Research ArticleGPCR SIGNALING

Differential β-Arrestin–Dependent Conformational Signaling and Cellular Responses Revealed by Angiotensin Analogs

See allHide authors and affiliations

Science Signaling  24 Apr 2012:
Vol. 5, Issue 221, pp. ra33
DOI: 10.1126/scisignal.2002522


The angiotensin type 1 receptor (AT1R) and its octapeptide ligand, angiotensin II (AngII), engage multiple downstream signaling pathways, including those mediated by heterotrimeric guanosine triphosphate–binding proteins (G proteins) and those mediated by β-arrestin. Here, we examined AT1R-mediated Gαq and β-arrestin signaling with multiple AngII analogs bearing substitutions at position 8, which is critical for binding to the AT1R and its activation of G proteins. Using assays that discriminated between ligand-promoted recruitment of β-arrestin to the AT1R and its resulting conformational rearrangement, we extend the concept of biased signaling to include the analog’s propensity to differentially promote conformational changes in β-arrestin, two responses that were differentially affected by distinct G protein–coupled receptor kinases. The efficacy of AngII analogs in activating extracellular signal–regulated kinases 1 and 2 correlated with the stability of the complexes between β-arrestin and AT1R in endosomes, rather than with the extent of β-arrestin recruitment to the receptor. In vascular smooth muscle cells, the ligand-induced conformational changes in β-arrestin correlated with whether the ligand promoted β-arrestin–dependent migration or proliferation. Our data indicate that biased signaling not only occurs between G protein– and β-arrestin–mediated pathways but also occurred at the level of the AT1R and β-arrestin, such that different AngII analogs selectively engaged distinct β-arrestin conformations, which led to specific signaling events and cell responses.


Angiotensin II (AngII) signaling primarily through the angiotensin type 1 receptor (AT1R) is implicated in various cardiovascular and renal disorders including hypertension, cardiac hypertrophy, myocardial infarction, congestive heart failure, and nephrotic syndrome (13). The AT1R, a member of the heterotrimeric guanosine triphosphate–binding protein (G protein)–coupled receptor (GPCR) family, couples to numerous signaling pathways including G protein (Gαq, Gαi, Gα12/13)– and β-arrestin–dependent signaling as well as receptor tyrosine kinase transactivation (4). AT1R primarily couples to Gαq, which is responsible for the downstream activation of phospholipase C (PLC), resulting in inositol 1,4,5-trisphosphate (IP3) production followed by calcium-dependent protein kinase C (PKC) activation (4). G protein–coupled receptor kinase (GRK)–mediated receptor phosphorylation increases the receptor’s avidity for β-arrestin (5), which leads to functional uncoupling of G protein signaling. Mitogen-activated protein kinase (MAPK) signaling can be initiated by both receptor-mediated G protein activation and β-arrestin complex formation with the receptor (6). GPCR-mediated transactivation of receptor tyrosine kinases occurs through both G protein– (7) and β-arrestin–dependent (8) mechanisms. Multiple studies indicate that different ligands for a GPCR can engage distinct signaling pathways, a phenomenon referred to as ligand-directed signaling. This involves the differential stimulation of G protein– and β-arrestin–dependent versus G protein–independent and β-arrestin–dependent signaling. For instance, biased peptide ligands stimulating only Gαs–protein kinase A (PKA)–mediated activation of extracellular signal–regulated kinases 1 and 2 (ERK1/2) or G protein–independent, β-arrestin–dependent activation of ERK1/2 have been developed for the type 1 parathyroid hormone–related protein receptor (PTH1R) (9). There are also numerous studies highlighting ligand-directed signaling for β-adrenergic receptors (1015).

For the octapeptide AngII, Tyr4 and Phe8 are essential for both AT1R ligand binding and G protein–dependent signaling (1618). Substituting residues at position 4 and/or 8 in AngII such as in the G protein–signaling antagonist SII [(Sar1,Ile4,Ile8)-Ang]—where Sar is N-methylglycine—and the AngII analog TRV120027 [(Sar1,d-Ala8-OH)-AngII] (19, 20) produces β-arrestin–selective signaling ligands. SII mediates β-arrestin–dependent signaling downstream of the AT1R (2123). Moreover, β-arrestin–biased ligands for AT1R have recently been demonstrated in vitro (20, 24) and in vivo (20, 25) to have beneficial effects, suggesting that the effects of such biased ligands are likely dependent on the pluridimensionality of their efficacies toward distinct and/or common sets of signaling modalities, in particular the β-arrestin–dependent ones. Given our limited knowledge on how AT1R-selective biased ligands function, understanding the structure-signaling relationship for AngII analogs and the specific molecular mechanisms underlying their biased activities may help in the development of more efficacious AngII analogs.

Here, using different AngII analogs, we show that differential conformations of β-arrestin promoted by these ligand-bound AT1R complexes affect the signaling capabilities of β-arrestin. Moreover, we demonstrate that these analogs, which selectively, but differentially, activate β-arrestin–dependent signaling, have distinct functional outcomes.


As a prelude to characterizing the signaling pathways promoted by AngII and its analogs (Fig. 1A), we first assessed the affinities of each ligand for the hemagglutinin (HA)–tagged AT1R stably expressed in human embryonic kidney (HEK) 293 cells. The inhibition constant (Ki) of unlabeled AngII for displacing 125I-labeled AngII was in the low nanomolar range (~1 nM, Fig. 1B), whereas all tested analogs had lower affinities with Ki ranging from the low to medium nanomolar range (~7 to 18 nM) for SI [(Sar1,Ile8)-AngII], SVdF [(Sar1,Val5,d-Phe8)-AngII], SBpA [(Sar1,Bpa8)-AngII, where Bpa is benzoylphenylalanine], and DVG [(Asp1,Val5,Gly8)-AngII] to the higher nanomolar range (~213 nM) for SII [(Sar1,Ile4,Ile8)-AngII]. We next examined receptor-mediated signaling events promoted by AngII and its analogs. AT1R-mediated production of inositol phosphate (IP1) was assessed, and activation of PKCβ1–green fluorescent protein (GFP) and translocation of β-arrestin–monomeric red fluorescent protein (mRFP) were simultaneously monitored in living cells by confocal microscopy. AngII (1 μM) and its analogs (10 μM) were used at concentrations occupying >95% of receptors. As expected, AngII induced a rapid and sustained recruitment of PKC to the plasma membrane (Fig. 2). Of the five analogs, SBpA and, to a lesser extent, SVdF induced PKC recruitment. Consistent with the PKC activation assay, we found that SBpA induced significant increases in IP1 accumulation above baseline—about 28% of that of AngII (fig. S1A and table S1)—although with a potency 18 times lower than that of AngII. SVdF was as potent as SBpA in inducing IP1 production, but only to 4% of the amounts seen with AngII. SI, SII, and DVG induced less than 1% (0.6%, 0.5%, and 1%, respectively) of the inositol accumulation induced by AngII. In contrast to the activation of PLC (through IP1 production) and PKC that was observed only for AngII, SBpA, and SVdF, all AngII analogs promoted the formation of complexes between AT1R and β-arrestin2 in endosomes, which were prominent at 15 min after treatment, although to varying degrees (Fig. 2 and fig. S1B). Immunoprecipitation experiments confirmed that β-arrestin was recruited to the AT1R after treatment with AngII and any of the analogs (fig. S1C).

