Discovery of β-arrestin-biased β₂-adrenoceptor agonists from 2-amino-2-phenylethanol derivatives

Abstract

β-Arrestins are a small family of proteins important for signal transduction at G protein-coupled receptors (GPCRs). β-Arrestins are involved in the desensitization of GPCRs. Recently, biased ligands possessing different efficacies in activating the G protein- versus the β-arrestin-dependent signals downstream of a single GPCR have emerged, which can be used to selectively modulate GPCR signal transduction in such a way that desirable signals are enhanced to produce therapeutic effects while undesirable signals of the same GPCR are suppressed to avoid side effects. In the present study, we evaluated agonist bias for compounds developed along a drug discovery project of β₂-adrenoceptor agonists. About 150 compounds, including derivatives of fenoterol, 2-amino-1-phenylethanol and 2-amino-2-phenylethanol, were obtained or synthesized, and initially screened for their β-adrenoceptor-mediated activities in the guinea pig tracheal smooth muscle relaxation assay or the cardiomyocyte contractility assay. Nineteen bioactive compounds were further assessed using both the HTRF cAMP assay and the PathHunter β-arrestin assay. Their concentration-response data in stimulating cAMP synthesis and β-arrestin recruitment were applied to the Black–Leff operational model for ligand bias quantitation. As a result, three compounds (L-2, L-4, and L-12) with the core structure of 5-(1-amino-2-hydroxyethyl)-8-hydroxyquinolin-2(1H)-one were identified as a new series of β-arrestin-biased β₂-adrenoceptor agonists, whereas salmeterol was found to be G_s-biased. These findings would facilitate the development of novel drugs for the treatment of both heart failure and asthma.

Introduction

G protein-coupled receptors (GPCRs) represent the largest class of druggable targets [1]. In recent years, functional selectivity has emerged as a major concept for the redevelopment of many traditional GPCR drug targets, such as β-adrenoceptors [2,3,4,5,6]. This concept suggests that a so-called biased ligand could be used to selectively modulate GPCR signal transduction in such a way that desirable signals are enhanced to produce therapeutic effects, whereas undesirable signals of the same GPCR are suppressed to avoid side effects. The terms “functional selectivity” and “biased agonism” are often used interchangeably. Although “biased agonism” is sometimes used in a stricter sense to refer to “β-arrestin-biased agonism” [7], it generally describes the disparity of the efficacies of ligands in activating signals mediated by different downstream effectors, for example, different G protein isoforms, G protein versus β-arrestin or biases from many other signaling readouts. In some cases, the biased ligand could act as an antagonist or an inverse agonist for G protein-dependent signaling but as an agonist for β-arrestin-dependent signaling in a single GPCR.

Ligand biases at both β₁- and β₂-adrenoceptors have been described [6]. For example, carvedilol is an antagonist of both β-adrenoceptors for the G_s pathway, but it is also mildly active in triggering β-arrestin-dependent signaling [8, 9]. It has been suggested that the superior efficacy of carvedilol compared with other β-blockers in heart failure may be due to the activation of the β-adrenoceptor-mediated β-arrestin-dependent signaling [7, 8].

Although it is well-established that long-term stimulation of the β₁-adrenoceptor would lead to cardiomyocyte death and predisposes the heart to failure [10,11,12,13], the role of the β₂-adrenoceptor is still unclear. As the β₂-adrenoceptor couples to both G_s and G_i proteins [14] and β₂-adrenoceptor-G_i signaling becomes exaggerated in heart failure [15,16,17], causing various adverse structural and functional consequences in the heart [17,18,19,20], based on the concept of functional selectivity, a G_s-biased β₂-adrenoceptor agonist, fenoterol (FEN), was discovered from screening and subsequently tested on an animal model of heart failure [21]. The efficacy of FEN, used alone or in combination with a β₁-adrenoceptor antagonist, in treating heart failure has subsequently been demonstrated in follow-up studies [22,23,24,25,26]. Therefore, we have proposed the use of G_s-biased β₂-adrenoceptor agonists to treat heart failure [20, 21, 27].

Our group is committed to developing β₂-adrenoceptor agonists for clinical use. Characterization of the signal transduction properties of β₂-adrenoceptor agonists is one of our major objectives because it can provide us with experimental compounds with different pharmacological properties, from which we can choose compounds for further research and development. We have synthesized over 90 derivatives of FEN [28,29,30,31]. Several G_s-biased β₂-adrenoceptor agonists, such as (R,R’)-FEN (Supplementary Fig. S1), have been discovered from this cohort [32, 33]. We have synthesized other β₂-adrenoceptor agonists including compounds with the classical 2-amino-1-phenylethanol core structure or those containing a novel 2-amino-2-phenylethanol core structure [34,35,36,37]. These compounds exhibit β₂-adrenoceptor agonistic activities and tracheal smooth muscle relaxant effects, and many of them are also more selective towards the β₂-adrenoceptor over the β₁-adrenoceptor [34,35,36,37]. The aim of the present study is to characterize the β-arrestin-biased agonism at the β₂-adrenoceptor for some of these novel compounds.

Materials and methods

Chemistry

The ¹H and ¹³C spectra (Supplementary Fig. S2) were recorded on a Bruker (Billerica, MA, USA) ARX-300, ARX-400, or ARX-600 NMR spectrometer using DMSO-d₆ as the solvent. Chemical shifts (δ) were reported in parts per million (ppm) relative to tetramethylsilane (internal standard), and coupling constants (J) were reported in Hz. High-resolution mass spectrometry (HRMS) was performed on a Bruker SolariX 7.0T. All reactions were monitored by thin layer chromatography (TLC) using TLC plates (Silica gel60 GF254, Merck) and UV light visualization. When appropriate, crude products were purified by column chromatography using silica gel (200–300 mesh) purchased from the Qingdao Haiyang Chemical Co. Ltd (China). The purity of the final products was determined by high performance liquid chromatography (Waters Corp., Milford, MA, USA) on a C18 column (250 mm × 4.6 mm, 5 µm bead size; Thermo Fisher Scientific, Waltham, MA, USA) with acetonitrile/H₂O (33/67 v/v, the pH was adjusted to 5.30 with H₃PO₄ and NaH₂PO₄) as the mobile phase at a 1.0 mL/min flow rate and a detection wavelength of 259 or 254 nm. The structures of newly synthesized compounds were confirmed by ¹H-NMR, ¹³C-NMR, and HRMS, whereas the structures of other compounds were confirmed by ¹H-NMR and low-resolution mass spectrometry (MS).

