Angiotensin receptor conformations stabilized by biased ligands differentially modulate β-arrestin interactions

Figure 1. Formation of 7TM core-mediated AT1R/β-arrestin complexes.A, schematic of complex formation strategy. Purified, non-phosphorylated AT1R bound to various orthosteric ligands is incubated with βarr1∗, a purified complex of β-arrestin1, V2Rpp, and the active-state-specific β-arrestin1 binders dscFv30 (or Fab30) and Nb24. B and C, competition binding between the radiolabeled antagonist [3H]-olmesartan and unlabeled (B) AngII or (C) losartan at purified AT1R in the absence or presence of βarr1∗ (2.5 μM). Log IC50 values are provided in Table S1. Data represent the mean ± S.D. of three independent experiments. D, SEC of deglycosylated AT1R/βarr1∗ complexes formed in the presence of AngII or candesartan. E and F, gel electrophoresis analysis (Coomassie staining) of SEC fractions for (E) AngII and (F) candesartan complexes shown in panelD. Data in panelsD–F are representative of three independent experiments.

Figure 2. DEER analysis of AT1R activation and βarr1∗ core interaction.A, spin label positions and pairwise combinations (dashed lines) used to record distances in this study. Spin labels are shown as sticks; clouds indicate the possible positions of the unpaired electron. B, probability distance distributions between the indicated spin pairs showing the effects of AngII (green), and AngII + βarr1∗ (blue) on apo AT1R (gray). The area under the curve for each distribution is normalized to 1; shaded areas indicate 95% confidence intervals.

Figure 3. Formation of AT1R/βarr1∗ complexes by β-arrestin-biased ligands.A, ligand-induced β-arrestin1 recruitment to the cell plasma membrane. Data are normalized to the AngII maximum response and represent the mean ± S.D. of three independent experiments. Curve analysis parameters are provided in Table S2. B, TM1-TM6 probability distance distributions for indicated ligands in the presence (blue) or absence (other colors) of βarr1∗ using a Gaussian mixture model. The area under the curve for each distribution is normalized to 1; shaded areas indicate 95% confidence intervals. The associated raw DEER data and model-based analysis are shown in Fig. S3. C, median populations ± 95% confidence intervals of Gaussian populations centered around 23, 32, 35, and 42 Å for TM1-TM6 distance distributions. Populations are provided in Table S3.

Figure 4. Ligand-dependent stability of AT1R/βarr1∗ complexes.A, AlphaScreen assay detecting the interaction between FLAG-tagged AT1R (100 nM) and βarr1∗ (33 nM) (via His6-tagged Fab30) in the presence of saturating concentrations of various ligands. Data are normalized to the no ligand/AngII signals and represent the mean ± S.D. of three independent experiments. B and C, Co-immunoprecipitation of βarr1∗ with AT1R (glycosylated) in the presence of saturating concentrations of various ligands. B, shows a representative Coomassie-stained gel; (C) shows quantitation of the β-arrestin1 band (mean ± S.D. of three independent experiments). D–G, stability of deglycosylated AT1R/βarr1∗ complexes formed with (D and E) TRV023 and (F and G) TRV026 as assessed by (D and F) SEC and (E and G) gel electrophoresis (Coomassie staining) of the resulting fractions. Data shown in panels D–G are representative of three independent experiments.

Figure 5. MD analysis of AT1R bound to TRV026.A, peak distance of the modeled TM1-TM6 distance distribution over simulation time (left). The bold line represents the sliding mean, and the semi-transparent gray area indicates the standard deviation. After about 5 μs, the inactive-like peak position (black line) shifts toward an intermediate distance (red line). The Kernel Density Estimate plots (right) show the peak-distance distributions from the early (black) and late (red) parts of the trajectory. These distributions differ significantly (Kolmogorov-Smirnov test, α = 0.05). B, intracellular and side views of the AT1R structural model with the intermediate TM6 position (red). The active (light gray, PDB: 6OS2) and inactive (dark gray, PDB: 4YAY) AT1R structures are shown for comparison. Backbone flexibility is visualized by mapping the RMSF to the width of the cartoon representation. C, superposition of GPCR–arrestin core complexes with the model of the TRV026-bound intermediate conformation. All observed poses of visual arrestin and β-arrestin1 clash with TM6 and H8 of the AT1R model.

Figure 6. Model of ligand-dependent conformational equilibria of AT1R/β-arrestin complexes. AngII and TRV023 promote the interaction of β-arrestin with both phosphorylated AT1R regions and the 7TM core. TRV026 promotes the interaction of β-arrestin with phosphorylated sites but not with the 7TM core.