Development and Characterization of Pepducins as Gs-biased Allosteric Agonists*♦

Background: A Gs-biased agonist for the β2-adrenergic receptor (β2AR) has yet to be reported. Results: A screen of β2AR pepducins identified receptor-dependent and receptor-independent pepducins that selectively activate Gs. Conclusion: Receptor-dependent pepducins promote a Gs-biased conformation of the β2AR, whereas receptor-independent pepducins directly activate Gs. Significance: Gs-biased pepducins provide a valuable tool for the continued study of β2AR function and may prove useful as next-generation asthma therapeutics. The β2-adrenergic receptor (β2AR) is a prototypical G protein-coupled receptor that mediates many hormonal responses, including cardiovascular and pulmonary function. β-Agonists used to combat hypercontractility in airway smooth muscle stimulate β2AR-dependent cAMP production that ultimately promotes airway relaxation. Chronic stimulation of the β2AR by long acting β-agonists used in the treatment of asthma can promote attenuated responsiveness to agonists and an increased frequency of fatal asthmatic attacks. β2AR desensitization to β-agonists is primarily mediated by G protein-coupled receptor kinases and β-arrestins that attenuate receptor-Gs coupling and promote β2AR internalization and degradation. A biased agonist that can selectively stimulate Gs signaling without promoting receptor interaction with G protein-coupled receptor kinases and β-arrestins should serve as an advantageous asthma therapeutic. To identify such molecules, we screened ∼50 lipidated peptides derived from the intracellular loops of the β2AR, known as pepducins. This screen revealed two classes of Gs-biased pepducins, receptor-independent and receptor-dependent, as well as several β-arrestin-biased pepducins. The receptor-independent Gs-biased pepducins operate by directly stimulating G protein activation. In contrast, receptor-dependent Gs-biased pepducins appear to stabilize a Gs-biased conformation of the β2AR that couples to Gs but does not undergo G protein-coupled receptor kinase-mediated phosphorylation or β-arrestin-mediated internalization. Functional studies in primary human airway smooth muscle cells demonstrate that Gs-biased pepducins are not subject to conventional desensitization and thus may be good candidates for the development of next generation asthma therapeutics. Our study reports the first Gs-biased activator of the β2AR and provides valuable tools for the study of β2AR function.

␤-arrestin-mediated internalization. Functional studies in primary human airway smooth muscle cells demonstrate that G sbiased pepducins are not subject to conventional desensitization and thus may be good candidates for the development of next generation asthma therapeutics. Our study reports the first G s -biased activator of the ␤ 2 AR and provides valuable tools for the study of ␤ 2 AR function.
The ␤ 2 -adrenergic receptor (␤ 2 AR) 3 is a G protein-coupled receptor (GPCR) responsible for hormonal signal transduction in functions such as cardiac muscle contraction, airway smooth muscle relaxation, and blood vessel dilation. The ␤ 2 AR has served as a prototypical model for understanding GPCR signaling and regulation (1). Crystallographic and biophysical characterization has provided insight into the structure of the basal state of the receptor as well as the conformational changes associated with agonist-stimulated receptor activation and G protein binding (2)(3)(4)(5)(6)(7)(8). A diverse set of ligands for the ␤ 2 AR have also been developed that are now mainstays in the clinic (9 -13). ␤-Antagonists have been used extensively in the treatment of hypertension, and some inverse agonists such as carvedilol have been used in the treatment of congestive heart failure (11,13). ␤ 2 AR agonists, including salbutamol and formoterol, are com-monly prescribed drugs for the treatment of asthma and chronic obstructive pulmonary disease (9,12).
Asthma is a chronic condition by which airway inflammation and bronchoconstriction promote peak airflow restriction. Bronchotone, determined by the contractile state of airway smooth muscle (ASM), is the product of differential signaling through a number of GPCRs. These include the ␤ 2 AR, which is a critical regulator of airway smooth muscle relaxation. ␤-Agonists stimulate G s activation leading to an increase in intracellular cAMP and ASM relaxation (9). Iterative ␤ 2 AR signaling is regulated by GPCR kinase (GRK)-mediated phosphorylation and subsequent ␤-arrestin recruitment that promotes receptor desensitization, internalization, and degradation (1,14). Current therapeutic strategies to combat airway constriction include the use of both short acting and long acting ␤-agonists (15). Treatment with long acting ␤-agonists has been linked to an increased incidence of an asthmatic episode that results in fatality (16). Prolonged ␤ 2 AR stimulation promotes recalcitrance to ␤-agonists by increasing receptor desensitization and degradation. The recruitment of ␤-arrestins may play an important role in the mechanism behind the severe side effect of long acting ␤-agonist use, as a murine model demonstrated the ability of ␤-arrestin2 to attenuate ␤-agonist-stimulated cAMP production, whereas a ␤-arrestin knockdown in human ASM partially reduced ␤ 2 AR desensitization (17).
Traditional ␤-agonists operate through the extracellular ligand-binding pocket to propagate intracellular signaling (1). One strategy to potentially modulate ␤ 2 AR interaction with G proteins, GRKs, and ␤-arrestins is to target the intracellular surface of the receptor using pepducins. Pepducins are cellpenetrating palmitoylated peptides derived from the intracellular loops of a GPCR (18). Pepducins have been generated from many GPCRs, including PAR1, PAR2, PAR4, sphingosine 1-phosphate receptor-3, formyl peptide receptor 2, melanocortin-4 receptor, Smoothened, CXCR1, CXCR2, and CXCR4, and have been shown to function as allosteric agonists or antagonists of their cognate receptor (19 -25). A recent study also found that pepducins might function in a biased manner as the CXCR4 pepducin ATI-2341 selectively promoted interaction with G i over G 13 , GRKs and ␤-arrestins that are typically associated with CXCR4 stimulation (26). Although the mechanism of action is unclear, pepducins are proposed to directly interact with a receptor and allosterically modulate receptor signaling (18).
In this study, we focused on determining whether pepducins derived from the ␤ 2 AR could function as biased modulators. We identified multiple G s -biased pepducins that stimulated cAMP production without the recruitment of ␤-arrestins to the ␤ 2 AR as well as several ␤-arrestin-biased pepducins. The G sbiased pepducins did not promote ␤ 2 AR desensitization, GRKmediated phosphorylation, or ␤-arrestin-mediated internalization over an extended time course. These pepducins fell into two classes with receptor-independent pepducins promoting cAMP production by direct activation of G s , whereas receptordependent pepducins induced a ␤ 2 AR conformation that selectively activated G s . These pepducins are the first reported G sbiased molecules operating through the ␤ 2 AR and show promise in the development of next generation asthma therapeutics.
cAMP Measurement-HEK 293 cells were grown to confluency in 24-well plates at 37°C in Dulbecco's modified Eagle's medium (DMEM, Cellgro) supplemented with 10% fetal bovine serum (FBS). Cells were stimulated with 1 M isoproterenol, 5 M salbutamol, or 10 M pepducin for various times at 37°C in the presence of 0.5 mM 3-isobutyl-1-methylxanthine (IBMX). For the initial screen of all ICL1 and ICL3-1 to ICL3-11 pepducins, cells were lysed by adding 270 l of 0.1 M HCl followed by 20 min at room temperature on an orbital shaker. Lysates were cleared by centrifugation at 1000 ϫ g for 10 min. cAMP levels were measured using the cyclic AMP EIA kit following the manufacturer's instructions (Cayman Chemical). In all other cAMP measurements, stimulation was stopped on ice by aspirating the media, adding 500 l of ice-cold ethanol, and incubating for 2 h at room temperature on an orbital shaker. Samples were lyophilized until dry and resuspended in 300 l of assay buffer (50 mM sodium acetate, pH 6.2). cAMP was measured by radioimmunoassay using an anti-cAMP antibody (a generous gift from Dr. Mario Ascoli, University of Iowa) and 125 I-labeled cAMP tracer (Biomedical Technologies, Inc., and PerkinElmer Life Sciences) as described (27).
