A Novel Biased Allosteric Compound Inhibitor of Parturition Selectively Impedes the Prostaglandin F2α-mediated Rho/ROCK Signaling Pathway*

The prostaglandin F2α (PGF2α) receptor (FP) is a key regulator of parturition and a target for pharmacological management of preterm labor. However, an incomplete understanding of signaling pathways regulating myometrial contraction hinders the development of improved therapeutics. Here we used a peptidomimetic inhibitor of parturition in mice, PDC113.824, whose structure was based on the NH2-terminal region of the second extracellular loop of FP receptor, to gain mechanistic insight underlying FP receptor-mediated cell responses in the context of parturition. We show that PDC113.824 not only delayed normal parturition in mice but also that it inhibited both PGF2α- and lipopolysaccharide-induced preterm labor. PDC113.824 inhibited PGF2α-mediated, Gα12-dependent activation of the Rho/ROCK signaling pathways, actin remodeling, and contraction of human myometrial cells likely by acting as a non-competitive, allosteric modulator of PGF2α binding. In contrast to its negative allosteric modulating effects on Rho/ROCK signaling, PDC113.824 acted as a positive allosteric modulator on PGF2α-mediated protein kinase C and ERK1/2 signaling. This bias in receptor-dependent signaling was explained by an increase in FP receptor coupling to Gαq, at the expense of coupling to Gα12. Our findings regarding the allosteric and biased nature of PDC113.824 offer new mechanistic insights into FP receptor signaling relevant to parturition and suggest novel therapeutic opportunities for the development of new tocolytic drugs.

The prostaglandin F2␣ (PGF2␣) receptor (FP) is a key regulator of parturition and a target for pharmacological management of preterm labor. However, an incomplete understanding of signaling pathways regulating myometrial contraction hinders the development of improved therapeutics. Here we used a peptidomimetic inhibitor of parturition in mice, PDC113.824, whose structure was based on the NH 2 -terminal region of the second extracellular loop of FP receptor, to gain mechanistic insight underlying FP receptor-mediated cell responses in the context of parturition. We show that PDC113.824 not only delayed normal parturition in mice but also that it inhibited both PGF2␣and lipopolysaccharide-induced preterm labor. PDC113.824 inhibited PGF2␣-mediated, G␣ 12 -dependent activation of the Rho/ROCK signaling pathways, actin remodeling, and contraction of human myometrial cells likely by acting as a non-competitive, allosteric modulator of PGF2␣ binding. In contrast to its negative allosteric modulating effects on Rho/ROCK signaling, PDC113.824 acted as a positive allosteric modulator on PGF2␣mediated protein kinase C and ERK1/2 signaling. This bias in receptor-dependent signaling was explained by an increase in FP receptor coupling to G␣ q , at the expense of coupling to G␣ 12 . Our findings regarding the allosteric and biased nature of PDC113.824 offer new mechanistic insights into FP receptor signaling relevant to parturition and suggest novel therapeutic opportunities for the development of new tocolytic drugs.
Premature birth due to preterm delivery is the most important cause of neonatal mortality and morbidity in industrialized countries (1,2). A common cause of preterm labor is a spontaneous increase in uterine contraction. However, little is known regarding the factors that either maintain uterine quiescence or initiate spontaneous uterine contraction. Pharmacological interventions that aim at preserving or inducing uterine quiescence remain the most attractive strategy for managing preterm labor to date.
Prostaglandins, whose synthesis is under the control of cyclooxygenases and specific prostaglandin synthases, play important roles during pregnancy and parturition (3,4). They are initiators of physiological labor and exert their effects through different G protein-coupled receptors (GPCRs). 6 For instance, prostaglandin F2␣ (PGF2␣) promotes myometrial contraction through activation of PGF2␣ (FP) receptors (5,6). Moreover, FP receptor null mice fail to deliver at term and are unresponsive to induced labor mediated by either PGF2␣ or the uterogenic hormone oxytocin (7,8). At the molecular level, activation of FP receptor leads to inositol phosphate accumulation, protein kinase C (PKC) activation, and intracellular calcium release, consistent with coupling of FP receptor to the G␣ q family of G proteins (9 -12). Activation of FP receptor has also been shown to promote Rho-dependent reorganization of the cytoskeleton (13). Both signaling pathways are believed to contribute to phasic and tonic myometrial smooth muscle contraction. However, their relative contributions to uterine tissue contraction during parturition remain poorly understood. Thus, a better comprehension of PGF2␣-mediated signaling mechanisms through FP receptor is not only essential to understanding parturition, but also for the potential development of drugs suppressing labor (tocolytics).
