The Oxytocin Receptor Antagonist Atosiban Inhibits Cell Growth via a “Biased Agonist” Mechanism*

In human myometrial cells, the promiscuous coupling of the oxytocin receptors (OTRs) to Gq and Gi leads to contraction. However, the activation of OTRs coupled to different G protein pathways can also trigger opposite cellular responses, e.g. OTR coupling to Gi inhibits, whereas its coupling to Gq stimulates, cell proliferation. Drug analogues capable of promoting a selective receptor-G protein coupling may be of great pharmacological and clinical importance because they may target only one specific signal transduction pathway. Here, we report that atosiban, an oxytocin derivative that acts as a competitive antagonist on OTR/Gq coupling, displays agonistic properties on OTR/Gi coupling, as shown by specific 35S-labeled guanosine 5′-3-O-(thio) trisphosphate ([35S]GTPγS) binding. Moreover, atosiban, by acting on a Gi-mediated pathway, inhibits cell growth of HEK293 and Madin-Darby canine kidney cells stably transfected with OTRs and of DU145 prostate cancer cells expressing endogenous OTRs. Notably, atosiban leads to persistent ERK1/2 activation and p21WAF1/CIP1 induction, the same signaling events leading to oxytocin-mediated cell growth inhibition via a Gi pathway. Finally, atosiban exposure did not cause OTR internalization and led to only a modest decrease (20%) in the number of high affinity cell membrane OTRs, two observations consistent with the finding that atosiban did not lead to any desensitization of the oxytocin-induced activation of the Gq-phospholipase C pathway. Taken together, these observations indicate that atosiban acts as a “biased agonist” of the human OTRs and thus belongs to the class of compounds capable of selectively discriminating only one among the multiple possible active conformations of a single G protein-coupled receptor, thereby leading to the selective activation of a unique intracellular signal cascade.

In human myometrial cells, the promiscuous coupling of the oxytocin receptors (OTRs) to G q and G i leads to contraction. However, the activation of OTRs coupled to different G protein pathways can also trigger opposite cellular responses, e.g. OTR coupling to G i inhibits, whereas its coupling to G q stimulates, cell proliferation. Drug analogues capable of promoting a selective receptor-G protein coupling may be of great pharmacological and clinical importance because they may target only one specific signal transduction pathway. Here, we report that atosiban, an oxytocin derivative that acts as a competitive antagonist on OTR/G q coupling, displays agonistic properties on OTR/G i coupling, as shown by specific 35

S-labeled guanosine 5-3-O-(thio) trisphosphate ([ 35 S]GTP␥S) binding. Moreover, atosiban, by acting on a G i -mediated pathway , inhibits cell growth of HEK293 and Madin-Darby canine kidney cells stably transfected
with OTRs and of DU145 prostate cancer cells expressing endogenous OTRs. Notably, atosiban leads to persistent ERK1/2 activation and p21 WAF1/CIP1 induction, the same signaling events leading to oxytocin-mediated cell growth inhibition via a G i pathway. Finally, atosiban exposure did not cause OTR internalization and led to only a modest decrease (20%) in the number of high affinity cell membrane OTRs, two observations consistent with the finding that atosiban did not lead to any desensitization of the oxytocin-induced activation of the G q -phospholipase C pathway. Taken together, these observations indicate that atosiban acts as a "biased agonist" of the human OTRs and thus belongs to the class of compounds capable of selectively discriminating only one among the multiple possible active conformations of a single G protein-coupled receptor, thereby leading to the selective activation of a unique intracellular signal cascade.
Oxytocin, a nonapeptide secreted by the neurohypophysis, exerts its biological effects by binding to and activating the oxytocin receptor (OTR), 1 a membrane receptor belonging to the G protein-coupled receptor family (1). As recently reviewed by Gimpl (2), OTRs were originally described in the uterus, mammary gland, and central nervous system. In the uterus, the level of OTR expression increases during pregnancy and peaks immediately before the onset of labor, when receptor activation promotes uterine contractility. The OTRs in the mammary gland are located on myoepithelial cells, where they stimulate contractility and promote milk ejection. In the central nervous system, where they modulate hippocampal synaptic plasticity during pregnancy (3), OTRs are involved in regulating a number of reproductive and non-reproductive behaviors.
