Structural insights into biased G protein-coupled receptor signaling revealed by fluorescence spectroscopy

Reconstitution of Purified V2R in Amphipol.

The Flag-tagged V2R (Flag-V2R) was expressed in insect cells by using recombinant baculovirus technology. Flag-V2R is properly expressed at the plasma membrane of Sf9 cells, binds [³H]-AVP with an affinity similar to receptor expressed in mammalian cells (K_d = 13.5 ± 0.7 nM) (Fig. S1 A and B), and is able to couple to Gs protein (Fig. S1C). The Flag-V2R was purified by immunochromatography using an M1 Flag antibody affinity resin (Fig. 1A) and reconstituted by exchanging detergents for NAPols (1:10 protein/NAPol weight ratio), which are known to enhance the stability of membrane proteins (20). To obtain a homogeneous fraction of the V2R, the NAPol-reconstituted receptors were then loaded onto a size exclusion chromatography column. A receptor population was eluted at 15.5 mL close to the 75-kDa protein standards (conalbumin) (Fig. 1B). Nondenaturating blue native-PAGE analysis (Fig. 1B) revealed NAPols reconstituted V2R at around 75–80 kDa. These results are consistent with the molecular mass of a monomeric V2R (42 kDa) that would be in complex with at least three molecules of NAPol, on average (19). The enriched monomeric fraction was then used throughout the study. The particles containing the V2R were then visually inspected by negative staining EM. Image analyses from electron micrographs of the monomeric V2R-containing sample are shown in Fig. 1C. They reveal the presence of an apparently homogeneous population of particles having a round shape with a diameter ranging from 7 to 10 nm. Finally, we analyzed the pharmacological properties of V2R reconstituted in NAPols using [³H]-AVP saturating experiments (Fig. 1D). We determined an affinity constant of 674 ± 90 nM. Using the calculated B_max value, we evaluated the active fraction of receptor at about 90% of the total protein amount (SI Text).

Effect of Ligands on G Protein Activation and Recruitment of Arrestin-2 by NAPol-Reconstituted V2R.

To determine the functionality of NAPol-reconstituted V2R, we first measured ligand-induced incorporation of [³⁵S]GTPγS to purified Gαs proteins (1:2.5 receptor/G protein molar ratio). This assay was done using two full agonists for the Gs pathway, the endogenous hormone AVP and the biased ligand MCF14 (18), as well as with a Gs inverse agonist (SR121463) (21) (Fig. 2A). The effects of ligands (10 μM) were compared with basal signaling (i.e., basal [³⁵S]-GTPγS binding to the subunit Gαs). As shown in Fig. 2B, V2R induced a 1.42-fold increase in [³⁵S]GTPγS incorporation over basal, suggesting constitutive activity toward Gαs. The full agonists elicited similar GTPγS incorporation (3.22- and 2.95-fold over basal), whereas the inverse agonist treatment reduced the constitutive activity of the untreated V2R/Gαs complex by 50% (Fig. 2B). These results demonstrate that NAPol-reconstituted V2R interacts with and efficiently activates Gs protein in a ligand-dependent manner. Importantly, the efficacy profiles of the ligands correlated well with those observed in living cells (18, 21).

It has been shown that AVP activation of the V2R induces a stable interaction with both arrestin-2 and arrestin-3 (22). In addition, SR121463, which is a V2R inverse agonist for the Gs protein pathway, has been described as a partial agonist for the arrestin pathway (12), defining this compound as an arrestin-biased ligand. To determine the ability of ligands to modulate the recruitment of arrestin-2 by the NAPol-reconstituted V2R, we measured direct interactions between the receptor and a monobromobimane-labeled purified arrestin-2 (1:1 arrestin/receptor molar ratio, 10 μM). Bimane is a small-sized fluorophore with a high sensitivity to the polarity of its molecular environment that can be used as a sensor to detect interactions between the arrestin and its protein partners, as observed for visual arrestin and rhodopsin (23). As illustrated in Fig. 2C, incubation of arrestin-2 with NAPol-reconstituted V2R led to a quenching of 54 ± 5% of the bimane fluorescence intensity, suggesting high constitutive interaction of the two proteins. Typical fluorescence spectra of the bimane-labeled arrestin are provided in Fig. S2 in the absence or presence of V2R and after AVP treatment. In the presence of the full agonist AVP and the partial agonist (SR121463), we observed a fluorescence quenching of 77 ± 3% and 65 ± 4%, respectively. On the contrary, the Gs-biased MCF14, an agonist for the Gs pathway and an antagonist for the V2R-arrestin pathway (18), did not induce a change in fluorescence intensity (Fig. 2C). Moreover, we found that the AVP-induced fluorescence quenching can be reversed by MCF14 as expected for an antagonist. To rule out the possibility that the NAPol by itself had an effect on the quenching of the fluorophore, the receptor-free amphipol was incubated with bimane-labeled purified arrestin-2 with or without AVP. No change in the fluorescence of bimane was recorded in these conditions (Fig. 2C).