At present, the use of tocolytics to delay preterm labor is often contraindicated due to significant maternal and fetal side effects (14,15). A new FP receptor ligand, THG113, corresponding to a peptide derived from the sequence of the second extracellular loop of FP receptor (Fig. 1A), was recently reported to inhibit preterm labor in a mouse model (16). However, the exact mechanism underlying its action on myometrial contraction remained unclear. In refining THG113 as a potential tocolytic compound specific for FP receptor with enhanced efficacy toward myometrial contraction, a peptidomimetic compound, PDC113.824, was synthesized ( Fig. 1B and  supplemental Fig. S1A). We used this compound to develop a better understanding of FP receptor signaling in the context of myometrial cell contraction and the regulation of parturition. We report here, that PDC113.824 is a potent tocolytic agent, selective for FP receptor. Interestingly, our results suggest that this compound acts as a functional allosteric modulator for FP receptor, because it exerts its effects at a site distinct from the orthosteric binding site, and further biases PGF2␣-bound receptors toward increased G␣ q -PKC-MAPK signaling, while blocking cell contraction and cytoskeleton reorganization through inhibition of the G␣ 12 -Rho-ROCK signaling pathway.  (17). PGF2␣ is from Cayman. Rabbit polyclonal anti-G␣ q (C-19) and anti-G␣ 12 (S-20) antibodies were from Santa Cruz Biotechnology. PD98059, Gö6983, and LY294002 were from Calbiochem. Y27632 was from Ascent Scientific. C3 exoenzyme was from Cytoskeleton. Fluor488-Phalloidin was from Molecular Probes. Type I rat collagen, oxytocin, and angiotensin II (AngII) were from Sigma. Mouse monoclonal anti-p-ERK and rabbit polyclonal anti-total ERK antibodies are from Cell Signaling. Puromycin was from InvivoGen. MEM was from HyClone. DMEM/F-12, heat-inactivated fetal bovine serum, and gentamycin were from Invitrogen. Analytical reversed-phase high-performance liquid chromatography for PDC113.824 was performed on a C 18 Gemini column (5 m, 4.6 mm ϫ 50 mm, flow rate 0.5 ml/min in 15 min) using a linear gradient from 20 to 80% of acetonitrile or methanol in water (each solvent containing 0.1% trifluoroacetic acid). Elution products were detected at 214 nm. N-tert-butoxycarbonyl-3-pyridylalanine was from GL-Biochem Shanghai, and other peptide mimic synthesis products, such as coupling reagents were purchased from Sigma. N-tert-butoxycarbonyl-␤-phenylalanine was synthesized from N-tert-butoxycarbonyl-phenylalanine according to an established method (18).
FP Receptor Antibody Generation-The peptide corresponding to the first extracellular loop of FP receptor was synthesized by using Fmoc (N-(9-fluorenyl)methoxycarbonyl) synthesis with Ͼ95% purity (Biosynthesis, Lewisville, TX) and included an additional residue (cysteine) at the NH 2 terminus (NH 2 -CSMNNSKQLVS-COOH). The cysteine-containing peptide was conjugated to the sulfhydryl-reactive carrier protein keyhole limpet hemocyanin (Pierce). The keyhole limpet hemocyanin-conjugated peptide in complete Freund's adjuvant (Pierce) was injected intraperitoneally and subcutaneously (total 25 g, in 100 l) in female BALB/c mice. Subsequent immunizations were performed every 10 -13 days with keyhole limpet hemocyanin-conjugated peptide in phosphate-buffered saline. Serial dilutions of mouse serum were screened for reactivity using a solid-phase enzyme-linked immunosorbent assay, testing the original peptide (unconjugated). Several other peptides or bovine serum albumin were used as negative controls (not shown). Four days after the last immunization, splenocytes were fused with SP2/0 mouse myeloma cells following established protocols. Hybridoma supernatants were screened by enzyme-linked immunosorbent assay against unconjugated peptide or controls as above. Supernatants were further screened for binding to cell surface FP receptor by FACScan assays, using live HEK293 cells stably transfected with FP receptor or parental HEK293 cells as a background control. Wells containing hybridomas secreting monoclonal antibodies reactive to FP receptor were subcloned twice by limiting dilution and expanded for monoclonal antibody purification. A hybridoma line producing antibody 3E12/2B2 was used in this study.
Cell Culture and Transfection-A stable HEK293 cell line expressing the human FP receptor was generated by transfection with pIRESP-HA-hFP. Stable lines were selected in 0.7 g/ml puromycin. All HEK293-derived cell lines were grown at 37°C in 5% CO 2 in MEM supplemented with 10% (v/v) heatinactivated fetal bovine serum and gentamycin (100 g/ml). hTERT-C3 myometrial cells were grown in DMEM/F-12 medium as described previously (19). For transient transfection, cells seeded at a density of 1 ϫ 10 6 cells per 100-mm dish, or 1 ϫ 10 5 per well in a 6-well plate, were transfected using a conventional calcium phosphate co-precipitation method. All experiments were performed 48 h post-transfection.