OTRs have more recently been detected in a vast number of tumor cells of various origin (breast and endometrial carcinomas, neuroblastomas, glioblastomas, Kaposi sarcomas, smallcell lung carcinomas), and their stimulation variably affects cell growth (4). Moreover, it has recently been suggested that OTRs are involved in the differentiation of cardiomyocytes (5) and myoblasts (6).
Despite the biological relevance of OTR signaling, the identity of signaling pathways activated to accomplish these different functions has only been partially clarified. In particular, although human OTRs have been shown to associate to G q , G i, and, to a minor extent, to G s (7), only the signaling events induced by G q activation have been extensively investigated and exploited to develop drugs already entered in the clinical use. In myometrial and breast myoepithelial cells, OTR coupling to G q is responsible for phospholipase C (PLC) activation followed by inositol phosphate (InsP) and diacylglycerol production, increased intracellular calcium, and increased contractility (8). Accordingly, selective oxytocin analogues capable of blocking this pathway are employed in the treatment of preterm labor (9). The physiological role played by OTRs coupled to G i in myometrial cells is far less clear, although it was shown some years ago to also lead to an increase of cell contractility (10). Recently, a specific role for OTR coupled to G i has emerged in MDCK and HEK293 cells stably transfected with the human OTR, where OT inhibits cell growth in a pertussis toxin (PTx)-sensitive manner, thus suggesting a key role of OTR-G i coupling in mediating an anti-proliferative effect (11).
A review of our previous work indicated that atosiban, an OTR antagonist that is currently used in the treatment of preterm labor because of its ability to block the G q /PLC/calcium signaling pathway in myometrial cells (12)(13)(14)(15), inhibits the proliferation of some cancer cells (16). As we have recently shown that the growth of MDCK and HEK293 cells is inhibited via G i -coupled OTRs (11), we hypothesized that atosiban inhibited cell growth via a biased agonist mechanism by selectively promoting OTR coupling to G i . Biased agonists have been defined as analogues displaying, at a single promiscuous GPCR, agonist properties at one coupling pathway and antagonist properties at another one (17,18). To check the biased agonist properties of atosiban at the human OTR, we examined atosiban signaling in MDCK and HEK293 cell clones stably transfected with human OTRs, as well as in DU145 human prostate cancer cells endogenously expressing OTRs.
Cell Growth Assay-Cells were seeded in 48-multiwell plates (10,000 cells/well) and allowed to adhere for 24 -48 h. Peptides were then added to the culture medium at the specified concentrations for up to 72 h. DU145 and MDCK cells were then fixed in 2.5% (v/v) glutaraldehyde, stained with 0.1% (w/v) crystal violet in 20% (v/v) methanol, and solubilized in 10% (v/v) acetic acid prior to cell growth evaluation through measurement of absorbance at 590 nm in a microplate reader (Wallac Victor2, PerkinElmer Life Sciences). Cell growth of HEK293 was determined by a methanethiosulfonate-based assay (Promega CellTiter 96® AQueous one solution assay) according to the manufacturer's instructions. All treatments were performed in quadruplicate. Equal numbers of cells were assayed, and a linear correlation between absorbance and cell counts was established up to 1 ϫ 10 5 cells. Statistical analysis was carried out by Student's t test.
[ 35 S]GTP␥S Binding Studies-[ 35 S]GTP␥S binding studies were carried out on transiently transfected cells. HEK293 cells, plated in 10-cm dishes at a 50% confluence, were transfected with 30 g of a pEGFP-N3 Clontech plasmid encoding human OTR⅐EGFP using calcium phosphate co-precipitation. Forty-eight h after transfection, cells were collected in PBS without Ca 2ϩ and Mg 2ϩ (PBS Ϫ ), centrifuged, and resuspended in 50 mM Tris-HCl, pH 7.4, supplemented with 2 mM EDTA and a protease inhibitor mixture (Sigma). After cell rupture with a glass homogenizer and removal of nuclei by low speed centrifugation (800 ϫ g), cell extracts were ultracentrifuged at 100,000 ϫ g for 1 h. The resulting pellets were resuspended in 20 mM HEPES buffer, pH 7.4, containing 5 mM Mg 2 Cl and 100 mM NaCl, to a final concentration of 1-2 mg/ml and stored at Ϫ80°C. When required, cells were exposed to PTx (Sigma) for 16 h at a final concentration of 150 ng/ml before harvesting. mM Mg 2 Cl) and filtration through Whatman GF/C filters followed by three quick washes with washing buffer. After air-drying filters and adding 5 ml of Ultima Gold TM XR scintillation fluid (PerkinElmer Life Sciences), the bound radioactivity was determined by liquid scintillation counting.