Materials

All solvents and reagents were obtained from commercial suppliers and were used as received. All biochemical reagents were obtained from Sigma-Aldrich (St. Louis, MO, USA) unless otherwise noted.

Compounds

(R)-Epinephrine ((R)-EPI), FEN, and (R)-isoproterenol ((R)-ISO) were purchased from Sigma-Aldrich. Salmeterol (SAL) was purchased from Adamas-Beta (Shanghai, China). The synthesis of (R,R’)-FEN and (R,R’)-4-methoxy-1-naphthylfenoteol ((R,R’)-MNF) has been described previously [29,30,31]. They were kindly provided by Dr. Joseph Kozocas from SRI International in California, USA (Supplementary Fig. S1). Compounds with a 2-amino-1-phenylethanol core structure including A-17, A-18, A-23, A-31, A-32, A-33, and A-35, as well as those with a 2-amino-2-phenylethanol core structure including B-24, B-30, L-2, L-4, L-6, and L-12 (Table 1) were designed and synthesized at Shenyang Pharmaceutical University by methods adopted from our previous studies [34, 36, 37]. The steps for the syntheses of the target compounds are illustrated in Scheme 1 for A-17, A-18, A-23, A-31, A-32, and A-33; Scheme 2 for B-24 and B-30; and Scheme 3 for L-2, L-4, L-6, and L-12. Compounds A-33, B-24, L-2, L-4, L-6, and L-12 are novel. The purities of all of the biologically evaluated compounds exceeded 95%. All test compounds are phenylethanolamines. Phenylethanolamine is the basic structure of β-agonists. Stock solutions of the compounds were prepared in dimethylsulfoxide (DMSO). Dilutions were made with buffer solutions. The final concentration of DMSO in the assays did not exceed 1%, and at this concentration, DMSO did not interfere with the assays.

Table 1 Structures of the compounds synthesized and tested

Scheme 1

Synthesis of 2-amino-1-phenylethanol derivatives. Reagents and conditions: (i) amines, ethanol, 5 °C to room temperature (rt), 5 h; (ii) NaBH₄, ethanol and water, rt

Scheme 2

Synthesis of B-24 and B-30. Reagents and conditions: (i) isopropylamine, ethanol, reflux, 13 h

Scheme 3

Synthesis of L-2, L-4, L-6, and L-12. Reagents and conditions: (i) mCPBA, DCM, 0 °C to rt, 30 min; (ii) Ac₂O, 100 °C, 3 h; (iii) CH₃COCl, AlCl₃, DCE, 0 °C to reflux; (iv) C₆H₅CH₂Br, K₂CO₃, in N,N-dimethylformamide, 40 °C, 3 h; (v) Br₂, BF₃Et₂O, DCM, reflux, 15 min; (vi) NaBH₄, CH₃OH, DCM, 0 °C, 30 min; (vii) KOH, CH₃OH, DCM, 0 °C, 30 min; (viii) isopropylamine, N-propylamine, tert-butylamine or N-hexylamine, ZnCl₂, CH₃CN, reflux, 12 h; (ix) H₂, 10% Pd/C, CH₃OH, rt, 2 h

General procedures for 2-amino-1-phenylethanol (A-series) and 2-amino-2-phenylethanol (B-series) compounds

A-series compounds were prepared by combining different substituted 1-(4-aminophenyl)-2-bromoethanone derivatives with various amines. The carbonyl group in each intermediate was then reduced by sodium borohydride to afford the 2-aminoethanol product (Scheme 1). B-series compounds were synthesized by coupling different substituted 2-bromo-1-phenylethanol compounds with various amines after reduction of the ethanone (Scheme 2). Finally, all of the target compounds were treated with a saturated solution of hydrogen chloride in isopropanol to form hydrochlorides.

1-(4-Amino-3-chloro-5-cyano-phenyl)-2-(3-ethoxy)propylamino-ethanol hydrochloride (A-33)

To a stirred solution of 2-amino-5-(2-bromoacetyl)-3-chlorobenzonitrile (5.00 g, 0.018 mol) in 50 mL of ethanol, 3-ethoxypropylamine (5.63 g, 0.054 mol) was added slowly under an atmosphere of nitrogen at 5 °C. The reaction mixture was stirred at room temperature for 5 h. Following the addition of 4 mL of water, sodium borohydride (0.68 g, 0.018 mol) was added portion-wise, and the reaction mixture was further stirred for 5 h. Aqueous 2 mol/L HCl was then added to adjust the pH to 3, followed by stirring for 30 min. Ammonia water was added to adjust the pH to 10, followed by evaporation under vacuum. The residue was mixed with 50 mL of water, and ammonia water was added again until the pH = 10. The mixture was extracted with ethyl acetate (3 × 50 mL). Organic layers were combined, and the ethyl acetate was evaporated under reduced pressure. The residue was extracted with aqueous 2 mol/L HCl (3 × 60 mL). Aqueous layers were combined, and activated carbon was added. The mixture was heated to reflux for 20 min and then filtered. After cooling to room temperature, the pH was adjusted to 10 with ammonia water. The mixture was extracted with diethyl ether (3 × 50 mL). The combined organic layers were washed with water, dried over anhydrous sodium sulfate, and filtered. A saturated solution of hydrogen chloride in isopropanol was added to the filtrate until the pH = 3. Filtration of the resulting mixture gave A-33 as a white solid (0.80 g, 14.8%). ¹H-NMR (400 MHz, DMSO-d₆, ppm) δ: 8.93 (br, 1H), 8.69 (br, 1H), 7.56 (s, 1H), 7.46 (s, 1H), 6.30 (s, 2H), 6.23(s, 1H), 4.83 (d, J = 3.18 Hz, 1H), 3.38–3.45 (m, 4H), 2.98–3.11 (m, 4H), 1.89 (s, 2H), 1.11 (d, J = 4.17, 3H). ¹³C-NMR δ: 147.2, 132.5, 130.7, 129.9, 118.9, 117.6, 95.6, 67.5, 67.1, 65.8, 53.1, 45.6, 26.1, 15.5. MS (EI) m/z 298, 206, 116, 44. HRMS ([M + H]⁺) m/z calcd for C₁₄H₂₁ClN₃O₂ 298.1322; found 298.1329.

