β-arrestin recruitment facilitates a direct association with G proteins

Abstract

G protein-coupled receptors (GPCRs) are targets for almost a third of all FDA-approved drugs. GPCRs are known to signal through both heterotrimeric G proteins and β-arrestins. Traditionally these pathways were viewed as largely separable, with G proteins primarily initiating downstream signaling while β-arrestins modulate receptor trafficking and desensitization in addition to regulating their own signaling events. Recent studies suggest an integrated role of G proteins and β-arrestins in GPCR signaling, however the cellular and biochemical requirements for G protein: β-arrestin interactions remain unclear. Here we show that G proteins and β-arrestins can directly interact. Through utilization of β-arrestin-biased receptors and artificially enforced β-arrestin relocalization, we demonstrate that recruitment of β-arrestin to the plasma membrane is sufficient to interact with the G protein Gαi. Using purified proteins, we show that Gαi directly interacts with β-arrestin. In addition, we find that Gαi family members differ in their degree of association with β-arrestin, and that a large degree of this selectivity resides within the alpha helical domain of Gαi. These findings delineate the cellular and biochemical conditions that drive direct interactions between G proteins and β-arrestins and illuminate the molecular basis for how they work together to effect GPCR signaling.

Figure 1:. M3R activation promotes Gαi:β-arrestin association.

(A) Cartoon of split luciferase assay. (B) HEK 293T cells transiently transfected with the M3R, the indicated Gα-LgBiT, and SmBiT-β-arrestin-2 were treated with the muscarinic agonist iperoxo (1 μM). (C) Cells transiently transfected with the M3R, Gαi1-LgBiT, and SmBiT-β-arrestin-2 were treated with the indicated muscarinic agonist iperoxo (1 μM), carbachol (100 μM), or oxotremorine-M (10 μM). (D) Cells were pretreated with either tiotropium (10 μM) or vehicle for 10 minutes and subsequently treated with the indicated iperoxo (1 μM). (E) Cells transiently transfected with the M3R, Gαi1-LgBiT, and SmBiT-β-arrestin-2 were treated with iperoxo at the indicated concentration. N=3 replicates per condition. Data shown are mean ± SEM.

Figure 2.. β-arrestin-biased GPCRs promote Gαi:β-arrestin association

(A) HEK 293T cells transiently transfected with the indicated β-arrestin-biased receptor, the indicated Gα-LgBiT, and SmBiT-β-arrestin-2 (A) AT1R-AAY stimulated with 1 μM AngII (B) AT1R-AAY stimulated with 10 μM TRV023. (C) D2R-ARB stimulated with 1 μM Quinpirole. (D) ACKR3 stimulated with 100nM of its endogenous ligand, CXCL11 (C-X-C Ligand 11), 1 μM Bam22, 1 μM WW36, and 1 μM WW38. For panel A-B *P<0.0001 by two-tailed t test comparing area under curve (AUC) for Gαi-AngII versus Gαq-AngII (A) and Gαi-TRV023 versus Gαq-TRV023 (B). For panel C *P<0.005 Gαi versus Gαs by two-tailed t test comparing AUC. n=3 for all conditions and represents independent plate replicates. Graphs depict the percent increase over vehicle treatment normalized to max Gαi signal (A-C) or percent ACKR3 max WW38 signal (D). Data shown are mean ± SEM.

Figure 3.. G protein-biased ligands do not promote Gαi:β-arrestin complex formation

HEK293T cells overexpressing the Gαi TRUPATH BRET sensor with (A) balanced ligands DAMGO, and Fentanyl, (n=3) or (B) G protein biased ligands Oliceridine and PZM21 (n=3). The antagonist Naloxone was used as a negative control. (C) Illustration of by-stander NanoBiT recruitment assay used to monitor β-arrestin-2 recruitment upon MOR activation. Briefly, LgBiT is localized to the plasma membrane by a CAAX tag. Upon activation of MOR, the β-arrestin-2 translocates to the plasma membrane to produce luminescence. (D) Balanced ligands, DAMGO (n=6) and Fentanyl (n=3), promote β-arrestin-2 recruitment to MOR. (E) G biased ligands, Oliceridine (n=5) and PZM21(n=5), do not recruit β-arrestin-2 to MOR. Naloxone was used as a negative control (n=6) in D-E. (F-H) Split luciferase Gαi-LgBiT:smBiT-β-arrestin-2 association upon stimulation with DAMGO (n=9) (F), Fentanyl (n=7) (G) but not G protein biased ligands Oliceridine and PZM21 (n=4) (H). Naloxone (n=7) was used as a negative control (F-H). For panel F,G, and H, a two-way ANOVA with Bonferroni post hoc test was performed to compare the effect of DAMGO, Fentanyl, Oliceridine, or PZM21 with the negative control, Naloxone. *P<0.05 (F-G), ns P>0.05 (H). All graphs denote the mean signal ± SEM and are normalized to max DAMGO signal.