Ligand  12 . On the day of the experiment, cells were serum-starved prior to treatment with or without 0.5 M PDC113.824 at 37°C for 30 min. Cells were collected in ice-cold buffer (10 mM Tris-HCl, pH 7.4, 5 mM EDTA) and homogenized on ice by 20 strokes with a Teflon potter. The homogenate was centrifuged at 23,000 rpm at 4°C and resuspended in TME buffer (50 mM Tris-HCl, pH 7.4, 2 mM EDTA, 4.8 mM MgCl 2 , 100 mM NaCl). GDP (final concentration of 1 M) was added to 50 g of membranes, and the mixture was incubated on ice for 10 min. The reaction was moved to 30°C and incubated for 5 min before the addition of [ 35 S]GTP␥S (1250 Ci/mmol) to a final concentration of 5 nM. PGF2␣ was added 30 s later, and the reaction was allowed to proceed for 5 min. Reactions were stopped with cold immunoprecipitation buffer (50 mM Tris-HCl, pH 7.5, 20 mM MgCl 2 , 150 mM NaCl, 0.5% Nonidet P-40, protease inhibitors, 100 M GDP and GTP), and membranes were solubilized for 30 min at 4°C. To immunoprecipitate specific complexes, 1.2 g of anti-G␣ q (clone C-19, Santa Cruz Biotechnology) or anti-G␣ 12 (clone S-20, Santa Cruz Biotechnology) antibodies were added for 2 h at 4°C. Protein G-agarose beads were added, and the mixture was incubated for an additional 60 min at 4°C. Beads were then washed three times with immunoprecipitation buffer, and incorporated [ 35 S]GTP␥S was measured by liquid scintillation spectrometry.
PKC␤I-GFP Translocation-FP receptor cells co-transfected with PKC␤I-GFP were serum-starved for 30 min and pretreated with vehicle (water) or PDC113.824 (2 M) for 30 min followed by treatment with increasing concentrations of PGF2␣ (from 10 Ϫ11 to 10 Ϫ7 M) for 10 min each. Images were collected every 30 s using live-cell microscopy at 37°C on a Zeiss LSM-510 Meta laser scanning microscope equipped with XL-3 temperature chamber with a 63ϫ glycerol/water immersion lens in single track mode using excitation at 488 nm for GFP and emission measured with the LP505 filter set. Translocation was determined by calculating the fluorescence level at the membrane (area under the curve) divided by the fluorescence level in the cytosol, using Metamorph (Universal Imaging Corp.).
Raichu-RBD Experiments-The Raichu-RBD probe was used as described previously (20). Briefly, Raichu-RBD is composed of a YFP moiety in the NH 2 -terminal domain (m2Venus), a central Rho-binding domain (RBD) of Rhotekin and a CFP moiety in the COOH-terminal domain. The FP receptor or AT 1 R cells were plated on 35-mm microscopy dishes at a density of 50,000 cells. 24 h later, cells were transfected with Raichu-RBD alone or co-transfected with G␣ 12 DN. 48 h later, cells were serum-starved for 30 min, pre-treated or not with PDC113.824 (1 M) for 30 min or C3 exoenzyme (1 g/ml) for 3-4 h, followed by stimulation with 1 M PGF2␣ for 25 min. Images were collected every 30 s for the first 6 min, then every minute until 12 min, and at 15, 20, and 25 min using live-cell microscopy at 37°C on a Zeiss LSM-510 Meta laser scanning microscope as described above with excitation at 405 nm for CFP and the FRET channel, 514 nm for YFP and with emission measured with BP420 -480 for CFP, and BP530 -600 for YFP and FRET. Energy transfer efficiency between the CFP (donor) and YFP (acceptor) was determined by calculating the ratio of the YFP over CFP fluorescence from three different regions of each cell, and corrected for background signal using Metamorph. FRET images were translated into colored gradient images using Rainbow2 visualization in Zen Light software (Zeiss). This function translates each pixel of the image into intensity values and reports them using a color code. 24 -36 cells/condition were quantified in five to eight independent experiments.
Collagen Contraction Assay-Collagen contraction assays were performed as described previously (19). Briefly, 14,000 to 17,500 hTERT-C3 myometrial cells were plated in DMEM/ F-12 medium in 0.5% (v/v) fetal bovine serum on the collagen lattice and left for 2 h at 37°C. Cells were then pre-treated with either vehicle, C3 exoenzyme (1 g/ml), or PDC113.824 (2 M) for 2 h. To allow contraction, each collagen lattice was detached from the bottom of the well with a small spatula and left overnight at 37°C in the absence or presence of 1 M PGF2␣ or oxytocin. Contraction was stopped by fixing lattices in phosphate-buffered saline with 4% paraformaldehyde. The plate containing collagen lattices was then photographed using the Alpha Imager System. Percentage of contraction of collagen lattices was then calculated using Metamorph, using the following equation: % contraction ϭ 100 Ϫ (area of lattice*100/area of the well).
Murine Preterm Labor Models and Ex Vivo Myometrial Contraction Assay-Timed-pregnant CD-1 mice at 16 days gestational (normal term is 19.2 days) were anesthetized with isoflurane (2%). Primed osmotic pumps (Alzet pump, Alzet, Cupertino, CA) containing either saline or PDC113.824 (10 mg/day/animal) were subcutaneously implanted on the backs of the animals; infusion of PDC113.824 was immediately pre-ceded by bolus injection of PDC113.824 (0.1 mg/animal intraperitoneally). Within 15 min after placement of the pumps, animals were injected with PGF2␣ or lipopolysaccharide (LPS) Escherichia coli endotoxin (50 g/animal intraperitoneally) to mimic the inflammatory/infectious component of human preterm labor. In a separate group of animals, PGF2␣ or LPS was injected 4 -7 h prior to administration of PDC113.824. Animals were inspected every hour for the first 18 h and every 2 h thereafter to document the timing of birth. All experiments were approved by the Animal Care Committee of Centre Hospitalier Universitaire Sainte-Justine (Montreal, Quebec, Canada). Ex vivo myometrial contraction assay was performed as previously described (16). Briefly, uteri from mice were obtained from animals immediately following term delivery. Myometrial strips (2-3 mm wide and 1-2 cm long) from both were suspended in organ baths containing Krebs buffer equilibrated with 21% oxygen at 37°C with an initial tension at 2 g. After 1 h of equilibration, changes in mean basal tension, as well as peak, duration, and frequency of spontaneous contraction in the absence or presence of PGF2␣ and PDC113.824 were recorded with a Kent digital polygraph system.
