Fig. 4D). In contrast, in ^LNS/EAEGα_s-rescued cells, Alexa Fluor 647-EGF localized predominantly to the limiting membranes of enlarged endosomes, with most endosomes exhibiting little intraluminal fluorescence (Fig. 4D). Quantification of EGFR distribution across the endosomes by line scan analysis of confocal cross-sections (4, 33) (Fig. 4E) revealed a 40% reduction of Alexa Fluor 647-EGF fluorescence in the endosomal lumen of ^LNS/EAEGα_s-rescued cells compared with ^wtGα_s-rescued cells (Fig. 4F). These results clearly demonstrate that mutations in the UIM of Gα_s cause defects in the sorting of EGFR in ILVs of MVBs.

Finally, to gain some mechanistic insight into the role of Gα_s-UIM in EGFR endosomal sorting, we next examined whether the mutation of UIM altered Gα_s interaction with the ESCRT machinery, a central player in EGFR sorting into ILVs of MVBs. Because Gα_s has been previously shown to interact with HRS, an ESCRT-0 component (4, 5), and because many ESCRT components, including HRS and tumor susceptibility gene 101 (TSG101), can themselves be ubiquitinated (10, 34, 35), we hypothesized that Gα_s interacts with the ubiquitinated form of HRS via its UIM. To support this idea, the level of ubiquitinated HRS interacting with Gα_s was assessed by sequential double-immunoprecipitation (IP). This procedure consists in an initial IP of either Gα_s or HRS, followed by elution of bound proteins and a second round of IP with antibodies against either ubiquitin to determine the amount of ubiquitinated HRS in the eluate or against HRS to determine the total amount of HRS in the eluate (Fig. 4G). Higher levels of ubiquitinated HRS were found in Gα_s co-IP compared with HRS IP (Fig. 4G). Indeed, quantitative analysis (Fig. 4H) indicated that the ratio of ubiquitinated HRS/total HRS in the Gα_s IP eluate (94.6 ± 11.8%) is 6.8-fold greater than the one found in the HRS IP eluate (14 ± 3.2%), demonstrating that Gα_s binds preferentially to ubiquitinated HRS. We next investigated whether the mutation of Gα_s UIM altered the interaction with HRS. ^wtGα_s, but not the UIM-deficient ^LNS/EAEGα_s, coimmunoprecipitated with endogenous HRS (Fig. 4I). However, neither ^wtGα_s nor ^LNS/EAEGα_s coimmunoprecipitated with the ESCRT-I component TSG101. Together, these results indicate that Gα_s interacts preferentially with ubiquitinated HRS and that the UIM in Gα_s is key for this interaction. This is consistent with the idea that Gα_s UIM is involved in intermolecular interaction with specific ubiquitinated ESCRT components, further supporting that Gα_s is an integral component of the ubiquitin-dependent endosomal sorting machinery.

In summary, these findings demonstrate that mutations in the UIM of Gα_s prevent the interaction of Gα_s with HRS, which delays the sorting of EGFR in ILVs of MVBs and subsequent lysosomal degradation. Interestingly, these results on EGFR degradation are phenotypically similar to those observed when Gα_s was depleted (5), suggesting that the interaction of Gα_s with ubiquitin, via its UIM, is crucial for its interaction and action on EGFR endosomal sorting machinery.

Conclusions

In this report, a conserved and bona fide UIM has been identified at the N-terminal extremity of Gα_s. Like other known UIMs, the Gα_s UIM is a single-sided motif that interacts with the classical hydrophobic surface patch of ubiquitin with low affinity. Among several Gα proteins tested, only Gα_s binds ubiquitin and contains a UIM (Fig. S4B), which is consistent with the finding that Gα_s can regulate endosomal sorting. Whereas other UBDs have been reported in close proximity to the lipid-binding domain or to have the ability to bind both ubiquitin and phospholipids (36, 37), the Gα_s UIM is a UBD containing a palmitoylation site (Cys3) in its sequence. This posttranslational modification influences the capacity of Gα_s to bind to ubiquitin. Indeed, an attenuation of the degree of Gα_s palmitoylation in cells treated with the palmitoylation inhibitor 2-bromopalmitate increases Gα_s binding to ubiquitin (Fig. S9). Interestingly, activation of Gα_s has been reported to cause its rapid depalmitoylation (38), which could explain our observation that the active form of Gα_s binds preferentially to ubiquitin. However, in the in vitro binding assays using nonpalmitoylated Gα_s, ubiquitin binds preferentially to the active form of Gα_s (Fig. 1E), indicating that the conformational change induced by activation has additional effects, perhaps through a better exposure of the N-terminal region containing the UIM. Of note, the insertion of a tag (GST or GFP) at the N terminus of Gα_s alters both its ability to properly interact with ubiquitin and the capacity of its active form to influence ubiquitin binding (Fig. S10). We believe this explains why the interaction between HRS and Gα_s-GFP or GST-Gα_s reported in our previous work (4) was weak and independent from the Gα_s activation state.