These results demonstrate that the monomeric NAPol-reconstituted V2R efficiently interacts with arrestin-2 and that the efficacy profiles of the ligands correlate with those observed in living cells.

Effect of Ligands on the Intrinsic Tryptophan Fluorescence of the V2R.

The fluorescent properties of tryptophans are very sensitive to their local chemical environment. There are 11 tryptophans distributed primarily in the TM segments, including TM2, TM4, TM5, TM6, and TM7. Therefore, tryptophan fluorescence can provide information about ligand-specific changes in the transmembrane core of the V2R. As shown in Fig. 3A, addition of SR121463 (Gs inverse agonist/Arr partial agonist) induced a 21 ± 1.7% decrease of the intrinsic tryptophan fluorescence intensity. On the other hand, AVP (Gs agonist/Arr agonist) led to an increase of 8 ± 0.9% (Fig. 3B). Similarly, MCF14 (Gs agonist/Arr antagonist) induced an increase of 10 ± 2% (Fig. 3B). As a control, we did not observe any changes in tryptophan intrinsic fluorescence when we added SR121463, AVP, and MCF14 to the unfolded receptor (0.61 ± 1%, 0.6 ± 2%, and 1 ± 0.6%, respectively, compared with the untreated receptor). We also determined the ligand concentration dependence on the tryptophan fluorescence signals and determined EC₅₀s for AVP (1.13 ± 0.5 μM), MCF14 (2.28 ± 0.8 μM), and SR121463 (2.86 ± 0.7 μM) as described in SI Text and Fig. S3.

Our results thus show that ligands with different efficacies toward Gs and arrestin signaling pathways induce opposite changes in intrinsic tryptophan fluorescence, suggesting that these ligands induce distinct conformational changes within the receptor that are necessary for pathway-selective signaling. However, intrinsic tryptophan fluorescence only gives information about global conformational changes in the receptor but cannot report on molecular movements from specific structural domains.

LRET Reveals Two Basal V2R States.

FRET has been used to study conformational changes in the β₂AR (24). This approach requires site-specific labeling of the protein with two different fluorescent probes. We developed a related approach based on resonance energy transfer, known as LRET or luminescence resonance energy transfer (SI Text). We used two mutants for the LRET experiments (depicted schematically in Fig. 4) for which the site-specific labeling and biochemistry were carefully characterized (SI Text and Figs. S4 and S5). For clarity, the Flag-V2R-A267C-C358A-Fluorescein Arsenical Helix binder (FlAsH) mutant will be referred to as the TM6 sensor and the Flag-V2R-S330C-C358A-FlAsH mutant will be referred to as the TM7-H8 sensor. Based on ligand binding and Gs-dependent cAMP accumulation in insect cells, the function of both mutants appeared equivalent to that of the WT receptor (Fig. S1C).

LRET measurements were performed as described in Materials and Methods. Our results show that the fluorescence decay of the donor-only species is adequately fit by a one-component exponential function for both sensors (Fig. 4 A and B). In contrast, the fluorescence decay of the donor in the presence of acceptor is best fit with a two-component exponential function (Fig. 4 A and B), suggesting the presence of two distinct lifetimes. This measurement is made by monitoring the decay of the acceptor-sensitized emission so that only the donor and acceptor engaged in LRET are detected (SI Text). We observed no intermolecular LRET in samples containing receptor labeled only with donor mixed with equivalent amounts of receptor labeled only with acceptor (Fig. S6).

These LRET data suggest the existence of at least two distinct basal states of the V2R characterized by distinct lifetime constants of the acceptor-sensitized emission (τ_AD fast and τ_AD slow, fast population and slow population) in both the TM6 and TM7-H8 sensors.

Effect of Ligands on the Conformational States of V2R Revealed by LRET.