Statistical Analysis-Statistical tests were performed with GraphPad Prism 4.3 software. Assumptions of normality and equal variance were met for all data analyzed. One-way analysis of variance with Dunnett's correction was used in Fig. 3 (C-F)) was used when the data were normalized, and basal levels were considered with respect to the hypothetical value (1 or 100). The Fisher's exact test was used in Fig. 2 (A, B, and D). A two-tailed p value of Ͻ0.05 was considered significant. All results are expressed as means Ϯ S.E. Sample size (n) and p values are given in the figure legends.

RESULTS
Design and Optimization of the Peptidomimetic PDC113.824-Conversion of the THG113 sequence (Ile-Leu-Gly-His-Arg-Asp-Tyr-Lys) into the peptide mimic ( Fig. 1 and supplemental Fig. S1A) involved, in brief, a systematic analysis of the sequence using alanine and enantiomeric amino acid scans, which highlighted the importance of the Arg and Asp side chains; replacement of the hydrophobic termini with hydrocarbon pharmacophores and the Gly-His residue by different   AUGUST 13, 2010 • VOLUME 285 • NUMBER 33 indolizidinone turn mimics (21)(22)(23)(24); and refinement near the Arg-Asp residue using different amino acid substitutions to arrive at the pyridylalanine-␤-homophenylalanine surrogate. The 3-phenylacetamido indolizidin-2-one 9-carboxyl and the pyridinylalaninyl-␤-homophenylalanine sections of PDC113.824 (Fig. 1B) are thus believed to mimic the active ␤-turn geometry about the Gly residue and the signaling pharmacophore of the Arg-Asp-Tyr triad in the parent peptide, respectively.

PDC113.824: A New Biased Allosteric FP Ligand
Tocolytic Effects of PDC113.824 in Normal and Preterm Labor Models-We first verified that the peptide mimetic, PDC113.824, acted as a tocolytic agent in vivo in normal parturition. Mice near term (gestational day 17.5) were treated or not with PDC113.824, and delivery was assessed in the animals ( Fig.  2A). Results showed that PDC113.824 significantly delayed delivery compared with untreated animals who all delivered at term (day 19). Indeed, at day 19 only 50% of PDC113.824treated animals had delivered. Delivery of all PDC113.824treated animals was delayed to day 20.
We also tested if PDC113.824 would also block provoked preterm labor using LPS, known to promote a general inflammatory state, which results in prostaglandin synthesis, and induce premature delivery (16). Results showed that for all the animals tested, delivery occurred within 12 h following LPS injection into mice at gestational day 16 (Fig. 2, B and C). Pretreatment with PDC113.824 prior to LPS injection significantly delayed delivery, such that by gestational day 17, only 20% of treated animals delivered. As was the case for normal term delivery (e.g. saline treatment), PDC113.824-treated mice did not deliver until day 19 even when treated with LPS. Hence, PDC113.824 significantly extended the average time of delivery following LPS treatment by ϳ20 h as compared with untreated animals (Fig. 2C). We also verified whether PDC113.824 interfered specifically with PGF2␣-induced labor. Animals were treated or not with PDC113.824 prior to injection with PGF2␣ at gestational day 15.5 ( Fig.  2D). All PGF2␣-treated animals delivered rapidly by 2 h post-injection. Again, PDC113.824 treatment markedly delayed delivery in PGF2␣injected animals, as around 40% of the animals had delivered by days 16 -17 post-injection (Fig. 2D). Accordingly, the mean time of delivery was significantly increased in the presence of PDC113.824 (Fig. 2E).
To ensure that the observed labor-delaying effect of our synthetic tocolytic was mediated through its actions on the uterus, and to dissociate its potential effects on luteolysis, which would decrease levels of the natural tocolytic progesterone produced in the ovary (7), we isolated myometrium from spontaneous post-partum mice and assessed the direct effects of PDC113.824 on spontaneous and PGF2␣-induced contraction (supplemental Fig. S2). Results showed that PDC113.824 significantly reduced both the strength and duration of both PGF2␣induced and spontaneous myometrial contraction in a dose-dependent manner, consistent with the increased expression of FP in the uterus during labor, which occurs even in rodents (6). Taken together, our results suggest that PDC113.824 delays both term and preterm labor, at least in part through the inhibition of uterine contraction.