Fig. 1

Binding and affinities of AngII analogs at the AT1R. (A) AngII ligands used in this study. Bold indicates important residues that are different from the AngII sequence. (B) Affinities of AngII analogs for AT1R. Nonlinear regression, one-site competition curves were generated from competitive radioactive ligand binding assays with values normalized to percent based on specific binding. Ki values were calculated from the IC50. Average values for all compounds are represented by six independent experiments done in triplicate.

Fig. 2

G protein– and β-arrestin–dependent signaling of AngII analogs at the AT1R. HA-AT1R cells transfected with PKCβ1-GFP and βarr2-mRFP were treated with either AngII or analog. Images of living cells were taken over 30 min. Images after 30 s and 15 min are shown. Images are representative of at least five cell clusters in three independent experiments. Scale bars, 10 μm. Insets show magnified views of the indicated regions of the cell.

To characterize the potency and efficacy of β-arrestin recruitment to the AT1R, we used a bioluminescence resonance energy transfer (BRET) approach that monitors the recruitment of β-arrestin1 to the AT1R in living cells after ligand stimulation. AngII was the most potent and efficacious at promoting the recruitment of β-arrestin (fig. S2A). SI, SVdF, and SBpA recruited β-arrestin with similar potency and efficacy to each other, whereas DVG was less potent but equally efficacious as these three analogs. SII displayed the lowest potency and efficacy of all ligands. Because recruitment of β-arrestin to receptors leads to its conformational rearrangement that can be monitored with a BRET-based double-brilliance β-arrestin biosensor (26), we next assessed whether the different AngII analogs differentially modulated such changes in conformation. For this purpose, we monitored ligand-induced BRET changes with a new, optimized double-brilliance β-arrestin1, which positions β-arrestin between GFP10 and RlucII. Because BRET depends on the distance and the orientation between RlucII and GFP10, changes in the signal would reflect a modification in the structure of β-arrestin, resulting in an alteration in the relative position of the enzyme and the fluorophore. Results showed that the analogs have a wide array of potencies and efficacies in promoting changes in β-arrestin conformation with AngII again being the most potent, whereas SI, SVdF, SBpA, and DVG had similar efficacies but lower potencies (fig. S2B). SII had the lowest potency and efficacy. Because GRK2- and GRK6-mediated phosphorylation plays differential roles in β-arrestin recruitment to AT1R (27) and thus could play important roles in the conformational rearrangement of β-arrestin after receptor binding, we next assessed the effect of selectively depleting GRK2 and GRK6 on both responses. Analysis of dose-response BRET curves for β-arrestin recruitment to AT1R and conformational rearrangement of β-arrestin in the presence and absence of small interfering RNA (siRNA) selectively targeting the individual kinases (fig. S3) reveals that depletion of GRK2 had relatively little effect on AngII and all analogs on promoting these responses (Fig. 3, A to D). In contrast, GRK6 depletion led to readily detectable but distinct changes in both recruitment and conformational responses for many ligands (Fig. 3, E and F). To better appreciate these changes, a Cartesian plot of the GRK knockdown–promoted changes in both compounds’ efficacy and potency to elicit β-arrestin recruitment and conformational changes is presented in fig. S4. This representation confirms that depletion of GRK2 led to modest increases (~20%) in AngII, SVdF, DVG, and SII efficacies to promote β-arrestin conformational rearrangement (fig. S4, A and B). The potencies of the compounds in either the recruitment or the conformation assay were only modestly affected by the GRK2 depletion (less than 0.2 log units). When the effect of GRK6 on β-arrestin recruitment (fig. S4C) is considered, no change in efficacies and only modest reductions (less than 0.2 log units) in the potencies of AngII, SBpA, and SVdF were observed. In contrast, both the potencies and the efficacies of SI and DVG were diminished, whereas for SII, the efficacy but not the potency was affected. With the exception of SBpA, the effects observed in the recruitment assays were amplified for conformational rearrangement (fig. S4D) as illustrated by a further reduction in the compounds’ potency and/or efficacies, which are particularly evident for SI and SII. In addition to showing that GRK6 has a greater influence than GRK2 on both β-arrestin recruitment and conformational rearrangement, our data indicate that the β-arrestin responses promoted by distinct ligands are differentially affected by knockdown of GRKs.

Fig. 3

Role of GRK2 and GRK6 in β-arrestin recruitment to AT1R and conformational changes. (A to F) Cells transfected with AT1R-YFP, βarr1-RlucII, and control siRNA (A), GRK2 siRNA (C), and GRK6 siRNA (E) were stimulated with increasing concentrations of AngII and analogs. Data are the means ± SEM of three independent experiments. Dose-dependent agonist-promoted decrease of β-arrestin intramolecular BRET. Cells were transfected with HA-AT1R, GFP10-βarr1-RlucII, and control siRNA (B), GRK2 siRNA (D), and GRK6 siRNA (F) and treated as in (A), (C), and (E). “x” represents the baseline signal from vehicle-treated cells. Data are the means ± SEM of three independent experiments done in triplicate.