2-(4-Amino-3-cyano-phenyl)-2-isopropylamino-ethanol hydrochloride (B-24)

To a stirred solution of 2-amino-5-(2-bromo-1-hydroxyethyl)-benzonitrile (2.45 g, 0.016 mol) in 50 mL of ethanol, isopropylamine (2.14 g, 0.036 mol) was slowly added. The mixture was heated to reflux for 13 h. Following the evaporation of most of the ethanol, the residue was extracted with aqueous 2 mol/L HCl (3 × 50 mL). Aqueous layers were combined and washed with methyl benzene. Activated carbon was then added. The mixture was heated to reflux for 20 min and then filtered. After cooling to room temperature, ammonia water was added to adjust the pH to 10. The mixture was extracted with diethyl ether (3 × 50 mL). The combined organic layers were washed with water, dried over anhydrous sodium sulfate, and filtered. A saturated solution of hydrogen chloride in isopropanol was added to the filtrate until the pH = 3. Filtration of the resulting mixture gave B-24 as a white solid (0.50 g, 15.2%). ¹H-NMR (300 MHz, DMSO-d₆, ppm) δ: 9.10 (br, 2H), 7.71 (d, J = 1.65 Hz, 1H), 7.56-7.60 (dd, J = 8.69 Hz, 1.73 Hz, 1H), 6.82 (d, J = 8.69 Hz, 1H), 6.28 (s, 2H), 5.55 (s, 1H), 4.20 (d, J = 0.56 Hz, 1H), 3.85 (s, 1H), 3.73 (s, 1H), 2.95–3.00 (m, 1H), 1.27 (d, J = 6.42 Hz, 3H), 1.18 (d, J = 6.42 Hz, 3H). ¹³C-NMR δ: 152.39, 134.65, 133.51, 121.13, 118.26, 115.94, 93.64, 62.61, 59.66, 47.83, 19.74, 18.00. MS (ESI) m/z 219.9 [M + H]^-. HRMS ([M + H]⁺) m/z calcd for C₁₂H₁₈N₃O 220.1450; found 219.1441.

General procedures for L-2, L-4, L-6, and L-12

Synthesis of the target compounds L-2, L-4, L-6, and L-12 was achieved through the pathway illustrated in Scheme 3. The starting material 8-hydroxyquinoline (commercially purchased) was reacted with meta-chloroperoxybenzoic acid (mCPBA) in dichloromethane (DCM) to give 8-hydroxyquinoline-1-oxide 2, which was subsequently acetylated to give 8-acetoxyquinolin-2(1H)-one 3. 5-Acetyl-8-hydroxyquinolin-2(1H)-one 4 was synthesized from intermediate 3 by a Fries rearrangement with aluminum trichloride and acetyl chloride as catalysts along with 1,2-dichloroethane (DCE) as the solvent. The phenolic hydroxyl group of intermediate 4 was benzyl protected to give 5-acetyl-8-benzyloxyquinolin-2(1H)-one 5, and then, the keto-carbonyl group was brominated to give 8-benzyloxy-5-(2-bromoacetyl)quinolin-2(1H)-one 6. Reduction of 6 with sodium borohydride afforded 8-benzyloxy-5-(2-bromo-1-hydroxyethyl)quinolin-2(1H)-one 7, which was subjected to intramolecular nucleophilic substitution to afford the epoxide 8-benzyloxy-5-(2-oxiranyl)quinolin-2(1H)-one 8. Detailed methods for the synthesis of compounds 1–8 can be found in Supplementary Methods. Intermediates 8-benzyloxy-5-[2-hydroxy-1-(isopropylamino)ethyl]quinolin-2(1H)-one hydrochloride 9a, 8-benzyloxy-5-[2-hydroxy-1-(tert-butylamino)ethyl]quinolin-2(1H)-one hydrochloride 9b, 8-benzyloxy-5-[2-hydroxy-1-(N-propylamino)ethyl]quinolin-2(1H)-one hydrochloride 9c and 8-benzyloxy-5-[2-hydroxy-1-(N-hexylamino)ethyl]quinolin-2(1H)-one hydrochloride 9d were prepared by refluxing 8 with different amines in the presence of zinc chloride as the catalyst and acetonitrile as the solvent. The target compounds L-2, L-4, L-6, and L-12 were synthesized from 9a, 9b, 9c, and 9d by hydrogenation with Pd/C in methanol.

Procedure A: 8-Benzyloxy-5-[2-hydroxy-1-(isopropylamino)ethyl]quinolin-2(1H)-one hydrochloride (9a)

To a stirred solution of intermediate 8 (3.0 g, 0.0102 mol) and zinc chloride (0.35 g, 0.0026 mol) in 150 mL of acetonitrile, isopropylamine (2.63 mL, 0.0307 mol) was added drop-wise. The reaction mixture was heated slowly under reflux for 12 h. Then, the solvent was evaporated. The product was purified by column chromatography (CH₂Cl₂:CH₃OH:NH₄OH = 350:10:1). After evaporation, the solid was dissolved in 50 mL of acetone, followed by acidification with a saturated solution of hydrogen chloride in isopropanol to a pH of 2. Filtration of the resulting mixture gives 9a as a yellow solid (0.82 g, 20.6%). ¹H-NMR (400 MHz, DMSO-d₆, ppm) δ: 10.84 (s, 1H), 9.45 (s, 1H), 9.23 (s, 1H), 8.32–8.33 (d, J = 10.02 Hz, 1H), 7.59–7.61 (m, 3H), 7.38–7.41 (m, 2H), 7.32–7.34 (m, 2H), 6.62–6.63 (d, J = 9.90 Hz, 1H), 5.53 (s, 1H), 5.34 (s, 2H), 4.89–4.92 (m, 1H), 3.87–3.94 (m, 1H), 3.74–3.82 (m, 1H), 3.05–3.12 (m, 1H), 1.29 (d, J = 5.76 Hz, 3H), 1.14 (d, J = 6.12 Hz, 3H). ¹³C-NMR δ: 161.2, 144.8, 137.0, 136.2, 130.1, 128.9, 128.5, 128.4, 123.7, 123.3, 121.6, 118.5, 112.7, 70.3, 63.3, 55.3, 48.4, 19.6, 18.6. HRMS ([M + H]⁺) m/z calcd for C₂₁H₂₅N₂O₃ 353.1865; found 353.1854.