PDC113.824 Negatively Modulates PGF2␣-mediated Myometrial Cell Contraction and Rho/ROCK Signaling-The putative inhibitory effects PDC113.824 on PGF2␣-mediated contraction were next examined in myometrial cells. Human myometrial smooth muscle cells (hTERT-C3) were used, because they have been previously shown to respond to uterotonic factors and to retain their contractile properties (19). As a prelude to these experiments, the expression of endogenous FP receptor in these cells was confirmed using a monoclonal antibody raised against the receptor (clone 3E12/2B2, Fig. 3A). Antibody labeling was shown to be specific for FP receptor, because immunofluorescent signals were not detected in HEK293 cells, which do not express endogenous FP receptor, but were detected at the cell surface in HEK 293 cells transfected with FP receptor. Strong labeling of FP receptor at the cell surface was also detected in myometrial cells (Fig. 3A). We also quantified the levels of endogenous FP receptor at the plasma membrane in myometrial cells using [ 3 H]PGF2␣ binding studies (Fig. 3B). Both PGF2␣ and the selective FP receptor antagonist AL-8810 displaced bound [ 3 H]PGF2␣ from myometrial cells, and FP receptor expression was ϳ5-10 fmol/mg total protein.
Next, PGF2␣ and PDC113.824 regulation of myometrial cell contraction was tested in vitro using a cell-induced collagen lattice contraction assay (Fig. 3C). Cells were layered on top of collagen and grown for a period of 16 h, which resulted in a decrease in the diameter of the collagen matrix due to selfcontraction of the cells. Addition of PGF2␣ further increased contraction of hTERT-C3 cells. PDC113.824 alone had no effect on hTERT-C3 contraction per se. On the other hand, it inhibited PGF2␣-mediated cell contraction responses (Fig. 3, C   and D). Moreover, PDC113.824 exhibited no effect on the myometrial contraction induced by a distinct uterotonic agent, oxytocin (Fig. 3D, OT), demonstrating the specificity of the action of the compound for FP receptor. Because the contractile function of FP receptor on smooth muscle has been shown to involve Rho-kinase activation (25,26), we also verified its contribution on myometrial cell contraction. As shown in Fig. 3 (C and D), both basal and agonist-induced cell contractions were dependent on Rho GTPase activation, because C3 exoenzyme blocked both responses.
To assess how PGF2␣-dependent activation of Rho was regulated by PDC113.824, we next used a biosensor for Rho activation expressed in HEK293 cells, because these cells are considerably easier to transfect than myometrial cells. We first characterized the binding properties of FP receptor in these cells by stably expressing HA-tagged receptors (thereafter referred to as FP receptor cells, see "Experimental Procedures"). No specific [ 3 H]PGF2␣ binding to untransfected cells was detected (Fig. 4A, inset). However, radiolabeled PGF2␣ binding on FP receptor-expressing cells was robustly displaced by both unlabeled PGF2␣ as well as AL-8810 (Fig. 4A). The effect of PDC113.824 on [ 3 H]PGF2␣ binding was also tested in these cells. PDC113.824, at concentrations that inhibited myometrial contraction (1 M) or higher (10 M), displaced no more than 15% of PGF2␣ binding to the FP receptor.
Rho family GTPases are known FP receptor effectors (13), whose activities can be regulated by G␣ 12/13 (27). Using FP receptor cells, we next tested PGF2␣-dependent activation of Rho GTPases by imaging the fluorescent FRET-based biosensor Raichu-RBD, which consists of the RBD of Rhotekin flanked by the FRET pair YFP and CFP (20). Under basal conditions, intramolecular interaction of YFP and CFP in the Rho biosensor generated a detectable FRET signal (Fig. 4B). Binding of endogenous, activated GTP-bound Rho to this biosensor following FP receptor stimulation induced a decrease in FRET signal. Although PDC113.824 alone had no effect on Rho activation (data not shown), the response to PGF2␣ stimulation was decreased in its presence (Fig. 4B). Quantification of the FRET signal was assessed by measuring changes in the YFP/ CFP emission ratio (20, 28) and revealed a time-dependent, agonist-mediated activation of Rho (Fig. 4C, i.e. decrease in FRET signal). PGF2␣-stimulated Rho activity was greatly  AUGUST 13, 2010 • VOLUME 285 • NUMBER 33 reduced at all time points in the presence of PDC113.824. PGF2␣-mediated activation of Rho, like contraction in myometrial cells was sensitive to C3 exoenzyme (Fig. 4D).

PDC113.824: A New Biased Allosteric FP Ligand
We next assessed the contribution of G␣ 12 in Rho activation by expressing a dominant negative version of this G␣ subunit (G␣ 12 DN, Q231L/D299N, Fig. 4D, left panel). An inhibition of FP receptor-mediated Rho activation was observed in a cell line expressing the G␣ 12 DN. This response was specific to FP receptor, because PDC113.824 had no significant effect on the angiotensin II (AngII) type 1 receptor (AT 1 R), another GPCR known to activate Rho (29) (Fig. 4D, right panel).
Rho GTPases can engage downstream targets, including the protein kinase ROCK (30), as well as promoting actin cytoskeletal rearrangement (31). We therefore assessed the effect of PGF2␣ and PDC113.824 on reorganization of the actin cytoskeleton manifested by membrane ruffling using phalloidin staining. Cell ruffling, as characterized by morphological rounding of cell edges, was detected after 5 min of PGF2␣ stimulation (Fig. 5A) and persisted for more than 60 min (data not shown). This response was dependent on FP receptor-mediated activation of G␣ 12 , because co-expression of G␣ 12 DN strongly inhibited PGF2␣induced cell ruffling (Fig. 5B). Pretreatment of cells with PDC113.824 again had a significant inhibitory effect on PGF2␣-mediated cell ruffling (Fig. 5, D and E), while treatment of cells with either selective Rho GTPase (C3 exoenzyme) or ROCK (Y27632) inhibitors, both blocked completely PGF2␣-mediated cell ruffling (Fig. 5, C and E).