Collectively, our data show that some AngII analogs differentially regulate AT1R-mediated responses in distinct ways. To quantitatively assess the biases of the ligands toward the different responses, we used the operational model method (15, 28, 29). This method allows one to calculate the analogs’ efficacy relative to their affinities for engaging a specific signaling pathway (σ factor, which corresponds to the signaling efficiency of the pathway) and to determine whether this pathway is favored over another (bias factor, β) as compared to AngII, which is set to be balanced for both signaling events. The operational model also allows for the comparison of pathways that may be subjected to differential amplification (for example, G protein or β-arrestin). Analysis of the G protein–dependent response as assessed by IP1 generation compared to the recruitment of β-arrestin reveals a significant bias of SVdF and SBpA toward the latter response (Table 1). The weak IP1 signals (fig. S1A and table S1) generated by SI, SII, and DVG preclude calculation of a formal bias factor because a σ value could not be calculated from the marginal IP1 signals generated by these ligands. Although a bias factor could not be quantified, it nonetheless reflects bias because only one of the two pathways (β-arrestin) is readily activated by these ligands. We also analyzed the bias of the ligands to promote β-arrestin conformational rearrangement compared to recruitment to AT1R. In control cells, SI, SII, and SBpA were significantly biased toward β-arrestin recruitment, whereas SVdF and DVG were not (in that they promoted both responses equally) (Table 2 and table S2). Thus, for the same magnitude of recruitment, SI, SII, and SBpA were less efficient at promoting a conformational rearrangement in β-arrestin compared to AngII. GRK2 depletion led to all analogs being biased toward recruitment except for SBpA, which became balanced at both responses. After GRK6 knockdown, only SI and SII remained significantly biased toward β-arrestin recruitment. The bias factor for SII was greater after GRK6 depletion, consistent with a role for this GRK in SII-mediated β-arrestin conformational rearrangement (fig. S4). To better appreciate how GRK depletion relatively affected the efficiency (as defined by the σ factor) of the analogs to promote each β-arrestin response, we plotted the σ factor of β-arrestin conformational change as a function of its recruitment’s σ factor (Fig. 4 and table S2). This analysis showed that GRK2 depletion had only minor effects on both responses induced by all analogs (Fig. 4, A and B). On the other hand, the efficiency of both signaling responses induced by SI and DVG were markedly decreased in GRK6-depleted cells (Fig. 4C). SBpA became more efficient at inducing β-arrestin conformational rearrangement, making it more balanced, which contrasted with the decreased efficiency at this response seen with SII. Little change in signaling efficiency of either pathway was seen with SVdF in GRK6-depleted cells.

Table 1

Sigma (σ) values and bias factors (β) comparing IP1 and β-arrestin recruitment to AT1R. The column “Bias” denotes whether a statistically significant difference between the two pathways is seen in reference to the control compound, AngII. Such qualitative bias is nonetheless indicative of functional selectivity. n.d., not determined.

View this table:
Table 2

Sigma (σ) values and bias factors (β) comparing β-arrestin conformational change to recruitment to AT1R. The column “Bias” denotes whether a statistically significant difference between the two pathways is seen in reference to the control compound, AngII. n.s., not significant.

View this table:
Fig. 4

Sigma plots examining the effects of GRK2 and GRK6 on the signaling efficiencies of β-arrestin. (A to C) Comparison of effective signaling (σ) in β-arrestin conformation change and β-arrestin recruitment for the analogs compared with the reference agonist, AngII, with control siRNA (A), GRK2 siRNA (B), and GRK6 siRNA (C). The broken line indicates theoretically balanced signaling.

Given the differences observed in β-arrestin recruitment and conformational rearrangements among the AngII analogs, we wondered how the complex between β-arrestin and the receptor in endosomes was affected by the different ligands. We used fluorescence recovery after photobleaching (FRAP) to monitor the avidity between β-arrestin and receptors in live cells (30, 31). This avidity is inversely proportional to the recovery rate because the exchange of a bleached, receptor-bound β-arrestin on the endosome by an unbleached free one is limited by the strength of the former complex. We observed that the β-arrestin recovery rate on endosomes varied, with AngII and DVG being at opposite ends of the spectrum (Fig. 5A). β-Arrestin recovery on AT1R-containing endosomes was significantly faster with DVG than with AngII, indicating that the DVG-promoted interaction between β-arrestin and AT1R was more labile than that induced by AngII. We determined recovery half-times on endosomes with linear regression analysis of the recovery curves (Fig. 5B). After AngII treatment, 2.2 min was required for 50% recovery of the fluorescence, whereas the analogs displayed faster recovery times with half-lives ranging from 0.8 to 1.7 min (Fig. 5C). β-Arrestin–containing endosomes induced by SII were small and motile, making it difficult to quantitatively assess β-arrestin recovery (fig. S1B). Nonetheless, it appeared that the fluorescence recovered rapidly after bleaching of the endosomes in SII-treated cells.

Fig. 5

Avidity of β-arrestin for AT1R varies dependent on the ligand as revealed by FRAP. (A) Representative images of FRAP recovery experiment for AngII and DVG. HA-AT1R cells were transfected with βarr2-YFP. In cells treated with AngII or analog, endosomes were selected and bleached, and their fluorescence recovery rate was monitored. Scale bars, 10 μm. (B) Linear regression analysis of the recovery data representing an average of at least seven endosomes per ligand. (C) Recovery half-life was established on the basis of the linear regression analysis. The slope of AngII was set as 100%.

Because it represents a point of convergence downstream of many AT1R-mediated signaling pathways (including G protein– and β-arrestin–dependent), we next examined the ability of AngII and the analogs to promote AT1R-mediated activation of ERK1/2. The analogs activated ERK1/2 with lower potencies and efficacies than AngII (Fig. 6A and fig. S5), revealing that the analogs were partial agonists of this pathway. AngII, SBpA, and SVdF, which activated both the Gαq protein–dependent (IP1 production and PKC activation) and β-arrestin–dependent pathways, stimulated larger ERK1/2 responses. To determine whether this ERK1/2 activation was β-arrestin–dependent, short hairpin RNA (shRNA) targeting β-arrestin1 and β-arrestin2 was used. After β-arrestin1 or β-arrestin2 depletion, ERK1/2 activation was significantly reduced for each ligand (Fig. 6B), indicating a contribution of β-arrestin–mediated signaling for all ligands. Depleting β-arrestin1 and β-arrestin2 had the greatest effect on SI-, SII-, and DVG-mediated ERK1/2 activation, consistent with the primary involvement of β-arrestins in this response. Because β-arrestin is involved in endosomal MAPK signaling (32), we examined the relationship between ERK1/2 activation and the avidity of the interaction between β-arrestin and AT1R. We found a linear relationship between the degree of interaction of β-arrestin with AT1R induced by the different AngII analogs and the activity of ERK1/2 at 15 min after agonist treatment (Fig. 6C). Furthermore, we found no correlation between the efficacy of ERK1/2 activation and the affinity of the ligands for the receptor, their efficacy to recruit or alter β-arrestin’s conformation (fig. S6, A to C). Together, these results indicate that although all analogs tested (including some that cannot activate Gαq) can promote β-arrestin recruitment, they vary in their ability to induce conformational rearrangements and the formation of high-avidity complexes with the receptors. Among all parameters assessed, the ability to promote stable complexes between AT1R and β-arrestin represents the best predictor of MAPK activation response.