Analysis of ␤-Arrestin2 Binding to the ␤ 2 AR Using Bioluminescence Resonance Energy Transfer (BRET)-␤-Arrestin2 recruitment was monitored following the protocol of Hamdan et al. (28). HEK 293 cells were grown in 6-well plates to 80% confluence in DMEM with 10% FBS. Cells were co-transfected with pcDNA3-␤-arrestin2-GFP10 (energy acceptor) and pcDNA3-␤ 2 AR-RLucII (energy donor) using Lipofectamine 2000 (Invitrogen) for 4 h in serum-free OptiMEM (Invitrogen). Cells were allowed to recover overnight in growth media and then replated in poly-L-ornithine (Sigma)-coated opaque 96-well plates (Optiplate, PerkinElmer Life Sciences) at a density of 100,000 cells per well. After overnight incubation at 37°C in DMEM with high glucose (Invitrogen), cells were washed three times with PBS plus glucose (Invitrogen) and incubated with PBS plus glucose. Coelenterazine 400a was added to 2.5 M final concentration and incubated at 37°C for 2 min. BRET was measured at 510 nm following addition of ␤-agonist or pepducin using a Tecan Infinite F500 microplate reader. BRET ratios were calculated as the light emitted by the GFP10 acceptor (510 nm) divided by the total light emitted by the donor RLucII (400 nm). ⌬BRET was calculated by subtracting the BRET ratio of the unstimulated trials from the stimulated trials.
Detection of ␤ 2 AR Phosphorylation Using Phosphospecific Antibodies-HEK 293 cells stably overexpressing FLAG-␤ 2 AR (a generous gift from Dr. Mark von Zastrow, University of Cal-ifornia, San Francisco) were grown to confluency in 10-cm dishes at 37°C in DMEM supplemented with 10% FBS and 500 g/ml G418 sulfate (Cellgro). Cells were stimulated with 1 M isoproterenol, 5 M salbutamol, or 10 M pepducin for given time points at 37°C. Media were removed, and cells were washed on ice three times with PBS (Cellgro). Cells were lysed on ice by the addition of 500 l of lysis buffer (20 mM Tris-HCl, pH 7.5, 100 mM NaCl, 2 mM EDTA, 1% Triton X-100, 1 Complete mini protease inhibitor tablet, and 1 PhosSTOP phosphatase inhibitor tablet (Roche Applied Science)). Cells were scraped, briefly sonicated, and cleared by centrifugation at 1000 ϫ g for 10 min. Equal protein concentrations were immunoprecipitated using rabbit polyclonal anti-FLAG (Sigma) and protein A-agarose beads (Roche Applied Science) for the detection of PKA phosphorylation. For detection of GRK phosphorylation, cell lysates were immunoprecipitated using mouse monoclonal M2 anti-FLAG (Sigma) and protein G-agarose PLUS beads (Santa Cruz Biotechnology). Samples were incubated overnight at 4°C and briefly centrifuged to pellet beads from immunodepleted lysate. Pelleted beads were washed with lysis buffer three times, and the washed pellets were resuspended in 60 l of 2ϫ Laemmli buffer. Immunoprecipitated proteins were separated by SDS-PAGE on a 10% polyacrylamide gel and receptor phosphorylation was analyzed by Western blotting. GRK phosphorylation was detected using a phosphospecific antibody (1:500) against ␤ 2 AR phosphoserines 355 and 356 (Ser(P) 355/6 ) (Santa Cruz Biotechnology). Equal receptor loading was confirmed by blotting using rabbit polyclonal anti-FLAG (Sigma), at 1:1000 in Tris-buffered saline with Tween 20 (TBST, 20 mM Tris-HCl, pH 7.5, 150 mM NaCl, and 0.1% Tween 20) plus 5% BSA, and rabbit anti-human ␤ 2 AR (Santa Cruz Biotechnology) at 1:500 (in TBST with 5% BSA). PKA phosphorylation was detected by blotting using a phosphospecific antibody, 2G3, against phosphoserine 262 (a generous gift from Dr. Richard Clark, University of Texas at Houston) at a 1:1000 dilution (29). Equal receptor loading was confirmed by blotting using mouse monoclonal M2 anti-FLAG (Sigma) at 1:1000. Chemiluminescence was measured using Pico chemiluminescent substrate (Thermo Scientific).
Receptor Internalization-HEK 293 cells stably overexpressing FLAG-␤ 2 AR were seeded into 24-well plates precoated with poly-L-lysine (Sigma) at a density of 150,000 cells per well and grown at 37°C in DMEM supplemented with 10% FBS and 500 g/ml G418 sulfate (Cellgro). At confluency, cells were washed once with warm DMEM and then treated with 1 M isoproterenol, 5 M salbutamol, or 10 M pepducin in complete media for given time points at 37°C. The media were removed, and cells were fixed on ice with 3.7% paraformaldehyde in Tris-buffered saline (TBS) for 10 min. Cells were washed twice with TBS and blocked for 45 min with blocking buffer (TBS, 1% BSA, and 1 mM CaCl 2 ) at room temperature. Cell surface FLAG-␤ 2 AR was detected by ELISA as described previously (30). Briefly, cells were incubated with rabbit polyclonal anti-FLAG (Sigma) for 1 h at room temperature, anti-rabbit HRP secondary antibody (Vector Laboratories) for 1 h at room temperature, and then washed twice with cold blocking buffer, developed by adding a one-step 2,2Ј-azinobis(3-ethylbenzothiazoline-6-sulfonic acid) (ABTS) substrate (Thermo Scientific), and incubated at room temperature for 25 min. 100 l of the developed solution was transferred to a 96-well plate, and the absorbance was measured on a plate reader (Bio-Rad) at 405 nm.
Functional Desensitization-Primary human airway smooth muscle cells were isolated from donors with no chronic illness or medication use. ASM cell cultivation and characterization were described previously (31,32). Passages 4 -7 ASM cells were maintained in Ham's F-12 medium supplemented with 10% FBS. Use of human ASM cells does not constitute research of human subjects because all donor tissue was harvested anonymously and de-identified. For assays measuring total cAMP, primary human ASM cells were seeded into 24-well plates precoated with poly-L-lysine in DMEM and 10% FBS and grown to confluence at 37°C. Media were removed, and wells were washed twice with Dulbecco's phosphate-buffered saline (DPBS, Cellgro), and 500 l of DPBS with calcium and magnesium was added to each well. Cells were then treated with 1 M isoproterenol, 5 M salbutamol, or 10 M pepducin for given time points. Stimulation was stopped on ice by adding 750 l of ice-cold ethanol and incubating for 2 h at room temperature on an orbital shaker. cAMP was measured using the 125 I-labeled cAMP radioimmunoassay protocol described above. For assays measuring intracellular cAMP levels, the same procedure was performed as above except the DPBS was removed before the addition of ice-cold ethanol.
Pepducin Specificity-CHO-K1 cells were seeded into 6-well plates and grown to 60% confluency at 37°C in Ham's F-12 media supplemented with 10% FBS. CHO-K1 cells were transfected with pcDNA3-FLAG-␤ 1 AR (a generous gift from Dr. Robert Lefkowitz, Duke University), pcDNA3-FLAG-␤ 2 AR and pcDNA3-FLAG-EP 2 R (a generous gift from Dr. Raymond Penn, Thomas Jefferson University), or pcDNA3 for 4 h with Lipofectamine 2000 (Invitrogen). The cells were allowed to recover for 48 h and then treated with 1 M isoproterenol, 10 M PGE 2 , or 10 M pepducin for 10 min in the presence of IBMX. Stimulation was stopped on ice by aspirating the media, adding 500 l of ice-cold ethanol, and incubating for 2 h at room temperature on an orbital shaker. cAMP was measured by the radioimmunoassay described above.