Inositol Phosphate Determination-Inositol phosphate (InsP) accumulation was measured as described previously (20,21). Briefly, cells grown in 6-well dishes were labeled for 24 h with myo-[2-3 H]inositol at a final concentration of 2 Ci/ml in a serum-and inositol-free medium (Invitrogen). The cells were washed twice in Krebs buffer (146 mM NaCl, 4.2 mM KCl, 0.5 mM MgCl 2 , 1.0 mM CaCl 2 , 10 mM HEPES base, 1% (w/v) glucose, pH 7.4) and preincubated for 10 min at 37°C in the same buffer supplemented with 10 mM LiCl. After incubation for 15 min in the absence (baseline) or presence of peptide at a final concentration 10 Ϫ7 M, the reaction was stopped with 5% (v/v) perchloric acid, and the [ 3 H]InsPs were extracted and separated using a strong anionic exchange column (Dowex AG1X8, formate form, 200 -400 mesh; Bio-Rad). The fraction containing the [ 3 H]inositol mono-, bis-, and tris-phosphates was collected, and its radioactivity content was determined by liquid scintillation counting. This fraction is referred to as total InsP and expressed as DPM/well.
cAMP Accumulation Assay-Cells grown in 10-cm Petri dishes for 48 h were washed twice with PBS Ϫ and incubated for 15 min at 37°C in PBS Ϫ supplemented with 4 mM EDTA. After scraping with a rubber policeman, cells were centrifuged at low speed, and resulting pellets were resuspended at a density of 10 6 cells/90 l in PBS Ϫ supplemented with 5.5 mM 3-isobutyl-1-methylxanthine. Cell suspensions (total volume of 90 l) were equilibrated at 37°C for 15 min and further incubated with peptides at a final concentration of 10 Ϫ7 M, or buffer (basal accumulation), for additional 10 min. cAMP accumulation was stopped by placing the tubes in liquid nitrogen and boiling for 5 min. The samples were then centrifuged for 8 min at 12,000 rpm in an Eppendorf minifuge, and the supernatants were immediately used to measure cAMP using a competitive binding assay (cyclic AMP [ 3 H] assay kit, Amersham Biosciences) according to the manufacturer's instructions.
Preparation of Cell Lysates and Western Blot Analysis-After 1 h of starvation in serum-free medium, cells were stimulated with OT or atosiban at the final concentration of 10 Ϫ7 M for the indicated periods of time. Cells were then washed once in ice-cold PBS Ϫ and lysed in 50 mM Tris-HCl, pH 6.8, containing 2% (w/v) SDS, preheated at 100°C. After 4 -5 cycles of freezing in dry ice and boiling for 2 min, aliquots of the lysates were assayed for protein content using the BCA protein assay reagent (Pierce). Cellular proteins (30 g) were resolved by Laemmli SDS-PAGE system using 11% acrylamide (45) and then blotted onto nitrocellulose membranes (Amersham Biosciences). Blots were incubated overnight at 4°C in Tris-buffered saline (20 mM Tris-HCl, pH 7.4, 150 mM NaCl), containing 5% (w/v) skimmed milk. The membranes were then incubated for 2 h with primary antibodies diluted in Trisbuffered saline/milk and for 2 h with horseradish peroxidase-conjugated goat anti-mouse/rabbit IgG (Pierce). When phospho p42/44 MAPK antibody was used, blocking was performed in 5% (w/v) bovine serum albumin/0.2% (v/v) Tween 20 for 1 h at 37°C, and incubation was performed overnight in Tris-buffered saline/bovine serum albumin at 4°C. Proteins were detected using the SuperSignal® chemiluminescent substrate (Pierce). For quantification, unsaturated bands were acquired by means of an Arcus II Scanner (Agfa-Gevaert) and analyzed with the NIH Image program Version 1.61 (National Technical Information Service, Springfield, VA).