8-Benzyloxy-5-[2-hydroxy-1-(tert-butylamino)ethyl]quinolin-2(1H)-one hydrochloride (9b)

In the presence of zinc chloride (0.35 g, 0.0026 mol), intermediate 8 (3.0 g, 0.0102 mol) was reacted with tert-butylamine (2.27 g, 0.0307 mol) in 150 mL of acetonitrile according to procedure A. The crude product was purified by column chromatography (CH₂Cl₂:CH₃OH:NH₄OH = 350:10:1). Evaporation of the solvent and acidification of the product gives 9b as a white solid (0.64 g, 17.11%). ¹H-NMR (600 MHz, DMSO-d₆, ppm) δ: 10.86 (s, 1H), 9.31 (s, 1H), 8.65 (s, 1H), 8.39 (s, J = 7.5 Hz, 1H), 7.60 (d, J = 4.8, 2H), 7.39 (t, J = 4.88 Hz, 2H), 7.32 (t, J = 4.84 Hz, 2H), 6.63 (d, J = 6.60 Hz, 1H), 5.52 (s, 1H), 5.34 (s, 2H), 4.91 (s, 1H), 3.89 (s, 1H), 3.80–3.82 (m, 1H), 1.23 (s, 9H). ¹³C-NMR δ: 160.1, 143.6, 135.9, 135.3, 128.9, 127.8, 127.4, 127.3, 124.4, 122.5, 121.2, 116.9, 111.5, 69.1, 63.0, 58.1, 53.0, 25.6. HRMS ([M + Cl]^-) m/z calcd for C₂₂H₂₆ClN₂O₃ 401.1632; found 401.1653.

8-Benzyloxy-5-[2-hydroxy-1-(N-propylamino)ethyl]quinolin-2(1H)-one hydrochloride (9c)

In the presence of zinc chloride (0.35 g, 0.0026 mol), intermediate 8 (3.0 g, 0.0102 mol) was reacted with N-propylamine (1.8 g, 0.0307 mol) in 150 mL of acetonitrile according to procedure A. The crude product was purified by column chromatography (CH₂Cl₂:CH₃OH:NH₄OH = 350:10:1). Evaporation of the solvent and acidification of the product gives 9c as a white solid (0.47 g, 13.09%). ¹H-NMR (600 MHz, DMSO-d₆, ppm) δ: 10.90 (s, 1 H), 9.47 (s, 1 H), 9.15 (s, 1H), 8.24 (d, J = 10.1 Hz, 1H), 7.60 (d, J = 7.3 Hz, 2H), 7.50 (d, J = 8.5 Hz, 1H), 7.39 (t, J = 7.5 Hz, 2H), 7.33 (s, 2H), 6.63 (d, J = 9.9 Hz, 1H), 5.74–5.46 (m, 1H), 5.35 (s, 2H), 4.99–4.81 (s, 1H), 3.86 (dd, J = 11.6, 6.2 Hz, 1H), 3.79 (dd, J = 11.6, 4.9 Hz, 1H), 2.88 (s, 1H), 2.67 (s, 1H), 1.65 (ddd, J = 27.9, 15.5, 7.4 Hz, 2H), 0.81 (t, J = 7.4 Hz, 3H). ¹³C-NMR δ: 160.6, 144.2, 136.4, 135.9, 129.5, 128.3, 127.9, 127.8, 123.1, 122.9, 120.8, 188.0, 112.1, 69.6, 62.1, 47.0, 18.8, 10.9. HRMS ([M + H]⁺) m/z calcd for C₂₁H₂₅N₂O₃ 353.1865; found 353.1860.

8-Benzyloxy-5-[2-hydroxy-1-(N-hexylamino)ethyl]quinolin-2(1H)-one hydrochloride (9d)

In the presence of zinc chloride (0.35 g, 0.0026 mol), intermediate 8 (3.0 g, 0.0102 mol) was reacted with N-hexylamine (2.64 mL, 0.0307 mol) in 150 mL of acetonitrile according to procedure A. The crude product was purified by column chromatography (CH₂Cl₂:CH₃OH:NH₄OH = 350:10:1). Evaporation of the solvent and acidification of the product gave 9d as a white solid (0.60 g, 13.6%). ¹H-NMR (400 MHz, DMSO-d₆, ppm) δ: 10.84 (s, 1H), 9.52 (s, 1H), 9.20 (s, 1H), 8.24 (d, J = 10.02 Hz, 1H), 7.51–7.60 (m, 3H), 7.29–7.41 (m, 4H), 6.62 (d, J = 9.87 Hz, 1H), 5.52 (br, 1H), 5.34 (s, 2H), 4.89 (t, J = 7.68, 1H), 3.75–3.91 (s, 2H), 2.86–2.89 (m, 1H), 2.70 (m, 1H), 1.59–1.62 (m, 2H), 1.17–1.24 (m, 6H), 0.79–0.83 (m, 3H). ¹³C-NMR δ: 161.2, 144.7, 137.0, 136.5, 130.0, 128.8, 128.4, 128.4, 123.7, 123.4, 121.4, 118.6, 112.6, 70.2, 62.7, 57.7, 46.0, 31.1, 26.1, 25.6, 22.2, 14.2. HRMS ([M + H]⁺): m/z calcd for C₂₄H₃₁N₂O₃ 395.2335, found 395.2325.

Procedure B: 8-Hydroxy-5-[2-hydroxy-1-(isopropylamino)ethyl]quinolin-2(1H)-one hydrochloride (L-2)

To a solution of intermediate 9a (150 mg, 0.386 mmol) in 50 mL of methanol, 10% Pd/C (20.0 mg) was added. The reaction mixture was stirred under an atmosphere of hydrogen at room temperature for 2 h. The resulting mixture was filtered and concentrated to give L-2 as a white solid (60 mg, 52.1%). ¹H-NMR (400 MHz, DMSO-d₆, ppm) δ: 10.81 (s, 1H), 10.63 (s, 1H), 9.33 (s, 1H), 9.14 (s, 1H), 8.29 (d, J = 9.90 Hz, 1H), 7.50 (d, J = 8.04 Hz, 1H), 7.09 (d, J = 8.10 Hz, 1H), 6.59 (d, J = 9.90 Hz, 1H), 5.58 (s, 1H), 4.93 (s, 1H), 3.89–3.91 (m, 1H), 3.75–3.77 (m, 1H), 3.10 (t, J = 3.88 Hz, 1H), 1.29 (d, J = 3.88 Hz, 3H), 1.14 (d, J = 3.88 Hz, 3H). ¹³C-NMR δ: 161.1, 144.5, 136.2, 129.2, 123.4, 121.7, 121.3, 118.7, 114.7, 63.5, 48.3, 19.6, 18.6. HRMS ([M + Cl]^-) m/z calcd for C₁₄H₁₈ClN₂O₃ 297.1006; found 297.1014.