Our demonstration of a functional UIM in Gα_s provides concrete molecular insights for its unconventional role in receptor endosomal sorting. Ubiquitin has emerged as a key signal for cargo sorting to the lysosomal pathway, and UIMs are recurring motifs found in ESCRT components, which are indispensable for interaction with ubiquitinated cargo to coordinate their transfer to ILVs. Furthermore, many components of the ESCRT sorting machinery, such as HRS, are also subject to monoubiquitination (34, 39). Monoubiquination mediates intermolecular interactions with UIMs present in other ESCRT components or regulatory proteins for proper assembly and coordination of the sorting machinery (10). It can also induce autoinhibitory conformation, such that the ubiquitin moiety interacts with its own UIM, thus preventing ESCRT components from recognizing ubiquitinated cargo and ESCRT from performing its sorting function (10, 33). In this study, we demonstrated that Gα_s UIM is crucial for Gα_s interaction with HRS and for EGFR sorting into ILVs of MVBs for lysosomal degradation. The possible mechanism by which Gα_s could regulate HRS sorting activity is through its interaction with ubiquitinated HRS, which would prevent its autoinhibitory conformation, or by acting as an adaptor for the recruitment of other regulatory proteins, such as ubiquitin ligases or deubiquitinases. These possibilities remain to be tested to further understand the role of Gα_s as a component of the ubiquitin-dependent endosomal sorting machinery.

Our findings also highlight the dual role of Gα_s in receptor signaling and trafficking, providing an additional piece to our understanding of the intertwined molecular network between these processes. It is becoming increasingly apparent that Gα_s has an important function in endosomal signaling, either by eliciting a second wave of G protein activation after GPCR internalization or, as we show herein, by regulating receptor sorting to lysosomes, and thus limiting the time spent on endosomes. Given that the active state of Gα_s binds preferentially to ubiquitin, an attractive possibility is that the stimulation of Gα_s-coupled GPCRs will increase the endosomal sorting of receptors to lysosomes leading to their down-regulation, thereby modulating the cellular response. Finally, with increasing reports revealing how ubiquitin acts as a versatile scaffold for the recruitment of trafficking and signaling proteins, this UIM motif in Gα_s may open new avenues to better understand Gα_s recruitment on organelles and its role in nonconventional signaling pathways.

Materials and Methods

Antibodies, Reagents, and Plasmid Constructs.

A detailed list of antibodies, DNA constructs, and other reagents is provided in SI Materials and Methods.

Cell Culture, Transfections, IP, Pull-Down, and Western Blot Analysis.

HeLa cells and HEK293T cells were maintained and transfected as described by Rosciglione et al. (4). Co-IP, pull-down, and immunoblotting assays were performed as described by Rosciglione et al. (4). The EGFR degradation assay was performed as described by Zheng et al. (5). Western blot quantification was performed using ImageJ software (NIH). Experiments were performed in triplicate, and the results are presented as the mean ± SD. The statistical significance of the differences between the samples was assessed using the Student’s t test; a value of P < 0.05 was considered significant. Experimental details are provided in SI Materials and Methods.

Immunofluorescence and Quantification of Alexa Fluor 647-EGF Localization in the Endosomal Lumen.

Immunofluorescence assays were performed as described by Rosciglione et al. (4). For the uptake of Alexa Fluor 647-EGF, HeLa cells treated with control or Gα_s siRNA duplex were transfected with siRNA-resistant forms of ^wtGα_s or ^LNS/EAEGα_s, along with GFP-tagged Rab5-Q79L, to create enlarged endosomes. Cell surface receptors were labeled with Alexa Fluor 647-EGF and internalized for 30 min in DMEM at 37 °C, and then fixed and processed for immunofluorescence. The quantification of Alexa Fluor 647 signal in the endosomal lumen was determined as previously described (4).

NMR Spectroscopy and Modeling.

The ¹⁵N- or ¹³C, ¹⁵N-double–labeled peptides were purified from Escherichia coli BL21(DE3) transformed with pGEX-KG-Gα_s^1-34 and incubated with ¹⁵N ammonium chloride and ¹³C glucose. All NMR experiments were performed at 298 K on a Varian INOVA 600-MHz spectrometer equipped with a triple-resonance probe with z-axis pulsed-field gradients. NMR data were processed using NMRPipe (40) and analyzed with CCPNmr Analysis (41). Gα_s^1-34 peptide was modeled using Swiss-PdbViewer, in conjunction with SWISS-MODEL (25). The HADDOCK2.2 flexible docking server (42) was used for the modeling of biomolecular complexes between Gα_s^1-34 and ubiquitin (Protein Data Bank ID code 1D3Z) according to residues presenting significant CSD in both structures. Experiment details are described in SI Materials and Methods.