Internalization Assays-Cells were washed twice in serum-free medium and left for 30 min to equilibrate at 37°C. Peptides were then added at a final concentration of 10 Ϫ7 M. At fixed time intervals (0, 5, 15, 30, and 60 min), cells were processed for fluorescence microscopy or binding assays.
For fluorescence microscopy, cells grown on glass coverslips placed in 6-well dishes were washed twice with 10 mM sodium phosphate buffer, pH 7.4, containing 150 mM NaCl and fixed for 20 min at room temperature with 4% (w/v) paraformaldehyde in PBS. Fixed cells were rinsed with LS buffer and mounted on glass slides with 90% (v/v) glycerol in PBS. Slides were observed under an MRC1024 Bio-Rad laser scanning confocal microscope.
To determine cell surface binding, cells subcultured into 24-well dishes were washed twice with ice-cold binding buffer (146 mM NaCl, 4.2 mM KCl, 0.5 mM MgCl 2 , 1.0 mM CaCl 2 , 10 mM HEPES base, 1% (w/v) glucose, 0.018% (w/v) L-tyrosine, 1% (w/v) bovine serum albumin, pH 7.4) and placed on ice.[ 3 H]OT at a final concentration of 4 nM was added to the wells with or without 10 Ϫ7 M unlabeled OT in a final volume of 200 l. After incubation at 4°C for 2 h, a time required to reach the equilibrium, cells were washed three times with binding buffer to remove the unbound radioactivity and then solubilized with 0.5 N NaOH. The samples were transferred to scintillation vials and counted in a ␤ counter after the addition of 3.5 ml of scintillation fluid. Specific surface binding at each time was calculated as percentage of specific binding at time 0.

Atosiban-induced Inhibition of MDCK and HEK29 Cells
Expressing Human OTRs-To analyze the ability of atosiban to inhibit the growth of cells expressing the human OTR, we checked its effects on MDCK and HEK293 cells stably transfected with the human OTR fused to EGFP (11,19). As shown in Fig. 1 (A and D), cell growth was significantly inhibited after 48 h of treatment with OT and atosiban. This effect was concentration-dependent with calculated IC 50 values of 20.8 Ϯ 0. Signaling Pathways Involved in the Atosiban-mediated Inhibition of Cell Growth-As it is known that OTRs couple to G q and G i (7) and that receptor coupling to G i is responsible for the OT-induced inhibition of HEK293 cell growth (11), we investigated the signaling events leading to the atosiban-induced inhibition of cell growth, and particularly, whether G i is involved. To determine whether atosiban could directly promote OTR coupling to heterotrimeric G proteins, we measured its ability to promote the binding of [ 35 S]GTP␥S, according to the paradigm that receptor activation promotes the GDP/GTP exchange on G protein ␣ subunits. This assay utilizes [ 35 S]GTP␥S resistance to hydrolysis, its high affinity for the G␣ subunits, and its relatively high specific activity to determine receptorinduced G␣ activation (22)(23)(24). As shown in Fig. 2A  These data indicate that atosiban is capable of promoting OTR coupling to G protein, thus displaying agonist properties toward the human OTR. In addition, in OTR-expressing cells pretreated with PTx, the capability of OT and atosiban to increase [ 35 S]GTP␥S binding was completely abolished, indicating a coupling to G i -like proteins. The lack of any residual [ 35 S]GTP␥S binding even in membranes treated with OT indicates that, in our experimental conditions, OTR coupling to G q did not significantly contribute to total [ 35 S]GTP␥S binding. We then checked the ability of atosiban to modulate proliferation of PTx-treated cells (Fig. 2B); under these conditions, the atosiban-induced inhibition of cell growth was completely abolished, thus indicating that a PTx-sensitive G protein of the G i family mediates the inhibitory effect.
