Increasing evidence indicates that G protein-coupled receptor (GPCR) signaling can achieve specificity in different ways. The previously accepted linear signaling concept has evolved into a network concept that allows for more possibilities with respect to the fine tuning of various pathways. The differential activation or inactivation of G protein-dependent or -independent signaling pathways can occur at various levels (Hermans, 2003; Swaminath et al., 2005; Michel and Alewijnse, 2007). First, receptors can adopt different active and inactive conformational states depending on the ligands they bind, resulting in multiplicity of coupling (Kenakin 2003). Second, GPCRs have the propensity to homo- or heterodimerize, leading to variations in their pharmacological properties (Terrillon and Bouvier, 2004; Park and Palczewski, 2005; Milligan, 2006). Finally, the receptor environment, in particular the localization of receptors within specialized microdomains of the plasma membrane such as rafts or caveolae, can profoundly modify their coupling properties and cause them to switch from one signaling pathway to another (Gimpl and Fahrenholz, 2000; Guzzi et al., 2002; Rimoldi et al., 2003). All of these aspects can contribute to the interaction of GPCRs with different protein complexes and can thus explain cell-dependent signaling variability and how signaling can be channeled into specific routes in particular cells (Schulte and Levy, 2007). Accordingly, such aspects must be taken into consideration to have a complete understanding of a given receptor's cellular lifespan and to develop new selective therapeutic compounds.

We have focused our work on the arginine-vasopressin (AVP) V_1b receptor. This GPCR, which is expressed in numerous peripheral tissues (e.g., pituitary gland and adrenal gland) and various areas of the nervous system (Lolait et al., 1995; Hurbin et al., 1998; Vaccari et al., 1998; Hernando et al., 2001; Young et al., 2006), is involved in stress behavior and represents an interesting therapeutic target for psychiatric disorders such as anxiety or depression (Serradeil-Le Gal et al., 2002). In this context, studying its functional activation properties is of importance.

Previous studies have suggested that the V_1b receptor can exhibit differential coupling, as has been shown for other members of the AVP/oxytocin (OT) receptor family (V_1a, V₂, and OT receptors). Indeed, other members of the family signal either through pathways involving different G proteins (Abel et al., 2000; Zhou et al., 2007) or even through G protein-independent pathways (Rimoldi et al., 2003; Charest et al., 2007). With respect to the V_1b receptor, previous work has shown that it activates the IP pathway via coupling to the G_q/11 protein (Jard et al., 1986). Signaling through the cAMP pathway has been reported in stably transfected CHO cells, although it was interpreted as nonspecific coupling to G_s caused by overexpression (Thibonnier et al., 1997). In addition, an atypical response of the V_1a and/or V_1b receptors toward the cAMP pathway was suspected in the vasopressinergic magnocellular neurons of the supraoptic nucleus (Hurbin et al., 1998). Specifically, AVP triggered an adenylyl cyclase-mediated intracellular response in these cells through a mechanism that is not yet understood because the G_s-coupled V₂ receptor is not expressed (Hurbin et al., 1998; Sabatier et al., 1998; Gouzénes et al., 1999).

To investigate the potential multiple coupling properties of the rat V_1b receptor, we sought to determine whether 1) this receptor could activate both the IP and cAMP signaling pathways at low levels of expression and in different cell systems, 2) a link could be established between the nature of the ligand on the one hand and the signaling pathway activated or intensity of the response on the other hand, 3) the receptor could form dimers at the cell surface, and 4) the response depended on the localization of the receptor within specialized plasma membrane microdomains.