Expression and Purification of G s Heterotrimer-Bovine G␣ s short, His 6 -rat G␤ 1 , and bovine G␥ 2 were expressed in High Five insect cells (Expression Systems Inc.) grown in ESF921 media (Expression Systems Inc.). Cultures were grown to a density of 3 million cells/ml and then infected with two separate viruses containing the G␣ s and G␤␥ cDNAs at a 1:1 multiplicity of infection. After 48 h of incubation, the infected cells were harvested by centrifugation and resuspended in 200 ml of lysis buffer (20 mM HEPES, pH 7.5, 100 M MgCl 2 , 5 mM ␤-mercaptoethanol, 10 M GDP, 2.5 g/ml leupeptin, and 160 g/ml benzamidine) per liter of cell culture for 30 min. Lysates were centrifuged for 10 min at 18,000 ϫ g and then resuspended in 100 ml of solubilization buffer (20 mM HEPES, pH 7.5, 100 mM NaCl, 1% sodium cholate, 0.05% dodecylmaltoside (DDM), 5 mM MgCl 2 , 5 l of calf intestinal alkaline phosphatase (Sigma), 5 mM ␤-mercaptoethanol, 10 M GDP, 5 mM imidazole, 2.5 g/ml leupeptin, and 160 g/ml benzamidine). Samples were Dounce-homogenized for 20 strokes and stirred for 1 h at 4°C followed by centrifugation at 18,000 ϫ g for 30 min. 2 ml of pre-equilibrated nickel-nitrilotriacetic acid resin per liter of cell culture was added to the solubilized supernatant and stirred for 1.5 h at 4°C. Bound G s was collected by centrifugation at 4000 ϫ g for 10 min and washed with solubilization buffer three times in a 50-ml conical tube. The G s -bound resin was washed for 30 min with 50% E1 buffer (solubilization buffer plus 15 mM imidazole) plus 50% E2 buffer (20 mM HEPES, pH 7.5, 50 mM NaCl, 0.1% DDM, 1 mM MgCl 2 , 5 mM ␤-mercaptoethanol, 20 M GDP, 20 mM imidazole) followed by washes with 25% E1/75% E2, 10% E1/90% E2, and 5% E1/95% E2. G s was eluted with E2 buffer supplemented with 200 mM imidazole. The eluate was passed through a 0.2-m filter and applied to ion exchange chromatography as described previously (6) to separate G s heterotrimer from free G␤␥ complex. The fractions containing pure G s heterotrimer were pooled and dialyzed against 20 mM HEPES, pH 7.5, 100 mM NaCl, 0.1% DDM, 1 mM MgCl 2 , 100 M tris(2-carboxyethyl)phosphine (TCEP), and 20 M GDP. The protein was concentrated to ϳ15 mg/ml using a 100-kDa cutoff concentrator (Millipore). and glycerol was added to 15%. The final sample was aliquoted, flash-frozen, and stored at Ϫ80°C.
Expression and Purification of ␤ 2 AR from Baculovirusinfected Sf9 Cells-Recombinant baculovirus was prepared using Bestbac expression system (Expression Systems Inc.) with pVL1392 as vector. The full-length ␤ 2 AR (termed "PN1") was expressed by infecting Sf9 cells at a density of 4.5 million cells/ml with second passage baculovirus stock using 20 ml of virus stock per liter of cell culture. 1 M alprenolol was added to stabilize the receptor during expression. The infected cells were harvested after 48 h of incubation at 27°C. Cell pellets were lysed by stirring in lysis buffer for 20 min (20 mM HEPES, pH 7.5, 5 mM EDTA, 1 M alprenolol, 2.5 g/ml leupeptin, 160 g/ml benzamidine; 10 ml of buffer/g of cell pellet). The receptor was then extracted from the membrane using Dounce homogenization in solubilization buffer (20 mM HEPES, pH 7.5, 100 mM NaCl, 1% DDM, 1 M alprenolol, 2.5 g/ml leupeptin, 160 g/ml benzamidine) for 1 h at room temperature. 10 ml of solubilization buffer was added per g of cell pellet. After addition of 2 mM CaCl 2 , the solubilized receptor was clarified by high speed centrifugation at 18,000 ϫ g for 30 min. The N-terminal FLAG-tagged receptor was then captured by M1 antibody affinity chromatography (Sigma). The column was extensively washed with HMS-CHS buffer (20 mM HEPES, pH 7.5, 350 mM NaCl, 0.1% DDM, 0.01% cholesterol hemisuccinate) plus 2 mM CaCl 2 to remove impurities and alprenolol. The receptor was then eluted with HMS-CHS buffer supplemented with 5 mM EDTA and 200 g/ml free FLAG peptide.
Analysis of G s Activation by [ 35 S]GTP␥S Binding-Purified G s was diluted to 3 M in 10 mM HEPES, pH 8, 1 mM EDTA, and 0.1% Lubrol and then diluted 2-fold in 50 mM HEPES, pH 8, 1 mM EDTA, 125 mM MgCl 2 , and 200 mM NaCl. Pepducins or 0.5% DMSO were incubated with G s for 15 min on ice, and the binding reaction was initiated by addition of 10 M GTP␥S (cold plus hot, ϳ1300 cpm/fmol). Samples were incubated for 15 min at 4°C and then quenched by addition of cold GTP␥S wash buffer (20 mM Tris-HCl, pH 8, 25 mM MgCl 2 , 100 mM NaCl) followed by rapid filtration through BA85 filters (Millipore). The filters were washed four times with 4 ml of cold GTP␥S wash buffer, and [ 35 S]GTP␥S binding was quantified by liquid scintillation counting.
We also evaluated G s activation in lipid bicelles. Purified G s (18 M) in 2% 3:1 dimyristoyl phosphatidylcholine (DOPC)/ CHAPSO bicelles with 1.13 mM CHS, 20 mM HEPES, pH 7.5, and 100 mM NaCl was incubated in the presence or absence of ␤ 2 AR (1.26 M) for 2 h on ice to allow protein incorporation into the lipid bicelles. 2 l of reconstituted ␤ 2 AR-G s or reconstituted G s alone was diluted 200-fold in 20 mM HEPES, pH 7.5, 150 mM NaCl, 1 mM MgCl 2 , and 38.5 nM [ 35 S]GTP␥S (Perkin-Elmer Life Sciences). 20-l reactions were initiated by the addition of 1 M isoproterenol or 10 M pepducin and incubated for the indicated times at room temperature. Non-pepducin trials included 0.05% DMSO. Bound [ 35 S]GTP␥S was collected by rapid filtration on GF/B filters (Whatman), washed four times with 4 ml of cold GTP␥S wash buffer, and analyzed by liquid scintillation counting.
Analysis of G␣ s Engagement to the ␤ 2 AR Using BRET-G␣ s interaction with the ␤ 2 AR was assayed by BRET using a ␤ 2 AR construct tagged with GFP10 at the C terminus of the receptor and G␣ s constructs tagged with RLucII either at the N terminus (RLucII-G␣ s ) or at residue 67 of G␣ s (G␣ s 67-RLucII). For the G␣ s 67-RLucII studies, HEK 293T cells were cultured in DMEM supplemented with 10% FBS, 0.2 units/ml penicillin, 100 g/ml streptomycin (Wisent Inc.) and were seeded in 6-well plates at 600,000 cells/well 24 h before transfection. Transient transfections with ␤ 2 AR-GFP10 and G␣ s 67-RLucII in the presence of untagged G␤ 1 and G␥ 2 were performed using linear polyethyleneimine, 25-kDa ( Polysciences, Inc.), as transfecting agent at a 3:1 ratio of polyethyleneimine/DNA. Two h after transfection, culture medium was replaced with fresh media, and the cells were then maintained in culture for 48 h before BRET experiments. The expression level of the acceptor was determined as total fluorescence, using a FlexStationII fluorometer (Molecular Devices) with 400-nm excitation and 510-nm emission filters. The expression level of the donor was measured as total luminescence, using a Mithras LB940 Multimode Microplate Reader (Berthold Technologies), following the addition of 2.5 M coelenterazine 400a. Cells were washed once with Hanks' balanced salt solution (Invitrogen) containing 20 mM HEPES (HBSS) and detached in HBSS supplemented with 0.1% BSA (HBSS/BSA) (Sigma) at room temperature. 100,000 cells/well were then distributed in a white 96-well microplate (Greiner). Cells were then treated with or without different concentrations of ligand, and BRET values were collected using the Mithras LB940 Reader equipped with BRET400-GFP10 filter set (acceptor, 515 Ϯ 20-nm, and donor, 400 Ϯ 70-nm filters), following the addition of coelenterazine 400a. BRET signals were determined as the ratio of the light emitted by the acceptor over donor. The specific BRET signal (net BRET) was determined by subtracting the background signal detected in cells transfected with the luciferase donor alone from the BRET obtained in cells expressing both energy donor and acceptor. The ligand-promoted BRET signal (⌬BRET) was calculated by subtracting the BRET values obtained in the vehicle condition from the one measured in the presence of ligand.