As we have shown that the OT-induced growth inhibition of MDCK and HEK293 cells is associated with a long-lasting activation of MAPK (11), we investigated whether atosiban can also induce MAPK phosphorylation. To this end, we applied atosiban for an increasing length of time and analyzed ERK1/2 activation by means of Western immunoblotting of cell lysates probed with phospho-specific anti-ERK1/2 antibody. As shown in Fig. 2C, atosiban induced a persistent increase in ERK1/2 phosphorylation, a pattern consistent with inhibition of cell growth, as demonstrated previously in HEK293 and MDCK cells (11).
We then verified whether atosiban can promote OTR coupling to G q by measuring the atosiban-induced accumulation of total InsPs; as shown in Fig. 2D, whereas OT caused a concentration-dependent enhancement in the production of InsPs, no effect was observed after atosiban treatment, even at very high concentrations, thus indicating that atosiban is completely devoid on any agonist properties on OTR/G q coupling. In add-  's t test). B, effect of PTx on atosiban-induced inhibition of cell growth. Atosiban-induced effects on cell growth were determined in cells pretreated or not with PTx. In cells treated with PTx, the atosiban-induced variations in cell number was determined with respect to control cells maintained in the presence of PTx for the complete duration of the experiment. C, MAPK activation by OT and atosiban. The cells were treated with 10 Ϫ7 M OT or atosiban for the indicated periods of time. Then cells were lysed, and 30 g of proteins/lane were subjected to separation by SDS-PAGE (11% acrylamide), Western immunoblotting with polyclonal anti-phosphoERK1/ERK2 antibody, and chemiluminescence detection. A caveolin-1 (cav-1) antibody was used to check the quantities of protein loaded in the different lanes. After image acquisition, the band densities were quantitated using the NIH Image Program version 1.61. The results are expressed as -fold increase in phospho-ERK1/ERK2 over basal (Bas). The blot is representative of two independent experiments. D, dose-response curves of OT-and atosiban-induced accumulation of InsP. OTR coupling to G q was assayed by measuring total InsP production induced by cell exposure for 15 min to increasing concentrations of OT or atosiban. Each point, performed in triplicate, represents the mean of total amount of InsPs produced (expressed in dpm/well). The data are representative of two independent experiments diverging for Ͻ20%. E, dose-response curves of OT-induced InsP accumulation in the absence and presence of different concentrations of atosiban. Total InsP production induced by OT (from 10 Ϫ10 to 10 Ϫ5 M) was measured in the presence of a fixed concentration of atosiban (ϩATO 10 Ϫ8 to 10 Ϫ5 M). The amount of InsP produced at each point is expressed as the percentage of InsP produced at the maximal OT concentration (10 Ϫ5 M) in the absence of atosiban (ϪATO) fixed at 100%. The data are representative of two independent experiments diverging for Ͻ20%. Each concentration was assayed in triplicate determinations; error bars were not included for the sake of clarity. F, agonist-induced cAMP production. The coupling of OTRs to G s was assayed by measuring cAMP levels after stimulation with OT (10 Ϫ7 M), atosiban (10 Ϫ7 M), or forskolin (2 M). The data are representative of two independent experiments diverging for Ͻ20%. **, p Ͻ 0.01 versus untreated cells (Student's t test). ition, we verified whether atosiban acted as an antagonist on the OTR/G q /PLC signaling pathway. As shown in Fig. 2E, increasing concentrations of atosiban shifted to the right the dose-response curve of OT-induced InsP production, as expected for a competitive antagonist. This experiment also confirmed that atosiban does not behave as a partial agonist on the human OTR as, if this were the case, a very different family of curves, characterized by an increasingly higher level of basal InsP production and unchanged EC 50 for OT, would have been obtained. Finally, the possibility that atosiban might act on OTRs coupled to G s was excluded by the lack of any increase in cAMP induced by the drug as well as by OT (Fig. 2F).