We also assessed the potential involvement of other signaling pathways downstream of the FP receptor on cell ruffling. Treatment of cells with selective inhibitors for PKC (Gö6983) or phosphatidylinositol 3-kinase (LY294002, data not shown) had no effect on membrane ruffling. The extent to which ERK1/2 MAPKs were involved in regulating cell ruffling was assessed, because they have been proposed to modulate Rho signaling (32). Treatment of cells with the MEK1 inhibitor (PD98059) blocked only weakly PGF2␣-mediated cell ruffling (Fig.  5E). Taken together our results suggested that PDC113.824 acts as an allosteric modulator of FP receptormediated myometrial contraction and cytoskeletal reorganization through inhibition of the Rho/ ROCK signaling pathway. Potentiation of FP Receptor-mediated PKC and MAPK Signaling by PDC113.824-Because our results suggested that PDC113.824 acts as a negative allosteric modulator of the Rho/ ROCK signaling pathway, we investigated whether it also antagonized other FP receptor-mediated signaling pathways. As FP receptor has been shown to promote ERK1/2 activation (33), we next tested PDC113.824 effects on PGF2␣-mediated MAPK activation. PGF2␣-dependent stimulation of FP receptor cells resulted in a time-dependent increase in ERK1/2 activation that reached maximal levels following 5-15 min of agonist stimulation (Fig. 6A, top panel). However, PDC113.824 alone had no effect on the ERK1/2 response (Fig. 6A, bottom  panel). Rather than inhibiting PGF2␣-dependent activation of ERK1/2, the treatment of cells with PDC113.824 potentiated the response. Pretreatment with PDC113.284 significantly augmented the activation of ERK1/2 induced by PGF2␣, increasing both the efficacy and potency of the response by 2-fold (Fig. 6B, EC 50 vehicle ϭ 0.19 nM versus EC 50 PDC113.824 ϭ 0.10 nM). The effects of PDC113.824 on MAPK activation were again FP receptor-specific, because AngII-induced ERK1/2 activation in AT 1 R cells was not affected by pre-treatment with the peptidomimetic (Fig. 6C; EC 50 vehicle ϭ 11.1 nM versus EC 50 PDC113.824 ϭ 16.8 nM). This effect of PDC113.284 on PGF2␣-dependent activation of ERK1/2 was also recapitulated in myometrial cells, because it potentiated the response by 1.5-fold relative to that of cells stimulated with PGF2␣ alone (Fig. 6D). Because PDC113.824 treatment increased PGF2␣-dependent activation ERK1/2, we also investigated the extent to which MAPK signaling contributed to myometrial cell contraction inhibition. As shown in Fig. 7A, pretreatment of cells with PD98059 caused no significant effect on PGF2␣-mediated cell contraction. Consistent with the notion that FP receptor-dependent activation of Rho and MAPK are two independent signaling pathways, the inhibition of Rho with C3 exoenzyme did not affect ERK1/2 activation in FP receptor cells (Fig. 7B).
For G␣ q -coupled receptors like the FP receptor, activation of ERK1/2 can be mediated through activation of PKC. As shown in Fig.  8A, treatment of FP receptor cells with the PKC inhibitor Gö6983 totally prevented PGF2␣-dependent activation of ERK1/2. Thus, we also tested whether PDC113.824 potentiated PGF2␣-dependent activation of PKC by measuring PKC␤I-GFP recruitment to the plasma membrane (Fig. 8, B and C). PKC recruitment to the plasma membrane following PGF2␣ stimulation of FP receptor cells was evident at concentrations of 1 nM agonist (Fig.  8, B and D); while treatment of the cells with PDC113.824 alone had again no effect (Fig. 8D). Quantification of this response revealed a dosedependent increase in PKC activation upon PGF2␣ stimulation, with an EC 50 of 2.5 nM (Fig. 8E, vehicle). Pre-treatment of cells with   17 nM; Fig. 8E). The specificity of PDC113.824 for FP receptor was again demonstrated, because it had no effect on AngII-dependent PKC activation (Fig. 8F). Together, these results suggest that PDC113.824 acts as positive allosteric modulator (PAM) on the PKC-MAPK signaling pathway induced by PGF2␣.
PGF2␣-mediated G Protein Coupling to FP Receptor Is Differentially Regulated by PDC113.824-Our ligand binding experiments suggested that PDC113.824 allosterically regulates FP receptor binding to PGF2␣ and/or its coupling to G proteins. Allosteric modulators are known to affect the off-rate of ligand binding to the orthosteric site on receptors (34,35).
We first investigated how PDC113.824 affected PGF2␣ binding to FP receptor by measuring the kinetics of dissociation of PGF2␣. Upon exposure to excess cold agonist, the dissociation rate of [ 3 H]PGF2␣ in the presence of PDC113.824 was increased by Ͼ1.5-fold as compared with control treatment with vehicle (Fig. 9A). On the other hand, no significant effect of PDC113.824 was observed on binding off-rates for AT 1 R (Fig. 9B).