For the RLucII-G␣ s studies, HEK 293 cells were grown in 6-well plates to 80% confluence in DMEM with 10% FBS. Cells were co-transfected with pcDNA3.1-RLucII-G␣ s (donor) and pGFP-␤ 2 AR-GFP10 (acceptor) using Lipofectamine 2000 (Invitrogen) for 4 h in serum-free OptiMEM (Invitrogen). Cells were allowed to recover overnight in growth media and then replated in poly-L-ornithine (Sigma)-coated opaque 96-well plates (Optiplate, PerkinElmer Life Sciences) at a density of 100,000 cells per well. After overnight incubation at 37°C in DMEM with high glucose (Invitrogen), cells were washed three times with PBS plus glucose (Invitrogen) and incubated with PBS plus glucose. Coelenterazine 400a was added to 2.5 M final concentration and incubated at 37°C for 2 min. BRET was measured at 510 nm following addition of 1 M isoproterenol or 10 M pepducin using a Tecan Infinite F500 microplate reader. BRET ratios were calculated as the light emitted by the GFP10 acceptor (510 nm) divided by the total light emitted by the RLucII donor (400 nm). ⌬BRET was calculated by subtracting the BRET ratio of the unstimulated trials from the stimulated trials.
[ 125 I]Iodocyanopindolol Binding-HEK 293 cells stably expressing FLAG-␤ 2 AR were isolated and washed three times with assay buffer (HBSS with calcium and magnesium, 0.1% BSA, pH 7.4), diluted to 50,000 cells/ml, and incubated with 1 nM [ 125 I]iodocyanopindolol in the presence or absence of pepducin or propranolol for 1 h at 25°C. Incubations were terminated by the addition of 4 ml of cold assay buffer and rapid filtration on GF/B filters. Filters were washed four times with 4 ml of cold assay buffer, and [ 125 I]iodocyanopindolol binding was quantitated by gamma emission counting.
Monobromobimane Labeling of ␤ 2 AR-Purified FLAG-␤ 2 AR and 20 M monobromobimane (Invitrogen) were incubated for 1 h on ice for labeling. The monobromobimane-labeled receptor was then purified by affinity chromatography using alprenolol-Sepharose as described previously to select functional receptors (4). 300 M alprenolol was used to elute the receptor to a tandemly linked M1 FLAG column. The column was washed with HMS-CHS buffer for removal of alprenolol to prepare unliganded receptor. The receptor was then eluted from M1 resin with HMS-CHS buffer supplemented with 5 mM EDTA, 200 g/ml free FLAG peptide. Size-exclusion chromatography on a Superdex-200 column (GE Healthcare) equilibrated in HMS-CHS buffer was used to increase the purity. The receptor was concentrated to 125 M with purity greater than 95% as assessed by SDS-PAGE.
Analysis of Monobromobimane-␤ 2 AR Fluorescence-Monobromobimane-labeled ␤ 2 AR (mBB-␤ 2 AR) was incorporated into 2% DOPC/CHAPSO (3:1) with 1.13 mM CHS lipid bicelles by incubating for 30 min on ice. Lipid bicelles containing 50 nM mBB-␤ 2 AR were incubated for 15 min at 25°C in 20 mM HEPES, pH 7.5, 100 mM NaCl with isoproterenol or pepducin. Isoproterenol samples also contained 0.1 or 0.5% DMSO to account for the pepducin solvent. In experiments using G s , 200 nM G s was incubated for 20 min at 25°C alone or post-agonist addition depending on the experimental setup. mBB-␤ 2 AR fluorescence was measured by excitation at 370 nm and recording emission from 430 to 490 nm at 1-nm increments with 1 nm s Ϫ1 integration on a Spex FluoroMax-3 spectrofluorometer (Jobin Yvon Inc.) in photon counting mode set at a 4-nm emission bandwidth pass. Background fluorescence contributed by the assay buffer and ligand was subtracted from the experimental spectra.

Characterization of a Library of ␤ 2 AR Pepducins-Recent
studies have shown that a pepducin from the first intracellular loop (ICL) of CXCR4 can effectively activate G i without promoting appreciable coupling to G 13 , GRKs, or ␤-arrestins (26).
In an effort to identify G s -biased pepducins, we synthesized a library of 51 pepducins corresponding to sequences derived from ICL1, ICL2, and ICL3 of the human ␤ 2 AR, a GPCR that primarily couples to G s (Fig. 1). These pepducins were then screened for their ability to promote cAMP production in HEK 293 cells. This screen yielded multiple pepducins that promote cAMP production with four demonstrating efficacy comparable with the partial agonist salbutamol (Fig. 2A). The majority of these pepducins were from the proximal portion of ICL3 (ICL3-2, ICL3-7,and ICL3-8), and others were from the central region of ICL3 (ICL3-9) or from ICL1 (ICL1-15).
To gain a more complete understanding of the diverse signaling profiles from the ␤ 2 AR pepducin library, all pepducins were also analyzed for their ability to promote ␤-arrestin recruitment to the ␤ 2 AR using BRET. This assay involved treating HEK 293 cells co-expressing a ␤ 2 AR-Renilla reniformis luciferase II fusion (␤ 2 AR-RLucII) and GFP10-tagged ␤-arres-tin2 with the various pepducins ( Fig. 2B) (28). Several ICL1derived pepducins effectively promoted ␤-arrestin2 interaction with the ␤ 2 AR, whereas ICL2 and ICL3 pepducins had no effect ( Fig. 2C and data not shown). Thus, multiple pepducins derived from ICL3 activate cAMP accumulation without promoting ␤-arrestin binding to the ␤ 2 AR and therefore appear to function as G s -biased allosteric agonists, whereas several pepducins from ICL1 (ICL1-4, ICL1-11, and ICL120) appear to be ␤-arrestin-biased allosteric agonists as they effectively stimulate ␤-arrestin2 engagement with the ␤ 2 AR without promoting any cAMP production (Fig. 2D).
Because the primary goal of this study was to develop G sbiased agonists, further characterization was limited to two of the candidate G s -biased pepducins, ICL3-8 and ICL3-9. ICL3-8 is representative of a family of sequence-related pepducins from the proximal portion of ICL3, whereas ICL3-9 is primarily from the central portion of ICL3 (Fig. 1). Both pepducins promoted ϳ40% cAMP production compared with the full agonist isoproterenol with ICL3-8 having an EC 50 of 577 Ϯ 14 nM and ICL3-9 an EC 50 of 4.7 Ϯ 0.1 M (Fig. 2E).
G s -biased Pepducins Do Not Induce Receptor Desensitization-Agonist-specific desensitization of the ␤ 2 AR is primarily mediated by GRK phosphorylation of the receptor, which promotes high affinity binding of ␤-arrestins and attenuates G protein coupling (1). Phosphorylation of Ser 355 and Ser 356 (Ser 355/6 ) on the C-terminal tail of the ␤ 2 AR has been attributed to GRK5/6 and is partially responsible for ␤-arrestin recruitment and receptor internalization (Fig. 3A) (29,33,34). Ligand-promoted phosphorylation of the ␤ 2 AR was monitored using phosphospecific antibodies targeting Ser 355/6 . Stimulation with either isoproterenol or salbutamol induced rapid and robust phosphorylation at Ser 355/6 , whereas treatment with either ICL3-8 or ICL3-9 did not induce appreciable receptor phosphorylation at this site (Fig. 3B).
Distinct phosphorylation sites for the cAMP-dependent protein kinase (PKA) have been identified at Ser 261 and Ser 262 in ICL3 of the ␤ 2 AR (Fig. 3A) as well as in the C-terminal tail (Ser 345 and Ser 346 ). To monitor ligand-promoted phosphorylation by PKA, we used a phosphospecific antibody against Ser 262 (29). Isoproterenol and salbutamol promoted an increase in phosphorylation that was observed at 5-10 min after stimulation (Fig. 3C). ICL3-8 and ICL3-9 also promoted effective phosphorylation of Ser 262 with kinetics and efficacy similar to that observed with isoproterenol (Fig. 3C). This result correlates well with the ability of ICL3-8 and ICL3-9 to promote cAMP production and thereby activate PKA.