Materials and Methods

cDNA constructs. Rat AVP V_1b receptor cDNA was obtained by RT-PCR. Total RNA was extracted from carefully dissected rat pituitary glands using the RNeasy kit (QIAGEN, Courtaboeuf, France). Forward (5′-CACCTCTAAACCTTTCTCTCTCATTCC-3′) and reverse (5′-GGATTGAGTGCTCTGATTTCCAAC-3′)V_1b-specific primers were synthesized by Sigma Genosys (St-Quentin-Fallavier, France). Reverse transcription was performed on RNA extracts previously digested with deoxyribonuclease I (Invitrogen, Cergy-Pontoise, France), using SuperScript II reverse transcriptase (Invitrogen), RNAsin (Promega Corp, Charbonnières, France) as an RNase inhibitor, and the reverse primer. DNA amplification was performed using a PTC-150/16 thermal controller (MJ Research, Inc., Watertown, MA), Platinum Pfx DNA polymerase (Invitrogen), the reverse primer previously used for reverse transcription, and the forward primer. N-terminal tags (hexahistidine, 6×His tag; or hemagglutinin, HA tag) were introduced in a second PCR step using specific primers designed to also add two restriction sites (5′-BamHI and 3′-XhoI) to control the direction of fragment insertion. The PCR products were electrophoresed on a 1% agarose gel in Tris-acetate/EDTA buffer (40 mM Tris-acetate and 1 mM EDTA, pH 8) and purified using GenElute columns (Sigma). Double digestion of the DNA ends with BamHI and XhoI was performed, and the main fragment containing the V_1b coding sequence was purified and ligated with T4 DNA Ligase (Roche, Meylan, France) into pcDNA3.1(+) vector (Invitrogen) that had been digested with the same restriction enzymes and dephosphorylated using alkaline phosphatase (Roche). The sequence was verified by sequencing of both strands (Genome Express, Meylan, France). Using the same procedure, rat AVP V_1a receptor cDNA was prepared using specific forward (5′-CCTCAGGACCAGACAGAAGTAGG-3′) and reverse (5′-CCACATAAACACATCTGCTCTTACG-3′) primers and RNA extracted from liver. The N-terminal 6×His-tag and restriction sites (5′-BamHI and 3′-XhoI) were then added by PCR, and the sequence was inserted by T4 ligation into the pcDNA3.1(+) vector as indicated above.

Cell Culture and Transfection. Cell lines were obtained from the American Type Culture Collection (Manassas, VA). Cell culture media, fetal bovine serum (FBS), and additives were provided by Invitrogen.

CHO, HEK-293, and COS-7 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% FBS, 100 U/ml penicillin/streptomycin, and 2 mM l-glutamine, at 37°C in a humidified atmosphere of 95% air and 5% CO₂. For the CHO cell line, nonessential amino acids (Invitrogen) were also added to the media.

Transient transfection of the cell lines was performed using electroporation in a 300-μl volume with a total of 10 μg of DNA (V_1b plasmid up to 1 μg plus pcDNA3.1 as carrier DNA to reach 10 μg) containing 10⁷ cells in electroporation buffer (50 mM K₂HPO₄, 20 mM CH₃COOK, 20 mM KOH, and 26 mM MgSO₄, pH 7.4). After electroporation (260-280 V, 1 mF, GeneZapper 450/2500; IBI, New Haven, CT), cells were suspended in complete medium and seeded into 96-well culture plates at a density of 10⁵ (ELISA and cAMP pathway test) or 8 × 10⁴ (IP pathway test) cells per well. Ninety-six-well culture plates were first coated with polyornithine diluted in PBS, incubated at 37°C for 1 h, and then rinsed with PBS before seeding. Twenty-four hours after transfection, receptor expression was measured by radioligand binding assay or ELISA, and the experiments on receptor dimerization and signaling pathways were performed. To improve the expression of the transfected receptors, CHO and HEK-293 cells were incubated overnight with 3 mM sodium butyrate before the experiments. This treatment was not necessary with the COS-7 cells.

ELISA. To measure the expression of the transfected receptors, cells were fixed with 4% paraformaldehyde in PBS for 5 min and rinsed three times with PBS. A blocking step of 30 min with PBS + 1% decomplemented FBS was performed before incubation with the primary antibodies (0.5 μg/ml) for 30 min. The cells were then rinsed four times for 5 min in PBS + 1% FBS and incubated for 30 min with an anti-mouse antibody conjugated with horseradish peroxidase (1/1000; Amersham, Orsay, France). The cells were rinsed three times with PBS + 1% FBS and three times with PBS. Afterward, 60 μl of PBS and 20 μl of Supersignal ELISA Femto (Perbio-Pierce, Brebières, France) were added to the wells. The luminescence was read using a Wallac Victor² (PerkinElmer Life and Analytical Sciences, Courtaboeuf, France).