Figure 4.. Translocation of β-arrestin to plasma membrane is sufficient to promote Gαi:β-arrestin association.

(A) Schematic of NanoBiT assay monitoring Gαi:β-arrestin association upon induced translocation β-arrestin-2 to the plasma membrane. Upon addition rapamycin (500 nM), FKBP and FRB heterodimerize, forcing translocation of β-arrestin to the plasma membrane independent of receptor activation. Gαi:β-arrestin complexes are assessed via production of luminescence from reconstituted luciferases. (B) Gαi:β-arrestin complex formation following addition of rapamycin. (C) AUC analysis of (B). *P < 0.05 by two-tailed t-test with Gαi vs Gαs. n=5 independent replicates for both graphs. Data shown are mean ± SEM.

Figure 5:. Purified Gαi and β-arrestin directly interact.

(A) Coomassie gel of a pull down of purified N-terminally protein-c tagged β-arrestin-1–393X with purified non-lipidated Gαi1 incubated either with buffer or various nucleotide conditions (GDP 20 μM, GTPγS 20 μM, or apyrase for the nucleotide-free state). (B) [35S]-GTPγS incorporation into non-lipidated Gαi1 or myristoylated Gαi2 in the presence or absence of 4x molar excess of purified β-arrestin-1–393X and β-arrestin-2–393X. For panel A, pull-down representative of three independent experiments. For panel B, N=3 replicates are plotted as maximal GTPγS binding. Error bars are the mean ± S.D.

Figure 6:. Mutations within Gα helical domain impairs interactions with β-arrestin

(A) HEK293T cells were transiently transfected with V2R, SmBiT-β-arrestin-2, and either LgBiT-tagged Gαi with Gαs α5 helix or LgBiT-tagged Gαs with Gαi α5helix and treated with AVP (500 nM) (B) LgBiT-tagged Gαo, Gαi1, Gαi2 and Gαi3 were compared with their ability to interact with SmBiT-β-arrestin-2 in cells overexpressing the V2R and treated with AVP (500 nM). (C) Area under the curve graph (AUC) made from the kinetic tracing data in panel B. *P <0.05 by one-way ANOVA, Dunnett post hoc test of Gαi1 versus Gαi2, Gαi3, or Gαo. (D) Sequence alignment of human Gαi1, Gαi2, Gαi3 with differing residues highlighted with red asterisks. Alignment performed by GproteinDb66,67 (E) Structure of inactive Gαi1 (PDB:1BOF) residues distinct between Gαi1 and Gαi2 and Gαi3 are highlighted in green. (F) Diagram of the NanoBRET assay for Gαi:β-arrestin complex formation. (G) Gαi1 A59D, S98A (M1), Gαi1 A113S, F118V, A121P, P165S (M2), or Gαi1 K132R S134G (M3) were compared to Gαi:β-arrestin by NanoBRET assay. HEK293T cells were transiently transfected with Nluc-tagged Gαi mutants (WT, M1, M2, or M3), β-arrestin-2-mKO, and V2R. Upon activation of V2R with AVP (500 nM), cells were monitored for changes in BRET ratio indicating Gαi:β-arrestin complex formation. (H) Area under the curve graph (AUC) made from the kinetic tracing data in panel G and represents Δ Net BRET between AVP and vehicle treated cells. *P<0.05 by one-way ANOVA with Dunnett’s post hoc comparing Gαs versus Gαi1 and Gαi1–3 M3 versus Gαi1. Data in A-B demonstrate % increase luminescence of AVP above vehicle signal. All data shown are from n=3 replicates. Data shown are mean ± SEM.