8-Hydroxy-5-[2-hydroxy-1-(tert-butylamino)ethyl]quinolin-2(1H)-one hydrochloride (L-4)

In the presence of 10% Pd/C (20.0 mg), intermediate 9b (150 mg, 0.386 mmol) was reacted with H₂ in 50 mL of methanol according to procedure B to give L-4 as a white solid (117.1 mg, 97.0%). ¹H-NMR (600 MHz, DMSO-d₆, ppm) δ: 10.70 (s, 1H) 10.53 (s, 1H), 9.05 (s, 1H), 8.54 (s, 1H), 8.32 (d, J = 9.99 Hz, 1H), 7.45 (d, J = 8.28 Hz, 1H), 7.05 (d, J = 8.28 Hz, 1H), 6.59 (d, J = 9.90 Hz, 1H), 5.50 (s, 1H), 4.87 (s, 1H), 3.79–3.84 (m, 2H), 1.23 (s, 9H). ¹³C-NMR δ: 160.0, 143.3, 135.3, 128.0, 125.5, 122.3, 121.3, 117.0, 113.4, 63.0, 58.0, 53.2, 25.6. HRMS ([M + H]⁺) m/z calcd for C₁₅H₂₁N₂O₃ 277.1552; found 277.1559.

8-Hydroxy-5-[2-hydroxy-1-(N-propylamino)ethyl]quinolin-2(1H)-one hydrochloride (L-6)

In the presence of 10% Pd/C (20.0 mg), intermediate 9c (150 mg, 0.386 mmol) was reacted with H₂ in 50 mL of methanol according to procedure B to give L-6 as a white solid (113.0 mg, 98.0%). ¹H-NMR (600 MHz, DMSO-d₆, ppm) δ: 10.74 (s, 1H) 10.54 (s, 1H), 9.36 (s, 1H), 9.08 (s, 1H), 8.20 (d, J = 10.08 Hz, 1H), 7.391 (d, J = 8.28 Hz, 1H), 7.06 (d, J = 8.28 Hz, 1H), 6.59 (d, J = 10.08 Hz, 1H), 5.53 (s, 1H), 4.86 (s, 1H), 3.75–3.90 (m, 2H), 2.84 (s, 1H), 2.67 (s, 1H), 1.59–72 (m, 2H), 0.82 (t, d = 7.47 Hz, 3H). ¹³C-NMR δ: 161.1, 144.5, 136.5, 129.2, 123.2, 121.7, 121.5, 118.9, 114.7, 62.8, 57.7,47.5, 19.4, 11.5. HRMS ([M + H]⁺) m/z calcd for C₁₄H₁₉N₂O₃ 263.1396; found 263.1392.

8-Hydroxy-5-[2-hydroxy-1-(N-hexylamino)ethyl]quinolin-2(1H)-one hydrochloride (L-12)

In the presence of 10% Pd/C (20.0 mg), intermediate 9d (150 mg, 0.386 mmol) was reacted with H₂ in 50 mL of methanol according to procedure B to give L-12 as a white solid (70 mg, 59.0%). ¹H-NMR (400 MHz, DMSO-d₆, ppm) δ: 10.74 (s, 1H) 10.54 (s, 1H), 9.38 (s, 1H), 9.09 (s, 1H), 8.19–8.22 (d, J = 9.99 Hz, 1H), 7.38–7.41 (d, J = 8.25 Hz, 1H), 7.05–7.08 (d, J = 8.25 Hz, 1H), 6.56–6.60 (d, J = 9.84 Hz, 1H), 5.52 (s, 1H), 4.86 (s, 1H), 3.75–3.87 (m, 2H), 2.86 (s, 1H), 2.71 (s, 1H), 1.61 (s, 2H), 1.17 (s, 6H), 0.78–0.83 (t, J = 7.80 Hz, 3H). ¹³C-NMR δ: 161.1, 144.5, 136.4, 129.2, 123.2, 121.7, 121.5, 118.9, 114.6, 62.8, 31.1, 26.1, 25.7, 22.2, 14.25. HRMS ([M + Cl]^-) m/z calcd for C₁₇H₂₄ClN₂O₃ 339.1475; found 339.1487.

Cell lines

The HEK293 cell line was obtained from Thermo Fisher Scientific and was cultured in Dulbecco’s modified Eagle’s medium (Thermo Fisher Scientific) supplemented with 10% fetal bovine serum (Beijing YuanHeng ShengMa Biotechnology Research Institute, Beijing, China). The PathHunter Chinese Hamster Ovary-K1 β₂-adrenoceptor β-arrestin (CHO-β₂-β-arr) cell line was purchased from Discoverx (CA, USA) and was cultured in a proprietary medium supplied by Discoverx. All cells were maintained in a humidified incubator at 37 °C in 5% CO₂.

cAMP assay

The homogeneous time-resolved fluorescence (HTRF) cAMP assay was conducted according to the manufacturer’s protocol for the cAMP Dynamic 2 kit (Cisbio Bioassays, Codolet, France). Briefly, HEK293 cells, expressing endogenous β₂-adrenoceptors, were grown to < 80% confluence. Dissociated cells were resuspended in Hank’s balanced salt with 20 mmol/L 4-(2-hydroxyethyl)piperazine-1-ethanesulfonic acid solution (HBSS) supplemented with 0.1% bovine serum albumin and 1 mmol/L 3-isobutyl-1-methylxanthine (IBMX), and then, they were dispensed into 384-well low volume plates (Greiner #784076) at 2000 cells/5 μL per well. The cells were stimulated with compounds diluted in HBSS supplemented with 0.1% bovine serum albumin (5 μL/well) for 30 min at room temperature. Reactions were stopped by the addition of the cAMP-d2 conjugate in lysis buffer (5 μL/well), followed by the addition of the anti-cAMP cryptate conjugate in lysis buffer (5 μL/well). After incubation for 1 h at room temperature, the plates were read in a PerkinElmer (Waltham, MA, USA) 2300 EnSpire multilabel reader for time-resolved fluorescence resonance energy transfer detection with an excitation wavelength of 337 nm and emission wavelengths of 620 nm and 665 nm, with a lag time of 100 μs and an integration time of 400 μs for each emission wavelength. The resultant cAMP concentrations were calculated using Prism 4 (GraphPad Software, CA, USA) by applying the 620/665 nm fluorescence ratios to a standard curve of known cAMP concentrations.