Fig. 6

Regulation of ERK1/2 activation by the AngII analogs. (A) Effects of AngII analogs on MAPK activation. HA-AT1R cells were treated with AngII or analog for the times indicated. Lysates were immunoblotted for phosphorylated ERK (pERK) and total ERK (tERK). pERK amounts were normalized to tERK and pERK at 5-min treatment of AngII was set at 100%. Shown is a representative blot. Quantification of results is the mean ± SEM and is representative of five independent experiments. Two-way ANOVA followed by a Bonferroni post hoc test (with AngII as reference) was performed. (B) Role of β-arrestin in MAPK activation. As in (A), except ERK1/2 activation was measured in cells transfected with shRNA against β-arrestin1 or β-arrestin2. Quantification is the mean ± SEM and is representative of three independent experiments done in duplicate. Two-way ANOVA followed by a post hoc Bonferroni test. **P < 0.01; ***P < 0.001. Inset represents percent inhibition of ERK1/2 signal relative to shRNA control transfected cells. (C) Correlation plot demonstrating the linear relationship between ligand-mediated ERK1/2 activation at 15 min and the avidity of the β-arrestin and receptor complex. R2 = 0.8862, P < 0.05.

Finally, we wanted to study the consequences of ligand-mediated functional selectivity in a physiologically relevant cell line, rat aortic vascular smooth muscle cells (VSMCs). We first assessed the ability of AngII and analogs to induce cell migration, a well-characterized AT1R-dependent event (33). Treatment of cells with AngII led to a doubling in cell migration, which was significantly decreased by knockdown of β-arrestin2 (Fig. 7A). Except for DVG, all analogs stimulated the migration in a largely β-arrestin2–dependent manner (Fig. 7A). Given the apparent uniqueness of DVG, we sought to further monitor the conformation status of β-arrestin after treatment with DVG compared to that of AngII. The signal produced by a fluorescence resonance energy transfer (FRET)–based intramolecular β-arrestin2 sensor (which evaluated the conformational changes in β-arrestin) was increased by AngII but decreased by DVG, suggesting ligand-mediated conformational differences in β-arrestin (Fig. 7B). AngII also induces VSMC growth [reviewed in (34)]; therefore, we examined the analogs’ ability to induce this response. Cell proliferation, as measured by a [3H]thymidine incorporation assay, was increased by AngII and DVG, but not with the other analogs (Fig. 7C). As was the case for the migration response, AngII- and DVG-mediated cell growth was blocked only by β-arrestin2 knockdown (Fig. 7D).

Fig. 7

AngII analog signaling in VSMCs reveals β-arrestin–dependent signaling bias at the subreceptor level. (A) VSMC migration assay in the presence of siRNA against β-arrestin. Cells were transfected with control, β-arrestin1, or β-arrestin2 siRNAs. Knockdown was verified with β-adaptin as a control. TCL, total cell lysate. Wb, Western blot. VSMCs were seeded into Boyden chambers, and cells migrating to the lower chamber were counted. Results are representative of four independent experiments, presented as means ± SEM. One-way ANOVA followed by a Dunnett’s post hoc test (untreated for each condition as control) was performed. (B) β-Arrestin conformational change by FRET in VSMCs. VSMCs transfected with AT1R-HA and CFP-βarr2-Venus were treated with buffer, AngII, or DVG. Readings were taken every minute for 16 min. Signal from untransfected cells was subtracted, and fold was determined by division of the prestimulated signal. Quantification of six independent experiments done in triplicate. Two-way repeated-measures ANOVA followed by a Bonferroni post hoc test was performed. (C) Cell growth by [3H]thymidine incorporation in VSMCs. [3H]Thymidine was added to VSMCs concurrent with stimulation, and thymidine incorporation was then assessed. Quantification of six independent experiments. One-way ANOVA followed by a Dunnett’s post hoc test (untreated as control) was performed. (D) Role of β-arrestin in cell growth by [3H]thymidine incorporation in VSMCs. As in (C), except cells were transfected with siRNA. Quantification of three independent experiments. Two-way ANOVA followed by a Bonferroni post hoc test was performed (control siRNA compared to three experimental conditions). *P < 0.05; **P < 0.01; ***P < 0.001.


The data presented here underscore the multifaceted regulation of β-arrestin’s change in conformation and directed signaling by providing new evidence that distinct complexes between β-arrestin and AT1R promoted by AngII analogs lead to different β-arrestin activated states and cellular outcomes. Furthermore, we reveal that ligands have different propensities to induce β-arrestin conformational rearrangement upon recruitment to the receptor, most likely as a result of distinct receptor conformations that are promoted or stabilized by the analogs, as well as selective GRK actions. We also show that the stability of the complex between β-arrestin and AT1R in endosomes depends on the nature of the ligand-bound receptor complex and that this parameter correlates well with the degree of MAPK activation induced by the ligands. Our study further unveils novel high-affinity biased β-arrestin AngII ligands, which are able to engage selective subsets of AT1R-mediated responses (specifically, cell growth or migration). Our findings confirm the notion that distinct β-arrestin–dependent signaling events can be selectively induced and that functional segregation of cell responses can be achieved by the use of specific AngII analogs.