Second Messenger (IP₁ and cAMP) Accumulation. Activation/inhibition of the IP and cAMP pathways by AVP receptor agonists or antagonists, respectively, was determined using the IP-One and cAMP dynamic kits (Cisbio Bioassays, Bagnols-sur-Cèze, France). In brief, after transfection, 8 × 10⁴ (IP pathway test) or 10⁵ (cAMP pathway test) cells were distributed in 100 μl of complete medium into a 96-well assay plate (Greiner Bio-One, Courtaboeuf, France). Twenty-four hours later, the medium was removed and replaced with 50 μl of incubation medium containing the agonist and/or antagonist at the appropriate concentrations. The IP-One test is based on the accumulation of IP₁, a downstream metabolite of the IP pathway that is produced by phospholipase C activated by the G_q/11 protein; IP₁ is stable in the presence of LiCl. The cAMP dynamic test is based on the accumulation of cAMP produced by adenylyl cyclase activated by G_s protein; a phosphodiesterase inhibitor, Ro-20-1724 (Calbiochem, Darmstadt, Germany), prevents cAMP degradation. The homogeneous time-resolved fluorescence-fluorescence resonance energy transfer (HTRF-FRET) assay was performed as described previously (Maurel et al., 2004). This assay involves the transfer of energy from a europium cryptate pyridine-bipyridine donor fluorophore to a d2 acceptor fluorophore. The assay is an immunoassay that measures competition between native IP₁ or cAMP produced by the cells and IP₁- or cAMP-labeled with the d2 acceptor, as revealed by a monoclonal antibody against IP₁ or cAMP labeled with europium cryptate pyridine-bipyridine. Twenty-five microliters of antibody and 25 μl of competitor diluted in lysis buffer provided in the kits were added to the wells after 30-min incubation at 37°C with the agonist. As a negative control, some wells only received the donor fluorophore-labeled antibody. After 1 h incubation at room temperature, fluorescence emissions were measured at both 620 and 665 nm on a RubyStar fluorometer (BMG Labtechnologies, Offenburg, Germany) equipped with a nitrogen laser as the excitation source (337 nm). A 400-μs reading was recorded after a 50-μs delay to eliminate the short-lived fluorescence background from the acceptor fluorophore-labeled antibody. The fluorescence intensities measured at 620 nm and 665 nm correspond to the total europium cryptate emission and to the FRET signal, respectively. The specific FRET signal was calculated using the following equation: ΔF% = 100 × (R_pos - R_neg)/(R_neg), with R_pos being the fluorescence ratio (665/620 nm) calculated in wells incubated with both donor- and acceptor-labeled antibodies, and R_neg being the same ratio for the negative control incubated only with the donor fluorophore-labeled antibody. The FRET signal (ΔF%), which is inversely proportional to the concentration of IP₁ or cAMP in the cells, was then transformed into IP₁ or cAMP accumulation using a calibration curve prepared on the same plate. It is worth noting that all comparisons of agonist or antagonist effects were done on the same day, on the same culture and plate, and were made against the AVP effect. The experiments were repeated at least three times on different cultures. Normalization was performed as indicated in the figure legends, either as a percentage of the maximal value or as a percentage of the maximal value for AVP when comparisons were necessary. Values corresponding to the low basal activities, determined in unstimulated cells, were first subtracted. Activation/inhibition curves were plotted to the log of agonist or antagonist concentrations and fitted to the Hill equation to extract the EC₅₀, Hill coefficient, and minimal/maximal values.

Overnight preincubation of cells with either cholera toxin (Ctx) (20 μg/ml culture medium) or pertussis toxin (Ptx) (100 ng/ml culture medium) was used to test the involvement of the G_s and G_i proteins in V_1b receptor cAMP signaling in response to AVP. When used to directly activate adenylyl cyclase, forskolin was added for 30 min at 50 μM.

The inhibitory effect of the specific nonpeptidic V_1b antagonist, SSR149415 (Serradeil-Le Gal et al., 2002), on IP₁ and cAMP accumulations induced by AVP was studied according to Arunlakshana and Schild (1959). Preincubation for 10 min with the antagonist was followed by 30-min incubation with the antagonist and AVP. Concentrations of AVP greater than 1 μM were not tested here and more generally in this study because, in our experimental conditions, they often led to nonspecific effects. Therefore, we cannot formally insure that the inhibition by SSR149415 is strictly competitive.

Incubation of cells with methyl-β-cyclodextrin (MβCD; Sigma) was used to study how removing cholesterol from the plasma membrane affects V_1b receptor signaling. Preincubation at 37°C for 10 min with 10 mM MβCD in incubation medium was followed by 30 min incubation with 10 mM MβCD plus the agonist serial dilutions. IP₁ and cAMP accumulations were then measured as described above.

RT-PCR. We performed RT-PCR on RNA extracts of COS-7 cells (which are derived from Chlorocebus aethiops, African green monkey) to detect the suspected expression of endogenous AVP V₂ receptor mRNA. Forward (5′-GCTAGTGATTGTGGTCGTCTATGTGC-3′) and reverse (5′-CACGCTGCTGCTGAAAGATGC-3′) primers were designed based on the human V₂ receptor sequence to amplify a 178-bp band from the three primate V₂ receptor sequences available to date (Homo sapiens, Pan troglodytes, Macaca mulatta). An intron lies between the two primer sites to ensure the rejection of undesired amplification of any residual genomic DNA. RT-PCR was performed as described above.

