Structural studies of phosphorylation-dependent interactions between the V2R receptor and arrestin-2

Arrestins recognize different receptor phosphorylation patterns and convert this information to selective arrestin functions to expand the functional diversity of the G protein-coupled receptor (GPCR) superfamilies. However, the principles governing arrestin-phospho-receptor interactions, as well as the contribution of each single phospho-interaction to selective arrestin structural and functional states, are undefined. Here, we determined the crystal structures of arrestin2 in complex with four different phosphopeptides derived from the vasopressin receptor-2 (V2R) C-tail. A comparison of these four crystal structures with previously solved Arrestin2 structures demonstrated that a single phospho-interaction change results in measurable conformational changes at remote sites in the complex. This conformational bias introduced by specific phosphorylation patterns was further inspected by FRET and 1H NMR spectrum analysis facilitated via genetic code expansion. Moreover, an interdependent phospho-binding mechanism of phospho-receptor-arrestin interactions between different phospho-interaction sites was unexpectedly revealed. Taken together, our results provide evidence showing that phospho-interaction changes at different arrestin sites can elicit changes in affinity and structural states at remote sites, which correlate with selective arrestin functions.

The synthetic fluorescent amino acid L-(7-hydroxycoumarin-4-yl) ethylglycine (Cou) was genetically incorporated into arrestin2 and served as a fluorescent acceptor for the emission light at 330 nm, and tryptophan served as the donor, which was excited at 280 nm. Cou produced its characteristic emission light at 450 nm. b). Analysis of the titration experiments monitored by FRET assay. The dissociation constant was determined by fitting to the nonlinear regression equation y = Bmax[X]/(Kd + [X]) + NS[X]+Background as previously described 4 . c). Schematic of thermal stability shift analysis of arrestin2-V2Rpp complexes. Typical fluorescence intensity versus temperature for protein (arrestin2, arrestin2-Phospho-peptide or arrestin2-Phospho-peptide-fab30 complex) unfolding in the presence of BODIPY. In the presence of a basic protein (at the baseline of the curve), a basic fluorescence intensity was excited under 504 nm light (depicted schematically by green curved arrows). As the temperature rose, the protein unfolded and exposed hydrophobic moieties or Cys residues (in grey), which were labelled covalently by the fluorescent dye. The fluorescent light of 511 nm (depicted by orange curved arrows) emitted by the dye molecules was recorded.
Then, a gradual fluorescence decrease could be observed, which mainly indicated the removal of the protein due to precipitation or aggregation. Left panel: the finger loop appeared at nearly 90° with respect to inactive arrestin2 (PDB:1G4M, grey). Right panel: residue K77 turned approximately 120° to T347 of V2Rpp-1. The hydrophobic interaction of F75 and F244 was broken, and F75 rotated approximately 150° to form cation-π interactions with R65 in the arrestin2-V2Rpp-1 complex. Meanwhile, both Y63 and R65 reorganized to orient towards V2Rpp-1 but did not form strong interactions with T347 and the polar interaction between Y63 and N245 of C loop was broken. Plots of the distance root mean square deviations (RMSDs) for individual residues between the newly solved arrestin2-phospho-peptide complexes and previously reported arrestin2-V2Rpp-FP complexes were shown. Important residues with significant conformational change are highlighted. a,c,e,g): The vertical axis shows all heavy-atom RMSDs per arrestin residue, whereas the horizontal axis represents the position of each residue in arrestin2. Residue Cα deviations of 1.5 Å and above are marked with stars, and the related functional regions are annotated over the corresponding residues. 8 residues of NLS, clathrin, c-Raf-1 and MEK1 functional regions in the arrestin2-V2Rpp-1 complex (a-b), 14 residues of NLS, clathrin, receptor and MEK1 binding regions in the arrestin2-V2Rpp-3 complex (c-d), 12 residues of NLS, P1, c-Raf-1 and MEK1 functional regions in the arrestin2-V2Rpp-6-7 complex (e-f) and 14 residues related to clathrin, c-Raf-1, MEK1 and SRC binding in the arrestin2-V2Rpp-4 complex (g-h) have a main chain Cα coordinate change between 1.5 Å and 4 Å. b, d, f, h): residues highlighted in a, c, e, g were also highlighted in the surface representation. Supplementary Fig. 9 Supplementary Fig. 9. Examination of the average plots of the distance root mean square deviations (RMSDs) for residues between four newly solved arrestin2 complexes and previously reported arrestin2-V2Rpp-FP complexes.
a-d). The vertical axis shows average RMSDs per arrestin residue including all atoms, whereas the horizontal axis represents the position of each residue in arrestin2. Residue average RMSD deviations of 1.5 Å and above are marked with diamond or stars, in which stars represent residues overlapping to those found by heavy-atom RMSD analysis ( Supplementary Fig. 8) whereas diamonds indicated residues that are not identified by heavy-atom RMSD analysis. The related functional regions are annotated. The 2Fo-Fc annealing omits maps of arrestin2-V2Rpp-1, arrestin2-V2Rpp-3, arrestin2-V2Rpp-4 and arrestin2-V2Rpp-6-7 clearly show the electron density of the functional related region in arrestin2 that were mentioned in the main text. All maps were contoured at 1.0 σ.