Previous studies have shown that some AT1R ligands are biased toward β-arrestin over Gαq signaling (20, 21, 24). Here, we add to the list of such biased ligands by characterizing four AngII analogs that all favor the β-arrestin response over the production of IP1 and PKC activation. In some cases, the analogs behaved as strong partial agonists for the arrestin pathway and were essentially unable to promote Gαq signaling (SI and DVG). Our study using two distinct BRET-based assays that monitored the ligand-mediated interaction between AT1R and β-arrestin, and the ensuing β-arrestin conformational rearrangements upon receptor binding, revealed additional biased signaling modalities for the AngII ligands. These included the differential propensity of ligands to promote β-arrestin recruitment to AT1R and β-arrestin conformational changes and unveiled differential contributions of GRKs in these bias responses (see below). Although the extent to which the changes in β-arrestin’s conformation are structurally similar is still unclear (namely, whether all compounds promote a similar conformational rearrangement of β-arrestin upon recruitment to the ligand-bound receptor), we nonetheless observed differences in both the recruitment and the conformational rearrangement of β-arrestin among the ligands. For some ligands, the extent of β-arrestin’s conformational rearrangement correlated well with its recruitment to the receptor (Table 2). However, for other ligands, we observed a clear disconnect between the recruitment and the conformational change. For instance, for SI, SII, and SBpA, there was less β-arrestin conformational change for an equivalent degree of recruitment to the receptor, resulting in a clear bias in the efficiency toward β-arrestin recruitment (Fig. 4A and Table 2). Furthermore, the observation that DVG promoted conformational changes distinct from that of AngII opens the possibility for ligand-specific β-arrestin conformations. Indeed, a previous study with a β-arrestin2 BRET-based biosensor demonstrated that AngII activation elicited an increased BRET signal compared to the decreased signal seen with SII, suggesting different β-arrestin conformations (35). Together, these results imply that the conformational change of β-arrestin promoted by a ligand is dependent on its recruitment and binding to the receptor but that other receptor-dependent events that may be differentially triggered by distinct ligands can also contribute to the structural change in β-arrestin.

The propensity of different analogs to promote distinct β-arrestin recruitment and conformational rearrangement has been suggested in some cases to be imparted by GRK-mediated phosphorylation of the receptor (36), but independent of this process in another study (35). For the β2-adrenergic receptor, ligand promotes phosphorylation of distinct sites by different GRKs, differentially affecting the conformation of β-arrestin after its stimulation (36). Our findings on the role of GRK2 and GRK6 in the efficiency (σ factor) of β-arrestin conformational rearrangement and recruitment to AT1R further imply distinct functions of the kinases in these responses. Overall, GRK2 depletion had only relatively modest effects on either recruitment or conformational changes, whereas GRK6 depletion had more noticeable effects on both the potencies and the efficacies of various ligands to promote the two responses. Our findings that the efficacy of recruitment (Emax) of β-arrestin to AT1R by AngII or SII was more affected by GRK6 than GRK2 depletion are in agreement with a recent report (24). However, they differ from a study implying that phosphorylation by GRK or other second-messenger–dependent kinases does not contribute in β-arrestin conformational changes promoted by either AngII or SII (35). This discrepancy might arise from the use of different intramolecular β-arrestin BRET sensors, which allowed in our case the detection of such response. Our data also underscore the importance of GRK6 in regulating both the potency and the efficacy of SI-, SII-, and DVG-mediated β-arrestin conformational changes. The minimal effect of GRK2 depletion suggests that either other kinases play redundant roles or GRK2 is not required for these responses in the cells studied. For the effects of GRK6 depletion, compounds can be divided into two groups: (i) modestly affected compounds (AngII, SBpA, and SVdF) and (ii) compounds whose propensity to promote β-arrestin recruitment and conformational rearrangement was considerably reduced (SI, SII, and DVG). The relative effect of GRK6 depletion on the efficacies and potencies of SI, SII, and DVG differed among the ligands as a function of the assay considered, indicating different roles of the kinase in promoting the recruitment and stabilizing discrete conformations of β-arrestin upon binding of different ligands. For example, the observation that GRK6 depletion had a particularly important effect on the potency of SII to elicit a conformational rearrangement compared to its effect on the recruitment suggests an important role for the kinase in promoting the structural change of β-arrestin after its recruitment to the receptor. This important effect on the β-arrestin conformation resulted in a significant increase in the bias factor for this ligand in the GRK6-depleted cells (Table 2). This may indicate a greater propensity of the SII-bound AT1R to act as a substrate for GRK6 and/or a greater GRK6 dependency of the SII-bound AT1R (for example, through receptor phosphorylation) to promote a β-arrestin conformational rearrangement. The present results indicate that GRK6 is an important regulator of β-arrestin engagement by the AT1R that may contribute to the ligand-biased signaling observed for this receptor.

In addition to the different propensity of the ligands to promote the engagement and the conformational rearrangement of β-arrestin, the ligands were also different in their ability to promote a stable receptor–β-arrestin complex. We found that the stability of the complex between β-arrestin and AT1R in endosomes correlated with the degree of ERK1/2 activation. In contrast, the extent of recruitment and the conformational changes in β-arrestin induced by the ligands did not correlate with their effectiveness at ERK1/2 activation (fig. S6, B and C). These results demonstrate that factors affecting the strength of the receptor and β-arrestin complex are more likely to affect ERK1/2 signaling rather than those altering the amount of β-arrestin recruitment to the receptor. Longer-lived complexes between AT1R and β-arrestin in the endosomes may enable the formation of more stable MAPK scaffolds [for example, with the kinases MEK (mitogen-activated or extracellular signal-regulated protein kinase kinase) and Raf] at this complex, thus allowing more efficient signaling. Although the use of biased ligands favoring such signaling modes should help us better appreciate the biological and physiological importance of receptor-mediated, β-arrestin–dependent ERK1/2 signaling in endosomes, more work is needed to understand the underlying mechanisms regulating such complex formation in this cellular compartment.