Radioligand Binding Assays to Intact Cells. The first binding assay was carried out with [³H]AVP to determine the amount of functional receptor present at the plasma membrane of transfected CHO and HEK-293 cells. After electroporation, cells were seeded in 24-well plates at a density of 4 × 10⁵ cells/well. Binding assays were performed after 24 h. Cells were washed twice in ice-cold Tris/Krebs buffer (20 mM Tris-HCl pH 7.4, 118 mM NaCl, 5.6 mM glucose, 1.2 mM KH₂PO₄, 1.2 mM MgSO₄, 4.7 mM KCl, and 1.8 mM CaCl₂) with 0.1% bovine serum albumin (BSA), after which each well received 0.2 ml of ice-cold Tris/Krebs buffer + 0.1% BSA + 5 mM phenylalanine and the appropriate dilution of [³H]AVP (0.25 to 8 nM). Plates were incubated for 4 h in the cold room before removal of the binding mixture by aspiration. After quickly rinsing three times with ice-cold Krebs buffer + 0.1% BSA, 0.3 ml of 0.1 N NaOH was added to each well to lyse the cells and extract the radioactivity. The fluid from the wells was then neutralized, transferred to scintillation vials, and counted on a β-counter (Tri-Carb 2100TR; PerkinElmer Life and Analytical Sciences). Nonspecific binding was determined under the same conditions in the presence of 10 μM unlabeled AVP. Protein levels were determined using the BCA protein assay (Pierce Biotechnology, Rockford, IL). The experiment was repeated three times on different cultures. Maximal binding capacity and dissociation constants were determined from the Scatchard transformation of the saturation binding curves.

The second binding assay was done to determine, through a competition experiment, the inhibitory dissociation constant (K_i) of arginine-vasotocin (AVT) for the rat AVP V_1b receptor. Cells were incubated with a fixed concentration of [³H]AVP (2 nM) that was displaced using increasing concentrations of AVT (10^-12 to 10^-6 M). The experiment was repeated three times on different cultures. Displacement curves were fitted to determine the IC₅₀ values, which were then converted to K_i using the Cheng-Prusoff equation: K_i = IC₅₀/(1 + L/K_d), where L represents the concentration of [³H]AVP, and K_d represents the dissociation constant of AVP for the rat V_1b receptor in intact CHO cells determined as indicated above.

Receptor Dimerization at the Plasma Membrane. Dimerization was measured using an HTRF-FRET assay. Monoclonal anti-6×His and/or anti-HA antibodies labeled with a europium cryptate pyridine-bipyridine donor fluorophore or a d2 acceptor fluorophore (Cisbio Bioassays) were used. After transfection, 10⁵ COS-7 cells/well were transferred to a black 96-well assay plate (Greiner Bio-One) in 100 μl of complete medium. Twenty-four hours later, COS-7 cells expressing the 6×His- or HA-tagged rat receptors were incubated at 4°C in 100 μl of Tris/Krebs buffer supplemented with 0.1% BSA and containing 1 nM europium cryptate pyridine-bipyridine- and 1 nM d2-labeled antibodies. To obtain the negative control value, some COS-7 cells were incubated with the donor fluorophore-labeled antibody only. Fluorescence intensities were measured at 620 and 665 nm and the FRET signal (ΔF%), reflecting dimerization, calculated as described above. The experiments were repeated at least three times on different cultures.

Statistical Analysis. Statistical significance of the differences between experimental groups was determined by one-way or two-way analysis of variance followed by a post hoc Duncan's multiple range test to make pairwise comparisons between means. Student's t test was also used.

Results

The Rat V_1b Receptor Signals through Both the IP and cAMP Pathways. AVP stimulation of CHO, HEK-293, and COS-7 cells transfected with 1 μg of rat V_1b receptor-encoding plasmid led to the activation of both the IP and cAMP pathways measured as accumulation of IP₁ and cAMP, respectively (Fig. 1). The responses were specific for V_1b receptor expression, because mock CHO and HEK-293 cells (transfected with an empty pcDNA3 vector) exhibited no response to AVP (Fig. 1, A and B). Similar patterns of IP₁ and cAMP accumulation were observed with transiently transfected COS-7 cells (Fig. 1C), despite the presence of endogenous V₂ receptor, which was responsible for approximately 20% of the cAMP production as demonstrated in mock cells. The presence of this endogenous receptor was confirmed by RT-PCR performed using primers specific for primate V₂ receptor sequences, revealing the endogenous V₂ receptor mRNA in COS-7 cell extracts (data not shown).