␤-Arrestin recruitment couples the ␤ 2 AR to the internalization machinery leading to a loss of cell surface receptors and further propagating receptor desensitization (35). Agonist-promoted ␤ 2 AR internalization was analyzed post-stimulation with isoproterenol, salbutamol, ICL3-8, and ICL3-9 in HEK 293 cells stably overexpressing FLAG-␤ 2 AR by cell surface ELISA. Although both isoproterenol and salbutamol induced rapid internalization of the receptor, the pepducins did not induce any internalization over a 1-h period (Fig. 4A).
The long term use of ␤-agonists in the treatment of asthma has been implicated in chronic airway desensitization (15, 16, 36 -39). To evaluate whether the G s -biased pepducins induce functional desensitization of the ␤ 2 AR, we studied cAMP production in primary human airway smooth muscle cells. Receptor desensitization was observed post-␤-agonist stimulation as noted by the decreasing rate of total cAMP production over a 2-h time course (Fig. 4B). In contrast, the pepducins promoted a steady rate of total cAMP production, suggesting that signaling through the pepducins is not subject to the conventional desensitization mechanisms (Fig. 4B). The root of the linearity was further characterized by monitoring intracellular cAMP production over time. Isoproterenol and salbutamol stimulated a rapid peak in cAMP levels that decreased over time as desensitized receptors were likely unable to maintain G s activation (Fig. 4C). In contrast, the pepducins slowly achieved a steady state level of cAMP production (Fig. 4C). Taken together, these results reveal that the ICL3-8 and ICL3-9 induce less desensitization compared with isoproterenol and salbutamol.
Mechanism of Pepducin-mediated Activation of G s -Stimulation of the ␤ 2 AR by ␤-agonists promotes rapid engagement of G s to the receptor which, in turn, promotes GDP dissociation, GTP binding, and G protein activation. To assess whether the pepducins promote G s binding to the ␤ 2 AR, we used BRET to measure association of G␣ s 67-RLucII and ␤ 2 AR-GFP10 (Fig.  5A) (40). Upon isoproterenol treatment, G s was rapidly engaged by the ␤ 2 AR as indicated by the change in BRET ratio (Fig. 5B). ICL3-9 also promoted G s interaction with the ␤ 2 AR, although this occurred on a slower time scale (Fig. 5B). The EC 50 value for ICL3-9-promoted ␤ 2 AR-G s interaction was 3.3 Ϯ 0.4 M (Fig. 5C), which is comparable with the EC 50 value observed for cAMP production (Fig. 2E). In contrast, ICL3-8 was unable to stimulate ␤ 2 AR-G s coupling and therefore may be activating G s in a manner independent of receptor-mediated nucleotide exchange (Fig. 5B).
From the two-dimensional screen, it is possible for pepducins to produce a G s -biased profile in either a receptor-dependent or a receptor-independent manner. A receptor-dependent pepducin would stimulate G protein activation by promoting receptor-G protein coupling, whereas a receptorindependent pepducin might directly activate the G protein (Fig. 5D). Receptor dependence was initially assessed in lipid bicelles by monitoring activation of purified G s by GTP␥S exchange in the presence or absence of purified ␤ 2 AR. Both isoproterenol and ICL3-9 were able to promote G protein acti-  is recruited to ␤ 2 AR-RLucII. Upon ␤-arrestin2 binding to the ␤ 2 AR, GFP10 will be within the BRET radius of RLucII allowing GFP10 emission readout to be indicative of ␤-arrestin recruitment. C, HEK 293 cells co-transfected with GFP10-␤-arrestin2 and ␤ 2 AR-RLucII were preincubated with coelenterazine 400a for 2 min and stimulated with 10 M pepducin, 1 M isoproterenol, or 5 M salbutamol for the indicated times. BRET was monitored at the indicated times post-addition. Data are expressed as ⌬BRET as the background BRET has been subtracted. The data are represented by the means of four independent experiments Ϯ S.D. D, cAMP output (10 min) as a function of ␤-arrestin recruitment reveals multiple G s -biased and ␤-arrestin-biased pepducins. Balanced agonists, such as isoproterenol, can effectively promote both cAMP production and ␤-arrestin recruitment with similar efficacies. An agonist that promotes cAMP production more effectively than ␤-arrestin recruitment is a G s -biased agonist (i.e. ICL3-8 and ICL3-9), whereas agonists that couple ␤-arrestins more effectively than stimulating cAMP production are considered ␤-arrestin-biased. E, HEK 293 cells were stimulated with various concentrations of isoproterenol, ICL3-8, or ICL3-9 for 10 min in the presence of 500 M IBMX. ICL3-8 has an EC 50 of 577 Ϯ 14 nM, whereas ICL3-9 has an EC 50 of 4.7 Ϯ 0.1 M. cAMP production is represented as percentage normalized to maximal isoproterenol stimulation. The data are represented by the mean of three independent experiments Ϯ S.D.

Development of G s -biased Pepducins
vation only when the ␤ 2 AR was included in the assay (Fig. 5E). In contrast, ICL3-8 promoted effective GTP␥S binding to G s independent of whether the ␤ 2 AR was present (Fig. 5E). To confirm ICL3-8 as a bona fide direct activator of G s , we also evaluated GTP␥S exchange on purified detergent-solubilized G s . ICL3-8 was found to rapidly and robustly stimulate GTP␥S binding, whereas ICL3-9 had no effect (Fig. 5F). Higher concentrations of pepducin (100 M) were necessary to observe maximal efficacy as the N-terminal lipidation of the pepducin was unable to contribute to its potency in the detergent-solubilized assay. Overall, these studies demonstrate that ICL3-8 directly activates G s , whereas ICL3-9 activates G s in a ␤ 2 AR-dependent manner.
Receptor Specificity of ICL3-9 -The ability of an agonist to promote receptor-dependent activation of downstream signaling is critical in drug targeting and predicting off-target effects. As multiple GPCRs can couple to G s , it is important to define the receptor specificity of ICL3-9. To evaluate this, we transfected CHO-K1 cells with the G s -coupled ␤ 1 AR, ␤ 2 AR, or prostaglandin E2 receptor (EP 2 R). Control-transfected CHO-K1 cells lack endogenous expression of these receptors as an agonist-induced increase in cAMP was only observed in the cells transfected with the specific receptor (Fig. 6). As expected, ICL3-8 activated cAMP production similarly in all of the cell lines further corroborating its receptor-independent activity. In contrast, ICL3-9 effectively stimulated cAMP production in cells expressing either the ␤ 1 AR or ␤ 2 AR but had no effect in EP 2 R-or control-transfected cells (Fig. 6). Thus, ICL3-9 is able to utilize both the ␤ 1 AR and ␤ 2 AR to activate cAMP production suggesting that it can function on closely related family members. ICL1, ICL2, and the proximal and distal portions of ICL3 are highly conserved in ␤ 1 AR and ␤ 2 AR, although the central portion of the ␤ 1 AR ICL3 contains a proline-rich insert that interrupts the partially conserved ICL3-9 sequence. The significant homology between the ␤ 1 AR and ␤ 2 AR along with the same G protein coupling profile may help to explain the dualspecificity of ICL3-9.
Mutagenesis of ICL3-9 -To understand the residues critical for ICL3-9 function, truncation and triple-alanine substitution variants were synthesized and assessed for their ability to stimulate cAMP production (Fig. 7A). As expected, the peptide palmitoylation and amidation were essential for ICL3-9 activity as removing these modifications markedly reduced functionality (Fig. 7B). This is likely due to the inability of the peptides to access the inner leaflet of the cell membrane. Similarly, both the N and C termini also seem critical in ICL3-9 activity as any truncation at these locations fully abrogated its ability to stimulate cAMP accumulation (Fig. 7B).
Triple-alanine substitutions through the central portion of ICL3-9 appeared to have different effects on ICL3-9 functionality. Substitutions in the N-terminal half of ICL3-9, as represented by ICL3-9A1 and ICL3-9A2, reduced the efficacy by ϳ70% (Fig. 7C). Interestingly, the ICL3-9A2 substitution displayed an ϳ25-fold increase in potency despite the reduced efficacy (Fig. 7D). Alanine substitutions in the C-terminal half of ICL3-9 yielded activity-null variants. The loss in efficacy could be attributed to the exchange of critical residues necessary for the pepducin functionality or residues that are participating in the interaction with the ␤ 2 AR. Pepducins that lack residues critical for activity might still have the ability to interact with the receptor and act as an antagonist in competition with ICL3-9, whereas binding-defective mutants would lack the ability to compete with ICL3-9. ICL3-9A3, ICL3-9A4, and ICL3-9A5 were unable to modulate ICL3-9-promoted GTP␥S binding in a ␤ 2 AR-G s binding assay, although ICL3-9A1 and ICL3-9A2 partially reduced G protein activation (Fig. 7E). It is unknown whether our results are assessing the necessity of the substituted residues or whether modulating the secondary structure of the pepducin contributes to the change in efficacy and potency.