Our data provide further evidence for the role of amino acid 8 of AngII in determining AT1R functional signaling selectivity (37, 38). We showed that the aromatic and bulky nature of Phe8 of AngII helps to maintain G protein–dependent signaling, because SVdF and SBpA still retained some Gαq-PLC-PKC–mediated signaling activity in HEK293 cells. Replacing Phe8 in AngII with smaller nonaromatic residues (such as Ile and Gly in SI, SII, or DVG) generated analogs lacking G protein–dependent signaling capabilities, consistent with what was previously reported (38). Such substitutions produced AT1R ligands that selectively promoted functionally distinct conformations in β-arrestin, affecting the stability of the signaling complex generated by the association of AT1R and β-arrestin and the cellular responses. Our results imply that SI and SII are more selective for engaging β-arrestin–dependent signaling than SVdF and SBpA, which also promoted G protein signaling. DVG demonstrated negligible amounts of ERK1/2 activation and Gαq activation, although it efficiently recruited β-arrestin to AT1R and promoted changes in β-arrestin’s conformation. Surprisingly, this analog was as potent as AngII in promoting β-arrestin–dependent cell growth in VSMCs, yet was unable to induce migration. In contrast, the other AngII analogs were as effective as AngII in promoting β-arrestin–dependent migration, yet lacked the ability to potentiate cell growth. Given the FRET data in VSMCs demonstrating that AngII induced a conformation different from that of DVG, these results suggest that β-arrestin–biased signaling depends on the selective conformation of β-arrestin imposed by the ligand-receptor complex. DVG would thus allow β-arrestin–dependent selective engagement of adaptors and/or signaling effectors common to AngII for VSMC growth, but would not allow the engagement of effectors involved in migration. Resolving these β-arrestin–mediated conformationally dependent signaling structures and understanding their underlying mechanisms of regulation remain important challenges.

The results presented here highlight a β-arrestin conformationally dependent mode of regulation and also have the potential to connect specific signaling pathways with their physiological outputs. Use of AngII-biased ligands should help further our understanding of the role of β-arrestin–mediated signaling in normal and pathophysiological conditions and to develop the appropriate therapy. However, a future challenge will be to determine the profile of drug action on biased signaling and to which extent the biases, such as the ones observed in this study, affect clinical efficacy.

Materials and Methods


AngII, trichloroacetic acid, trifluoroacetic acid (TFA), collagen (type I/calf), and poly-l-lysine are from Sigma Chemical Co. SI is from MP Biomedicals. SII, SVdF, SBpA, and DVG were synthesized by us. Dimethyl sulfoxide (DMSO) was purchased from BioShop. Antibodies against pERK1/2, total ERK1/2, and GRK2 were from Cell Signaling Technology. The IP-One HTRF assay was from Cisbio. Anti-phosphotyrosine (4G10) was from Millipore. The polyclonal antibody against the C-terminal domain of β-arrestins (3978) was described elsewhere (39). Anti-HA antibodies coupled to agarose beads and mouse anti-HA (clone 12CA5) were purchased from Roche. GRK6 antibody was from Santa Cruz. Anti–β-adaptin antibody was from BD Biosciences. Anti-mouse and anti-rabbit horseradish peroxidase–conjugated immunoglobulin G were from Sigma. Western Lightning Plus-ECL, [3H]thymidine, and iodine-125 were obtained from PerkinElmer. Scrambled siRNA and siRNA against rat β-arrestin1 (rUrGrCrCrArCrArGrUrArUrCrArUrCrArGrCrUrUrCrCrUrCrCrArU and rGrCrUrCrUrArUrGrArGrArUrUrGrGrUrGrUrCrUrArCrUrGrGrArG) and β-arrestin2 (rGrCrUrCrArGrArArArCrGrArArGrArUrArUrGrGrCrCrUTG, rGrGrArGrUrArGrArCrUrUrUrGrArGrArUrUrCrGrArGrCCT, and rArGrArCrCrGrUrCrArArGrArArGrArUrCrArGrArGrUrGTC) were purchased from IDT. Control siRNAs and siRNAs against human GRK2 and GRK6 (siGENOME Non-Targeting siRNA pool#1, siGENOME SMARTpool human ADRBK1, and siGENOME SMARTpool human GRK6, respectively) were purchased from Dharmacon. Boyden chambers (Corning), precoated iodination tubes (Thermo Scientific), and acetonitrile were purchased from Fisher Scientific. Sep-Pak Plus C18 cartridges were from Waters.

Plasmids and constructs

Plasmids encoding HA-AT1R, PKCβ1-GFP, βarr2-YFP (yellow fluorescent protein), and βarr2-mRFP have been previously described (3941). The GFP10-βarr1-RlucII and βarr1-RlucII constructs were derived from a previously published GFP10-EPAC-RlucII construct (42) by excising the EPAC coding sequence with Acc 65I–Hind III and Nhe I–Hind III restriction enzymes, respectively, and replacing it with a polymerase chain reaction (PCR)–amplified coding sequence of human β-arrestin1. The cyan fluorescent protein (CFP)–βarr2–Venus construct was generated from a CFP-EPAC-Venus construct, which was derived from a previously published CFP-EPAC-RlucII construct (43) by replacing EPAC by PCR-amplified human β-arrestin2 between Acc 65I and Hind III restriction sites. All constructs were verified by DNA sequencing (Sequencing Service, Genome Quebec Innovation Centre, McGill University, Quebec, Canada).

Cell culture and transfection

HEK293 cells stably expressing the AT1R (293HA-AT1R) were grown in minimum essential medium (MEM, Hyclone) supplemented with 10% fetal bovine serum (FBS, Gibco), 2 mM l-glutamine (Gibco), and gentamicin (100 μg/ml; Gibco). Cells were seeded in six-well plates (1.5 × 105 cells per well), 35-mm glass-bottomed culture dishes (Mattek Corp.) (1.2 × 105 cells per dish), or 10-cm dishes (1.5 × 106 cells per dish). Transfections were performed by a calcium phosphate coprecipitation method as previously described (44). Rat aortic VSMCs were a gift from M. Servant (Université de Montréal). They were grown in low-glucose Dulbecco’s modified Eagle’s medium (DMEM) (Hyclone) supplemented with 10% fetal calf serum (FCS, Gibco), 2 mM l-glutamine, and gentamicin (100 μg/ml). Cells were seeded in 24-well plates at a density of 1.0 × 104 cells per well ([3H]thymidine experiments), 8.8 × 105 cells per 10-cm dish (IP1 experiments), and 1.5 × 104 cells per well (96-well plate) for AlphaScreen experiments. For experiments with siRNA, cells were transfected by Lipofectamine 2000 (Invitrogen) with 20 nM siRNA (control or anti–β-arrestin) in serum-free medium. Four hours later, an equal amount of medium containing 20% FCS was added to return the cells to 10% FCS overnight. The next day, cells were split into 24-well plates (104 cells per well), allowed to adhere for 4 hours, and then serum-starved for 48 hours, at which point the experiment was performed.