The dual coupling was not due to overexpression of V_1b receptors (448 × 10³ receptors/cell compared with 36 × 10³ receptors/cell in corticotrophs of the adenohypophysis; Gaillard et al., 1984) because it was still observed when lowering V_1b receptors expression (31 × 10³ receptors/cell) to a level similar to that of the physiological one (Supplemental Fig. 1).

Differential Activation of IP and cAMP Pathways after AVP Stimulation. Although AVP stimulated both the IP and cAMP signaling pathways, it is worth noting that in the three cell lines tested AVP stimulated the accumulation of IP₁ at lower concentrations than that of cAMP (Fig. 1, A-C). Specifically, as shown in Table 1, the EC₅₀ values for the IP accumulation, which were very close to the K_d values, were consistently three to five times lower than those obtained for the cAMP accumulation. This has been observed whatever the V_1b receptor expression level. Moreover, in both CHO and HEK-293 cells, where no endogenous V₂ receptor disturbs the response, the Hill coefficients of the activation curves were lower for the IP pathway than for the cAMP pathway. Reliable similar analyses of COS-7 cell parameters were not possible because of the presence of endogenous V₂ receptor.

To confirm that the cAMP pathway is directly stimulated by the V_1b receptor, we examined the involvement of G_s. Specifically, we treated transfected CHO cells overnight with Ctx to induce direct and sustained stimulation of G_s. As expected, under these conditions, cAMP accumulation was much higher, but AVP had no additive effect (Fig. 2A). This absence of effect of AVP under Ctx treatment was not due to saturation of adenylyl cyclase activity, because simultaneous treatment of the cells with Ctx and forskolin further increased cAMP accumulation (by 53.0 ± 4.2 times; data not shown). It can thus be inferred from these two observations that the stimulation of the V_1b receptor by AVP requires the G_s protein to activate the cAMP pathway. By contrast, the V_1b receptor did not stimulate the G_i or G_o protein, because an overnight treatment with Ptx did not affect the AVP-induced cAMP accumulation (Fig. 2B). As a positive control, such a treatment inhibited the serotonin-induced response of the G_i-coupled human 5HT_1a receptor (Supplemental Fig. 2).

AVP Analogs Activate the IP and cAMP Pathways in Different Manners. To address whether the two signaling pathways exhibit different pharmacological properties, we examined the effects on both pathways of one specific nonpeptidic V_1b antagonist, SSR149415 (Serradeil-Le Gal et al., 2002), and two peptidic agonists of AVP, arginine-vasotocin (AVT), the nonmammalian vertebrate ortholog of AVP, and d[Cha⁴]AVP, a specific V_1b receptor agonist (Guillon et al., 2006).

IP₁ and cAMP accumulations in V_1b-transfected CHO cells stimulated with increasing AVP concentrations was studied without or with increasing amounts of SSR149415. The resulting Arunlakshana-Schild plots showed that both the IP and cAMP pathways were turned off (Fig. 3). It is noteworthy that the antagonist affected the cAMP accumulation at lower concentrations (pA₂ = 9.99; i.e., K_inact = 0.10 nM) than the IP₁ accumulation (pA₂ = 9.14; i.e., K_inact = 0.72 nM) (Fig. 3; Table 2).

Although the two agonists tested, AVT and d[Cha⁴]AVP, stimulated both pathways, the curves differed from those obtained with AVP (Fig. 4). As with AVP, AVT activated both the IP and cAMP pathways, but did it with lower EC₅₀ values than AVP (Fig. 4A; Table 2). Again, it is worth noting that the EC₅₀ of the IP₁ accumulation curve was lower than that of the cAMP one and was very close to the affinity of AVT for the V_1b receptor measured in binding experiments (Table 2). The stimulation of IP signaling by AVT was similar to that by AVP with respect to the slopes of the curves (Table 2) and the maximal stimulation values (Table 2). By contrast, cAMP stimulation by AVT differed from that by AVP in that it had a steeper slope (Fig. 4A; Table 2) and showed a 36% higher level of maximal stimulation (Table 2). AVT is thus a more potent and more efficacious agonist than AVP on the cAMP pathway.