Supplementary Fig. 11
Supplementary Fig. 11. Incorporation of TMSiPhe at functionally relevant motifs of arrestin2. a). Schematic representation of the strategy for incorporating TMSiPhe into arrestin2 as previously described. The codons that encode K49, F244, R307 and K357 were mutated to TAG, and then, E. coli cells cotransformed with plasmids encoding the arrestin2 mutants and the specific M. jannaschii tyrosyl amber suppressor tRNA/tyrosyl-tRNA synthetase mutants (TMSiPheRS) were cultured in Luria−Bertani (LB) medium containing 0.5 mM TMSiPhe, and the expressed arrestin2 was purified by Ni-NTA affinity chromatography. b). Frontal view of the TMSiPhe incorporation sites depicted by spheres in the active arrestin2 crystal structure (PDB: 4JQI). The red ball: K49 in the NLS region; The green ball: F244 in the C loop; The purple ball: R307 in the c-Raf-1 binding region; The orange ball: K357 in the MKE1 interaction area. c). Coomassie-stained gel was used to analyze the purified wild-type arrestin2 or arrestin2 mutants with TMSiPhe incorporation. This experiment was repeated three times (n=3) independently with similar results. +: Presence of 1 mM TMSiPhe in the culture medium; -: Absence of 1 mM TMSiPhe in the culture medium. d). 1D 1 H NMR spectra of arrestin2 labelled described in (Supplementary Fig. 11b). The spectra were recorded with 10 μM TMSiPhe-incorporated arrestin2, and the total recording time per spectrum was 15 min. The chemical shift for analysis was less than 0.55 ppm. The pentagrams: the position of NMR signal peak for the arrestin2 a-b). Elisa experiments to assay the expression levels of the wild type and indicated mutants of V2R or V2R-Rluc in HEK293 cells. Data from three independent experiments (n=3) are presented as mean ± SD. Statistical differences between WT and mutations were determined by two-sided one-way ANOVA with Tukey test. "ns": the mutants showed no significant difference compared to V2R-WT. c). The vasopressin-induced arrestin2 recruitment to different V2R phospho-deficient mutants were compared with V2R-WT, examined by BRET assay in a concentration dependent manner. n = 3 biologically independent replicates were examined for three independent experiments. Data (means ± SEM) of each mutant were normalized to maximal response of V2R-WT. d). Statistical analysis of the Emax and pEC50 of vasopressin induced arrestin2 recruitment to V2R WT or V2R mutants. Supplementary Fig. 15 Supplementary Fig. 15