To study ligand-dependent conformational changes in the V2R, we monitored the effects of AVP (Gs agonist/Arr agonist), SR121463 (Gs inverse agonist/Arr partial agonist), and MCF14 (Gs agonist/Arr antagonist) on the LRET signals for the TM6 sensor and TM7-H8 sensor. Fig. 5 A and B shows typical acceptor-sensitized fluorescence decays on a linear scale obtained with the untreated TM6 sensor and TM7-H8 sensor, respectively (black trace), and after treatment with the full agonist AVP (red trace). In this case, the lifetime analysis revealed significant changes in τ_AD fast and τ_AD slow after AVP treatment (Fig. 5 C and D).

We then analyzed the effect of biased ligands on the LRET signals. For clarity and as discussed in SI Text, only the effect on the largest population is shown in Fig. 6 (slow population). Importantly, the effects on both receptor populations follow a similar trend for both sensors (Tables S1–S4).

The results show a clear correlation between the ligands’ efficacy and their effect on lifetimes values [Fig. 6, Upper (representing the ligands’ efficacy) and Lower (representing the ligands’ effects on lifetime measurements)]. First, for the TM6 sensor, AVP and MCF14 (full agonists toward G protein; Fig. 6A, Upper) induced an increase in the lifetime τ_AD (26 ± 1% and 23 ± 3%, respectively; Fig. 6A, Lower), whereas treatment of the receptor with the Gs inverse agonist SR121463 (Fig. 6A, Upper) led to an opposite effect with a decreased lifetime (Fig. 6A, Lower).

For the TM7-H8 sensor, AVP (Arr full agonist) and SR121463 (Arr partial agonist) (Fig. 6B, Upper) induced an increase of 27 ± 7% and 31 ± 6% of τ_AD, respectively (Fig. 6B, Lower). Finally, treatment of the receptor with the Arr antagonist, MCF14 (Fig. 6B, Upper) did not induce a significant change in lifetime measurements (Fig. 6B, Lower). As a control, under the same conditions, we did not observe any changes in the lifetime τ_AD when we treated the receptor with leuprolide, an unrelated nonapeptide acting as an agonist for the GnRH receptor (25) (Tables S1–S4).

To address the contribution of the background labeling on the LRET results, we measured the amount of Lumi4-terbium maleimide (Lumi4-Tb) incorporated into the Flag-V2R-C358A-FlAsH receptor. The ratio of Lumi4-Tb labeling is 0.2 instead of 0.5–0.7 for TM6 and TM7-H8 sensors (SI Text). This labeling level did not allow us to measure any LRET when using a receptor concentration equivalent to that used for the TM6 or TM7-H8 sensor (0.1 μM). By increasing the amount of protein used in the LRET experiment (0.5 μM), we were able to detect significant LRET. In this condition, as for the two other sensors, we measured two lifetime components (Fig. S8A and Tables S1–S4). However, we cannot directly compare this background labeling with that of the TM6 and TM7/H8 sensors because in the absence of the most reactive cysteine, the probes will necessarily react with the low-reactive cysteines, thus increasing the amount of background. In the presence of the reactive cysteine, one would expect a ratio of background labeling much lower than 0.2. It is unlikely that this level of background labeling influences our results, however, because both SR and MCF induce opposite effects on receptor labeled specifically at the TM6 or TM7-H8 sensor.

Preparation of V2R/NAPol Complexes.

Construction, expression, solubilization, and purification of the receptor mutants from Sf9 insect cells, as well as the labeling kinetics methods, are described in SI Text. Purified detergent-soluble receptors were incubated at 4 °C in the presence of the amphipols at a 1:10 protein/NAPol weight ratio (19). After detergent removal with Bio-beads (Biorad), the sample was subjected to size exclusion chromatography (SI Text) to isolate the monomeric fraction of V2R. The V2R/NAPol complexes were then characterized by negative-stain transmission EM as described in SI Text.

[³⁵S]-GTPγS Binding and Arrestin-2 Recruitment Assays.

Gαs protein was produced and purified as described in SI Text and in Fig. S7. Binding experiments were performed as described in SI Text. The arrestin-2 mutant L68C-R169E was produced in Escherichia coli and purified by immobilized metal ion affinity chromatography (IMAC) as described in SI Text. The purified protein was then labeled with monobromobimane as described by Sommer et al. (23), and the arrestin recruitment assays were performed as described in SI Text.

Analysis of LRET Data.

Luminescence emission decays were measured at 620 nm and 520 nm and fitted as described in SI Text. For acceptor-donor, we calculated two lifetimes, defined as τ_AD fast and τ_AD slow. Ligand-induced changes in LRET and distances were measured as described in SI Text. The proportion of slow and fast populations was calculated as previously described (31), and details on the analysis are available in SI Text.