d[Cha⁴]AVP also activated both the IP and cAMP pathways, but did not show any differences in the activation of the two pathways, and the dose-response curves for IP₁ and cAMP accumulation were superimposed (Fig. 4B). In both cases, high concentrations of d[Cha⁴]AVP were necessary to elicit the response (Fig. 4B; Table 2), even though the affinity of this compound for the rat V_1b receptor is known to be 1.4 nM in CHO cells (Guillon et al., 2006). Both curves exhibited a slope factor of around 1 and lower maximal stimulation levels than those observed with AVP (-18 and -59% for the IP and cAMP pathways, respectively; Table 2). Therefore, d[Cha⁴]AVP was shown to be a partial agonist of the IP and cAMP pathways compared with AVP or AVT.

Dimers of Rat V_1b Receptors Detected at the Plasma Membrane. Slope factors greater than 1 (1.67 for AVT on cAMP accumulation) or less than 1 (AVP and AVT on IP₁ accumulation) were observed. Among other hypotheses, such as binding site heterogeneity of the ligand for the receptor and/or of the ligand-receptor complexes for G proteins, these slope factors different from 1 strongly suggested the existence of cross-talk between receptors, dimers, or oligomers. Because cross-talk between receptors has been related to dimerization in various reports (Terrillon and Bouvier, 2004; Park and Palczewski, 2005; Milligan, 2006; Urban et al., 2007), we tested whether the rat V_1b receptor is able to dimerize at the plasma membrane. DNA encoding the rat V_1b receptor, tagged with the 6×His epitope at the extracellular N terminus, was transiently transfected into COS-7 cells and HTRF technology used to quantify dimers at the cell surface. As a positive control for dimerization, COS-7 cells from the same culture were transfected with a rat V_1a receptor that had similarly been tagged with the 6×His epitope. In both cases, a FRET signal was observed for the V_1a receptor (ΔF = 236%) and the V_1b receptor (ΔF = 128%) (Fig. 5A) when expressed at similar levels (Fig. 5B). The specificity of the FRET signal was demonstrated by 1) the absence of signal in mock cells; 2) the relationship between the amount of V_1b receptor expressed at the membrane and the FRET signal: SSR149415 (1 μM, overnight), by acting as a pharmacochaperone on the receptor, thus facilitating its targeting to the cell membrane (Robert et al., 2005), increased membrane V_1b receptor by 61% (measured by ELISA; Fig. 5, B and D) and the FRET signal by 57% (Fig. 5, A and C); and 3) the weak FRET signal observed (ΔF = 16%) between an HA-tagged human GABA_B2 receptor (a member of the GPCR C family) and the 6×His-tagged V_1b receptor compared with the signal obtained with the HA-V_1b and 6×His-V_1b receptors (71%) (Fig. 5C). The weak signal was not due to low HA-GABA_B2 receptor expression, because the HA-GABA_B2 and HA-V_1b receptors were expressed at similar levels (Fig. 5D). Instead, it probably resulted from the simple promiscuity of the receptors.

Lowering Plasma Membrane Cholesterol Modifies V_1b Receptor Signaling. The OT receptor has been shown to activate both the G_q/11 and G_i pathways, and these dual properties are dependent on the localization of the receptor to specialized domains of the plasma membrane (Gimpl and Fahrenholz, 2000; Guzzi et al., 2002; Rimoldi et al., 2003). We thus decided to investigate whether the activation levels of the two signaling pathways triggered by stimulating the V_1b receptor would be similarly affected by alterations in receptor localization within different membrane compartments. Because we had observed that AVP-induced IP₁ accumulation was quite similar in the CHO and HEK-293 cell lines, whereas AVP-induced cAMP accumulation differed strongly between the two cell types, we compared the effects of MβCD, a compound known to rapidly remove more than half of the cholesterol from the plasma membrane (Kilsdonk et al., 1995), on these two cell lines.

CHO cells expressing the rat V_1b receptor were incubated with 10 mM MβCD. Under these conditions, the maximal AVP-induced IP₁ accumulation was significantly reduced (-30% at maximal effect), whereas the maximal cAMP accumulation was increased (+10% at maximal effect) (data not shown).