Mechanism of Receptor-dependent G s Bias of ICL3-9 -The traditional definition of an orthosteric receptor agonist is a ligand that binds within the ligand-binding pocket of the receptor and elicits a biological response that is subject to inhibition by a receptor antagonist (41,42). To assess whether ICL3-8 or ICL3-9 interact with the ligand-binding pocket of the ␤ 2 AR, we tested their ability to compete for binding of the ␤ 2 AR antagonist [I 125 ]iodocyanopindolol. Although [I 125 ]iodocyanopindolol binding to the ␤ 2 AR was effectively inhibited by propranolol, there was no effect of ICL3-8 or ICL3-9 (Fig. 8A). Thus, ICL3-9 does not appear to interact with the orthosteric ligand-binding pocket of the ␤ 2 AR.
Inverse agonists have the ability to occupy the ligand-binding pocket and, unlike receptor antagonists, attenuate spontaneous signal activation from a receptor (42,43). For the ␤ 2 AR, inverse agonists are proposed to constrict the receptor conformational plasticity needed for spontaneous receptor signaling (44). The G s -biased pepducins were evaluated for sensitivity to two different ␤ 2 AR inverse agonists. The weak inverse agonist propranolol effectively blocked isoproterenol-promoted cAMP production in HEK 293 cells but had no effect on ICL3-8-or ICL3-9-stimulated cAMP production (Fig. 8B). In contrast, the potent inverse agonist ICI118,551 effectively suppressed both isoproterenol-and ICL3-9-promoted cAMP production with an IC 50 of ϳ10 nM, whereas the responsiveness to ICL3-8 was unaffected (Fig. 8C). As ICL3-9 is not sensitive to a weak inverse agonist and is not operating through the ligand-binding pocket, its sensitivity to ICI118,551 likely stems from a competition between an ICL3-9-promoted G s -biased conformation and an ICI118,551-promoted inactive conformation of the ␤ 2 AR.   DECEMBER 26, 2014 • VOLUME 289 • NUMBER 52  ICL3-9 Antagonizes ␤-Agonist-promoted ␤ 2 AR Internalization-As both isoproterenol and ICL3-9 stimulate G s through the ␤ 2 AR, it is not possible to detect a functional difference between the ICL3-9-and isoproterenol-induced active states through monitoring cAMP production. Because isoproterenol promotes ␤-arrestin binding to the receptor, although ICL3-9 does not, if ICL3-9 stabilizes a conformation distinct from that promoted by isoproterenol, it might be able to modulate the efficacy of isoproterenol to promote receptor internalization. Consistent with this notion, ICL3-9 was able to inhibit isoproterenol-promoted ␤ 2 AR internalization in a dose-dependent manner (Fig. 9A). The ability of ICL3-9 to inhibit isoproterenolstimulated internalization can be attributed to reduced ␤-arrestin binding as increasing concentrations of ICL3-9 also attenuate ␤-arrestin2 recruitment to the ␤ 2 AR as monitored by BRET (Fig. 9B). These results suggest that the ␤ 2 AR conformation induced by ICL3-9 appears to be different from that promoted by the ␤-agonist isoproterenol. Alternatively, it is possible that ICL3-9 binding to the ␤ 2 AR might sterically hinder receptor interaction with GRKs and ␤-arrestin.

Development of G s -biased Pepducins
ICL3-9 Promotes a Unique Conformational Change in the ␤ 2 AR-The structure of the ␤ 2 AR-G s complex suggests a large outward movement of TM6 that is unique to the proposed active state of the receptor (6). Site-specific monobromobimane labeling of Cys 265 in TM6 of the ␤ 2 AR allows detection of TM6 movement because Cys 265 moves from a hydrophobic environment to a solvent-exposed position upon receptor activation (5,45,46). Monobromobimane is an environment-sensitive fluorophore that exhibits decreased fluorescence intensity and a red shift of peak emission in polar environments (45). Consequently, a decrease in fluorescence and an increase in max is indicative of a receptor conformational change at the proximal portion of TM6 (Fig. 10A). The addition of isoproterenol to purified Cys 265 monobromobimane-labeled ␤ 2 AR led to a dose-dependent decrease in fluorescence intensity demonstrating the effect of an orthosteric agonist on TM6 movement. In contrast, the addition of ICL3-9 did not promote any change in monobromobimane fluorescence suggesting that ICL3-9 does not promote significant movement of the proximal portion of TM6 (Fig. 10B). Similar results were observed for ICL3-8 (data not shown).  . ICL3-9 antagonizes ␤-agonist promoted ␤ 2 AR desensitization processes. A, 100 nM isoproterenol-induced ␤ 2 AR internalization was monitored by cell surface ELISA in HEK 293 cells stably overexpressing a FLAG-␤ 2 AR that were pretreated with various concentrations of ICL3-9 for 5 min (0.05% DMSO was included in non-pepducin-treated cells). The data are represented by the mean of three independent experiments Ϯ S.D. B, HEK 293 cells cotransfected with GFP10-␤-arrestin2 and ␤ 2 AR-RLucII (same pair as in Fig. 2B) were preincubated with coelenterazine 400a and various concentrations of ICL3-9 for 2 min and stimulated with 100 nM isoproterenol. BRET was monitored at the indicated times post-addition. Background BRET has been subtracted from the plotted points. The data are represented by the mean of four independent experiments Ϯ S.D.
Although G s binding to the ␤ 2 AR is enhanced by ␤-agonists, G s can also couple to unliganded ␤ 2 AR and promote TM6 movement similar to agonist-induced changes (46). Indeed, a 4-fold molar excess of G s led to an ϳ10% decrease in fluorescence intensity and an increase in max (Fig. 10C). Interestingly, a 10-min preincubation of the ␤ 2 AR with 20 M ICL3-9 inhibited G s -promoted TM6 movement (Fig. 10C), whereas addition of the inactive ICL3-9 variant ICL3-9A3 had no effect on G s coupling to the ␤ 2 AR (Fig. 10D). Biochemical analysis and BRET biosensors clearly demonstrate that ICL3-9 promotes ␤ 2 AR-G s coupling, but the inability of G s to induce full TM6 movement on ICL3-9-treated ␤ 2 AR suggests that the G protein may be coupling to the receptor in a different manner than that induced by isoproterenol.
To further monitor the conformational rearrangement between the ␤ 2 AR and G s , we took advantage of a distinct biosensor where the energy donor is positioned at the N terminus of G␣ s (RLucII-G␣ s ) rather than at position 67 (Fig. 11A). Rather than promoting an increase in BRET as is observed with G␣ s 67-RLucII, isoproterenol stimulation promotes a decrease in BRET between ␤ 2 AR-GFP10 and RLucII-G␣ s . Such a difference in the BRET signal orientation reflects the different position of the energy donor and acceptor in the signaling complex and thus provides a means to assess the conformation of the receptor-G s complex following activation from two reference points. Similar biosensors with different relative positions of the energy donor and acceptor have previously been used to probe the conformational changes occurring upon activation of G i by the ␣ 2 -adrenergic receptor (47). Surprisingly, ICL3-9 induced a steady increase in ⌬BRET, in striking contrast to the decreased ⌬BRET observed with isoproterenol treatment (Fig.  11B). This result clearly demonstrates that the ␤ 2 AR conformation promoted by the pepducin is different from that induced by isoproterenol.

