Identification of a β-arrestin-biased negative allosteric modulator for the β2-adrenergic receptor

Abstract

Catecholamine-stimulated β₂-adrenergic receptor (β₂AR) signaling via the canonical G_s-adenylyl cyclase-cAMP-PKA pathway regulates numerous physiological functions, including the therapeutic effects of exogenous β-agonists in the treatment of airway disease. β₂AR signaling is tightly regulated by GRKs and β-arrestins, which together promote β₂AR desensitization and internalization as well as downstream signaling, often antithetical to the canonical pathway. Thus, the ability to bias β₂AR signaling toward the G_s pathway while avoiding β-arrestin-mediated effects may provide a strategy to improve the functional consequences of β₂AR activation. Since attempts to develop G_s-biased agonists and allosteric modulators for the β₂AR have been largely unsuccessful, here we screened small molecule libraries for allosteric modulators that selectively inhibit β-arrestin recruitment to the receptor. This screen identified several compounds that met this profile, and, of these, a difluorophenyl quinazoline (DFPQ) derivative was found to be a selective negative allosteric modulator of β-arrestin recruitment to the β₂AR while having no effect on β₂AR coupling to G_s. DFPQ effectively inhibits agonist-promoted phosphorylation and internalization of the β₂AR and protects against the functional desensitization of β-agonist mediated regulation in cell and tissue models. The effects of DFPQ were also specific to the β₂AR with minimal effects on the β₁AR. Modeling, mutagenesis, and medicinal chemistry studies support DFPQ derivatives binding to an intracellular membrane-facing region of the β₂AR, including residues within transmembrane domains 3 and 4 and intracellular loop 2. DFPQ thus represents a class of biased allosteric modulators that targets an allosteric site of the β₂AR.

Keywords: G protein–coupled receptor; asthma; biased signaling; cell signaling; negative allosteric modulator.

Fig. 1.

Identifying small-molecule allosteric modulators of β-arrestin recruitment to the β₂AR. (A) Schematic of screening methodologies for cAMP production and β-arrestin binding. Cells were treated with ISO ± test compounds. The primary high throughput screen utilized a luciferase-based cAMP biosensor, GloSensor (Promega), and an enzyme complementation-based assay, PathHunter (DiscoveRx). Independent secondary screening utilized a cAMP ELISA and a BRET assay for β-arrestin recruitment. (B) Structure of DFPQ. (C) Secondary screen for cAMP production by ELISA. HEK 293 cells stably expressing β₂AR were preincubated with 0.1% DMSO (negative control) or 10 μM DFPQ for 30 min and then stimulated with or without 1 μM ISO for 10 min. Cells were lysed and cAMP production was measured. Data are normalized to 1 μM ISO and are the mean % ± SEM, n = 3. (D) Dose–response curve for DFPQ as measured by BRET. HEK 293 cells cotransfected with β-arrestin2-GFP10 and β₂AR-RlucII were preincubated with 0.1% DMSO (negative control) or the indicated concentrations of DFPQ for 30 min. Cells were incubated with Coelenterazine 400a for 2 min and then stimulated with 1 μM ISO. Data for the dose–response curve was taken 12 min post-ISO addition. Data are the mean ± SEM, n = 3. (E) HEK 293 cells cotransfected with β-arrestin2-GFP10 and β₂AR-RLucII were preincubated with the indicated concentrations of DFPQ for 30 min. Cells were then incubated with Coelenterazine 400a for 2 min and then stimulated with the indicated concentrations of ISO. Data for dose–response curves were taken 12 min post-ISO addition and are the mean ± SEM, n = 3. (F) HEK 293 cells stably expressing β₂AR were preincubated with 0.1% DMSO (negative control), 1 μM DFPQ, or 10 μM DFPQ and stimulated with the indicated concentrations of ISO for 10 min. Cells were lysed and cAMP production was measured by ELISA. Data are normalized to 1 μM ISO and are the mean % ± SEM, n = 3.

Fig. 2.

GPCR specificity of DFPQ. HEK 293 cells were cotransfected with β-arrestin2-GFP10 and either β₂AR-RlucII (A), β₁AR-RlucII (B), or CXCR4-RlucII (C) for 48 h and cells were preincubated with 0.1% DMSO (negative control) or 10 μM DFPQ for 30 min. Cells were then incubated with Coelenterazine 400a for 2 min and stimulated with 1 μM of the indicated agonist. Cells were read every 2 min post agonist addition. One-way ANOVA with multiple comparisons was performed to compare the means of each treatment group. Data are the mean ± SEM, n = 3. ****P < 0.0001; ns, not significant

Fig. 3.