We next assessed the effect of PDC113.824 on FP receptor coupling to G proteins by monitoring [ 35 S]GTP␥S loading onto G␣ q and G␣ 12 following PGF2␣ stimulation. Preincubation of FP receptor cells with PDC113.824 alone increased [ 35 S]GTP␥S loading onto G␣ q to levels similar to that of PGF2␣ (i.e. in the absence of PDC113.824). Consistent with the effects on PKC and ERK1/2, treatment with PDC113.824 also significantly  (Fig. 9D). This effect is shown again to be specific for the FP receptor as the agonist activation of AT 1 R, which also resulted in [ 35 S]GTP␥S load-ing onto G␣ q and G␣ 12 was not significantly influenced by PDC113.824 (Fig. 9, E and F).

DISCUSSION
Here, we describe the design and characterization of PDC113.824 as a new allosteric modulator with biased signaling properties on FP receptor, acting both as potent tocolytic agent in vivo and as an inhibitor of myometrial contraction in vitro and ex vivo. PDC113.824 actions were specific to FP receptor, because no significant effects were observed on two other GPCRs, AT 1 R and the oxytocin receptor. Functional studies revealed that PDC113.824 increased agonist-mediated activation of MAPK by FP receptor via G␣ q , whereas it inhibited cytoskeletal rearrangement and myometrial contraction through uncoupling of the receptor to the G␣ 12 -Rho-ROCK signaling pathway (summarized in Fig. 10). The functional selectivity of PDC113.824 toward two distinct and opposite G protein-dependent events supports the biased nature of this compound on PGF2␣-mediated, FP receptor-dependent signaling.
A hallmark of allosteric affinity modulators is their ability to promote conformational changes in the receptor, which mechanistically can translate into alterations in the dissociation kinetics of preformed orthosteric ligand-receptor complexes (36,37). These effects, however, are not seen if interacting ligands compete for the same orthosteric site. Our findings that PDC113.824 partially decreases [ 3 H]PGF2␣ binding to the FP receptor (Fig. 4A, albeit no more than 15% of maximum binding) suggested that it can either act as an allosteric modulator of orthosteric site affinity (i.e. through a conformational change in the orthosteric binding site) and/or as a weak (partial) competitive ligand. The increased rate of dissociation of PGF2␣ in the presence of PDC113.824 is, however, more consistent with a conformational change in the orthosteric binding site of the receptor promoted by the non-competitive nature of an allosteric ligand. Although, we cannot totally exclude that PDC113.824 may still  AUGUST 13, 2010 • VOLUME 285 • NUMBER 33 partially compete with PGF2␣ for the orthosteric binding site, our results, showing that PDC113.824 minimally decreases total binding of PGF2␣ while increasing agonist off-rate kinetics of ligand binding to the receptor, strongly suggest that PDC113.824 primarily acts as a negative affinity allosteric modulator of the FP receptor.

PDC113.824: A New Biased Allosteric FP Ligand
Although a number of studies have characterized the allosteric properties of synthetic compounds on different GPCRs (see Refs. 38 and 39 for review), their potential to bias receptor signaling remains largely unexplored. To our knowledge, PDC113.824 represents the first example of a synthetic allosteric modulator derived from a specific region of a GPCR (e.g. the NH 2 -terminal domain of the second extracellular loop of FP receptor), which promotes functional selectivity for two distinct G protein-mediated signaling events. Our study not only provides conceptual insights into FP receptor signaling relevant to myometrial contraction, but also for the potential development of new classes of tocolytic drugs and other allosteric GPCR modulators with biased signaling properties.
The extracellular loops of GPCRs have been demonstrated to be important for both ligand recognition and allosteric modulation of certain receptors (40 -43). Transmembrane domains involved in ligand recognition and receptor activation have also been shown to influence extracellular loop conformations (44,45). For instance, light-dependent activation of rhodopsin has been shown to induce changes in the conformation of the second extracellular loop. Moreover, a recent study on the ␤ 2 -adrenergic receptor also revealed that ligands known to differen-tially affect the conformation of transmembrane domains and subsequent receptor activity also stabilize distinct conformation of the second extracellular loop (45). Thus, PDC113.824, which mimics structural features of the second extracellular loop of the FP receptor, could constrain the conformation of agonist-bound receptor into selective coupling configurations, promoting more efficient G␣ q activation, while reducing coupling to G␣ 12 . Interestingly, differential G protein coupling with the FP receptor was modulated by PDC113.824 even in the absence of the orthosteric ligand, suggesting that it primes two distinct pre-existing receptor/G protein complexes of FP receptor/G␣ q and FP receptor/ G␣ 12 but in distinct ways. Alternatively, PDC113.824 may be capable of interacting with receptors that are constitutively active but not necessarily pre-coupled per se. Although PDC113.824 acted as a negative allosteric modulator for agonist binding, it seemed to affect how FP receptor coupled to G protein in the absence of agonist in a biased fashion.