The experiments were then repeated in HEK-293 cells transfected with the V_1b receptor, in which only a low level of cAMP accumulation by AVP was initially observed (Table 1). When these cells were stimulated with AVP in the presence of 10 mM MβCD, IP₁ accumulation fell by 32% at maximal effect, similar to what was observed in CHO cells (Fig. 6A; Table 3). The cAMP accumulation, in contrast, showed a much higher level of activation (+234% at saturation) than it did in the absence of MβCD (Fig. 6B; Table 3). The MβCD treatment of HEK-293 cells did not change significantly the EC₅₀ of the AVP-induced IP₁ accumulation (Table 3). It is noteworthy that the EC₅₀ of the AVP-induced cAMP accumulation shifted significantly from 32.7 to 13.1 nM after MβCD treatment. The same effects on IP₁ (-31%) and cAMP (+265%) accumulations were obtained when V_1b receptor-transfected HEK-293 cells were stimulated with AVT instead of AVP (Fig. 6, C and D; Table 3). When d[Cha⁴]AVP was used to stimulate the cells, cAMP accumulation was also highly increased by MβCD treatment; as in CHO cells, activation of both pathways were observed at higher similar concentrations (Fig. 6, E and F; Table 3). Together, these results clearly indicated that compartmentalization of the V_1b receptor in cholesterol-rich specialized domains modulates its G protein-dependent signalization.

Finally, to investigate the possibility that treatment of the cells with MβCD directly affected adenylyl cyclase activity, wild-type HEK-293 cells were stimulated with forskolin with or without treatment with MβCD. MβCD did not change the level of cAMP accumulation (Fig. 6G).

Discussion

We observed dual signaling activity for the V_1b receptor; it activated both the IP pathway, as expected (Jard et al., 1986), and the cAMP pathway through coupling to the G_s protein. Coupling to G_i, in contrast, was not observed. These results were not cell-line-dependent, in that they were obtained in the three different cell lines tested. Moreover, at variance with previously reported results (Thibonnier et al., 1997), we observed that the dual signaling of the V_1b receptor is not restricted to cells in which the receptor is overexpressed (>25 pmol/mg protein), but is also observable at receptor densities in the same range as those observed in native tissues. This dual signaling was observed with various agonists, albeit with some notable differences in the activation curves. For example, although the EC₅₀ values for AVP or AVT activation of the IP pathway were very close to their binding affinities, for cAMP pathway activation, the EC₅₀ values were three to five times higher. By contrast, the EC₅₀ values of d[Cha⁴]AVP for the IP and cAMP pathways were consistently higher than the affinity constants. Dual signaling has been observed previously for many GPCRs, and in some cases, such as the corticotropin-releasing hormone receptor or the luteinizing hormone receptor, coupling to even three G proteins has been described previously (Hermans, 2003). In most cases, and as observed here also, the coupling occurs at different concentrations (Ashkenazi et al., 1987). In rare cases, such as here with d[Cha⁴]AVP, activation of the two pathways occurs at the same ligand concentration (Cussac et al., 2002). Regarding the AVP/OT receptor family, dual G protein coupling has been reported for the V_1a (Abel et al., 2000) and OT receptors (Strakova and Soloff, 1997), both of which couple to G_q/11 and G_i. In addition, the V₂ receptor possibly activates G_s and G_q/11 (Liu and Wess, 1996).

Although the concomitant activation of different signaling pathways has been well established, the molecular mechanisms underlying differential coupling remain unclear. A few hypotheses have been formulated to explain this phenomenon. Dual signaling and a shift between two activation curves may be explained by a difference in the affinity of the ligand-receptor complex for each G protein, or by a difference in the relative concentrations of the G proteins (Kenakin, 2003). In addition, cross-talk between ligands, receptors, and G proteins can influence the conformations of the different partners and lead to significant differences in receptor signaling (Kenakin, 2003).

The first and probably simplest hypothesis is that the affinities of the receptor-G protein complex for the ligand are dependent on the nature of the G protein. For example, stimulation of the IP pathway by AVP and AVT could result from the binding of the ligand to the high-affinity binding sites, whereas the cAMP pathway could be activated by the binding of the ligand to the low-affinity binding sites. However, this hypothesis is not completely in accordance with different data: 1) the curves of IP₁ accumulation exhibit slope factors of around 0.7 to 0.8, indicating either negative cooperative binding or the coexistence of independent high- and low-affinity binding sites that both couple to the IP signaling pathway; 2) the d[Cha⁴]AVP-induced accumulation of both IP₁ and cAMP is shifted to the right of the binding curve, suggesting that the high-affinity binding sites are devoid of any activity and that low-affinity binding sites are involved in both coupling pathways; and 3) the AVT-induced cAMP pathway activation curves are steep, indicating a positive cooperative effect.