All together, these data indicate that atosiban is able to promote selective OTR coupling to G i , whereas being devoid of any agonist activity upon OTR-G q coupling, and inhibits cell growth through an intracellular cascade that involves a PTxsensitive G protein and MAPK activation. The ability of atosiban to promote the selective activation of a G i -mediated pathway while being, at the same time, a competitive antagonist of the OT-induced OTR/G q coupling, supports the hypothesis that it acts as a biased agonist, i.e. by promoting and/or stabilizing an OTR conformational state(s) that "prefers" G␣ i over G q .

Atosiban-induced Inhibition of DU145 Human Prostate Cancer Cell
Growth-The human DU145 prostate cancer cell line expresses OTRs but not the endogenous OT ligand, and, in these cells, exposure to OT leads to a significant decrease in cell proliferation, already detectable after 48 h and peaking after 96 h (25). Therefore, this cell line represents a suitable model for testing the ability of atosiban to affect the proliferation of cells endogenously expressing OTRs. When treated with 100 nM atosiban, DU145 cells responded with a significant decrease in proliferation, firstly observed after 48 h and reaching its maximum at 72 h (Fig. 3A). The extent of growth inhibition induced by atosiban was similar to that observed after treatment with equal concentrations of OT or the highly selective OTR agonist Thr 4 Gly 7 OT (Fig. 3B). The anti-proliferative effect of atosiban was concentration-dependent, with a calculated EC 50 of 3.9 ϫ 10 Ϫ8 M (Fig. 3C). Finally, no additive effect was observed upon combined exposure of cultures to OT and atosiban for 48 h (Fig. 3D).
Finally, we investigated whether atosiban stimulation of OTRs in DU145 cells leads to ERK1/2 phosphorylation. As shown in Fig. 4, both the highly selective OT analogue Thr 4 Gly 7 OT (26) and atosiban elicited a sustained ERK1/2 activation. Moreover, Western blot analysis showed that the levels of cell cycle inhibitor p21 WAF1/CIP1 started to increase after 24 h of exposure and reached the maximum after 48 h, when the cell growth inhibition by Thr 4 Gly 7 OT and atosiban became measurable. These data suggest that the sustained MAPK activation observed after OTR stimulation by Thr 4 Gly 7 OT and atosiban inhibits the growth of DU145 cells by inducing the expression of p21 WAF1/CIP1 .
OT and Atosiban Induce Different OTR Desensitization and Internalization Patterns-GPCR signaling is generally downregulated by agonist-induced internalization, a process that sequentially involves agonist binding to the receptor, and subsequent receptor phosphorylation, ␤-arrestin recruitment, and endocytosis. No data are yet available on the effects of biased agonists on GPCR internalization. To investigate whether atosiban can induce receptor internalization, we used MDCK and HEK293 cells stably transfected with OTR fused to EGFP to monitor receptor distribution by confocal microscopy. Thus, we examined the subcellular distribution of OTRs after incubation with 10 Ϫ7 M OT or atosiban for increasing lengths of time (Fig. 5, A and B). Before agonist exposure (time 0), confluent monolayers expressing OTRs were homogeneously fluorescent all around the cell edges. A punctate fluorescence (presumably associated with endocytotic vesicles) could be seen as early as 5 min after OT application and became more evident after 15 min. After 30 and 60 min, the fluorescent patches seemed to be mainly localized within intracellular compartments of both MDCK and HEK293 cells. In contrast, OTRs remained permanently localized at the cell surface even after 60 min of atosiban treatment, without any signs of endocytic vesicle formation.
We also estimated the number of cell surface OTRs by measuring the high affinity binding of [ 3 H]OT to intact cells after different times of peptide exposure. As shown in Fig. 6A, OT treatment led to a progressive and almost complete (75%) loss of high affinity sites in HEK293 cells after only 15 min, whereas atosiban treatment induced the loss of only 20% specific high affinity binding sites even after 120 min.
Finally, we investigated whether the surface OTRs remaining after prolonged atosiban exposure displayed any agonistinduced desensitization of the G q /PLC signaling pathway by measuring the OT-induced accumulation of total InsP in the presence or absence of atosiban. As shown in Fig. 6B, a 60-min atosiban treatment did not affect the subsequent accumulation of total InsP induced by OT, thus indicating that atosiban did not induce any measurable desensitization of the canonical G q /PLC pathway.