Labeling of tracer and affinity determination

125I-AngII was prepared with Iodogen, and its specific radioactivity (1051 Ci/mmol) was determined from self-displacement and saturation experiments as previously described (45). Dose-displacement experiments to determine the median inhibitory concentration (IC50) for the analogs were performed with slight modifications to a published protocol (45). Briefly, 293HA-AT1R cells were grown in 10-cm dishes to confluence. Cells were washed with phosphate-buffered saline (PBS) and then resuspended in PBS containing 2 mM EDTA. Cells were dosed by use of the Lowry-HS assay (Bio-Rad) so that 25 to 50 μg could be added per reaction. Cells were resuspended in 0.5 ml of binding buffer [2% bovine serum albumin, 25 mM tris-HCl (pH 7.4), 100 mM NaCl, 5 mM MgCl2, 1 mM phenylmethylsulfonyl fluoride, leupeptin (25 μg/ml), aprotinin (2.5 μg/ml), and 1 mM pepstatin]. About 90,000 cpm of 125I-AngII was added to each tube (~0.11 nM) and was displaced with varying concentrations of unlabeled ligand for 1 hour at room temperature. Bound radioactivity was separated from free ligand by filtration through GF/C filters presoaked in binding buffer. The radioligand content was evaluated by γ-counting on a PerkinElmer Wizard 1470 automatic γ counter. The Bmax (375 fmol/mg of total proteins) and dissociation constant (Kd) for 125I-AngII (0.45 nM) were determined from saturation binding experiments with 293HA-AT1R cells. Apparent affinities (Ki) of AngII and its analogs were calculated from the IC50 from radioligand displacement curves with the following equation: Ki = IC50/(1 + [ligand]/Kd).

IP1 assay

The assay was performed according to the manufacturer’s specifications. Briefly, 104 cells per well (384-well plate) were treated with increasing concentrations of AngII or AngII analog for 30 min. IP1-d2 and anti–IP1-cryptate were added for 2 hours. Plates were read on a Synergy 2 multimode microplate reader.

ERK1/2 phosphorylation

293HA-AT1R cells were seeded in six-well plates coated with poly-l-lysine. Forty-eight hours after seeding, cells were serum-starved in MEM containing 20 mM Hepes and then left untreated or treated with AngII (1 μM) or analogs (10 μM). Cells were lysed in Laemmli buffer [250 mM tris-HCl (pH 6.8), 2% (w/v) SDS, 10% (v/v) glycerol, 0.01% (v/v) bromophenol blue, and 5% (v/v) β-mercaptoethanol] and sonicated. Lysates were resolved on SDS–polyacrylamide gel electrophoresis (SDS-PAGE), transferred to nitrocellulose membranes, and blotted for pERK1/2 and ERK1/2.

AlphaScreen assay

ERK1/2 activation in 293HA-AT1R cells was monitored with PerkinElmer’s AlphaScreen SureFire assay as previously reported (46). Cells were seeded in 10-cm dishes, transfected with Lipofectamine 2000 with shRNA against β-arrestin1 or β-arrestin2, and 24 hours later reseeded in 96-well plates (104 cells). For experiment without transfection, cells were directly seeded in 96-well plates (104 cells). Forty-eight hours later, cells were serum-starved overnight. Cells were then stimulated for 5 or 15 min with 1 μM AngII or 10 μM AngII analog and lysed in 30 μl of lysis buffer. Cells were incubated at room temperature on a plate shaker for 10 min. Five microliters of lysate was used for analysis. Readings were performed on a PerkinElmer EnSpire 2300 multilabel reader.

Confocal microscopy

293HA-AT1R cells were plated in 35-mm glass-bottomed dishes and transfected with (i) β-arrestin2–mRFP and PKCβ1-GFP or (ii) β-arrestin2–pEYFP. Forty-eight hours after transfection, cells were serum-starved followed by stimulation with AngII (1 μM) or analogs (10 μM). For (i), PKC translocation and β-arrestin endosome formation were monitored for 30 min. For (ii), 15 min after ligand addition, endosomes were bleached and fluorescence recovery was monitored over a period of 5 min, with photos taken every 30 s. Images were collected on a Zeiss LSM 510 Meta laser scanning microscope with a 40× oil immersion lens using the following excitation/emission filter sets: 543 nm/560 nm long pass (LP) for mRFP, 488 nm/505 to 530 nm band pass (BP) for GFP, and 514 nm/530 to 600 nm BP for YFP.

Immunoprecipitation experiments

Coimmunoprecipitation of covalently cross-linked β-arrestin to AT1R was performed as previously described (41). Western blotting was performed with the anti–β-arrestin antibody (3978) and the anti-HA antibody (12CA5).

BRET assay

BRET experiments were performed as previously described (47, 48). Briefly, HEK293T cells were cultured in DMEM supplemented with 10% FBS, penicillin (100 U/ml), and streptomycin (Wisent Inc.). For experiments without siRNA, 3.0 × 106 cells were seeded in 10-cm dishes. Transient transfection was performed 24 hours later with polyethyleneimine (PEI; Polysciences) at a DNA/PEI ratio of 1:3. AT1R-YFP (5 μg) and 500 ng of βArr1-RlucII, or 5 μg of HA-AT1R and 200 ng of GFP10-βarr1-RlucII were cotransfected, and the total amount of DNA transfected in each plate was adjusted to 10 μg with salmon sperm DNA (Invitrogen). For experiments with siRNA, 1.0 × 107 cells were seeded in 10-cm dishes in serum-free medium 24 hours before transfection. Transient transfection was performed with Lipofectamine 2000 (Invitrogen) with the same amounts of DNA as described above and with 20 nM siRNA (Dharmacon). Twenty-four hours after transfection, cells were detached, seeded in pretreated poly-l-ornithine hydrobromide (Sigma-Aldrich) 96-well white plates at 50,000 cells per well, and reincubated at 37°C for an additional 24 hours before being processed. Cells were washed once with Tyrode’s buffer (140 mM NaCl, 1 mM CaCl2, 2.7 mM KCl, 0.49 mM MgCl2, 0.37 mM NaH2PO4, 5.6 mM glucose, 12 mM NaHCO3, and 25 mM Hepes, pH 7.5) directly in the 96-well plates and incubated in buffer with or without indicated concentrations of AngII and analogs for 25 min at room temperature. Coelenterazine h (Nanolight Technologies) or coelenterazine 400A (Biotium) was added to a final concentration of 5 μM in Tyrode’s buffer 5 min before reading. Readings were collected with a MITHRAS LB 940 multidetector plate reader (Berthold Technologies), allowing the sequential integration of the signals detected in the 480 ± 20– and 530 ± 20–nm windows for the RlucII Renilla luciferase and YFP light emissions, respectively, and in the 410 ± 40– and 515 ± 15–nm windows for the RlucII and GFP10 light emissions, respectively. The BRET signal was determined by calculating the ratio of the light intensity emitted by the YFP or GFP10 over the light intensity emitted by the RlucII. The values were corrected by subtracting the background BRET signals detected when RlucII was expressed alone.