β-Arrestin assay

The PathHunter β-arrestin assay (Discoverx) was carried out with the CHO-β₂-β-arr cell line expressing the Prolink-tagged human β₂-adrenoceptors and the β-arrestin-2-β-galactosidase enzyme fragment fusion proteins. An active β-galactosidase enzyme is formed when a β-arrestin interacts with an activated β₂-adrenoceptor in the cell. The assay was performed according to the manufacturer’s protocol (Discoverx). Briefly, CHO-β₂-β-arr cells were grown to 90% confluence. Dissociated cells were resuspended in a cell plating medium (Discoverx) and seeded at 10 000 cells/20 μL per well into 384-well tissue culture plates (with white walls and a clear bottom). After overnight culturing at 37 °C in 5% CO₂, compounds diluted in phosphate-buffered saline (pH = 7.4) supplemented with 0.1% bovine serum albumin (5 μL/well) were added to the cells. The plates were then incubated at 37 °C for 90 min. A β-galactosidase substrate (Discoverx; 12 μL/well) was added. After further incubation for 60 min at room temperature, the plates were read in a PerkinElmer 2300 EnSpire multilabel reader for luminescence detection.

Isolated guinea pig trachea relaxation assay

Male Dunkin–Hartley guinea-pigs weighing 250–350 g were provided by the Experimental Animal Center of Shenyang Pharmaceutical University. The experiment has been approved by the Animal Care and Use Committee of Shenyang Pharmaceutical University. The methods for the isolation of tracheal smooth muscle strips and muscle tone measurements have been described [36]. The strips were equilibrated in Krebs–Henseleit solution (composition in mM: NaCl 118, KCl 5.4, CaCl₂ 2.5, MgSO₄ 0.6, NaH₂PO₄ 2.94, NaHCO₃ 25 and D-glucose 11.7) and aerated with 95% O₂ and 5% CO₂ for 2 h. Histamine (10 μmol/L) was added to produce maximum contraction. The compound was then added to produce relaxation. The percentage of relaxation was calculated as follows:

$${\mathrm{Relaxation}}\,\% = \frac{{{\mathrm{tension}}\,{\mathrm{after}}\,{\mathrm{compound}} - {\mathrm{tension}}\,{\mathrm{after}}\,{\mathrm{histamine}}}}{{{\mathrm{tension}}\,{\mathrm{after}}\,{\mathrm{histamine}}}} \\ \times 100\%$$

Curve fitting and statistical analysis

Fitting of the concentration-response curves and statistical analysis of the data were performed using Prism 4. E_max and pEC₅₀ values were calculated from the curves fitted on the four-parameter logistic model. Ligand bias was quantified as described previously [38]. Briefly, the concentration-response data of all the compounds to be analyzed were fitted into the Black-Leff operational model: [39]

$${\mathrm{Response}} = \frac{{E_{\mathrm{m}}\left[ {\mathrm{A}} \right]^n\tau ^n}}{{\left[ {\mathrm{A}} \right]^n\tau ^n + (\left[ {\mathrm{A}} \right] + K_{\mathrm{A}})^n}} + {\mathrm{basal}}$$

(1)

where [A] is the concentration of the agonist, E_m is the maximal response of the system and n is the transducer slope. The two latter parameters and basal are common for the curve fitting of all of the concentration-response data. Fitting of the functional data to eq. 1 would yield some combinations of τ and K_A. The log transformation of eq. 1 would yield the transduction coefficient, log(τ/K_A), which represents the efficiency of a particular agonist in activating a given signaling pathway. The Δlog(τ/K_A) value would, therefore, describe the relative “efficacy” of an agonist for a particular pathway when normalized with a reference agonist, defined as (R)-ISO in the current analysis. The ligand bias or log bias values, ΔΔlog(τ/K_A), of a given agonist for different signaling pathways were then evaluated by statistical analysis using Bonferroni’s t-test against the reference ligand (R)-ISO. The standard errors (SE) of the ΔΔlog(τ/K_A) values were calculated as described previously [38].

Results

We initially screened a library of approximately 150 2-amino-1-phenylethanol (A-series) or 2-amino-2-phenylethanol (B-series and L-series) compounds for β-adrenoceptor-mediated activity using the tracheal smooth muscle relaxation assay or the cardiomyocyte contractility assay [32, 33, 36, 37]. Compounds with biological effects were short-listed for subsequent tests. To study ligand bias, the concentration-response profiles of (R)-ISO, (R)-EPI, SAL, FEN, (R,R’)-FEN, (R,R’)-MNF, A-17, A-18, A-23, A-31, A-32, A-33, A-35, B-24, B-30, L-2, L-4, L-6, and L-12 (Table 1 and Supplementary Fig. S1) in activating the β₂-adrenoceptor-G_s-cAMP signaling and the β₂-adrenoceptor-β-arrestin signaling were obtained from the two different assays (Supplementary Fig. S3 and S4). The HTRF cAMP assay (Cisbio Bioassays) was conducted on compound-stimulated HEK293 cells expressing a low level of an endogenous β₂-adrenoceptor [40]. The PathHunter β-arrestin assay (Discoverx) was conducted on compound-stimulated CHO-β₂-β-arr cells. Both of these cell lines have no background expression of other β-adrenoceptor isoforms.

The E_max and pEC₅₀ values of the concentration-response curves are shown in Table 2. The high E_{max cAMP} values of the respective compounds (approximately 90%) suggest that (R)-ISO, FEN and (R,R’)-FEN are full agonists of the β₂-adrenoceptor in the cAMP assay. The other compounds with lower E_{max cAMP} values are partial agonists in this assay. SAL and L-6 are the two compounds with the lowest E_{max cAMP} values (15% and 10%, respectively). Similarly, (R)-ISO, FEN and (R,R’)-FEN are also full agonists of the β₂-adrenoceptor in the β-arrestin assay, and SAL and L-6 also have the lowest efficacies in the β-arrestin assay (E_{max β-Arrestin} values are 16% and 14%, respectively). (R)-EPI has an E_{max cAMP} value of 61% and an E_{max β-Arrestin} value of 112%. L-4 has an E_{max cAMP} value of 54% and an E_{max β-Arrestin} value of 90%. The E_max value differences of these two compounds are the greatest among the compounds tested. They are considered full β₂-adrenoceptor agonists in the β-arrestin assay and partial β₂-adrenoceptor agonists in the cAMP assay. For (R)-ISO, the pEC_{50 cAMP} value is 7.16 and the pEC_{50 β-Arrestin} value is 7.28. The difference between these two pEC₅₀ values is negligible ( < 0.3). This result also holds true for most of the test compounds except SAL, L-12, L-4, A-35, and L-2 (with the absolute differences being 0.87, 0.76, 0.54, 0.53, and 0.40, respectively). The similarity in the efficacies and potencies of most of these compounds across the two assay systems suggests that these two assays are comparable in terms of signal amplification. As the efficacy of (R)-EPI is just 61% of (R)-ISO in cAMP induction, (R)-ISO is the more suitable unbiased β₂-agonist reference for subsequent ligand bias determination.