DISCUSSION
The results of this study show that atosiban, a peptidic compound used in the management of preterm labor because of its antagonistic activity on uterine contractility (9,12,27), can also inhibit cell growth by acting as a selective agonist of OTR/G i coupling. This effect is indeed mediated by OTRs as it was only observed in MDCK and HEK293 cells stably transfected with human receptors but not in OTR-negative parental MDCK and HEK293 cells.
Human OTRs promiscuously couple to G q and G i (7) and can inhibit or stimulate cell growth depending on their differential coupling to G proteins, occurring inside or outside plasma membrane lipid rafts (11,19). In particular, the OT-induced inhibition of cell growth is completely blocked by pretreatment with PTx, thus indicating the involvement of OTRs coupled to G i -like proteins.
Here, we show that atosiban inhibits cell growth via selective activation of OTR/G i coupling, being completely blocked by pretreatment with PTx. In addition, as is true of OT (11) treatment, atosiban persistently induced MAPK phosphorylation and expression of the cell cycle inhibitor p21 WAF1/CIP1 . It is worth noting that atosiban, although behaving as agonist of OTR/G i coupling, acts as a competitive antagonist of OTR/G q coupling. The ability of displaying agonist-like properties toward a specific receptor-G protein coupling (G i ) and, simultaneously, acting as an antagonist at another receptor-G protein coupling (G q ) identifies atosiban as an analogue endowed with biased agonist properties.
The term biased agonist was introduced in 1998 by Jarpe et al. (17) to define the behavior of a substance P derivative, (D-Arg 1 ,D-Phe 5 ,D-Trp 7 , 9 ,Leu 11 ) substance P, a neurokinin-1 receptor antagonist that, in small-cell lung carcinoma, inhibits the G q -mediated pathway and behaves as an agonist to activate the c-Jun N-terminal kinase. It was subsequently found that the latter effect is mediated by bombesin receptors, i.e. substance P activates bombesin receptor signaling through G i and simultaneously blocks receptor coupling to G q . Moreover, it was shown that substance P, via G i , stimulates the MAPK pathway, leading to cell growth inhibition (28). Another analogue characterized by a biased agonist activity is the bradykinin antagonist dimer CU201 that, in small-cell lung carcinoma and other cancer cell lines, blocks the G q pathways and activates the G 12-13 pathways, once again leading to apoptosis and inhibition of cell growth (29). Biased agonists can have enormous pharmacological and clinical importance because they can trigger, for each single receptor subtype, only the "appropriate" downstream signaling pathway, thus representing a novel class of highly selective compounds. The existence of biased agonists is consistent with a multistate model of receptor activation in which single ligands can promote specific receptor conformations capable of differentially activating distinct signaling partners. This phenomenon, termed "agonist-directed trafficking of receptor stimulus" by Kenakin (30,31), has recently gained further support by the finding that inverse agonists acting on ␤-adrenergic and vasopressin receptors can lead to ligand-selective receptor activation (32).
Agonist-directed trafficking of receptor stimulus should not be confused with an "agonist strength-based mechanism." In this case, a powerful agonist may produce a stimulus sufficient to simultaneously activate two intracellular pathways, whereas a weaker agonist may only stimulate the most sensitive one. As quoted from a recent review of Kenakin: "The most sound evidence for the demonstration of receptor-selective states with a ligand is either: (1) thorough clear differences in efficacy (maximal response in which affinity is not an issue) or (2) actual reversal of relative potency in which there can be no question of differences as a result of a different strength of signal" (18). Here, comparing the potencies of OT and atosiban would be questionable because potency at G q -coupled OTRs is evaluated as increased generation of InsPs, whereas potency at G i -coupled OTRs is measured as degree of cell growth inhibition, an effect downstream of the proximal second messenger generation step. The evidence of stimulus trafficking comes, instead, from the very different efficacies of OT and atosiban in promoting OTR/G q coupling, despite binding affinities differing by only 1 order of magnitude. Notably, atosiban was devoid of any agonist activity on OTR/G q coupling when compared with the full agonist oxytocin, demonstrating the induction of ligand-selective receptor states. Furthermore, atosiban was confirmed to be a competitive antagonist on OTR/G q coupling, indicating biased agonist properties.