Because positive and negative cooperative effects have been related to receptor dimerization (Durroux, 2005; Urizar et al., 2005; Springael et al., 2006), we investigated the ability of rat V_1b to undergo oligomerization. The AVP V_1a and V₂ receptors, as well as the OT receptor, have previously been shown to homo- and heterodimerize (Terrillon et al., 2003). Using the HTRF method and conjugated antibodies against extracellular epitopes, we have now been able to detect homodimers of V_1b receptors at the plasma membrane of transfected cells. Because saturation and homologous competition experiments performed on the human V_1b receptor using radioligands have clearly demonstrated negative cooperative binding (Albizu et al., 2006), the activation of the two signaling pathways could be related to ligand-receptor stoichiometry. Moreover, as with other GPCRs, changes in V_1b receptor conformation upon agonist binding may differ considerably with respect to the ligand structure, the cooperativity between the two protomers in the dimer, their affinities for different G proteins, and/or their abilities to activate bound G proteins (Michel and Alewijnse, 2007; Urban et al., 2007). In accordance with the receptor dimerization hypothesis, the IP pathway could be activated by the binding of either one or two AVP molecules to a single receptor dimer. Because of the negative cooperative binding, the activation curve would thus exhibit a slope of less than 1. By contrast, cAMP production could only be turned on upon the binding of two ligands to a dimer. Because of the negative cooperative binding, the cAMP activation curve would be observed only during the saturation of the low-affinity binding sites and would thus exhibit a slope factor of around 1; it would therefore be shifted to the right of the IP activation curve. Contrary to AVP and AVT, two d[Cha⁴]AVP molecules would need to bind to the receptor to induce an active receptor conformation, whatever the signaling pathway considered. Consistent with these hypotheses, the SSR149415 antagonist shuts down the cAMP pathway as soon as one AVP ligand is displaced but does not inactivate the IP pathway until both ligands are removed.

GPCR signaling has been shown to be related to the compartmentalization of receptors or G proteins within specialized microdomains of the plasma membrane. For example, the OT receptor is able to signal through G_q/11 or G_i depending on its localization within or outside of cholesterol-rich microdomains of the cell surface (Guzzi et al., 2002; Rimoldi et al., 2003). To address whether the interactions between V_1b receptors and G proteins are random interactions or whether they are related to compartmentalization, we investigated the effects of MβCD, a compound that rapidly removes more than half of the cholesterol content of the plasma membrane and disrupts rafts and caveolae (Kilsdonk et al., 1995). Previous work has shown that MβCD treatment does not prevent cAMP signaling (Miura et al., 2001); indeed, we found that it had no effect on forskolin-induced cAMP production. However, it did lower the maximal IP₁ accumulation and increased the maximal cAMP accumulation. Using AVT as the stimulating ligand in HEK-293 cells led to the same modifications. These modifications observed upon either AVP or AVT stimulation during MβCD treatment probably reflect the number of receptors available for each pathway. The difference between CHO and HEK-293 cells with respect to cAMP accumulation after MβCD treatment can most probably be attributed to a different compartmentalization and/or concentration of the transduction partners involved in the response to AVP, as previously reported for the β₂-adrenergic receptor (Huang et al., 2007) or the δ-opioid receptor (Ostrom et al., 2002). In fact, various interpretations can be proposed to explain the results: 1) the V_1b receptor could be localized to cholesterol-rich domains, where it could preferentially couple with G_q/11 and activate the IP pathway upon AVP or AVT binding, and/or to other (non-cholesterol-rich) domains, where it could interact with G_s and signal through the cAMP pathway; 2) G_s or other proteins of the cAMP signaling cascade could be sequestered to cholesterol-rich domains, and the disruption of these microdomains with MβCD could increase the available concentration of the sequestered proteins, facilitating the competition between the usual G_q coupling of the V_1b receptor and the newly available G_s coupling (Pontier et al., 2008).

The presence of V_1b receptors has now been demonstrated in various brain regions. Their function has been recently highlighted. Indeed, the only currently available specific antagonist, SSR149415, was shown to exert anxiolytic and antidepressant effects in animal models (Serradeil-Le Gal et al., 2002). However, nothing is known about how the V_1b receptor exerts its effects in the brain. The present work increases our understanding of V_1b pharmacology by showing that the receptor can activate two different signaling pathways depending on its localization within the plasma membrane. Furthermore, the observation that the relative intensities of agonist effects on the two signaling pathways vary strongly with their molecular structure opens new perspectives to the design of ligands that could be used to selectively activate or inhibit one or the other pathway and thereby specifically modulate particular V_1b receptor-regulated functions within the central nervous system.