Table 2 Efficacies (E_max) and potencies (pEC₅₀) of the compound-stimulated cAMP responses in HEK293 cells and β-arrestin responses in CHO-β₂-β-arr cells, respectively

The log(τ/K_A) values of the compounds for cAMP responses and β-arrestin responses (Table 3) were determined after fitting the concentration-response data to the Black–Leff operational model as previously described [38]. Ligand bias or the ΔΔlog(τ/K_A) values of the compounds (Table 3 and Fig. 1) were then calculated using (R)-ISO as the reference. Ligand bias is also expressed by the ratio values (bias factors or BFs) in Table 3. The bias factor of SAL is 0.14, or in other words, the activation of the G_s compared with the β-arrestin pathway is 7.35-fold. L-12, L-4, L-2, and A-35 activate the β-arrestin over the G_s pathway with bias factors equal 5.01, 4.32, 2.64, and 2.64, respectively. Statistical analyses of the ΔΔlog(τ/K_A) values suggest that SAL is a G_s-biased β₂-agonist, whereas L-2, L-4, and L-12 are β-arrestin-biased β₂-agonists (Fig. 1). The agonist bias of A-35 towards β-arrestin signaling does not reach statistical significance. Other compounds tested did not show ligand bias in the current analysis.

Table 3 Log(τ/K_A) and Δlog(τ/K_A) values for cAMP responses, log(τ/K_A) and Δlog(τ/K_A) values for β-arrestin responses, ΔΔlog(τ/K_A) values (ΔΔLog) and bias factors (BFs) of the compounds

Fig. 1

ΔΔLog(τ/K_A) values of the compounds for β-arrestin signaling over G_s signaling. The significance of the ligand bias was determined by comparing each value with that of (R)-ISO using Bonferroni’s t-test. *P < 0.05, ***P < 0.001 (Error bars are standard errors, n = 5)

Compounds were tested for their abilities to produce bronchodilation by the isolated guinea pig trachea relaxation assay. Relaxant effects on airway smooth muscles were observed with the L-series compounds tested at 5 μmol/L (Table 4). In line with the stimulatory effects of the compounds on cellular cAMP production (Table 2), compounds L-2, L-4, and L-12 but not L-6 produced relaxation responses greater than the contractile effect of histamine (10 μmol/L). These effects also appeared less than the relaxant effect produced by ISO.

Table 4 Effects of compounds (5 μmol/L) on relaxation of tracheal smooth muscle strips precontracted with histamine

Discussion

Several studies have implicated β-arrestin-biased signaling at the β-adrenoceptor to be cardioprotective [9, 41], leading to the hypothesis that β-arrestin-biased agonism at the β-adrenoceptor may be a novel therapeutic target for heart failure or other cardiovascular diseases [7,8,9, 41]. Interestingly, our own studies have shown that the G_s-biased β₂-agonist FEN is beneficial to survival, cardiac remodeling, and myocardial function in various animal models of heart failure [22,23,24,25,26]. This G_s-biased agonism of FEN at the β₂-adrenoceptor describes the ligand bias for preferential G_s signaling over G_s and G_i dual signaling because the β₂-adrenoceptor is capable of coupling to both G_s and G_i proteins [14]. FEN and (R,R’)-FEN have been shown to be G_s-biased [21, 32]. (R,R’)-MNF is unbiased in terms of β₂-adrenoceptor-G_s/G_i signaling [33]. Here, the signaling biases of FEN, (R,R’)-FEN, and (R,R’)-MNF for β-arrestin over G_s were investigated. Our results show that the Δlog(τ/K_A) values of FEN, (R,R’)-FEN, and (R,R’)-MNF for G_s are slightly larger than those for β-arrestin (Table 3), but the ligand biases for these compounds have no difference compared with that of (R)-ISO, a non-biased agonist of the β₂-adrenoceptor (P > 0.05, Fig. 1).

A research group has studied 65 β₂-agonists for biased agonism in the signaling pathways downstream of the β₂-adrenoceptor, namely, G_s, β-arrestin-1, and β-arrestin-2 [42,43,44]. These series of studies have included many FEN derivatives including (R,R’)-FEN and (R,R’)-MNF. In one of these studies [44], (R)-SAL and (S)-SAL have been shown to be strongly G_s-biased over both β-arrestin-1 and β-arrestin-2. Littmann et al. have also found that many FEN derivatives are G_s biased. Specifically, (R,R’)-FEN exhibits no bias in their study, whereas (R,R’)-MNF is weakly G_s-biased. Littmann et al. have performed no statistical analysis on their ligand bias data, likely because these data were pooled from different studies. Thus, the ligand bias data of the present study for SAL, (R,R’)-FEN, and (R,R’)-MNF are consistent with those reported in Littmann et al.

On the other hand, the study of Rajagopal et al. [45] on quantifying ligand bias at seven transmembrane receptors suggests that SAL is strongly β-arrestin-biased, whereas FEN exhibits no bias towards β-arrestin and G_s at the β₂-adrenoceptor. In addition, Van der Westhuizen et al. [38] have reported that SAL is biased towards the extracellular signal-regulated kinase 1/2 (ERK1/2) pathway over the cAMP pathway. ERK1/2 has been widely regarded as a downstream effector of β-arrestin in the field [46], but through G_i or G_s activation, the phosphorylation of ERK1/2 can also be induced [40, 47]. The results of Rajagopal et al. on SAL, and possibly those of Van der Westhuizen et al., are in direct contradiction with those of Littmann et al. and the present study. In Littmann et al., (R)-EPI rather than (R)-ISO was used as the reference unbiased agonist for ligand bias quantitation. However, this is not the main reason why Littmann et al. obtained a different result on the SAL stereoisomers because EPI has been shown to be a non-biased β₂-agonist in other studies including the present one. In addition, all of these studies have employed the Black–Leff operational model in quantifying ligand bias. Therefore, the discrepancy in results can only be due to the different techniques used in determining the G_s and β-arrestin activities of the compounds.