It is worth noting that a number of other peptidic antagonists have been found to interfere with cell proliferation in a complex and only partially understood manner, including the bradykinin B2 antagonist Hoe140 (33), the "dipeptoid" CCK-B/ gastrin agents (34), the neurotensin receptor antagonist SR48692 (35), the leukotriene B4 receptor antagonist LY293111 (36), and the somatostatin receptor antagonist cSSTA (37). It will be interesting to verify whether the complex effects on cell growth by these peptides may also be explained by a biased agonist action on their receptors. In this case, their classification as antagonists will need to be re-evaluated.
Atosiban acts on a GPCR activated by OT, the small peptide from which it was originally derived. Since a vast number of peptidic analogues of oxytocin have been developed over the last 40 years (38,39), it would be very interesting to evaluate the structural-functional relationships leading to biased agonist properties. The OTR has been subjected to molecular modeling by several groups (40), and recently, the atosiban-OTR interactions leading to putative high affinity binding have been investigated (41). A combined molecular modeling and sitedirected mutagenesis study could greatly improve our understanding of the molecular basis underlying receptor activation as well as the specificity of receptor-G protein interaction. It will also be interesting to determine the amino acid residues of OT that are responsible for the induction of a selective coupling state of the receptor, as done for the human parathyroid hormone, where the regions responsible for the dual coupling to adenylyl cyclase and PLC have been mapped (42).
As atosiban acts as an agonist to promote G i -mediated signaling, we investigated whether, like endogenous OT, it could induce agonist-induced receptor internalization (19,43). On the basis of our confocal microscopy data, we can conclude that atosiban, at the concentrations used, does not induce any receptor internalization. Likewise, no desensitization of G q /PLC signaling occurred. However, as atosiban treatment led to a small but significant 20% decrease in the number of high affinity binding sites, it is tempting to speculate that this OTR

FIG. 6. Agonist-induced desensitization of OTR in HEK293 cells.
A, time course of OTR internalization. HEK293 cells expressing the wildtype OTR were incubated in the presence of 10 Ϫ7 M OT or atosiban for the indicated times. Specific cell surface binding, determined at 4°C using [ 3 H]OT as a radioligand, is expressed as the percentage of [ 3 H]OT specific binding at time 0. The curve is representative of two independent assays; the data are expressed as means ϩ S.E. of triplicate determinations. B, agonist-induced desensitization of OTR. Desensitization was assayed by comparing the increase of total InsP in response to a 15-min OT application (final concentration of 10 Ϫ7 M) in basal (Bas) conditions and after a 60-min treatment with atosiban (final concentration of 10 Ϫ7 M). Total InsPs were measured after 60 min of atosiban treatment to exclude any change in basal InsP due to long-lasting atosiban application. The data are representative of two independent assays diverging for Ͻ20%. Each point represents the total amount of InsP and is expressed as dpm/well (means ϩ S.E. of triplicate determinations). population may represent the agonist-activated OTRs coupled to G i that undergo desensitization, a hypothesis that will require further investigation. It has been recently shown that recombinant human OTRs expressed in HEK293 cells undergo agonist-dependent phosphorylation by GRK2, an event that ultimately leads to receptor internalization (44), but it is still unclear whether receptor coupling to different G proteins affects the recruitment of the kinase and the subsequent events involved in receptor trafficking. To clarify this issue, atosiban could represent a very valuable tool.
Finally, it is worth noting that atosiban inhibits cell growth not only in cell lines transfected with the human OTR but also in DU145 human prostate cancer cells endogenously expressing OTRs (this study) and mammary carcinoma cells (16); this occurs in the absence of G q /PLC stimulation and thus represents a very specific downstream signaling event. Given that atosiban has already been approved as an antagonist for the treatment of preterm delivery, a pilot clinical trial of its use in the treatment of cancer cells expressing high human OTR levels warrants consideration.
In conclusion, our data indicate that atosiban, a peptidic derivative of the posterior pituitary OT hormone, possesses biased agonist activity on the human OTRs and represents a very valuable tool for investigating the molecular basis of ligand-induced coupling specificity.