Migration assay

This assay was previously described using HEK293 cells but was modified for VSMCs (49). Briefly, VSMCs were serum-starved overnight and seeded into Boyden chambers (24-well inserts with 8-μm-pore collagen-coated membranes). One hour after plating, cells were stimulated with AngII (100 nM), AngII analog (10 μM), or left untreated. After 4 hours, cells were fixed with paraformaldehyde (4%) for 20 min and incubated with crystal violet (0.1% in 20% methanol) overnight. Membranes were washed three times in distilled water, and cells were removed from the upper chamber, leaving those that migrated to the lower chamber. Cells were detached and counted.

FRET assay

VSMCs were transfected with Lipofectamine 2000 at a DNA/lipofectamine ratio of 1:3. HA-AT1R (2 μg) and β-arrestin2 (4 μg) tagged at both the N (CFP) and the C (Venus) termini were cotransfected. pcDNA3.1 zeo+ replaced β-arrestin in the mock condition. Twenty-four hours after transfection, cells were serum-starved overnight. Cells were detached the following morning in Tyrode’s buffer and seeded into 96-well plates (50,000 cells per well) and incubated at 30°C. Four baseline readings were taken on a PerkinElmer EnSpire 2300 multilabel reader with shaking between each read to prevent cells from settling. Cells were then treated with buffer alone, 100 nM AngII, or 10 μM DVG and monitored for 16 min with readings every minute. Values were corrected by subtracting the background signal from cells not expressing the β-arrestin sensor.

[3H]Thymidine incorporation assay

VSMCs were seeded in 24-well plates. Twenty-four hours later, the medium was replaced with low-glucose DMEM containing 0.2% FCS. Forty-eight hours later, cells were placed in serum-free medium with either DMSO or AG1478 (125 nM) and 0.5 μCi of [3H]thymidine (concentration of 1 μCi/ml). Cells were then treated with medium alone or medium containing AngII (final concentration, 100 nM) or analogs (final concentration, 10 μM). Twenty-four hours later, cells were put on ice and washed twice with cold PBS followed by 30-min incubation with 5% TFA. Finally, cells were washed with PBS, and radioactivity was recovered in 0.5 ml of 1 M NaOH, which was added to vials containing 5 ml of scintillation liquid, and counted with a PerkinElmer Tri-Carb 2800TR liquid scintillation analyzer.

Data analysis

Intensity of the signals from Western blots was determined by densitometric analysis with ImageJ and is presented as the means ± SEM of at least three independent experiments. Bias factors (β) and sigma values (σ) were calculated with the operational model as described previously (15, 28). Statistical analysis was performed with GraphPad Prism software with one- or two-way analysis of variance (ANOVA) when appropriate. Bonferroni (comparison between all groups) or Dunnett’s (comparison to control) post hoc tests were performed where necessary.

Supplementary Materials

Fig. S1. Analog-induced production of IP1 and recruitment of β-arrestin as analyzed by confocal microscopy and immunoprecipitation.

Fig. S2. Recruitment of β-arrestin to AT1R and conformational changes induced by AngII analogs.

Fig. S3. siRNA-mediated knockdown of GRK2 and GRK6.

Fig. S4. Role of GRK2 and GRK6 in the efficacy and potency on ligand-induced β-arrestin recruitment and conformational change.

Fig. S5. Dose-response relationship of ERK1/2 activation by AngII and analogs.

Fig. S6. Correlation plots between ligand affinity, BRET values, and ERK1/2 activation.

Table S1. IP1 dose-response curve data used to calculate σ values and β factors.

Table S2. BRET dose-response data used to calculate σ values and β factors.

References and Notes

Acknowledgments: We thank W. Stallaert and V. Lukashova for providing us with β-arrestin1–RlucII and CFP–β-arrestin2–Venus, respectively. We thank C. Le Gouill for helpful discussion and E. Goupil for help with the binding experiments. Funding: This work was supported by Canadian Institutes of Health Research (CIHR) Operating Grants to S.A.L. (MOP-74603, CTP-79848), A.C. (MOP-79470), M.B. (FRN-11215), and E.E. (MOP-69019); a CIHR Maintenance Grant to S.A.L. (PRG-82673); and a CQDM grant to S.A.L. and M.B. B.Z. holds a CIHR Banting and Best doctoral research award. A.B. was supported by postdoctoral fellowships from Groupe de Recherche Universitaire sur le Médicament and Fonds de la Recherche en Santé du Québec. S.A.L. and M.B. hold Canada Research Chairs in Molecular Endocrinology and in Signal Transduction and Molecular Pharmacology, respectively. Author contributions: B.Z. and S.A.L. designed the study. B.Z., A.B., B.A., and R.C. performed the experiments. B.Z., A.B., A.C., E.E., M.B., and S.A.L. analyzed the data. B.Z., A.B., B.A., and R.C. generated figures. E.E. designed the AngII analogs. B.Z., A.B., M.B., and S.A.L. wrote the paper. Competing interests: M.B. is the chief executive officer of a not-for-profit company (IRICoR) whose mission is to promote the commercial transfer of the technologies of the Institut de Recherche en Immunologie et Cancérologie. The other authors declare that they have no competing interests. The double-brilliance β-arrestin construct is patented (U.S. 7,932,080; authors that are inventors: A.B. and M.B.) and a continuation in part application is also pending (U.S. 2011/0275134; authors that are inventors: A.B., M.B., B.Z., and S.A.L.). The invention can be used without limitations for academic research but requires a license for commercial use. A standard academic materials transfer agreement (MTA) applies for the academic use of the double-brilliance β-arrestin.

Stay Connected to Science Signaling

Navigate This Article