DISCUSSION
Conventional stimulation of many GPCRs promotes interaction with heterotrimeric G proteins, GRKs, and ␤-arrestins to mediate G protein signaling, receptor desensitization and internalization, and ␤-arrestin-dependent signaling (1,9). It is now FIGURE 10. ICL3-9 promotes unique conformational changes in the ␤ 2 AR that promote G s coupling. A, monobromobimane is an environmentally sensitive fluorophore that when chemically conjugated to ␤ 2 AR-Cys 265 can indicate local conformational changes. When the ␤ 2 AR is in an inactive state, Cys 265 -monobromobimane is occupying a hydrophobic pocket and fluorescence is high. Upon receptor activation, a large outward movement of TM6 repositions Cys 265 to be solvent-exposed resulting in decreased fluorescence and an increase in max . B, lipid bicelles containing 50 nM monobromobimanelabeled ␤ 2 AR were incubated for 10 min at 25°C with isoproterenol (300 nM or 1 M) or ICL3-9 (20 or 100 M) in 20 mM HEPES, pH 7.5, 100 mM NaCl. Fluorescence spectra were gathered by excitation at 370 nm and scanning 430 -490 nm at 1.0 nm/s. 0.5% DMSO was included in non-pepducin-stimulated samples. C, lipid bicelles containing 50 nM monobromobimane-labeled ␤ 2 AR were incubated for 10 min at 25°C with 20 M ICL3-9. 200 nM G s was then incubated for 20 min at 25°C in co-treatment studies. 0.1% DMSO was included in non-pepducin-stimulated samples. D, lipid bicelles containing 50 nM monobromobimane-labeled ␤ 2 AR were incubated for 10 min at 25°C with 20 M ICL3-9A3. 200 nM G s was then incubated for 20 min at 25°C in co-treatment studies. 0.1% DMSO was included in non-pepducin-stimulated samples.
understood that GPCR ligands cannot be simply classified as agonists or antagonists. Functional studies of a diverse set of ligands for the ␤ 2 AR have demonstrated the complex spectrum of signaling profiles whereby the ␤ 2 AR can couple to downstream pathways with unbalanced efficacies. This concept of "pluridimensional efficacy" or "ligand-biased signaling" was first observed for the ␤ 2 AR when compounds that were previously characterized as receptor antagonists were reported to have the ability to stimulate ␤-arrestin-dependent MAPK signaling (48). Subsequent studies pioneered the re-classification of ␤ 2 AR ligands. Drake et al. (49) monitored the correlation between indicators of receptor desensitization with G protein activation for a subset of receptor agonists and found that most promoted cAMP production and ␤-arrestin recruitment. However, the ␤-agonists N-cyclopentylbutanephrine, isoetharine, and ethyl-norepinephrine demonstrated the ability to more efficiently stimulate ␤-arrestin-dependent pathways over G protein activation. This ␤-arrestin-biased profile was attributed to an agonist-promoted increased rate of GRK phosphorylation of the receptor (49). van der Westhuizen et al. (50) continued the "taxonomy" of ␤ 2 AR ligands by screening a number of full and partial ␤-agonists, neutral antagonists, and inverse agonists for their ability to modulate cAMP, Ca 2ϩ mobilization, ERK1/2 phosphorylation, and ␤ 2 AR internalization. Their findings revealed surprising diversity in the signaling profiles of closely related receptor ligands and further support the functional basis of ligand-biased signaling. Ligand biased signaling is believed to depend on an agonist's ability to induce a receptor conformation that either favors or disfavors interaction with downstream effector proteins. The net result of the interaction (or lack thereof) with these signaling molecules leads to the observed diverse signaling profiles. For the ␤ 2 AR, structural studies further corroborated this notion as distinct ligand-dependent receptor conformations have been observed (5,7,8,45,51).
Pepducin discovery efforts have yielded a variety of allosteric agonists and antagonists for a diverse set of GPCRs over the past decade (18 -25). Most recently, the CXCR4 pepducin ATI-2341 demonstrated receptor-dependent functional selectivity toward G i over G 13 , GRK, and ␤-arrestin (26). To test the concept that pepducins might have a general ability to bias GPCR signaling, we screened 51 pepducins derived from the intracellular regions of the human ␤ 2 AR for their ability to stimulate cAMP production and ␤-arrestin binding to the ␤ 2 AR. This screen identified four distinct classes of pepducins that had agonist-like properties as follows: 1) one that functioned like a partial agonist to activate G s and promote ␤-arrestin binding (ICL1-15); 2) a group that was ␤-arrestin-biased (ICL1-4, ICL1-11, and ICL1-20); 3) one that demonstrated a receptor-dependent G s -bias (ICL3-9); and 4) a group that was G s -biased and receptor-independent (e.g. ICL3-8). We further characterized the pepducins that promoted G s -biased signaling and found that ICL3-8 directly activated G s , whereas ICL3-9 induced conformational changes in the ␤ 2 AR that promoted G s activation.
ICL3-8 is derived from a region of the third intracellular loop that has been proposed to be a critical site of ␤ 2 AR interaction with G s . The crystal structure of the ␤ 2 AR-G s complex suggests extensive interaction between the proximal portion of the ␤ 2 AR ICL3 (QKIDKSEGF) and the ␣5-␤4 loop on G␣ s (6). Interestingly, all of the pepducins from the proximal portion of ICL3 that activated cAMP production, including ICL3-8, contain this contact region between the receptor and G protein.
Thus, this region likely plays a significant role in the guanine nucleotide exchange factor function of an activated receptor. However, it is also worth noting that this sequence alone was not sufficient to activate G s because ICL3-1 through ICL3-8 all contain the QKIDKSEGF motif but vary in their efficacy from no activation of cAMP production (ICL3-5) to high activation (ICL3-2, -7 and -8) ( Fig. 2A). This suggests that the surrounding sequence also contributes to regulating G s interaction and/or activation. More specifically, the adjacent HV residues following the contact motif (missing in the ␤ 2 AR-G s structure) seem to play a role in the activity as pepducins that lack these residues (ICL3-4 and ICL3-5) show a loss of efficacy. In addition, it is possible that the sequence proximal to the QKIDKSEGF motif also contributes to G s activation. Indeed, previous studies demonstrated that peptides from the proximal portion of the hamster ␤ 2 AR ICL3 are direct activators of nucleotide exchange on G s (52). The most effective peptide from these studies was VYS-RVFQVAKRQLQK, which is proximal to the QKIDKSEGF motif. It is interesting that this particular sequence is fully contained within the ICL3-2 pepducin, which was the most effective activator of cAMP production ( Fig. 2A). In the ␤ 2 AR-G s complex structure, this sequence makes extensive contact with the C-terminal tail of the G␣ s subunit. The observed increased efficacy of ICL3-2 may be a product of the substantial contact surface between the pepducin and the G␣ s subunit and the FIGURE 11. ICL3-9 promotes G s coupling to the ␤ 2 AR differently from a ␤-agonist. A, schematic of G s engagement analysis by BRET. Upon ␤-agonist stimulation, RLucII-G␣ s is engaged by the ␤ 2 AR-GFP10 in a manner that causes a decrease in BRET signal. B, HEK 293 cells co-transfected with RLucII-G␣ s and ␤ 2 AR-GFP10 were preincubated with coelenterazine 400a for 2 min and stimulated with 10 M pepducin or 10 M isoproterenol. Changes in BRET were monitored over the indicated time points. 0.05% DMSO was included in non-pepducin-stimulated trials. The data are represented by the mean of three independent experiments Ϯ S.D. In contrast to isoproterenol, ICL3-9 promoted an increase in BRET indicating a different mode of engagement of G s by the pepducin-activated receptor that results in a different conformation of the complex. ability of multiple regions within the pepducin to activate G s . Although previous studies used unmodified peptides that have a limited ability to cross the cell membrane (52), the N-terminal palmitoylation and C-terminal amidation of a pepducin enable membrane incorporation and effective delivery to the intracellular surface of the cell membrane (18). For ICL3-8, this provides a means of targeting it to the site of action as G s is localized to the intracellular surface of the plasma membrane.