. Functional analysis of c-Raf-1/MEK1 FlAsH biosensor on isoproterenol (ISO) induced arrestin2 complex formation.
HEK293 cells were co-transfected with plasmids encoded the YFP-arrestin2 and HA-c-Raf-1-WT or HA-c-Raf-1-CCPGCC (a); HA-MEK1-WT or HA-MEK1-CCPGCC (b). Forty hours after transfection, the cells were starved for 8 h and stimulated with 10 μM isoproterenol (ISO) for 15 min. HA-MEK1/HA-c-Raf-1 was immunoprecipitated by HA-antibody-conjugated agarose, and the arrestin2 were detected by western blotting using specific YFP antibodies. The results suggest that that these FlAsH biosensors of c-Raf-1 or MEK1 are able to recognize active arrestin. Full blot images of Supplementary Fig. 15 were shown in Supplementary Fig. 19 and Source Data file.
The western blot signals were quantified and are shown as columns in the right panel, +: with ISO; -: without ISO.
Statistics were determined by one-way ANOVA with Tukey's test. ***, p<0.001 (ISO stimulation was compared   a Data given as "-" mean that distance was beyond the critical value to form relevant interactions, e.g., H-bond interactions were not counted in case of distance > 3.5 Å; Anion-π interactions were not counted in case of distance > 6.0 Å; Polar interactions were not counted in case ofdistance > 4.0 Å. b Data given as "/" mean that residue could not be unambiguously assigned by electron density. Supplementary Table 4 Supplementary The fitting region of the S0, S1 and S2 were at 0.070~0.150 ppm, 0.160~0.210ppm, 0.210~0.240 ppm respectively, and then the area of the main peaks and corresponding residual error were obtained. Signal-to-noise ratio was measured using 'sinocal' routine within Topspin 4.0 (Bruker Biospin, Billerica MA), on the TMSiPhe signal of R307TMSiPhe arrestin2 at 0.116ppm, 0.195ppm or 0.219 ppm, using 2 ppm noise regions (centered around 11 ppm) for SNR calculations.
Supplementary Table 5 Supplementary a Data given as "-" mean that distance was beyond the critical value to form relevant interactions, e.g., H-bond interactions were not counted in case of distance > 3.5 Å; Anion-π interactions were not counted in case of distance > 6.0 Å; Polar interactions were not counted in case of distance > 4.0 Å. Charge-Charge interactions were not counted in case of distance > 6.0 Å.  The fitting region of the R0, R1 and R2 were at 0.060~0.120 ppm, 0.000~0.070 ppm, -0.030~0.060 ppm respectively, and then the area of the main peaks and corresponding residual error were obtained. Signal-to-noise ratio was measured using 'sinocal' routine within Topspin 4.0 (Bruker Biospin, Billerica MA), on the TMSiPhe signal of F244TMSiPhe arrestin2 at 0.091ppm, 0.026ppm or 0.019 ppm, using 2 ppm noise regions (centered around 11 ppm) for SNR calculations.

Supplementary
Supplementary Table 8 Supplementary -0.040~0.040 ppm respectively, and then the area of the main peaks and corresponding residual error were obtained.
Supplementary Table 9 Supplementary a Data given as "-" mean that distance was beyond the critical value to form relevant interactions, e.g., H-bond interactions were not counted in case of distance > 3.5 Å; Anion-π interactions were not counted in case of distance > 6.0 Å; Polar interactions were not counted in case of distance > 4.0 Å. Charge-Charge interactions were not counted in case of distance > 6.0 Å.

Supplementary Table 11
Supplementary The fitting region of the N1and N2 were at 0.100~0.150 ppm, 0.120~0.170 ppm respectively, and then the area of the main peaks and corresponding residual error were obtained. Signal-to-noise ratio was measured using 'sinocal' routine within Topspin 4.0 (Bruker Biospin, Billerica MA), on the TMSiPhe signal of K49TMSiPhe arrestin2 at 0.135 ppm or 0.149 ppm, using 2 ppm noise regions (centered around 11 ppm) for SNR calculations. a Data given as "-" mean that distance was beyond the critical value to form relevant interactions, e.g., H-bond interactions were not counted in case of distance > 3.5 Å; Anion-π interactions were not counted in case of distance > 6.0 Å; Polar interactions were not counted in case of distance > 4.0 Å. Charge-Charge interactions were not counted in case of distance > 6.0 Å.