DFPQ inhibits GRK-mediated phosphorylation of the β₂AR. (A) Representative time course of agonist promoted phosphorylation of the β₂AR. HEK 293 cells stably expressing FLAG-β₂AR were preincubated with 0.1% DMSO or indicated concentrations of DFPQ for 30 min and then stimulated with 1 μM ISO for 10 min. Cells were lysed and the FLAG-β₂AR was immunoprecipitated. Phosphorylation at serine 355 and 356 was analyzed by western blot using a pSer^355/356 antibody while total β₂AR was measured using a β₂AR C-terminal antibody. The western blot is representative of at least three experiments. (B) Densitometric quantification for concentration–activity curve-fitting from panel A is shown. Data are normalized to 1 μM ISO and are the mean % ± SEM, n = 3. (C) A representative autoradiograph of GRK5 autophosphorylation and GRK5-mediated phosphorylation of tubulin and the β₂AR in the presence of 1 μM ISO and the indicated concentration of DFPQ. The apparent small increase in GRK5-autophosphorylation was not consistent across the three replicates, as shown by the average quantification in panel D. (D) In vitro phosphorylation data from three experiments were quantified and are shown as a percentage of no DFPQ. Data are mean ± SEM, n = 3. Concentration–activity curves were generated by fitting the western blot densitometric data to the logistic equation log(inhibitor) vs. response (three parameters) while data from autoradiography were fit using the logistic equation log(inhibitor) vs. response (four parameters, variable slope). All curves were generated using GraphPad Prism.

Fig. 4.

DFPQ inhibits internalization of the β₂AR and protects from agonist-induced desensitization in cell and tissue models. (A) Effect of DFPQ on agonist-promoted internalization of the β₂AR. HEK 293 cells stably expressing FLAG-β₂AR were preincubated with 0.1% DMSO or 10 μM DFPQ for 30 min and then stimulated with 10 μM ISO for up to 60 min. Cells were fixed and receptor surface expression was measured by ELISA. These data represent the mean ± SD from three independent experiments. ***P < 0.001; **P < 0.01; *P < 0.05 (B) HEK 293 cells were desensitized by incubating with 1 μM ISO ± 10 μM DFPQ, washed with PBS, then stimulated with ISO at the indicated concentrations for 10 min. Cells were lysed and cAMP production was measured by ELISA. Cells incubated with DFPQ during desensitization were protected from reduced response to agonist after wash. Data are the mean ± SD, n = 3. (C) Comparison of the stiffness in HASM cells following a 30 min pretreatment with (n = 473 individual cell measurements) or without (n = 463 individual cell measurements) 1 μM DFPQ. Histograms are represented as geometric mean ± 95% CIs. (D) Functional desensitization of the β₂AR was studied by pretreating HASM cells with or without 1 μM DFPQ for 30 min, followed by relaxation over a 30 min period induced by 10 μM ISO. Data are collected at every 1.3 s (t, 0 to 1,800 s) and presented as estimated mean ± SE (n = 463 to 473 cell measurement per treatment). (E) Murine precision cut lung slices were contracted with 1 μM methacholine for 5 min and then incubated overnight with 1 μM ISO ± 1 μM DFPQ. Airway slices were washed, recontracted with methacholine, and incubated with ISO for 1 h (n = 4 airways). Airway contractility was measured by the luminal area of bronchioles by microscopy at various times. (F) Bar graphs represent normalized ISO-promoted relaxation from methacholine contraction on day 2 of treatment at the 25 min time point from 4 airways. Statistical significance (*) in panels E and F were assessed using an unpaired t test, P < 0.05.

Fig. 5.

SAR of quinazolines in inhibition of β-arrestin recruitment to the β₂AR. (A) Functional groups for which substitutions were present either in the screening or from medicinal chemistry. Representative structures for R1, R2, and R3 substitutions ranked from high to low activity for inhibition of β-arrestin recruitment to the β₂AR are shown. (B) Concentration–activity curves for inhibition of ISO-induced β-arrestin recruitment to the β₂AR by DFPQ, a bromo-derivative of DFPQ (AP-7-168), and a trifluoromethyl-derivative of DFPQ (AP-7-203). Data are normalized to 1 μM ISO and are the mean % ± SEM, n = 9. The chemical structures of AP-7-168 and AP-7-203 are shown.

Fig. 6.

Identification of a DFPQ binding site on the β₂AR. Concentration–activity curves for DFPQ-, AP-7-168- and AP-7-203-mediated inhibition of β-arrestin recruitment to the β₁AR and β₂AR compared with chimeras generated by swapping TM3 (A) or TM4 (B) from the β₁AR into the β₂AR. Data are normalized to 1 μM ISO and are the mean % ± SEM, n = 3. (C and D). Two views of the best docking model for DFPQ bound to the β₂AR at a site between TM3 and TM4. ISO and DFPQ are shown as a molecular surface.

Fig. 7.

Model of NAM interaction and effect of β₂AR mutation on NAM-mediated inhibition of β-arrestin recruitment. (A) Best molecular model of DFPQ showing key residue contacts in the β₂AR identified by mutagenesis. The all-atom consensus model of DFPQ is shown in gray. The β₂AR residue side chains are shown for the three mutations E122W, V129L, and M156T with the strongest effects on NAM activity. Concentration–activity curves for inhibition of ISO-induced β-arrestin recruitment by DFPQ, AP-7-168, and AP-7-203 at the β₁AR and β₂AR were compared with mutants of residues E122 (B); V129 (C) and M156 (D) of the β₂AR. Data are normalized to 1 μM ISO and are the mean % ± SEM, n = 3.