ICL3-9 targets the ␤ 2 AR and stimulates interaction with G s but no apparent interaction with GRKs or ␤-arrestins. Thus, ICL3-9 functions as a G s -biased allosteric agonist for the ␤ 2 AR. Multiple avenues of biophysical analysis have provided insight into the conformational changes that occur upon activation of the ␤ 2 AR. Crystallographic studies of the activated ␤ 2 AR-G s complex demonstrate conformational changes in the ␤ 2 AR proposed to be indicative of receptor activation. The most notable of these is a 14-Å outward movement of the proximal portion of TM6 and a helical extension of the distal portion of TM5 into the intracellular space (6). NMR analysis of the ␤ 2 AR confirms mobility in both of these regions (8), whereas monobromobimane labeling of Cys 265 of the ␤ 2 AR confirms movement of the cytoplasmic end of TM6 upon ␤-agonist treatment (5,45,46). In our studies, ICL3-9 did not affect the movement of TM6, and thus its mechanism of receptor-G s coupling may be different from that induced by conventional ␤-agonists. As ICL3-9 clearly promotes G protein engagement with the ␤ 2 AR (BRET biosensors) and receptor-mediated activation of G s (GTP␥S loading in ␤ 2 AR-containing bicelles) without detectable influence on TM6, conformational changes in TM6 of the ␤ 2 AR may not be a critical step in adopting the ICL3-9 promoted G s -biased receptor state. A helical extension of TM5 into the intracellular surface is also associated with agonistinduced ␤ 2 AR activation (6). This helical extension contains many of the residues in direct contact with G s in the ␤ 2 AR-G s structure and also shares similarity to the peptides that can directly activate G s . Plausibly, ICL3-9 may have the ability to modulate movement of TM5 and promote a unique active conformation utilizing the helical extension of TM5 to unconventionally activate G s . It should also be noted that molecular dynamics simulations of the ␤ 2 AR show a weak relationship between conformational changes in the ligand-binding pocket and TM5/6 movement (8,53). Thus, the ability of the pepducin to promote conformational changes in this region does not depend on operation through an orthosteric mechanism.
Corroborating the notion that ICL3-9 induces a distinct active state of the ␤ 2 AR, ICL3-9 appears to promote G s coupling to the receptor in a different manner than the ␤-agonist. Although G s can interact with unliganded mBB-␤ 2 AR and promote detectable TM6 movement (46), ICL3-9 possesses the ability to attenuate conventional G s -promoted TM6 movement, suggesting that ICL3-9 induces a unique coupling of the ␤ 2 AR and G s (Fig. 10C). Additionally, BRET analysis monitoring RLucII-G␣ s engagement to ␤ 2 AR-GFP10 demonstrated that the N-terminal region of G␣ s may be oriented differently when associated with the ICL3-9 activated receptor as opposed to a ␤-agonist occupied receptor (Fig. 11). Although not much is known about the conformational changes of the G␣ N terminus during G protein activation (54), the observed differences in the BRET signal orientation between RLucII-G␣ s and ␤ 2 AR-GFP10 upon activation with ICL3-9 (increase) versus isoproterenol (decrease) suggest that the position of the N terminus of G␣ s differs between the isoproterenol-and ICL3-9-activated receptor states. The loss in BRET upon isoproterenol stimulation reflects a conformational rearrangement of the pre-coupled ␤ 2 AR-G s complex (40) that results in an increase in the distance between the GFP at the C terminus of the ␤ 2 AR and the RLucII at the N terminus of G␣ s . In contrast, the ICL3-9stimulated functional engagement of G s promotes an increase in BRET and thus reflects a reduction in the distance between the N terminus of G␣ s and the C terminus of the ␤ 2 AR. When a different viewpoint of the complex is monitored using BRET between ␤ 2 AR-GFP10 and RLucII inserted in the linker 1 region of G␣ s between the helical and GTPase domains, comparable BRET changes in direction and efficiency between isoproterenol and ICL3-9 are observed. This suggests that this region of G␣ s may be oriented similarly in both the pepducinand ␤-agonist-stimulated states.
Although ICL3-9 appears to be the first reported G s -biased activator of the ␤ 2 AR, Staus et al. (55) previously reported stimulation of biased signaling from the ␤ 2 AR using intracellularly expressed nanobodies (intrabodies). In this report, intrabodies against agonist-activated or inactive ␤ 2 AR were selective for inhibiting G protein activation or GRK and ␤-arrestin engagement. The expression of intrabodies that block GRK phosphorylation shift the activation profile of a ␤-agonist from activating G proteins and ␤-arrestin to one that selectively stimulates G protein signaling. Essentially, these intrabodies transform a balanced agonist to a G s -biased ␤ 2 AR modulator, although they do not have the ability to activate signaling on their own. Moreover, it is unknown whether the intrabodies stabilize an agonist-bound conformation that favors G s activation or sterically hinders GRK interaction with the ␤ 2 AR, while not affecting G s activation. Although ICL3-9 can directly promote ␤ 2 AR-dependent G s -biased signaling, we also do not know whether this is due to the ICL3-9-induced conformation being unable to mediate GRK/␤-arrestin binding or whether ICL3-9 directly inhibits GRK/␤-arrestin binding to the ␤ 2 AR. Identification of the ICL3-9-binding site on the ␤ 2 AR will be critical in answering these questions.
Because pepducins have been historically thought to be specific for their cognate receptor, another interesting aspect of ICL3-9 was its ability to utilize both the ␤ 1 AR and ␤ 2 AR to mediate cAMP production. The ␤ 1 AR and ␤ 2 AR are closely related family members and share ϳ54% amino acid identity with the transmembrane domains and intracellular loops being the most conserved. For example, ICL1 of the ␤ 2 AR (Ile 55 -Leu 75 ) is ϳ71% identical with ICL1 of the ␤ 1 AR (Leu 78 -Ala 101 ), whereas ICL2 is 77% identical between the ␤ 2 AR (Val 129 -Leu 155 ) and ␤ 1 AR (Leu 154 -Leu 178 ). The ICL3 of the ␤ 2 AR (Phe 217 -Ile 288 ) and the ␤ 1 AR (Phe 241 -Ile 329 ) exhibit lower homology with ϳ38% identity, although this is partly due to a proline-rich insert between Leu 266 and Ala 302 in the ␤ 1 AR. Interestingly, the ICL3 of ␤ 1 AR lacks multiple residues in the ICL3-9 sequence. The commonality of sequence features on the intracellular surface of both the ␤ 1 AR and ␤ 2 AR may be the reason that ICL3-9 can activate G s through both receptors as highly homologous regions may be indistinguishable to the binding and/or function of ICL3-9. Conversely, uncommon regions between the two receptors (i.e. sequence divergence in ICL3) are unlikely to play a critical role in ICL3-9 operation. For example, the C-terminal tail of the ␤ 1 AR and ␤ 2 AR does not exhibit significant homology and thus is unlikely to contribute to the interaction or activity of ICL3-9.
Previous studies on the ␤ 2 AR also identified a few inverse agonists such as carvedilol and nebivolol that appear to function as ␤-arrestin-biased agonists by selectively promoting ␤-arrestin recruitment and signaling over G protein activation (56,57). It is worth noting that carvedilol is used in the treatment of heart failure, and its ability to function in a ␤-arrestinbiased manner appears to be cardioprotective (57)(58)(59). Although we were not specifically searching for ␤-arrestin-biased pepducins, we identified a few pepducins from ICL1 that had a clear ␤-arrestin bias with one being ϳ75% as effective as isoproterenol in promoting ␤-arrestin binding (Fig. 2, C and D). Characterizing the ability of the ␤ 2 AR to selectively activate ␤-arrestin-associated pathways (e.g. receptor internalization and ␤-arrestin-dependent signaling) using our ␤-arrestin-biased pepducins could be an interesting avenue of future investigation. Moreover, our results suggest that pepducins may prove particularly useful in dissecting the mechanisms involved in biased signaling and the potential links between ␤-arrestin activation and the treatment of heart failure.
G s -biased pepducins demonstrated an independence from classical receptor desensitization mechanisms as the induction of functional desensitization of the ␤ 2 AR was not observed in primary human airway smooth muscle cells. These data serve as an initial "proof-of-concept" that G s -biased agonists could serve as a potentially advantageous asthma therapeutic. Certainly, ICL3-9 is a potentially attractive lead drug candidate as its receptor dependence leads to a degree of drug targeting and specificity, although its potency must be improved. Direct activators of G s such as ICL3-2 and ICL3-8 might also prove advantageous in promoting airway smooth muscle relaxation, while targeting these molecules to the proper cell type will be critical. Beyond their application in the treatment of asthma, the pepducins provide additional tools to study ␤ 2 AR activation and the benefits of G s -biased signaling.