Careful examination of the data revealed that the reported E_{max cAMP} or E_{max Gs} values of SAL (or its stereoisomers) in Littmann et al., Rajagopal et al., and Van der Westhuizen et al. are equivalent to that of (R)-ISO, but the E_{max β-Arrestin} values of SAL are only 29% or less relative to (R)-ISO. The reported E_{max ERK1/2} value of SAL in Van der Westhuizen et al. is approximately 83% that of saturated (R)-ISO. A major finding of the present study different from those previous ones is that both the E_{max cAMP} and the E_{max β-Arrestin} values of SAL are approximately 16% of saturated (R)-ISO. SAL is regarded as a partial agonist of the β₂-adrenoceptor compared with the natural ligand (R)-EPI, a full β₂-agonist for G_s activation in relieving bronchoconstriction [48, 49]. In view of the low efficacies of SAL in our determination, the data reported in the present investigation may be more akin to data derived from physiological measurements compared with other studies.

Moreover, the incubation time of compound stimulation for the cAMP assay was 5 min and that for the Tango β-arrestin assay was 14–20 h in Rajagopal et al., whereas the incubation times of compound-stimulation for our cAMP and β-arrestin assays were 30 and 90 min, respectively. Therefore, kinetic differences between the assays would complicate the interpretation of our data to a lesser extent than those of Rajagopal et al. In addition, the incubation time of compound stimulation for the ERK1/2 assay in Van der Westhuizen et al. was <5 min. However, it has been shown that the G protein-dependent ERK1/2 activation at the β₂-adrenoceptor has a sharp peak between 2 and 5 min, whereas the β-arrestin-dependent ERK1/2 activation persists from 5 to 30 min [50]. Therefore, the ERK1/2 signals detected by Van der Westhuizen et al. may represent more of the signals of the G protein pathway rather than the signals of the β-arrestin pathway. More recent data have provided evidence that carvedilol signaling from the β₂-adrenoceptor is not merely β-arrestin-mediated but requires G_i [51]. Moreover, β-arrestins do not mediate ERK1/2 phosphorylation in the absence of functional G proteins [52]. These new findings suggest that ERK1/2 activity is not a good parameter for β-arrestin recruitment and cannot be used to access β-arrestin bias. In conclusion, our data and those of Littmann et al. support the notion that sustained bronchodilation with the chronic use of SAL is due to a defect in the desensitization of the SAL-stimulated β₂-adrenoceptor in a β-arrestin-dependent manner [49].

In the same manner as ERK1/2 activity is to β-arrestin recruitment, cAMP accumulation is widely used to represent the activity of the G_s pathway, as the G_s-adenylyl cyclase-cAMP-protein kinase A signaling cascade is accepted as a dogma in the field of receptor pharmacology. Applying the cAMP assay in ligand bias quantitation, however, is not necessarily error-free. One should note that most β₂-agonists including ISO and EPI stimulate the β₂-adrenoceptor to activate both G_s and G_i proteins, and the selectivity for the G_i versus G_s pathways to varying degrees may affect the determination of β-arrestin-biased agonism by altering the cAMP production. To address this issue, we tested whether pertussis toxin (PTX), a disruptor of G_i signaling, may cause a strong enhancement in the cAMP responses of the compounds. We used zinterol as the positive control compound because it has exhibited very substantial β₂-adrenoceptor-G_i signaling and a sensitivity towards PTX-treatment in our previous studies using rodent cardiomyocytes [14, 21, 33]. As expected, treatment with PTX increased the E_{max cAMP} values of zinterol (from 21.4% without PTX to 26.4% with PTX, 123% increase, P < 0.001) and (R)-ISO (from 33.7% without PTX to 36.3% with PTX, 110% increase, P > 0.05) without causing substantial changes in the EC₅₀ values of either compounds (Supplementary Fig. S5A, B). Treatment with PTX also increased the cAMP responses of most of the ligands tested at a saturated concentration and a sub-saturated concentration (Supplementary Fig. S5C). The greatest changes occurred in (R)-EPI (126% increase) and L-6 (127% increase) at their respective saturating concentrations, whereas the PTX-induced changes in the maximal cAMP responses for SAL, as well as the G_s-selective β₂-adrenoceptor agonists FEN and (R,R’)-FEN were 96%, 110%, and 116%, respectively. The PTX-induced changes in the compound-stimulated responses were generally small but definite. Statistical significance in some of the comparisons occurred only by chance and is not an indication of exceptional G_i activity for any particular compound(s). These results suggest that the HTRF cAMP assay is not particularly sensitive for detecting β₂-adrenoceptor-G_i signaling compared with other assays, such as the rat cardiomyocyte contractility assay. Thus, the selectivity of the β₂-adrenoceptor ligands to different G_s/G_i pathways should not be a confounding factor in the determination of β-arrestin bias in the present study.

The trachea relaxation assay results suggest that these compounds are bronchodilators like other β₂-adrenoceptor agonists and that their activities (presumably those of cAMP induction) rather than their ligand bias status are associated with the bronchodilator effects of the compounds. The correlation of E_max in the cAMP assay with relaxation is fairly good except for L-12, which shows more potent relaxation than L-2 and L-4, suggesting that the smooth muscle relaxant effect of L-12 might involve additional mechanisms other than stimulation of the β₂-adrenoceptor. One possible mechanism is the inhibition of phosphodiesterase, which causes an accumulation of cAMP in the cytosol. This effect, if it exists, will be masked in the cAMP assay because the cAMP assay buffer contained IBMX, a phosphodiesterase inhibitor, in an excess amount.

BI-167107 (Supplementary Fig. S1), a high affinity and selective β₂-adrenoceptor agonist previously used in X-ray crystallographic studies of the active conformations of the β₂-adrenoceptor [53], has been shown to be partially biased for β-arrestin over G_s [54]. As the L-series compounds possess a different head group and a different core structure compared with BI-167107, the present study has discovered a new scaffold for the design of novel β-arrestin-biased β₂-adrenoceptor agonists. Additionally, compounds L-2, L-4, and L-12 have much lower potencies in inducing cAMP compared with (R)-ISO (Table 2). This result may suggest a low binding affinity. In summary, these compounds may better serve as tool reagents or as lead compounds for further development.

Conclusion

In the present study, we characterized a cohort of phenylethanolamines for β-arrestin-biased agonism at the β₂-adrenoceptor. Our data show that three 2-amino-2-phenylethanol derivatives, namely, L-2, L-4, and L-12, are partial β₂-adrenoceptor agonists with weak ligand biases for β-arrestin over G_s. The present identification of β-arrestin-biased β₂-adrenoceptor agonists with a new 5-(1-amino-2-hydroxyethyl)-8-hydroxyquinolin-2(1H)-one core structure will facilitate the discovery of novel biased agonists with potential usefulness in the treatment of diseases of the heart and airway.