Our study also yields new information regarding FP receptor signaling and the mechanisms underlying myometrial contraction and cytoskeletal remodeling. We observed that blocking MAPK activation had only a marginal effect on cell ruffling, a response dependent on the G␣ 12 -Rho-ROCK signaling pathway (46,47), and no effects on myometrial cell contraction. Inhibiting Rho only affected cytoskeletal rearrangement and not FP receptor-dependent activation of ERK1/2. Our findings suggest an important role of the G␣ 12 -Rho-ROCK pathway in regulating myometrial cell contraction and that G␣ q , which is involved in controlling the PKC-MAPK signaling pathway, is independently regulated. However, we cannot exclude that other signaling events upstream of MAPKs, such as those seen for the PDC113.824dependent increase in FP receptor-mediated activation of PKC, contribute to blocking functional coupling of FP receptor to G␣ 12/13 .
Our findings underscore the importance of FP receptor signaling through the G␣ 12 -Rho-ROCK pathway as a pharmacological target in the management of parturition and preterm labor. Our in vitro and in vivo data are consistent with both the observations that RhoA activity is increased in the myometrium during pregnancy, and that inhibition of ROCK blocks both PGF2␣and LPS-induced preterm labor in mice (48 -50). However, the extent to which MAPK contributes to myometrial contraction and preterm labor remains an open question (51). That PDC113.824 sensitizes PGF2␣-dependent activation FIGURE 10. PDC113.824-mediated biased signaling effects through FP receptor. PGF2␣ induces ERK1/2 activation via G␣ q , and actin reorganization and contraction through G␣ 12 . PDC113.824 increases the coupling of the FP receptor to G␣ q , which increases PGF2␣-mediated activation of PKC␤I and ERK1/2 (i.e. acts as a positive allosteric modulator, PAM). In contrast, PDC113.824 reduces PGF2␣-induced cytoskeletal reorganization, modulation of cell ruffling, and myometrial cell contraction via decreased coupling of FP to G␣ 12 (i.e. acts as a negative allosteric modulator, NAM). FP receptor-mediated activation of phospholipase C␤ (PLC␤) and production of inositol trisphosphate (IP 3 ) promotes intracellular Ca 2ϩ release, which stimulates Ca 2ϩ -dependent, calmodulin-mediated activation of myosin light chain kinase and subsequent phosphorylation of myosin light chain to promote smooth muscle cell contraction. Activation of the Rho-ROCK pathway through G␣ 12 facilitates inhibitory regulation of the myosin light chain (MLC) phosphatase and the increase in myosin light chain phosphorylation, which maintains the cell in the contracted state (50).
of ERK1/2 in myometrial cells, while inhibiting myometrial contraction and parturition in mice, suggests that MAPK plays a minor role in uterine contraction. Development of biased FP receptor ligands that selectively engage the G␣ q -PKC-MAPK signaling pathway, without affecting the G␣ 12 -Rho-ROCK signaling pathway, will be of great value in addressing this issue.
To date, only a very few examples of biased allosteric modulators have been described that direct GPCR signaling toward distinct effector pathways. The metal ion Gd 3ϩ allosterically modulates the orthosteric ligand glutamate for the mGluR1 homodimer to promote differential coupling of the receptor to either G␣ q or G␣ s (52). It is unlikely, however, that Gd 3ϩ acts in a selective manner, because it could affect other class C GPCRs. A calcium-sensing receptor autoantibody has also been shown to potentiate the Ca 2ϩ /G␣ q response, while inhibiting the G␣ idependent activation of MAPK (53). Recently, a drug screen has identified a new allosteric antagonist of NK2 receptor, which biases the receptor toward increased Ca 2ϩ signaling, while inhibiting cAMP production (54). PDC113.824 represents a significant addition to this new "repertoire" of allosteric modulators that bias receptors toward distinct G protein-dependent signaling events. Our findings are also distinguishable, because they highlight the possibility of developing GPCR-specific synthetic allosteric and biased modulators by constructing mimics to particular regions of a given receptor. Ligands acting on orthosteric sites have also been shown to bias GPCR signaling (55,56). However, because allosteric sites on receptors are presumably more diverse than orthosteric sites, it is likely that many allosteric ligands described to act on GPCRs will also turn out to have unsuspected biased signaling properties.
The clinical use of allosteric compounds has recently attracted more attention (57)(58)(59). These modulators can be used at saturating concentrations, because their effects are only revealed in the presence of endogenous ligands (e.g. neutral allosteric ligands), potentially reducing adverse effects (60 -62). The design of allosteric ligands with biased signaling properties, as in the case of PDC113.824, offers not only the advantage of specificity for a single GPCR, but also selectivity for a specific subset of signaling pathways, further reducing unwanted side effects. At present, tocolytic drugs used in clinic have significant off target and/or non-selective actions (14,15). Sympathomimetics (e.g. ␤-agonists), or non-steroidal anti-inflammatory drugs (e.g. indomethacin) target multiple tissues and organs leading to unwanted responses in both the mother and fetus. Moreover, the benefit of oxytocin receptor blockade using antagonists (e.g. Atosiban) in preventing pre-term labor remains limited, because the oxytocin receptor, in contrast to FP receptor, is not involved in regulating the initial stages of preterm parturition (7,8). Thus, the future development of biased, allosteric compounds specific for FP receptor will not only help further our understanding of mechanisms underlying parturition, but may also contribute to the design of better and more selective tocolytic drugs.